diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,39037 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.999884664180539,
+  "global_step": 6502,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.115089514066497e-08,
+      "loss": 4.6228,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0230179028132994e-07,
+      "loss": 4.0014,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.534526854219949e-07,
+      "loss": 3.8888,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0460358056265988e-07,
+      "loss": 4.1256,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5575447570332484e-07,
+      "loss": 3.835,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.069053708439898e-07,
+      "loss": 4.2457,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.5805626598465475e-07,
+      "loss": 4.0011,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.0920716112531976e-07,
+      "loss": 3.6465,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.603580562659847e-07,
+      "loss": 3.5037,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.115089514066497e-07,
+      "loss": 3.8619,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.626598465473146e-07,
+      "loss": 3.7314,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.138107416879796e-07,
+      "loss": 4.1018,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.649616368286447e-07,
+      "loss": 4.5126,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.161125319693095e-07,
+      "loss": 4.0486,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.672634271099745e-07,
+      "loss": 3.8919,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.184143222506395e-07,
+      "loss": 4.2373,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.695652173913044e-07,
+      "loss": 4.1806,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.207161125319694e-07,
+      "loss": 4.1494,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.718670076726342e-07,
+      "loss": 4.0906,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0230179028132994e-06,
+      "loss": 4.2199,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0204081632653061e-07,
+      "loss": 3.7042,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0408163265306121e-07,
+      "loss": 4.0132,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.0612244897959183e-07,
+      "loss": 3.9304,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.0816326530612243e-07,
+      "loss": 4.4279,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.102040816326531e-07,
+      "loss": 3.8656,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.122448979591837e-07,
+      "loss": 3.8767,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.142857142857143e-07,
+      "loss": 4.0807,
+      "step": 27
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.163265306122449e-07,
+      "loss": 4.017,
+      "step": 28
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.183673469387756e-07,
+      "loss": 3.7091,
+      "step": 29
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0204081632653063e-06,
+      "loss": 3.877,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.122448979591837e-06,
+      "loss": 4.0755,
+      "step": 31
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2244897959183673e-06,
+      "loss": 3.8504,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3265306122448982e-06,
+      "loss": 3.8438,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4285714285714286e-06,
+      "loss": 3.8322,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5306122448979593e-06,
+      "loss": 3.6464,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6326530612244897e-06,
+      "loss": 4.1511,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7346938775510206e-06,
+      "loss": 3.6794,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8367346938775512e-06,
+      "loss": 3.6042,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.938775510204082e-06,
+      "loss": 3.6337,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0408163265306125e-06,
+      "loss": 3.756,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1428571428571427e-06,
+      "loss": 3.5674,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.244897959183674e-06,
+      "loss": 3.4821,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3469387755102044e-06,
+      "loss": 3.16,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4489795918367347e-06,
+      "loss": 2.9648,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5510204081632657e-06,
+      "loss": 2.9141,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6530612244897964e-06,
+      "loss": 2.9461,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7551020408163266e-06,
+      "loss": 3.1041,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8571428571428573e-06,
+      "loss": 2.9343,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.959183673469388e-06,
+      "loss": 2.9607,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0612244897959185e-06,
+      "loss": 2.7754,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.1632653061224496e-06,
+      "loss": 2.861,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2653061224489794e-06,
+      "loss": 2.626,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3673469387755105e-06,
+      "loss": 2.5885,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.469387755102041e-06,
+      "loss": 2.3972,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.5714285714285718e-06,
+      "loss": 2.264,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6734693877551024e-06,
+      "loss": 2.1644,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7755102040816327e-06,
+      "loss": 2.1415,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.877551020408164e-06,
+      "loss": 2.0875,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.979591836734694e-06,
+      "loss": 2.3436,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.081632653061225e-06,
+      "loss": 2.0473,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.183673469387755e-06,
+      "loss": 2.1611,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.2857142857142855e-06,
+      "loss": 1.9309,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.3877551020408165e-06,
+      "loss": 1.9239,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.489795918367348e-06,
+      "loss": 1.872,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.591836734693878e-06,
+      "loss": 1.8077,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.693877551020409e-06,
+      "loss": 2.0137,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.795918367346939e-06,
+      "loss": 2.1205,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.897959183673469e-06,
+      "loss": 1.8733,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5e-06,
+      "loss": 1.7783,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.1020408163265315e-06,
+      "loss": 1.8582,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.204081632653062e-06,
+      "loss": 1.9384,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.306122448979593e-06,
+      "loss": 1.9074,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.408163265306123e-06,
+      "loss": 1.7146,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.510204081632653e-06,
+      "loss": 1.9142,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.6122448979591834e-06,
+      "loss": 1.856,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.7142857142857145e-06,
+      "loss": 1.7465,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.816326530612246e-06,
+      "loss": 1.7528,
+      "step": 77
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.918367346938776e-06,
+      "loss": 1.7993,
+      "step": 78
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.020408163265307e-06,
+      "loss": 1.6797,
+      "step": 79
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.122448979591837e-06,
+      "loss": 1.8134,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.224489795918368e-06,
+      "loss": 1.7928,
+      "step": 81
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.326530612244899e-06,
+      "loss": 1.6557,
+      "step": 82
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.4285714285714295e-06,
+      "loss": 1.5643,
+      "step": 83
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.530612244897959e-06,
+      "loss": 1.6808,
+      "step": 84
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.63265306122449e-06,
+      "loss": 1.7959,
+      "step": 85
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.734693877551021e-06,
+      "loss": 1.8157,
+      "step": 86
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.836734693877551e-06,
+      "loss": 1.6966,
+      "step": 87
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.938775510204082e-06,
+      "loss": 1.6308,
+      "step": 88
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.0408163265306125e-06,
+      "loss": 1.6321,
+      "step": 89
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.1428571428571436e-06,
+      "loss": 1.6028,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.244897959183675e-06,
+      "loss": 1.6194,
+      "step": 91
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.346938775510205e-06,
+      "loss": 1.6288,
+      "step": 92
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.448979591836736e-06,
+      "loss": 1.5884,
+      "step": 93
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.551020408163265e-06,
+      "loss": 1.7593,
+      "step": 94
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.653061224489796e-06,
+      "loss": 1.5937,
+      "step": 95
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.755102040816327e-06,
+      "loss": 1.4913,
+      "step": 96
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.857142857142858e-06,
+      "loss": 1.7858,
+      "step": 97
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.959183673469388e-06,
+      "loss": 1.6268,
+      "step": 98
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.06122448979592e-06,
+      "loss": 1.5917,
+      "step": 99
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.16326530612245e-06,
+      "loss": 1.5547,
+      "step": 100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.26530612244898e-06,
+      "loss": 1.5545,
+      "step": 101
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.36734693877551e-06,
+      "loss": 1.6464,
+      "step": 102
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.469387755102042e-06,
+      "loss": 1.5398,
+      "step": 103
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.571428571428571e-06,
+      "loss": 1.8221,
+      "step": 104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.673469387755103e-06,
+      "loss": 1.759,
+      "step": 105
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.775510204081633e-06,
+      "loss": 1.5942,
+      "step": 106
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.877551020408163e-06,
+      "loss": 1.6189,
+      "step": 107
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.979591836734695e-06,
+      "loss": 1.4358,
+      "step": 108
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.081632653061225e-06,
+      "loss": 1.5923,
+      "step": 109
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.183673469387756e-06,
+      "loss": 1.6142,
+      "step": 110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.285714285714288e-06,
+      "loss": 1.6147,
+      "step": 111
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.387755102040818e-06,
+      "loss": 1.5154,
+      "step": 112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.489795918367348e-06,
+      "loss": 1.472,
+      "step": 113
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.591836734693878e-06,
+      "loss": 1.5385,
+      "step": 114
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.693877551020408e-06,
+      "loss": 1.575,
+      "step": 115
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.795918367346939e-06,
+      "loss": 1.444,
+      "step": 116
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.89795918367347e-06,
+      "loss": 1.377,
+      "step": 117
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1e-05,
+      "loss": 1.4541,
+      "step": 118
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0102040816326531e-05,
+      "loss": 1.5987,
+      "step": 119
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0204081632653063e-05,
+      "loss": 1.5139,
+      "step": 120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0306122448979591e-05,
+      "loss": 1.4052,
+      "step": 121
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0408163265306123e-05,
+      "loss": 1.4002,
+      "step": 122
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0510204081632654e-05,
+      "loss": 1.4798,
+      "step": 123
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0612244897959186e-05,
+      "loss": 1.42,
+      "step": 124
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0714285714285714e-05,
+      "loss": 1.4845,
+      "step": 125
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0816326530612246e-05,
+      "loss": 1.3027,
+      "step": 126
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0918367346938776e-05,
+      "loss": 1.5146,
+      "step": 127
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1020408163265306e-05,
+      "loss": 1.3157,
+      "step": 128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1122448979591838e-05,
+      "loss": 1.502,
+      "step": 129
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1224489795918367e-05,
+      "loss": 1.5395,
+      "step": 130
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1326530612244899e-05,
+      "loss": 1.4553,
+      "step": 131
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1428571428571429e-05,
+      "loss": 1.5775,
+      "step": 132
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1530612244897961e-05,
+      "loss": 1.4504,
+      "step": 133
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1632653061224491e-05,
+      "loss": 1.6017,
+      "step": 134
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1734693877551021e-05,
+      "loss": 1.5011,
+      "step": 135
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1836734693877552e-05,
+      "loss": 1.4512,
+      "step": 136
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1938775510204084e-05,
+      "loss": 1.4912,
+      "step": 137
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2040816326530614e-05,
+      "loss": 1.3928,
+      "step": 138
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2142857142857142e-05,
+      "loss": 1.243,
+      "step": 139
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2244897959183674e-05,
+      "loss": 1.3431,
+      "step": 140
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2346938775510204e-05,
+      "loss": 1.5565,
+      "step": 141
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2448979591836736e-05,
+      "loss": 1.2735,
+      "step": 142
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2551020408163267e-05,
+      "loss": 1.4029,
+      "step": 143
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2653061224489798e-05,
+      "loss": 1.3409,
+      "step": 144
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2755102040816327e-05,
+      "loss": 1.4538,
+      "step": 145
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2857142857142859e-05,
+      "loss": 1.4143,
+      "step": 146
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2959183673469389e-05,
+      "loss": 1.3945,
+      "step": 147
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3061224489795918e-05,
+      "loss": 1.5087,
+      "step": 148
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.316326530612245e-05,
+      "loss": 1.4053,
+      "step": 149
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.326530612244898e-05,
+      "loss": 1.5279,
+      "step": 150
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3367346938775512e-05,
+      "loss": 1.3159,
+      "step": 151
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3469387755102042e-05,
+      "loss": 1.3882,
+      "step": 152
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3571428571428574e-05,
+      "loss": 1.4087,
+      "step": 153
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3673469387755102e-05,
+      "loss": 1.5711,
+      "step": 154
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3775510204081634e-05,
+      "loss": 1.4261,
+      "step": 155
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3877551020408165e-05,
+      "loss": 1.3999,
+      "step": 156
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3979591836734696e-05,
+      "loss": 1.346,
+      "step": 157
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4081632653061225e-05,
+      "loss": 1.5246,
+      "step": 158
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4183673469387755e-05,
+      "loss": 1.4275,
+      "step": 159
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4285714285714287e-05,
+      "loss": 1.2309,
+      "step": 160
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4387755102040817e-05,
+      "loss": 1.4205,
+      "step": 161
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.448979591836735e-05,
+      "loss": 1.4981,
+      "step": 162
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.4591836734693878e-05,
+      "loss": 1.5111,
+      "step": 163
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.469387755102041e-05,
+      "loss": 1.3562,
+      "step": 164
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.479591836734694e-05,
+      "loss": 1.2303,
+      "step": 165
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.4897959183673472e-05,
+      "loss": 1.3216,
+      "step": 166
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 1.3718,
+      "step": 167
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.510204081632653e-05,
+      "loss": 1.3585,
+      "step": 168
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.5204081632653063e-05,
+      "loss": 1.4635,
+      "step": 169
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.530612244897959e-05,
+      "loss": 1.36,
+      "step": 170
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.5408163265306123e-05,
+      "loss": 1.178,
+      "step": 171
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.5510204081632655e-05,
+      "loss": 1.3373,
+      "step": 172
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.5612244897959187e-05,
+      "loss": 1.2223,
+      "step": 173
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.5714285714285715e-05,
+      "loss": 1.5029,
+      "step": 174
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.5816326530612247e-05,
+      "loss": 1.4571,
+      "step": 175
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.5918367346938776e-05,
+      "loss": 1.3603,
+      "step": 176
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6020408163265308e-05,
+      "loss": 1.3451,
+      "step": 177
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.612244897959184e-05,
+      "loss": 1.3517,
+      "step": 178
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6224489795918368e-05,
+      "loss": 1.3082,
+      "step": 179
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.63265306122449e-05,
+      "loss": 1.5138,
+      "step": 180
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.642857142857143e-05,
+      "loss": 1.2624,
+      "step": 181
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.653061224489796e-05,
+      "loss": 1.3437,
+      "step": 182
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6632653061224492e-05,
+      "loss": 1.5041,
+      "step": 183
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.673469387755102e-05,
+      "loss": 1.4104,
+      "step": 184
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6836734693877553e-05,
+      "loss": 1.3233,
+      "step": 185
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6938775510204085e-05,
+      "loss": 1.4402,
+      "step": 186
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7040816326530613e-05,
+      "loss": 1.4238,
+      "step": 187
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7142857142857142e-05,
+      "loss": 1.2496,
+      "step": 188
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7244897959183674e-05,
+      "loss": 1.3182,
+      "step": 189
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7346938775510206e-05,
+      "loss": 1.3495,
+      "step": 190
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7448979591836738e-05,
+      "loss": 1.3764,
+      "step": 191
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7551020408163266e-05,
+      "loss": 1.5462,
+      "step": 192
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7653061224489798e-05,
+      "loss": 1.248,
+      "step": 193
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7755102040816327e-05,
+      "loss": 1.3593,
+      "step": 194
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.785714285714286e-05,
+      "loss": 1.327,
+      "step": 195
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.795918367346939e-05,
+      "loss": 1.253,
+      "step": 196
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.806122448979592e-05,
+      "loss": 1.1221,
+      "step": 197
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.816326530612245e-05,
+      "loss": 1.2253,
+      "step": 198
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.826530612244898e-05,
+      "loss": 1.4232,
+      "step": 199
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.836734693877551e-05,
+      "loss": 1.3973,
+      "step": 200
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8469387755102043e-05,
+      "loss": 1.2387,
+      "step": 201
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8571428571428575e-05,
+      "loss": 1.2245,
+      "step": 202
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8673469387755104e-05,
+      "loss": 1.2159,
+      "step": 203
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8775510204081636e-05,
+      "loss": 1.4366,
+      "step": 204
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8877551020408164e-05,
+      "loss": 1.2574,
+      "step": 205
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8979591836734696e-05,
+      "loss": 1.3074,
+      "step": 206
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9081632653061225e-05,
+      "loss": 1.1391,
+      "step": 207
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9183673469387756e-05,
+      "loss": 1.248,
+      "step": 208
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.928571428571429e-05,
+      "loss": 1.1953,
+      "step": 209
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9387755102040817e-05,
+      "loss": 1.273,
+      "step": 210
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.948979591836735e-05,
+      "loss": 1.3648,
+      "step": 211
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9591836734693877e-05,
+      "loss": 1.2784,
+      "step": 212
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.969387755102041e-05,
+      "loss": 1.3103,
+      "step": 213
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.979591836734694e-05,
+      "loss": 1.3404,
+      "step": 214
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9897959183673473e-05,
+      "loss": 1.4429,
+      "step": 215
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2e-05,
+      "loss": 1.2227,
+      "step": 216
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.99999987590285e-05,
+      "loss": 1.2146,
+      "step": 217
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999995036114308e-05,
+      "loss": 1.4213,
+      "step": 218
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999988831258343e-05,
+      "loss": 1.3376,
+      "step": 219
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999998014446215e-05,
+      "loss": 1.3867,
+      "step": 220
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999968975727885e-05,
+      "loss": 1.3537,
+      "step": 221
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999955325058314e-05,
+      "loss": 1.3026,
+      "step": 222
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999939192456834e-05,
+      "loss": 1.3941,
+      "step": 223
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999920577927442e-05,
+      "loss": 1.3338,
+      "step": 224
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999899481474763e-05,
+      "loss": 1.244,
+      "step": 225
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999875903104028e-05,
+      "loss": 1.1823,
+      "step": 226
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999984984282109e-05,
+      "loss": 1.2183,
+      "step": 227
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999982130063243e-05,
+      "loss": 1.3825,
+      "step": 228
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9999790276545107e-05,
+      "loss": 1.2667,
+      "step": 229
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9999756770566842e-05,
+      "loss": 1.2104,
+      "step": 230
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9999720782705944e-05,
+      "loss": 1.3405,
+      "step": 231
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999968231297134e-05,
+      "loss": 1.4473,
+      "step": 232
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9999641361372583e-05,
+      "loss": 1.251,
+      "step": 233
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999959792791984e-05,
+      "loss": 1.1743,
+      "step": 234
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9999552012623887e-05,
+      "loss": 1.1662,
+      "step": 235
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9999503615496117e-05,
+      "loss": 1.185,
+      "step": 236
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999945273654855e-05,
+      "loss": 1.1613,
+      "step": 237
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9999399375793808e-05,
+      "loss": 1.2294,
+      "step": 238
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9999343533245134e-05,
+      "loss": 1.3355,
+      "step": 239
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9999285208916393e-05,
+      "loss": 1.2074,
+      "step": 240
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9999224402822055e-05,
+      "loss": 1.3376,
+      "step": 241
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9999161114977214e-05,
+      "loss": 1.2137,
+      "step": 242
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9999095345397582e-05,
+      "loss": 1.2914,
+      "step": 243
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9999027094099475e-05,
+      "loss": 1.2156,
+      "step": 244
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998956361099835e-05,
+      "loss": 1.2934,
+      "step": 245
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999888314641622e-05,
+      "loss": 1.2974,
+      "step": 246
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99988074500668e-05,
+      "loss": 1.2231,
+      "step": 247
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998729272070366e-05,
+      "loss": 1.1777,
+      "step": 248
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998648612446315e-05,
+      "loss": 1.4118,
+      "step": 249
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998565471214667e-05,
+      "loss": 1.3287,
+      "step": 250
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998479848396065e-05,
+      "loss": 1.4516,
+      "step": 251
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999839174401175e-05,
+      "loss": 1.2631,
+      "step": 252
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998301158083596e-05,
+      "loss": 1.179,
+      "step": 253
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999820809063408e-05,
+      "loss": 1.2139,
+      "step": 254
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998112541686308e-05,
+      "loss": 1.1978,
+      "step": 255
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998014511263987e-05,
+      "loss": 1.1915,
+      "step": 256
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997913999391456e-05,
+      "loss": 1.2919,
+      "step": 257
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997811006093654e-05,
+      "loss": 1.3169,
+      "step": 258
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999770553139615e-05,
+      "loss": 1.2714,
+      "step": 259
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997597575325115e-05,
+      "loss": 1.2704,
+      "step": 260
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997487137907346e-05,
+      "loss": 1.2543,
+      "step": 261
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997374219170254e-05,
+      "loss": 1.2984,
+      "step": 262
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997258819141864e-05,
+      "loss": 1.4022,
+      "step": 263
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997140937850822e-05,
+      "loss": 1.2931,
+      "step": 264
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999702057532638e-05,
+      "loss": 1.1851,
+      "step": 265
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999689773159841e-05,
+      "loss": 1.3687,
+      "step": 266
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996772406697406e-05,
+      "loss": 1.2644,
+      "step": 267
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999664460065447e-05,
+      "loss": 1.2361,
+      "step": 268
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996514313501323e-05,
+      "loss": 1.1225,
+      "step": 269
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996381545270303e-05,
+      "loss": 1.1036,
+      "step": 270
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999624629599436e-05,
+      "loss": 1.2167,
+      "step": 271
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996108565707066e-05,
+      "loss": 1.3216,
+      "step": 272
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995968354442602e-05,
+      "loss": 1.1645,
+      "step": 273
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995825662235763e-05,
+      "loss": 1.2672,
+      "step": 274
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995680489121975e-05,
+      "loss": 1.1246,
+      "step": 275
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999553283513726e-05,
+      "loss": 1.2242,
+      "step": 276
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999538270031827e-05,
+      "loss": 1.2049,
+      "step": 277
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995230084702268e-05,
+      "loss": 1.3348,
+      "step": 278
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999507498832713e-05,
+      "loss": 1.1879,
+      "step": 279
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999491741123135e-05,
+      "loss": 1.1809,
+      "step": 280
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994757353454034e-05,
+      "loss": 1.2028,
+      "step": 281
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994594815034912e-05,
+      "loss": 1.3352,
+      "step": 282
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999442979601433e-05,
+      "loss": 1.2507,
+      "step": 283
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994262296433236e-05,
+      "loss": 1.3814,
+      "step": 284
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994092316333206e-05,
+      "loss": 1.332,
+      "step": 285
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993919855756428e-05,
+      "loss": 1.0811,
+      "step": 286
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993744914745705e-05,
+      "loss": 1.2313,
+      "step": 287
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993567493344458e-05,
+      "loss": 1.2058,
+      "step": 288
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999338759159672e-05,
+      "loss": 1.1506,
+      "step": 289
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993205209547146e-05,
+      "loss": 1.196,
+      "step": 290
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993020347240995e-05,
+      "loss": 1.3487,
+      "step": 291
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992833004724153e-05,
+      "loss": 1.179,
+      "step": 292
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9992643182043118e-05,
+      "loss": 1.2446,
+      "step": 293
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9992450879245002e-05,
+      "loss": 1.1281,
+      "step": 294
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9992256096377534e-05,
+      "loss": 1.2629,
+      "step": 295
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9992058833489053e-05,
+      "loss": 1.3692,
+      "step": 296
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9991859090628526e-05,
+      "loss": 1.1317,
+      "step": 297
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9991656867845526e-05,
+      "loss": 1.2104,
+      "step": 298
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.999145216519024e-05,
+      "loss": 1.072,
+      "step": 299
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.999124498271348e-05,
+      "loss": 1.2935,
+      "step": 300
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.999103532046666e-05,
+      "loss": 1.284,
+      "step": 301
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.999082317850182e-05,
+      "loss": 1.1898,
+      "step": 302
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.999060855687162e-05,
+      "loss": 1.3403,
+      "step": 303
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9990391455629315e-05,
+      "loss": 1.1867,
+      "step": 304
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9990171874828795e-05,
+      "loss": 1.3082,
+      "step": 305
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998994981452456e-05,
+      "loss": 1.1509,
+      "step": 306
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998972527477172e-05,
+      "loss": 1.1532,
+      "step": 307
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998949825562601e-05,
+      "loss": 1.3084,
+      "step": 308
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9989268757143765e-05,
+      "loss": 1.1165,
+      "step": 309
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998903677938196e-05,
+      "loss": 1.3194,
+      "step": 310
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998880232239816e-05,
+      "loss": 1.1946,
+      "step": 311
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998856538625056e-05,
+      "loss": 1.1206,
+      "step": 312
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998832597099796e-05,
+      "loss": 1.2876,
+      "step": 313
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9988084076699788e-05,
+      "loss": 1.2888,
+      "step": 314
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987839703416077e-05,
+      "loss": 1.1133,
+      "step": 315
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998759285120748e-05,
+      "loss": 1.0826,
+      "step": 316
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998734352013527e-05,
+      "loss": 1.1439,
+      "step": 317
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987091710261323e-05,
+      "loss": 1.2536,
+      "step": 318
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986837421648137e-05,
+      "loss": 1.2333,
+      "step": 319
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986580654358828e-05,
+      "loss": 1.0931,
+      "step": 320
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986321408457122e-05,
+      "loss": 1.139,
+      "step": 321
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998605968400736e-05,
+      "loss": 1.2243,
+      "step": 322
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985795481074508e-05,
+      "loss": 1.3268,
+      "step": 323
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985528799724135e-05,
+      "loss": 1.0119,
+      "step": 324
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985259640022426e-05,
+      "loss": 1.2171,
+      "step": 325
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984988002036192e-05,
+      "loss": 1.0532,
+      "step": 326
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984713885832847e-05,
+      "loss": 1.1578,
+      "step": 327
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984437291480428e-05,
+      "loss": 1.1588,
+      "step": 328
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998415821904758e-05,
+      "loss": 1.2536,
+      "step": 329
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983876668603578e-05,
+      "loss": 1.1347,
+      "step": 330
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983592640218287e-05,
+      "loss": 1.2666,
+      "step": 331
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998330613396221e-05,
+      "loss": 1.1675,
+      "step": 332
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983017149906453e-05,
+      "loss": 1.1618,
+      "step": 333
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982725688122745e-05,
+      "loss": 1.1439,
+      "step": 334
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982431748683415e-05,
+      "loss": 1.2537,
+      "step": 335
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982135331661428e-05,
+      "loss": 1.2192,
+      "step": 336
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981836437130344e-05,
+      "loss": 1.0391,
+      "step": 337
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981535065164356e-05,
+      "loss": 1.222,
+      "step": 338
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981231215838255e-05,
+      "loss": 1.1631,
+      "step": 339
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998092488922746e-05,
+      "loss": 1.0636,
+      "step": 340
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980616085407992e-05,
+      "loss": 1.1935,
+      "step": 341
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980304804456504e-05,
+      "loss": 1.0963,
+      "step": 342
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997999104645025e-05,
+      "loss": 1.2887,
+      "step": 343
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99796748114671e-05,
+      "loss": 1.2237,
+      "step": 344
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997935609958555e-05,
+      "loss": 1.1726,
+      "step": 345
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979034910884692e-05,
+      "loss": 1.2001,
+      "step": 346
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997871124544425e-05,
+      "loss": 1.2477,
+      "step": 347
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978385103344556e-05,
+      "loss": 1.2841,
+      "step": 348
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997805648466655e-05,
+      "loss": 1.1797,
+      "step": 349
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99777253894918e-05,
+      "loss": 1.2693,
+      "step": 350
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997739181790248e-05,
+      "loss": 1.2907,
+      "step": 351
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997705576998138e-05,
+      "loss": 1.0352,
+      "step": 352
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976717245811907e-05,
+      "loss": 1.303,
+      "step": 353
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976376245478078e-05,
+      "loss": 1.3406,
+      "step": 354
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976032769064534e-05,
+      "loss": 1.1364,
+      "step": 355
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975686816656514e-05,
+      "loss": 1.1941,
+      "step": 356
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975338388339885e-05,
+      "loss": 1.0285,
+      "step": 357
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9974987484201125e-05,
+      "loss": 1.1822,
+      "step": 358
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.997463410432733e-05,
+      "loss": 1.1277,
+      "step": 359
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9974278248806204e-05,
+      "loss": 1.3741,
+      "step": 360
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9973919917726067e-05,
+      "loss": 1.0813,
+      "step": 361
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.997355911117586e-05,
+      "loss": 1.2238,
+      "step": 362
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9973195829245123e-05,
+      "loss": 1.0725,
+      "step": 363
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9972830072024033e-05,
+      "loss": 1.2662,
+      "step": 364
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.997246183960336e-05,
+      "loss": 1.0584,
+      "step": 365
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9972091132074504e-05,
+      "loss": 1.1498,
+      "step": 366
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9971717949529464e-05,
+      "loss": 1.0336,
+      "step": 367
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9971342292060868e-05,
+      "loss": 1.2075,
+      "step": 368
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.997096415976195e-05,
+      "loss": 1.0564,
+      "step": 369
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.997058355272656e-05,
+      "loss": 1.2663,
+      "step": 370
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.997020047104916e-05,
+      "loss": 1.1597,
+      "step": 371
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9969814914824838e-05,
+      "loss": 1.1787,
+      "step": 372
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9969426884149277e-05,
+      "loss": 1.3945,
+      "step": 373
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9969036379118788e-05,
+      "loss": 1.2097,
+      "step": 374
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9968643399830293e-05,
+      "loss": 1.2874,
+      "step": 375
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9968247946381325e-05,
+      "loss": 1.0676,
+      "step": 376
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9967850018870033e-05,
+      "loss": 1.1655,
+      "step": 377
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9967449617395182e-05,
+      "loss": 1.2598,
+      "step": 378
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9967046742056152e-05,
+      "loss": 1.384,
+      "step": 379
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966641392952928e-05,
+      "loss": 1.2443,
+      "step": 380
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966233570186116e-05,
+      "loss": 1.3047,
+      "step": 381
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9965823273856944e-05,
+      "loss": 1.1987,
+      "step": 382
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9965410504067238e-05,
+      "loss": 1.1483,
+      "step": 383
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964995260919445e-05,
+      "loss": 1.1158,
+      "step": 384
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964577544516627e-05,
+      "loss": 1.3446,
+      "step": 385
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996415735496246e-05,
+      "loss": 1.3159,
+      "step": 386
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996373469236123e-05,
+      "loss": 1.2641,
+      "step": 387
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9963309556817844e-05,
+      "loss": 1.1321,
+      "step": 388
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996288194843781e-05,
+      "loss": 1.1345,
+      "step": 389
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9962451867327267e-05,
+      "loss": 1.3351,
+      "step": 390
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9962019313592957e-05,
+      "loss": 1.2077,
+      "step": 391
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9961584287342234e-05,
+      "loss": 1.2169,
+      "step": 392
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996114678868307e-05,
+      "loss": 1.2962,
+      "step": 393
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996070681772405e-05,
+      "loss": 1.3075,
+      "step": 394
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9960264374574372e-05,
+      "loss": 1.1565,
+      "step": 395
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9959819459343847e-05,
+      "loss": 1.0678,
+      "step": 396
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9959372072142905e-05,
+      "loss": 1.1552,
+      "step": 397
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995892221308258e-05,
+      "loss": 1.0515,
+      "step": 398
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9958469882274526e-05,
+      "loss": 1.1255,
+      "step": 399
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9958015079831007e-05,
+      "loss": 1.0588,
+      "step": 400
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9957557805864906e-05,
+      "loss": 1.2171,
+      "step": 401
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9957098060489716e-05,
+      "loss": 1.0264,
+      "step": 402
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9956635843819542e-05,
+      "loss": 1.1807,
+      "step": 403
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99561711559691e-05,
+      "loss": 1.1294,
+      "step": 404
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995570399705373e-05,
+      "loss": 1.2536,
+      "step": 405
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9955234367189367e-05,
+      "loss": 1.2074,
+      "step": 406
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995476226649258e-05,
+      "loss": 1.4134,
+      "step": 407
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9954287695080543e-05,
+      "loss": 1.1263,
+      "step": 408
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9953810653071035e-05,
+      "loss": 1.173,
+      "step": 409
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9953331140582457e-05,
+      "loss": 1.0309,
+      "step": 410
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9952849157733825e-05,
+      "loss": 1.2331,
+      "step": 411
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995236470464476e-05,
+      "loss": 1.1827,
+      "step": 412
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9951877781435502e-05,
+      "loss": 1.1748,
+      "step": 413
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9951388388226903e-05,
+      "loss": 1.2454,
+      "step": 414
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9950896525140432e-05,
+      "loss": 1.2183,
+      "step": 415
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9950402192298156e-05,
+      "loss": 1.0589,
+      "step": 416
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9949905389822774e-05,
+      "loss": 1.0817,
+      "step": 417
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994940611783759e-05,
+      "loss": 1.0567,
+      "step": 418
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9948904376466513e-05,
+      "loss": 1.16,
+      "step": 419
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994840016583408e-05,
+      "loss": 1.1344,
+      "step": 420
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994789348606543e-05,
+      "loss": 1.1571,
+      "step": 421
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994738433728632e-05,
+      "loss": 1.0568,
+      "step": 422
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.994687271962311e-05,
+      "loss": 1.1559,
+      "step": 423
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.994635863320279e-05,
+      "loss": 0.9645,
+      "step": 424
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.994584207815295e-05,
+      "loss": 1.1431,
+      "step": 425
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99453230546018e-05,
+      "loss": 1.2563,
+      "step": 426
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9944801562678148e-05,
+      "loss": 1.1151,
+      "step": 427
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9944277602511437e-05,
+      "loss": 1.2635,
+      "step": 428
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9943751174231702e-05,
+      "loss": 1.1574,
+      "step": 429
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9943222277969606e-05,
+      "loss": 1.1309,
+      "step": 430
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9942690913856417e-05,
+      "loss": 1.2307,
+      "step": 431
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9942157082024014e-05,
+      "loss": 1.107,
+      "step": 432
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.994162078260489e-05,
+      "loss": 1.1441,
+      "step": 433
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9941082015732155e-05,
+      "loss": 1.2949,
+      "step": 434
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9940540781539526e-05,
+      "loss": 1.219,
+      "step": 435
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9939997080161335e-05,
+      "loss": 1.1883,
+      "step": 436
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9939450911732527e-05,
+      "loss": 1.1526,
+      "step": 437
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9938902276388652e-05,
+      "loss": 1.0498,
+      "step": 438
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9938351174265883e-05,
+      "loss": 1.0152,
+      "step": 439
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9937797605501003e-05,
+      "loss": 1.0395,
+      "step": 440
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9937241570231398e-05,
+      "loss": 1.2447,
+      "step": 441
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9936683068595078e-05,
+      "loss": 1.2001,
+      "step": 442
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993612210073066e-05,
+      "loss": 1.1068,
+      "step": 443
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9935558666777364e-05,
+      "loss": 1.0292,
+      "step": 444
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9934992766875044e-05,
+      "loss": 1.2299,
+      "step": 445
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9934424401164147e-05,
+      "loss": 1.1595,
+      "step": 446
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993385356978574e-05,
+      "loss": 1.2735,
+      "step": 447
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9933280272881496e-05,
+      "loss": 1.0637,
+      "step": 448
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9932704510593704e-05,
+      "loss": 1.1929,
+      "step": 449
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9932126283065268e-05,
+      "loss": 1.1345,
+      "step": 450
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9931545590439705e-05,
+      "loss": 1.1108,
+      "step": 451
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9930962432861133e-05,
+      "loss": 1.2778,
+      "step": 452
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9930376810474286e-05,
+      "loss": 1.2382,
+      "step": 453
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992978872342452e-05,
+      "loss": 1.1103,
+      "step": 454
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9929198171857788e-05,
+      "loss": 1.1918,
+      "step": 455
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9928605155920667e-05,
+      "loss": 1.3297,
+      "step": 456
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9928009675760342e-05,
+      "loss": 1.1172,
+      "step": 457
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9927411731524597e-05,
+      "loss": 1.1583,
+      "step": 458
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9926811323361848e-05,
+      "loss": 1.2474,
+      "step": 459
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992620845142111e-05,
+      "loss": 1.2394,
+      "step": 460
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992560311585201e-05,
+      "loss": 1.2265,
+      "step": 461
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9924995316804792e-05,
+      "loss": 1.0854,
+      "step": 462
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992438505443031e-05,
+      "loss": 1.0985,
+      "step": 463
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992377232888002e-05,
+      "loss": 1.1375,
+      "step": 464
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992315714030601e-05,
+      "loss": 1.1055,
+      "step": 465
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9922539488860953e-05,
+      "loss": 1.1155,
+      "step": 466
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992191937469815e-05,
+      "loss": 1.2003,
+      "step": 467
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9921296797971516e-05,
+      "loss": 1.1317,
+      "step": 468
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9920671758835565e-05,
+      "loss": 1.3758,
+      "step": 469
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992004425744543e-05,
+      "loss": 1.2244,
+      "step": 470
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991941429395685e-05,
+      "loss": 1.3872,
+      "step": 471
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9918781868526186e-05,
+      "loss": 1.1275,
+      "step": 472
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9918146981310394e-05,
+      "loss": 1.1007,
+      "step": 473
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991750963246706e-05,
+      "loss": 1.0842,
+      "step": 474
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9916869822154357e-05,
+      "loss": 1.1911,
+      "step": 475
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991622755053109e-05,
+      "loss": 1.1145,
+      "step": 476
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9915582817756667e-05,
+      "loss": 1.2263,
+      "step": 477
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9914935623991104e-05,
+      "loss": 1.1236,
+      "step": 478
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9914285969395034e-05,
+      "loss": 1.2748,
+      "step": 479
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99136338541297e-05,
+      "loss": 1.219,
+      "step": 480
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9912979278356944e-05,
+      "loss": 1.1269,
+      "step": 481
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991232224223924e-05,
+      "loss": 1.1904,
+      "step": 482
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9911662745939647e-05,
+      "loss": 1.1011,
+      "step": 483
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9911000789621857e-05,
+      "loss": 1.2003,
+      "step": 484
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9910336373450165e-05,
+      "loss": 1.1809,
+      "step": 485
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9909669497589473e-05,
+      "loss": 1.1622,
+      "step": 486
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9909000162205294e-05,
+      "loss": 1.2697,
+      "step": 487
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.990832836746375e-05,
+      "loss": 1.1622,
+      "step": 488
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.990765411353159e-05,
+      "loss": 1.11,
+      "step": 489
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9906977400576146e-05,
+      "loss": 1.2051,
+      "step": 490
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.990629822876538e-05,
+      "loss": 1.1487,
+      "step": 491
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.990561659826786e-05,
+      "loss": 1.2339,
+      "step": 492
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9904932509252756e-05,
+      "loss": 1.3189,
+      "step": 493
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9904245961889864e-05,
+      "loss": 1.2416,
+      "step": 494
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9903556956349576e-05,
+      "loss": 1.1675,
+      "step": 495
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.99028654928029e-05,
+      "loss": 1.0259,
+      "step": 496
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9902171571421455e-05,
+      "loss": 1.1767,
+      "step": 497
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9901475192377466e-05,
+      "loss": 1.3208,
+      "step": 498
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.990077635584377e-05,
+      "loss": 1.1583,
+      "step": 499
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9900075061993814e-05,
+      "loss": 1.1918,
+      "step": 500
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9899371311001664e-05,
+      "loss": 1.1492,
+      "step": 501
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9898665103041972e-05,
+      "loss": 1.1484,
+      "step": 502
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9897956438290027e-05,
+      "loss": 1.2357,
+      "step": 503
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9897245316921707e-05,
+      "loss": 1.1118,
+      "step": 504
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9896531739113513e-05,
+      "loss": 1.0362,
+      "step": 505
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9895815705042552e-05,
+      "loss": 1.1808,
+      "step": 506
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989509721488654e-05,
+      "loss": 0.8972,
+      "step": 507
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9894376268823793e-05,
+      "loss": 1.1433,
+      "step": 508
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9893652867033256e-05,
+      "loss": 1.3061,
+      "step": 509
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989292700969447e-05,
+      "loss": 1.1469,
+      "step": 510
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9892198696987586e-05,
+      "loss": 1.2635,
+      "step": 511
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9891467929093374e-05,
+      "loss": 1.2771,
+      "step": 512
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98907347061932e-05,
+      "loss": 1.1111,
+      "step": 513
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9889999028469047e-05,
+      "loss": 1.1513,
+      "step": 514
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9889260896103507e-05,
+      "loss": 1.2129,
+      "step": 515
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9888520309279782e-05,
+      "loss": 1.1466,
+      "step": 516
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9887777268181676e-05,
+      "loss": 1.2477,
+      "step": 517
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9887031772993614e-05,
+      "loss": 1.1457,
+      "step": 518
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988628382390062e-05,
+      "loss": 1.1954,
+      "step": 519
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9885533421088327e-05,
+      "loss": 1.0862,
+      "step": 520
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988478056474299e-05,
+      "loss": 1.2938,
+      "step": 521
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9884025255051456e-05,
+      "loss": 1.1606,
+      "step": 522
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988326749220119e-05,
+      "loss": 0.9742,
+      "step": 523
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9882507276380268e-05,
+      "loss": 1.0948,
+      "step": 524
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9881744607777368e-05,
+      "loss": 1.1093,
+      "step": 525
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988097948658178e-05,
+      "loss": 1.199,
+      "step": 526
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9880211912983407e-05,
+      "loss": 1.0232,
+      "step": 527
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987944188717275e-05,
+      "loss": 1.1251,
+      "step": 528
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987866940934093e-05,
+      "loss": 1.2695,
+      "step": 529
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9877894479679666e-05,
+      "loss": 1.0915,
+      "step": 530
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9877117098381298e-05,
+      "loss": 1.143,
+      "step": 531
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987633726563876e-05,
+      "loss": 1.2711,
+      "step": 532
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987555498164561e-05,
+      "loss": 1.2338,
+      "step": 533
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9874770246596003e-05,
+      "loss": 1.176,
+      "step": 534
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9873983060684706e-05,
+      "loss": 1.1671,
+      "step": 535
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987319342410709e-05,
+      "loss": 1.1318,
+      "step": 536
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9872401337059146e-05,
+      "loss": 1.2934,
+      "step": 537
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9871606799737458e-05,
+      "loss": 1.1866,
+      "step": 538
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987080981233923e-05,
+      "loss": 1.1377,
+      "step": 539
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987001037506227e-05,
+      "loss": 1.1906,
+      "step": 540
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986920848810499e-05,
+      "loss": 1.2589,
+      "step": 541
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986840415166642e-05,
+      "loss": 1.0893,
+      "step": 542
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9867597365946186e-05,
+      "loss": 1.1504,
+      "step": 543
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986678813114453e-05,
+      "loss": 1.1275,
+      "step": 544
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98659764474623e-05,
+      "loss": 1.0643,
+      "step": 545
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986516231510095e-05,
+      "loss": 1.2003,
+      "step": 546
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986434573426254e-05,
+      "loss": 1.1018,
+      "step": 547
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9863526705149746e-05,
+      "loss": 1.1984,
+      "step": 548
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9862705227965845e-05,
+      "loss": 1.135,
+      "step": 549
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986188130291472e-05,
+      "loss": 1.2493,
+      "step": 550
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9861054930200872e-05,
+      "loss": 1.1685,
+      "step": 551
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986022611002939e-05,
+      "loss": 1.2419,
+      "step": 552
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9859394842605996e-05,
+      "loss": 1.3064,
+      "step": 553
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9858561128136996e-05,
+      "loss": 1.1008,
+      "step": 554
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.985772496682932e-05,
+      "loss": 1.2328,
+      "step": 555
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.985688635889049e-05,
+      "loss": 1.1261,
+      "step": 556
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.985604530452865e-05,
+      "loss": 1.1419,
+      "step": 557
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9855201803952545e-05,
+      "loss": 1.0433,
+      "step": 558
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.985435585737152e-05,
+      "loss": 1.1745,
+      "step": 559
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9853507464995546e-05,
+      "loss": 1.216,
+      "step": 560
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.985265662703518e-05,
+      "loss": 1.01,
+      "step": 561
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.98518033437016e-05,
+      "loss": 1.1691,
+      "step": 562
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.985094761520658e-05,
+      "loss": 1.1437,
+      "step": 563
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9850089441762513e-05,
+      "loss": 1.0651,
+      "step": 564
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9849228823582392e-05,
+      "loss": 1.065,
+      "step": 565
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9848365760879816e-05,
+      "loss": 0.9671,
+      "step": 566
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.984750025386899e-05,
+      "loss": 1.3074,
+      "step": 567
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.984663230276473e-05,
+      "loss": 1.1176,
+      "step": 568
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.984576190778246e-05,
+      "loss": 1.2294,
+      "step": 569
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.98448890691382e-05,
+      "loss": 1.0166,
+      "step": 570
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9844013787048592e-05,
+      "loss": 1.1462,
+      "step": 571
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.984313606173087e-05,
+      "loss": 1.1469,
+      "step": 572
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9842255893402877e-05,
+      "loss": 1.0154,
+      "step": 573
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9841373282283073e-05,
+      "loss": 1.1859,
+      "step": 574
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9840488228590518e-05,
+      "loss": 1.0744,
+      "step": 575
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9839600732544873e-05,
+      "loss": 1.2567,
+      "step": 576
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9838710794366408e-05,
+      "loss": 1.0529,
+      "step": 577
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9837818414276006e-05,
+      "loss": 1.1262,
+      "step": 578
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9836923592495148e-05,
+      "loss": 1.0815,
+      "step": 579
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983602632924592e-05,
+      "loss": 1.1605,
+      "step": 580
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983512662475102e-05,
+      "loss": 1.1344,
+      "step": 581
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983422447923376e-05,
+      "loss": 1.1683,
+      "step": 582
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9833319892918027e-05,
+      "loss": 1.2938,
+      "step": 583
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983241286602835e-05,
+      "loss": 1.1233,
+      "step": 584
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9831503398789843e-05,
+      "loss": 1.2654,
+      "step": 585
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983059149142823e-05,
+      "loss": 0.9669,
+      "step": 586
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982967714416984e-05,
+      "loss": 1.1467,
+      "step": 587
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9828760357241613e-05,
+      "loss": 1.0326,
+      "step": 588
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9827841130871085e-05,
+      "loss": 1.2119,
+      "step": 589
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9826919465286406e-05,
+      "loss": 1.1503,
+      "step": 590
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982599536071633e-05,
+      "loss": 1.106,
+      "step": 591
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9825068817390213e-05,
+      "loss": 1.1653,
+      "step": 592
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9824139835538014e-05,
+      "loss": 1.2304,
+      "step": 593
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9823208415390306e-05,
+      "loss": 1.15,
+      "step": 594
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982227455717826e-05,
+      "loss": 1.025,
+      "step": 595
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982133826113365e-05,
+      "loss": 1.178,
+      "step": 596
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9820399527488868e-05,
+      "loss": 1.0668,
+      "step": 597
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.98194583564769e-05,
+      "loss": 1.1736,
+      "step": 598
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9818514748331334e-05,
+      "loss": 1.0605,
+      "step": 599
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981756870328637e-05,
+      "loss": 1.1112,
+      "step": 600
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9816620221576816e-05,
+      "loss": 1.0588,
+      "step": 601
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9815669303438077e-05,
+      "loss": 1.1009,
+      "step": 602
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9814715949106163e-05,
+      "loss": 1.1476,
+      "step": 603
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9813760158817692e-05,
+      "loss": 1.1584,
+      "step": 604
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9812801932809887e-05,
+      "loss": 1.1529,
+      "step": 605
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981184127132057e-05,
+      "loss": 1.0786,
+      "step": 606
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9810878174588185e-05,
+      "loss": 1.1983,
+      "step": 607
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9809912642851753e-05,
+      "loss": 1.2567,
+      "step": 608
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9808944676350916e-05,
+      "loss": 1.332,
+      "step": 609
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980797427532592e-05,
+      "loss": 1.1061,
+      "step": 610
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9807001440017615e-05,
+      "loss": 1.222,
+      "step": 611
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9806026170667453e-05,
+      "loss": 1.1316,
+      "step": 612
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9805048467517483e-05,
+      "loss": 1.1607,
+      "step": 613
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9804068330810376e-05,
+      "loss": 0.9948,
+      "step": 614
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9803085760789383e-05,
+      "loss": 1.219,
+      "step": 615
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980210075769839e-05,
+      "loss": 1.173,
+      "step": 616
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9801113321781855e-05,
+      "loss": 1.2159,
+      "step": 617
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9800123453284857e-05,
+      "loss": 1.0799,
+      "step": 618
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.979913115245308e-05,
+      "loss": 1.1536,
+      "step": 619
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97981364195328e-05,
+      "loss": 1.1311,
+      "step": 620
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9797139254770913e-05,
+      "loss": 1.1199,
+      "step": 621
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9796139658414907e-05,
+      "loss": 1.0921,
+      "step": 622
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.979513763071287e-05,
+      "loss": 1.2824,
+      "step": 623
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9794133171913503e-05,
+      "loss": 1.1883,
+      "step": 624
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.979312628226611e-05,
+      "loss": 1.1547,
+      "step": 625
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9792116962020594e-05,
+      "loss": 1.1677,
+      "step": 626
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.979110521142746e-05,
+      "loss": 1.1093,
+      "step": 627
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.979009103073782e-05,
+      "loss": 1.0341,
+      "step": 628
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.978907442020339e-05,
+      "loss": 1.2421,
+      "step": 629
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.978805538007649e-05,
+      "loss": 1.106,
+      "step": 630
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9787033910610026e-05,
+      "loss": 1.1244,
+      "step": 631
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.978601001205753e-05,
+      "loss": 1.1262,
+      "step": 632
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9784983684673135e-05,
+      "loss": 1.2483,
+      "step": 633
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9783954928711558e-05,
+      "loss": 1.1225,
+      "step": 634
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9782923744428136e-05,
+      "loss": 1.0119,
+      "step": 635
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9781890132078803e-05,
+      "loss": 1.0838,
+      "step": 636
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9780854091920093e-05,
+      "loss": 1.3687,
+      "step": 637
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9779815624209148e-05,
+      "loss": 1.219,
+      "step": 638
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977877472920371e-05,
+      "loss": 1.0434,
+      "step": 639
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9777731407162118e-05,
+      "loss": 1.0827,
+      "step": 640
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9776685658343323e-05,
+      "loss": 1.1337,
+      "step": 641
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9775637483006873e-05,
+      "loss": 1.0374,
+      "step": 642
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977458688141292e-05,
+      "loss": 1.0103,
+      "step": 643
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9773533853822218e-05,
+      "loss": 1.1105,
+      "step": 644
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977247840049612e-05,
+      "loss": 1.041,
+      "step": 645
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9771420521696584e-05,
+      "loss": 0.9783,
+      "step": 646
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9770360217686173e-05,
+      "loss": 0.9616,
+      "step": 647
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9769297488728042e-05,
+      "loss": 1.0361,
+      "step": 648
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9768232335085962e-05,
+      "loss": 1.0503,
+      "step": 649
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976716475702429e-05,
+      "loss": 1.1764,
+      "step": 650
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9766094754807997e-05,
+      "loss": 1.1018,
+      "step": 651
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9765022328702653e-05,
+      "loss": 1.1827,
+      "step": 652
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9763947478974425e-05,
+      "loss": 1.1302,
+      "step": 653
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9762870205890085e-05,
+      "loss": 1.1994,
+      "step": 654
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9761790509717008e-05,
+      "loss": 1.1544,
+      "step": 655
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976070839072317e-05,
+      "loss": 1.1283,
+      "step": 656
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975962384917714e-05,
+      "loss": 1.2602,
+      "step": 657
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9758536885348102e-05,
+      "loss": 1.1262,
+      "step": 658
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975744749950583e-05,
+      "loss": 1.1438,
+      "step": 659
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9756355691920703e-05,
+      "loss": 1.1049,
+      "step": 660
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9755261462863706e-05,
+      "loss": 1.1463,
+      "step": 661
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9754164812606414e-05,
+      "loss": 1.171,
+      "step": 662
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9753065741421017e-05,
+      "loss": 1.2895,
+      "step": 663
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975196424958029e-05,
+      "loss": 1.138,
+      "step": 664
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9750860337357622e-05,
+      "loss": 0.9916,
+      "step": 665
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9749754005026998e-05,
+      "loss": 1.1342,
+      "step": 666
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9748645252863e-05,
+      "loss": 1.1849,
+      "step": 667
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974753408114082e-05,
+      "loss": 1.001,
+      "step": 668
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9746420490136238e-05,
+      "loss": 1.2615,
+      "step": 669
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9745304480125644e-05,
+      "loss": 1.1003,
+      "step": 670
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9744186051386024e-05,
+      "loss": 1.2176,
+      "step": 671
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9743065204194966e-05,
+      "loss": 1.15,
+      "step": 672
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9741941938830658e-05,
+      "loss": 1.2315,
+      "step": 673
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9740816255571893e-05,
+      "loss": 1.1047,
+      "step": 674
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973968815469805e-05,
+      "loss": 1.2955,
+      "step": 675
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9738557636489122e-05,
+      "loss": 1.1038,
+      "step": 676
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97374247012257e-05,
+      "loss": 1.0105,
+      "step": 677
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9736289349188966e-05,
+      "loss": 1.2397,
+      "step": 678
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9735151580660712e-05,
+      "loss": 1.0884,
+      "step": 679
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9734011395923326e-05,
+      "loss": 0.9615,
+      "step": 680
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973286879525979e-05,
+      "loss": 1.019,
+      "step": 681
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9731723778953696e-05,
+      "loss": 1.2221,
+      "step": 682
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9730576347289236e-05,
+      "loss": 1.0244,
+      "step": 683
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.972942650055118e-05,
+      "loss": 1.097,
+      "step": 684
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.972827423902493e-05,
+      "loss": 1.172,
+      "step": 685
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9727119562996462e-05,
+      "loss": 1.1702,
+      "step": 686
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9725962472752357e-05,
+      "loss": 1.1991,
+      "step": 687
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9724802968579807e-05,
+      "loss": 1.1805,
+      "step": 688
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.972364105076659e-05,
+      "loss": 1.1426,
+      "step": 689
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9722476719601086e-05,
+      "loss": 1.024,
+      "step": 690
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.972130997537228e-05,
+      "loss": 1.1141,
+      "step": 691
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9720140818369743e-05,
+      "loss": 1.2232,
+      "step": 692
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9718969248883664e-05,
+      "loss": 1.1951,
+      "step": 693
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.971779526720481e-05,
+      "loss": 1.0859,
+      "step": 694
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9716618873624564e-05,
+      "loss": 1.1002,
+      "step": 695
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9715440068434892e-05,
+      "loss": 1.1003,
+      "step": 696
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9714258851928375e-05,
+      "loss": 1.1215,
+      "step": 697
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9713075224398182e-05,
+      "loss": 1.157,
+      "step": 698
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.971188918613808e-05,
+      "loss": 1.0928,
+      "step": 699
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9710700737442436e-05,
+      "loss": 1.2873,
+      "step": 700
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9709509878606222e-05,
+      "loss": 1.1619,
+      "step": 701
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9708316609924996e-05,
+      "loss": 1.2396,
+      "step": 702
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9707120931694925e-05,
+      "loss": 1.1746,
+      "step": 703
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970592284421277e-05,
+      "loss": 1.0436,
+      "step": 704
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9704722347775883e-05,
+      "loss": 1.084,
+      "step": 705
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9703519442682228e-05,
+      "loss": 1.3426,
+      "step": 706
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9702314129230354e-05,
+      "loss": 1.0406,
+      "step": 707
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9701106407719414e-05,
+      "loss": 1.1885,
+      "step": 708
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9699896278449162e-05,
+      "loss": 1.2555,
+      "step": 709
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969868374171994e-05,
+      "loss": 1.007,
+      "step": 710
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9697468797832694e-05,
+      "loss": 1.1248,
+      "step": 711
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9696251447088966e-05,
+      "loss": 1.048,
+      "step": 712
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9695031689790895e-05,
+      "loss": 0.9669,
+      "step": 713
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969380952624122e-05,
+      "loss": 1.0092,
+      "step": 714
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9692584956743275e-05,
+      "loss": 1.1906,
+      "step": 715
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9691357981600992e-05,
+      "loss": 1.1046,
+      "step": 716
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9690128601118893e-05,
+      "loss": 1.0271,
+      "step": 717
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968889681560211e-05,
+      "loss": 1.1048,
+      "step": 718
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9687662625356362e-05,
+      "loss": 1.128,
+      "step": 719
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968642603068797e-05,
+      "loss": 1.1148,
+      "step": 720
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9685187031903845e-05,
+      "loss": 1.1194,
+      "step": 721
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9683945629311503e-05,
+      "loss": 1.0584,
+      "step": 722
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9682701823219056e-05,
+      "loss": 1.0661,
+      "step": 723
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9681455613935204e-05,
+      "loss": 1.0493,
+      "step": 724
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9680207001769253e-05,
+      "loss": 1.1642,
+      "step": 725
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9678955987031098e-05,
+      "loss": 1.2478,
+      "step": 726
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9677702570031237e-05,
+      "loss": 1.1647,
+      "step": 727
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9676446751080763e-05,
+      "loss": 1.1137,
+      "step": 728
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9675188530491356e-05,
+      "loss": 1.0703,
+      "step": 729
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96739279085753e-05,
+      "loss": 1.0405,
+      "step": 730
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967266488564548e-05,
+      "loss": 1.1865,
+      "step": 731
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967139946201537e-05,
+      "loss": 1.1197,
+      "step": 732
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9670131637999037e-05,
+      "loss": 1.1426,
+      "step": 733
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966886141391115e-05,
+      "loss": 1.1351,
+      "step": 734
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9667588790066975e-05,
+      "loss": 1.0114,
+      "step": 735
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966631376678236e-05,
+      "loss": 1.0953,
+      "step": 736
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9665036344373768e-05,
+      "loss": 1.1067,
+      "step": 737
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9663756523158244e-05,
+      "loss": 1.0408,
+      "step": 738
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9662474303453432e-05,
+      "loss": 1.1591,
+      "step": 739
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9661189685577572e-05,
+      "loss": 1.1264,
+      "step": 740
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96599026698495e-05,
+      "loss": 1.184,
+      "step": 741
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965861325658865e-05,
+      "loss": 0.9564,
+      "step": 742
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9657321446115035e-05,
+      "loss": 1.0431,
+      "step": 743
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9656027238749287e-05,
+      "loss": 1.0792,
+      "step": 744
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9654730634812615e-05,
+      "loss": 1.1447,
+      "step": 745
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965343163462683e-05,
+      "loss": 1.1953,
+      "step": 746
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9652130238514334e-05,
+      "loss": 1.072,
+      "step": 747
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9650826446798132e-05,
+      "loss": 0.9713,
+      "step": 748
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9649520259801814e-05,
+      "loss": 1.1718,
+      "step": 749
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9648211677849567e-05,
+      "loss": 1.1544,
+      "step": 750
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9646900701266176e-05,
+      "loss": 1.2057,
+      "step": 751
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9645587330377017e-05,
+      "loss": 1.0951,
+      "step": 752
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.964427156550806e-05,
+      "loss": 1.1944,
+      "step": 753
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9642953406985873e-05,
+      "loss": 1.0012,
+      "step": 754
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.964163285513761e-05,
+      "loss": 1.0314,
+      "step": 755
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.964030991029103e-05,
+      "loss": 1.0294,
+      "step": 756
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.963898457277448e-05,
+      "loss": 1.1466,
+      "step": 757
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.96376568429169e-05,
+      "loss": 1.1065,
+      "step": 758
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9636326721047823e-05,
+      "loss": 1.2361,
+      "step": 759
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.963499420749738e-05,
+      "loss": 1.1083,
+      "step": 760
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9633659302596296e-05,
+      "loss": 1.1246,
+      "step": 761
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9632322006675875e-05,
+      "loss": 1.2012,
+      "step": 762
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9630982320068045e-05,
+      "loss": 1.0502,
+      "step": 763
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9629640243105292e-05,
+      "loss": 1.0504,
+      "step": 764
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.962829577612072e-05,
+      "loss": 1.0918,
+      "step": 765
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9626948919448018e-05,
+      "loss": 1.1046,
+      "step": 766
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9625599673421464e-05,
+      "loss": 1.1382,
+      "step": 767
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9624248038375934e-05,
+      "loss": 1.1115,
+      "step": 768
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.96228940146469e-05,
+      "loss": 1.0857,
+      "step": 769
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9621537602570422e-05,
+      "loss": 1.1108,
+      "step": 770
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.962017880248315e-05,
+      "loss": 1.2288,
+      "step": 771
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9618817614722333e-05,
+      "loss": 1.1132,
+      "step": 772
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9617454039625808e-05,
+      "loss": 0.998,
+      "step": 773
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9616088077532013e-05,
+      "loss": 1.3006,
+      "step": 774
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9614719728779964e-05,
+      "loss": 1.1791,
+      "step": 775
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9613348993709284e-05,
+      "loss": 1.2161,
+      "step": 776
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9611975872660176e-05,
+      "loss": 1.0992,
+      "step": 777
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9610600365973444e-05,
+      "loss": 1.1716,
+      "step": 778
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960922247399048e-05,
+      "loss": 0.9993,
+      "step": 779
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9607842197053268e-05,
+      "loss": 1.2164,
+      "step": 780
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9606459535504387e-05,
+      "loss": 1.0295,
+      "step": 781
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9605074489687005e-05,
+      "loss": 1.1225,
+      "step": 782
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9603687059944882e-05,
+      "loss": 1.1165,
+      "step": 783
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960229724662237e-05,
+      "loss": 1.0025,
+      "step": 784
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9600905050064415e-05,
+      "loss": 1.0302,
+      "step": 785
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959951047061655e-05,
+      "loss": 1.0917,
+      "step": 786
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9598113508624898e-05,
+      "loss": 0.993,
+      "step": 787
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9596714164436183e-05,
+      "loss": 1.0222,
+      "step": 788
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9595312438397712e-05,
+      "loss": 1.1784,
+      "step": 789
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959390833085739e-05,
+      "loss": 1.127,
+      "step": 790
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.95925018421637e-05,
+      "loss": 1.0678,
+      "step": 791
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959109297266573e-05,
+      "loss": 1.2166,
+      "step": 792
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9589681722713147e-05,
+      "loss": 0.9771,
+      "step": 793
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9588268092656225e-05,
+      "loss": 1.1633,
+      "step": 794
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9586852082845815e-05,
+      "loss": 1.0938,
+      "step": 795
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958543369363336e-05,
+      "loss": 1.0707,
+      "step": 796
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9584012925370898e-05,
+      "loss": 1.1805,
+      "step": 797
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9582589778411055e-05,
+      "loss": 1.0429,
+      "step": 798
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958116425310705e-05,
+      "loss": 1.0732,
+      "step": 799
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9579736349812684e-05,
+      "loss": 1.149,
+      "step": 800
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9578306068882364e-05,
+      "loss": 1.14,
+      "step": 801
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957687341067107e-05,
+      "loss": 1.1652,
+      "step": 802
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9575438375534384e-05,
+      "loss": 1.0923,
+      "step": 803
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957400096382847e-05,
+      "loss": 1.1286,
+      "step": 804
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957256117591009e-05,
+      "loss": 0.9655,
+      "step": 805
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9571119012136585e-05,
+      "loss": 1.1161,
+      "step": 806
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.95696744728659e-05,
+      "loss": 1.2071,
+      "step": 807
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956822755845655e-05,
+      "loss": 1.0253,
+      "step": 808
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956677826926766e-05,
+      "loss": 1.0741,
+      "step": 809
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9565326605658934e-05,
+      "loss": 1.0474,
+      "step": 810
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9563872567990666e-05,
+      "loss": 1.3344,
+      "step": 811
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956241615662374e-05,
+      "loss": 1.086,
+      "step": 812
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9560957371919625e-05,
+      "loss": 0.9903,
+      "step": 813
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9559496214240386e-05,
+      "loss": 1.0877,
+      "step": 814
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.955803268394868e-05,
+      "loss": 0.9842,
+      "step": 815
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9556566781407733e-05,
+      "loss": 1.1079,
+      "step": 816
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9555098506981386e-05,
+      "loss": 0.9983,
+      "step": 817
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9553627861034052e-05,
+      "loss": 1.1249,
+      "step": 818
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.955215484393074e-05,
+      "loss": 1.1338,
+      "step": 819
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9550679456037038e-05,
+      "loss": 1.1489,
+      "step": 820
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9549201697719137e-05,
+      "loss": 1.1026,
+      "step": 821
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9547721569343797e-05,
+      "loss": 1.1993,
+      "step": 822
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9546239071278388e-05,
+      "loss": 1.1984,
+      "step": 823
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9544754203890853e-05,
+      "loss": 1.1216,
+      "step": 824
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9543266967549727e-05,
+      "loss": 1.0158,
+      "step": 825
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9541777362624136e-05,
+      "loss": 1.2414,
+      "step": 826
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.954028538948379e-05,
+      "loss": 1.1054,
+      "step": 827
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.953879104849899e-05,
+      "loss": 0.9906,
+      "step": 828
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.953729434004062e-05,
+      "loss": 1.0984,
+      "step": 829
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9535795264480156e-05,
+      "loss": 1.1044,
+      "step": 830
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9534293822189662e-05,
+      "loss": 1.1445,
+      "step": 831
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9532790013541783e-05,
+      "loss": 1.1973,
+      "step": 832
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.953128383890976e-05,
+      "loss": 1.1188,
+      "step": 833
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9529775298667414e-05,
+      "loss": 1.0489,
+      "step": 834
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9528264393189156e-05,
+      "loss": 1.0576,
+      "step": 835
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9526751122849986e-05,
+      "loss": 1.2415,
+      "step": 836
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9525235488025493e-05,
+      "loss": 1.1829,
+      "step": 837
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9523717489091837e-05,
+      "loss": 1.0725,
+      "step": 838
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952219712642579e-05,
+      "loss": 0.9945,
+      "step": 839
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9520674400404688e-05,
+      "loss": 1.0537,
+      "step": 840
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9519149311406464e-05,
+      "loss": 0.9974,
+      "step": 841
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.951762185980964e-05,
+      "loss": 1.0839,
+      "step": 842
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9516092045993322e-05,
+      "loss": 1.1022,
+      "step": 843
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.95145598703372e-05,
+      "loss": 1.1161,
+      "step": 844
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9513025333221548e-05,
+      "loss": 1.2691,
+      "step": 845
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.951148843502723e-05,
+      "loss": 1.0963,
+      "step": 846
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9509949176135695e-05,
+      "loss": 1.2624,
+      "step": 847
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950840755692898e-05,
+      "loss": 1.1927,
+      "step": 848
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9506863577789708e-05,
+      "loss": 1.2051,
+      "step": 849
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9505317239101085e-05,
+      "loss": 1.2116,
+      "step": 850
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.95037685412469e-05,
+      "loss": 1.1334,
+      "step": 851
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950221748461153e-05,
+      "loss": 1.1687,
+      "step": 852
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9500664069579948e-05,
+      "loss": 1.0958,
+      "step": 853
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9499108296537693e-05,
+      "loss": 1.0974,
+      "step": 854
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9497550165870902e-05,
+      "loss": 1.0004,
+      "step": 855
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9495989677966297e-05,
+      "loss": 1.1226,
+      "step": 856
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9494426833211177e-05,
+      "loss": 1.1858,
+      "step": 857
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9492861631993435e-05,
+      "loss": 0.9694,
+      "step": 858
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9491294074701544e-05,
+      "loss": 1.0671,
+      "step": 859
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9489724161724563e-05,
+      "loss": 1.1464,
+      "step": 860
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.948815189345213e-05,
+      "loss": 1.0636,
+      "step": 861
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9486577270274486e-05,
+      "loss": 1.0625,
+      "step": 862
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.948500029258243e-05,
+      "loss": 1.0961,
+      "step": 863
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9483420960767366e-05,
+      "loss": 1.0888,
+      "step": 864
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9481839275221276e-05,
+      "loss": 1.0898,
+      "step": 865
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.948025523633672e-05,
+      "loss": 0.9651,
+      "step": 866
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.947866884450685e-05,
+      "loss": 1.2204,
+      "step": 867
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9477080100125402e-05,
+      "loss": 1.2096,
+      "step": 868
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.947548900358669e-05,
+      "loss": 1.0546,
+      "step": 869
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9473895555285618e-05,
+      "loss": 1.1375,
+      "step": 870
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9472299755617666e-05,
+      "loss": 1.1282,
+      "step": 871
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.947070160497891e-05,
+      "loss": 1.177,
+      "step": 872
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946910110376599e-05,
+      "loss": 1.0467,
+      "step": 873
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9467498252376154e-05,
+      "loss": 1.0752,
+      "step": 874
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946589305120721e-05,
+      "loss": 1.018,
+      "step": 875
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946428550065757e-05,
+      "loss": 1.2202,
+      "step": 876
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9462675601126215e-05,
+      "loss": 1.1772,
+      "step": 877
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9461063353012708e-05,
+      "loss": 1.0457,
+      "step": 878
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9459448756717205e-05,
+      "loss": 1.1148,
+      "step": 879
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9457831812640438e-05,
+      "loss": 1.04,
+      "step": 880
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.945621252118372e-05,
+      "loss": 0.9924,
+      "step": 881
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.945459088274896e-05,
+      "loss": 0.9582,
+      "step": 882
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9452966897738625e-05,
+      "loss": 1.0044,
+      "step": 883
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9451340566555792e-05,
+      "loss": 1.1951,
+      "step": 884
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.94497118896041e-05,
+      "loss": 1.056,
+      "step": 885
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9448080867287778e-05,
+      "loss": 1.2618,
+      "step": 886
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.944644750001164e-05,
+      "loss": 1.1228,
+      "step": 887
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.944481178818107e-05,
+      "loss": 1.0983,
+      "step": 888
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9443173732202053e-05,
+      "loss": 1.1262,
+      "step": 889
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.944153333248114e-05,
+      "loss": 1.1165,
+      "step": 890
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9439890589425474e-05,
+      "loss": 1.1459,
+      "step": 891
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.943824550344276e-05,
+      "loss": 1.1229,
+      "step": 892
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9436598074941315e-05,
+      "loss": 1.1425,
+      "step": 893
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9434948304330014e-05,
+      "loss": 1.1292,
+      "step": 894
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9433296192018322e-05,
+      "loss": 1.0659,
+      "step": 895
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9431641738416277e-05,
+      "loss": 1.0639,
+      "step": 896
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9429984943934517e-05,
+      "loss": 1.0802,
+      "step": 897
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9428325808984244e-05,
+      "loss": 1.1049,
+      "step": 898
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9426664333977244e-05,
+      "loss": 0.9895,
+      "step": 899
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.942500051932589e-05,
+      "loss": 1.0635,
+      "step": 900
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9423334365443126e-05,
+      "loss": 1.0797,
+      "step": 901
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9421665872742483e-05,
+      "loss": 1.2152,
+      "step": 902
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9419995041638073e-05,
+      "loss": 1.0354,
+      "step": 903
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9418321872544584e-05,
+      "loss": 0.9667,
+      "step": 904
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9416646365877297e-05,
+      "loss": 1.0877,
+      "step": 905
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9414968522052046e-05,
+      "loss": 1.2036,
+      "step": 906
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9413288341485276e-05,
+      "loss": 1.0919,
+      "step": 907
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9411605824593997e-05,
+      "loss": 1.0514,
+      "step": 908
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.940992097179579e-05,
+      "loss": 1.0095,
+      "step": 909
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.940823378350884e-05,
+      "loss": 1.1649,
+      "step": 910
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9406544260151887e-05,
+      "loss": 1.0949,
+      "step": 911
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9404852402144267e-05,
+      "loss": 1.1621,
+      "step": 912
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9403158209905883e-05,
+      "loss": 0.9386,
+      "step": 913
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.940146168385723e-05,
+      "loss": 1.1234,
+      "step": 914
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9399762824419376e-05,
+      "loss": 0.8643,
+      "step": 915
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9398061632013965e-05,
+      "loss": 1.1637,
+      "step": 916
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.939635810706322e-05,
+      "loss": 1.0842,
+      "step": 917
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9394652249989958e-05,
+      "loss": 1.1139,
+      "step": 918
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.939294406121755e-05,
+      "loss": 1.1367,
+      "step": 919
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.939123354116997e-05,
+      "loss": 1.214,
+      "step": 920
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.938952069027175e-05,
+      "loss": 1.0064,
+      "step": 921
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.938780550894801e-05,
+      "loss": 1.1504,
+      "step": 922
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9386087997624457e-05,
+      "loss": 1.0831,
+      "step": 923
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.938436815672736e-05,
+      "loss": 1.0498,
+      "step": 924
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.938264598668358e-05,
+      "loss": 1.1898,
+      "step": 925
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9380921487920543e-05,
+      "loss": 1.0505,
+      "step": 926
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.937919466086626e-05,
+      "loss": 1.1299,
+      "step": 927
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9377465505949323e-05,
+      "loss": 1.101,
+      "step": 928
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.93757340235989e-05,
+      "loss": 1.1078,
+      "step": 929
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9374000214244734e-05,
+      "loss": 1.1519,
+      "step": 930
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9372264078317143e-05,
+      "loss": 0.937,
+      "step": 931
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9370525616247025e-05,
+      "loss": 1.1587,
+      "step": 932
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9368784828465866e-05,
+      "loss": 1.0941,
+      "step": 933
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.936704171540571e-05,
+      "loss": 1.0332,
+      "step": 934
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.936529627749919e-05,
+      "loss": 0.9473,
+      "step": 935
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9363548515179516e-05,
+      "loss": 1.1054,
+      "step": 936
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9361798428880472e-05,
+      "loss": 1.16,
+      "step": 937
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.936004601903642e-05,
+      "loss": 1.0847,
+      "step": 938
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.935829128608229e-05,
+      "loss": 1.0169,
+      "step": 939
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.935653423045361e-05,
+      "loss": 1.1104,
+      "step": 940
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9354774852586463e-05,
+      "loss": 1.028,
+      "step": 941
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.935301315291752e-05,
+      "loss": 1.07,
+      "step": 942
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9351249131884022e-05,
+      "loss": 1.2341,
+      "step": 943
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.934948278992379e-05,
+      "loss": 1.1707,
+      "step": 944
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9347714127475222e-05,
+      "loss": 1.0277,
+      "step": 945
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9345943144977287e-05,
+      "loss": 1.0645,
+      "step": 946
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9344169842869532e-05,
+      "loss": 1.0905,
+      "step": 947
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9342394221592086e-05,
+      "loss": 0.9406,
+      "step": 948
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9340616281585644e-05,
+      "loss": 1.0043,
+      "step": 949
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.933883602329148e-05,
+      "loss": 0.9975,
+      "step": 950
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.933705344715144e-05,
+      "loss": 1.2166,
+      "step": 951
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.933526855360796e-05,
+      "loss": 0.996,
+      "step": 952
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9333481343104036e-05,
+      "loss": 0.9883,
+      "step": 953
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.933169181608324e-05,
+      "loss": 1.2324,
+      "step": 954
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.932989997298972e-05,
+      "loss": 1.0473,
+      "step": 955
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.932810581426821e-05,
+      "loss": 1.1151,
+      "step": 956
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9326309340364008e-05,
+      "loss": 1.0639,
+      "step": 957
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9324510551722982e-05,
+      "loss": 0.9717,
+      "step": 958
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9322709448791586e-05,
+      "loss": 1.008,
+      "step": 959
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9320906032016846e-05,
+      "loss": 1.0854,
+      "step": 960
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9319100301846357e-05,
+      "loss": 1.0728,
+      "step": 961
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9317292258728288e-05,
+      "loss": 1.1178,
+      "step": 962
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9315481903111387e-05,
+      "loss": 1.2257,
+      "step": 963
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9313669235444974e-05,
+      "loss": 1.0165,
+      "step": 964
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9311854256178947e-05,
+      "loss": 1.0615,
+      "step": 965
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9310036965763766e-05,
+      "loss": 0.9872,
+      "step": 966
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9308217364650478e-05,
+      "loss": 1.0435,
+      "step": 967
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9306395453290693e-05,
+      "loss": 1.0133,
+      "step": 968
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9304571232136605e-05,
+      "loss": 1.1061,
+      "step": 969
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.930274470164097e-05,
+      "loss": 1.0243,
+      "step": 970
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9300915862257126e-05,
+      "loss": 1.097,
+      "step": 971
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9299084714438975e-05,
+      "loss": 1.142,
+      "step": 972
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9297251258641004e-05,
+      "loss": 1.0346,
+      "step": 973
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9295415495318263e-05,
+      "loss": 1.1429,
+      "step": 974
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9293577424926382e-05,
+      "loss": 0.9403,
+      "step": 975
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.929173704792155e-05,
+      "loss": 1.1767,
+      "step": 976
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.928989436476055e-05,
+      "loss": 1.1384,
+      "step": 977
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9288049375900716e-05,
+      "loss": 1.1393,
+      "step": 978
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9286202081799968e-05,
+      "loss": 1.1774,
+      "step": 979
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9284352482916793e-05,
+      "loss": 1.1758,
+      "step": 980
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9282500579710254e-05,
+      "loss": 1.0961,
+      "step": 981
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9280646372639977e-05,
+      "loss": 1.1504,
+      "step": 982
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9278789862166172e-05,
+      "loss": 1.0246,
+      "step": 983
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9276931048749606e-05,
+      "loss": 1.1118,
+      "step": 984
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9275069932851635e-05,
+      "loss": 1.1114,
+      "step": 985
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.927320651493417e-05,
+      "loss": 1.0952,
+      "step": 986
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9271340795459705e-05,
+      "loss": 1.1563,
+      "step": 987
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9269472774891303e-05,
+      "loss": 1.122,
+      "step": 988
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.926760245369259e-05,
+      "loss": 1.1633,
+      "step": 989
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.926572983232777e-05,
+      "loss": 0.9387,
+      "step": 990
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.926385491126162e-05,
+      "loss": 1.1217,
+      "step": 991
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9261977690959486e-05,
+      "loss": 1.1101,
+      "step": 992
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.926009817188728e-05,
+      "loss": 0.9855,
+      "step": 993
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9258216354511486e-05,
+      "loss": 0.9299,
+      "step": 994
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.925633223929917e-05,
+      "loss": 0.9812,
+      "step": 995
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9254445826717946e-05,
+      "loss": 1.0909,
+      "step": 996
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.925255711723602e-05,
+      "loss": 1.0976,
+      "step": 997
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9250666111322154e-05,
+      "loss": 0.9763,
+      "step": 998
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9248772809445685e-05,
+      "loss": 1.1466,
+      "step": 999
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9246877212076525e-05,
+      "loss": 0.9784,
+      "step": 1000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9244979319685144e-05,
+      "loss": 1.0985,
+      "step": 1001
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9243079132742594e-05,
+      "loss": 1.2436,
+      "step": 1002
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9241176651720484e-05,
+      "loss": 1.1453,
+      "step": 1003
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9239271877091004e-05,
+      "loss": 1.0974,
+      "step": 1004
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9237364809326908e-05,
+      "loss": 1.0632,
+      "step": 1005
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.923545544890151e-05,
+      "loss": 1.216,
+      "step": 1006
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9233543796288717e-05,
+      "loss": 1.0798,
+      "step": 1007
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9231629851962984e-05,
+      "loss": 1.1072,
+      "step": 1008
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9229713616399342e-05,
+      "loss": 1.0234,
+      "step": 1009
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9227795090073384e-05,
+      "loss": 0.9929,
+      "step": 1010
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9225874273461284e-05,
+      "loss": 1.031,
+      "step": 1011
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9223951167039776e-05,
+      "loss": 1.2846,
+      "step": 1012
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9222025771286164e-05,
+      "loss": 1.0549,
+      "step": 1013
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.922009808667832e-05,
+      "loss": 1.1391,
+      "step": 1014
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.921816811369468e-05,
+      "loss": 1.0374,
+      "step": 1015
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.921623585281426e-05,
+      "loss": 1.018,
+      "step": 1016
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.921430130451663e-05,
+      "loss": 1.0041,
+      "step": 1017
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9212364469281942e-05,
+      "loss": 1.0751,
+      "step": 1018
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9210425347590897e-05,
+      "loss": 1.0045,
+      "step": 1019
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9208483939924778e-05,
+      "loss": 1.0755,
+      "step": 1020
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9206540246765434e-05,
+      "loss": 1.007,
+      "step": 1021
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9204594268595275e-05,
+      "loss": 1.0878,
+      "step": 1022
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9202646005897288e-05,
+      "loss": 1.0875,
+      "step": 1023
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9200695459155013e-05,
+      "loss": 1.029,
+      "step": 1024
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9198742628852565e-05,
+      "loss": 1.0377,
+      "step": 1025
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.919678751547463e-05,
+      "loss": 1.1402,
+      "step": 1026
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9194830119506456e-05,
+      "loss": 1.1081,
+      "step": 1027
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.919287044143385e-05,
+      "loss": 1.2334,
+      "step": 1028
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9190908481743204e-05,
+      "loss": 1.2242,
+      "step": 1029
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9188944240921458e-05,
+      "loss": 1.1979,
+      "step": 1030
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.918697771945613e-05,
+      "loss": 1.0006,
+      "step": 1031
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.918500891783529e-05,
+      "loss": 0.9906,
+      "step": 1032
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9183037836547595e-05,
+      "loss": 1.0934,
+      "step": 1033
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9181064476082247e-05,
+      "loss": 1.1323,
+      "step": 1034
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.917908883692903e-05,
+      "loss": 0.889,
+      "step": 1035
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9177110919578284e-05,
+      "loss": 1.1002,
+      "step": 1036
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.917513072452091e-05,
+      "loss": 1.024,
+      "step": 1037
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9173148252248396e-05,
+      "loss": 1.0384,
+      "step": 1038
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9171163503252766e-05,
+      "loss": 1.07,
+      "step": 1039
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.916917647802663e-05,
+      "loss": 1.0473,
+      "step": 1040
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.916718717706315e-05,
+      "loss": 1.102,
+      "step": 1041
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.916519560085607e-05,
+      "loss": 1.1739,
+      "step": 1042
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.916320174989968e-05,
+      "loss": 0.9466,
+      "step": 1043
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9161205624688844e-05,
+      "loss": 1.2008,
+      "step": 1044
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9159207225718993e-05,
+      "loss": 1.0916,
+      "step": 1045
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.915720655348611e-05,
+      "loss": 0.9209,
+      "step": 1046
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.915520360848676e-05,
+      "loss": 1.0544,
+      "step": 1047
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9153198391218056e-05,
+      "loss": 1.0869,
+      "step": 1048
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9151190902177682e-05,
+      "loss": 1.0924,
+      "step": 1049
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9149181141863888e-05,
+      "loss": 1.0605,
+      "step": 1050
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.914716911077548e-05,
+      "loss": 0.9215,
+      "step": 1051
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.914515480941184e-05,
+      "loss": 1.0468,
+      "step": 1052
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.91431382382729e-05,
+      "loss": 1.0085,
+      "step": 1053
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9141119397859165e-05,
+      "loss": 0.8895,
+      "step": 1054
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9139098288671697e-05,
+      "loss": 1.0457,
+      "step": 1055
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9137074911212128e-05,
+      "loss": 1.1352,
+      "step": 1056
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9135049265982646e-05,
+      "loss": 1.0328,
+      "step": 1057
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9133021353486004e-05,
+      "loss": 1.1211,
+      "step": 1058
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9130991174225518e-05,
+      "loss": 1.006,
+      "step": 1059
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9128958728705065e-05,
+      "loss": 0.9941,
+      "step": 1060
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9126924017429092e-05,
+      "loss": 1.0562,
+      "step": 1061
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9124887040902603e-05,
+      "loss": 1.1755,
+      "step": 1062
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9122847799631155e-05,
+      "loss": 1.029,
+      "step": 1063
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.912080629412089e-05,
+      "loss": 0.9359,
+      "step": 1064
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9118762524878484e-05,
+      "loss": 1.125,
+      "step": 1065
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9116716492411193e-05,
+      "loss": 1.0848,
+      "step": 1066
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9114668197226835e-05,
+      "loss": 1.0548,
+      "step": 1067
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9112617639833784e-05,
+      "loss": 1.1056,
+      "step": 1068
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9110564820740975e-05,
+      "loss": 1.0246,
+      "step": 1069
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9108509740457906e-05,
+      "loss": 1.0509,
+      "step": 1070
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9106452399494636e-05,
+      "loss": 0.9527,
+      "step": 1071
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.910439279836179e-05,
+      "loss": 1.2718,
+      "step": 1072
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.910233093757054e-05,
+      "loss": 1.1036,
+      "step": 1073
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9100266817632635e-05,
+      "loss": 1.0474,
+      "step": 1074
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9098200439060375e-05,
+      "loss": 1.1254,
+      "step": 1075
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9096131802366626e-05,
+      "loss": 1.034,
+      "step": 1076
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.909406090806481e-05,
+      "loss": 1.1564,
+      "step": 1077
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.909198775666891e-05,
+      "loss": 1.0855,
+      "step": 1078
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9089912348693477e-05,
+      "loss": 1.2211,
+      "step": 1079
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9087834684653604e-05,
+      "loss": 1.0823,
+      "step": 1080
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9085754765064963e-05,
+      "loss": 1.0889,
+      "step": 1081
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.908367259044378e-05,
+      "loss": 0.9916,
+      "step": 1082
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.908158816130683e-05,
+      "loss": 1.1344,
+      "step": 1083
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9079501478171466e-05,
+      "loss": 1.0226,
+      "step": 1084
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9077412541555588e-05,
+      "loss": 1.1344,
+      "step": 1085
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9075321351977654e-05,
+      "loss": 1.1377,
+      "step": 1086
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.907322790995669e-05,
+      "loss": 1.0359,
+      "step": 1087
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9071132216012272e-05,
+      "loss": 0.8762,
+      "step": 1088
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9069034270664546e-05,
+      "loss": 0.9342,
+      "step": 1089
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9066934074434204e-05,
+      "loss": 1.0511,
+      "step": 1090
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9064831627842507e-05,
+      "loss": 1.1524,
+      "step": 1091
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9062726931411267e-05,
+      "loss": 1.0833,
+      "step": 1092
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9060619985662856e-05,
+      "loss": 1.0063,
+      "step": 1093
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9058510791120212e-05,
+      "loss": 1.0964,
+      "step": 1094
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9056399348306822e-05,
+      "loss": 0.95,
+      "step": 1095
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9054285657746732e-05,
+      "loss": 1.075,
+      "step": 1096
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9052169719964553e-05,
+      "loss": 1.0865,
+      "step": 1097
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9050051535485442e-05,
+      "loss": 1.1657,
+      "step": 1098
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9047931104835126e-05,
+      "loss": 0.9458,
+      "step": 1099
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.904580842853988e-05,
+      "loss": 1.0446,
+      "step": 1100
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9043683507126546e-05,
+      "loss": 1.0099,
+      "step": 1101
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9041556341122506e-05,
+      "loss": 1.1347,
+      "step": 1102
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9039426931055723e-05,
+      "loss": 1.1387,
+      "step": 1103
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9037295277454696e-05,
+      "loss": 1.069,
+      "step": 1104
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9035161380848497e-05,
+      "loss": 1.0813,
+      "step": 1105
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9033025241766738e-05,
+      "loss": 1.0344,
+      "step": 1106
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9030886860739603e-05,
+      "loss": 1.0764,
+      "step": 1107
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9028746238297824e-05,
+      "loss": 1.0761,
+      "step": 1108
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.902660337497269e-05,
+      "loss": 1.1148,
+      "step": 1109
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9024458271296046e-05,
+      "loss": 1.0626,
+      "step": 1110
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.90223109278003e-05,
+      "loss": 1.1283,
+      "step": 1111
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.902016134501841e-05,
+      "loss": 1.1683,
+      "step": 1112
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9018009523483886e-05,
+      "loss": 0.9798,
+      "step": 1113
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9015855463730798e-05,
+      "loss": 0.9688,
+      "step": 1114
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.901369916629377e-05,
+      "loss": 1.0602,
+      "step": 1115
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9011540631707996e-05,
+      "loss": 0.982,
+      "step": 1116
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.900937986050919e-05,
+      "loss": 1.0513,
+      "step": 1117
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9007216853233662e-05,
+      "loss": 1.0117,
+      "step": 1118
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.900505161041825e-05,
+      "loss": 1.1315,
+      "step": 1119
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9002884132600355e-05,
+      "loss": 1.2387,
+      "step": 1120
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9000714420317932e-05,
+      "loss": 1.1309,
+      "step": 1121
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8998542474109497e-05,
+      "loss": 1.0532,
+      "step": 1122
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8996368294514104e-05,
+      "loss": 1.0683,
+      "step": 1123
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8994191882071384e-05,
+      "loss": 1.1066,
+      "step": 1124
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8992013237321497e-05,
+      "loss": 1.0756,
+      "step": 1125
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8989832360805186e-05,
+      "loss": 0.9869,
+      "step": 1126
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.898764925306372e-05,
+      "loss": 1.126,
+      "step": 1127
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8985463914638935e-05,
+      "loss": 1.0687,
+      "step": 1128
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8983276346073226e-05,
+      "loss": 1.1383,
+      "step": 1129
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8981086547909532e-05,
+      "loss": 1.0577,
+      "step": 1130
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8978894520691345e-05,
+      "loss": 1.0848,
+      "step": 1131
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8976700264962718e-05,
+      "loss": 1.0381,
+      "step": 1132
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.897450378126825e-05,
+      "loss": 1.2279,
+      "step": 1133
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8972305070153097e-05,
+      "loss": 1.0982,
+      "step": 1134
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8970104132162965e-05,
+      "loss": 1.1053,
+      "step": 1135
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.896790096784412e-05,
+      "loss": 1.0028,
+      "step": 1136
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8965695577743365e-05,
+      "loss": 0.9607,
+      "step": 1137
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8963487962408074e-05,
+      "loss": 0.9714,
+      "step": 1138
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8961278122386158e-05,
+      "loss": 1.0072,
+      "step": 1139
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8959066058226095e-05,
+      "loss": 1.213,
+      "step": 1140
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8956851770476897e-05,
+      "loss": 1.0835,
+      "step": 1141
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8954635259688147e-05,
+      "loss": 1.0723,
+      "step": 1142
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8952416526409966e-05,
+      "loss": 1.0573,
+      "step": 1143
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8950195571193026e-05,
+      "loss": 0.9361,
+      "step": 1144
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.894797239458856e-05,
+      "loss": 1.1175,
+      "step": 1145
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.894574699714835e-05,
+      "loss": 1.1283,
+      "step": 1146
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8943519379424727e-05,
+      "loss": 0.9509,
+      "step": 1147
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8941289541970566e-05,
+      "loss": 1.0001,
+      "step": 1148
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8939057485339306e-05,
+      "loss": 1.007,
+      "step": 1149
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8936823210084934e-05,
+      "loss": 1.2032,
+      "step": 1150
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8934586716761972e-05,
+      "loss": 1.0452,
+      "step": 1151
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.893234800592552e-05,
+      "loss": 1.0641,
+      "step": 1152
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8930107078131203e-05,
+      "loss": 1.2815,
+      "step": 1153
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8927863933935208e-05,
+      "loss": 0.9739,
+      "step": 1154
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8925618573894275e-05,
+      "loss": 1.0608,
+      "step": 1155
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8923370998565685e-05,
+      "loss": 0.9778,
+      "step": 1156
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8921121208507275e-05,
+      "loss": 0.9376,
+      "step": 1157
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.891886920427743e-05,
+      "loss": 1.1025,
+      "step": 1158
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8916614986435085e-05,
+      "loss": 1.0551,
+      "step": 1159
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8914358555539726e-05,
+      "loss": 0.988,
+      "step": 1160
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8912099912151382e-05,
+      "loss": 1.0907,
+      "step": 1161
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8909839056830635e-05,
+      "loss": 1.1114,
+      "step": 1162
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.890757599013862e-05,
+      "loss": 0.9561,
+      "step": 1163
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.890531071263702e-05,
+      "loss": 1.2335,
+      "step": 1164
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8903043224888057e-05,
+      "loss": 1.1563,
+      "step": 1165
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.890077352745451e-05,
+      "loss": 1.0233,
+      "step": 1166
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.889850162089971e-05,
+      "loss": 0.9747,
+      "step": 1167
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8896227505787528e-05,
+      "loss": 1.1285,
+      "step": 1168
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8893951182682382e-05,
+      "loss": 1.0631,
+      "step": 1169
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.889167265214925e-05,
+      "loss": 1.0367,
+      "step": 1170
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8889391914753646e-05,
+      "loss": 1.0338,
+      "step": 1171
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.888710897106164e-05,
+      "loss": 1.0375,
+      "step": 1172
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.888482382163984e-05,
+      "loss": 1.1474,
+      "step": 1173
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8882536467055406e-05,
+      "loss": 1.0286,
+      "step": 1174
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8880246907876057e-05,
+      "loss": 1.0069,
+      "step": 1175
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8877955144670034e-05,
+      "loss": 1.1015,
+      "step": 1176
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8875661178006152e-05,
+      "loss": 1.005,
+      "step": 1177
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8873365008453757e-05,
+      "loss": 1.0611,
+      "step": 1178
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.887106663658274e-05,
+      "loss": 1.1163,
+      "step": 1179
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8868766062963552e-05,
+      "loss": 1.0088,
+      "step": 1180
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8866463288167173e-05,
+      "loss": 1.0791,
+      "step": 1181
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8864158312765144e-05,
+      "loss": 0.9426,
+      "step": 1182
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.886185113732955e-05,
+      "loss": 1.0527,
+      "step": 1183
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.885954176243301e-05,
+      "loss": 1.0575,
+      "step": 1184
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8857230188648706e-05,
+      "loss": 1.0836,
+      "step": 1185
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8854916416550355e-05,
+      "loss": 1.1035,
+      "step": 1186
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8852600446712216e-05,
+      "loss": 1.0493,
+      "step": 1187
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.885028227970911e-05,
+      "loss": 1.0607,
+      "step": 1188
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8847961916116385e-05,
+      "loss": 0.9978,
+      "step": 1189
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8845639356509945e-05,
+      "loss": 1.0143,
+      "step": 1190
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8843314601466235e-05,
+      "loss": 1.0063,
+      "step": 1191
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.884098765156225e-05,
+      "loss": 0.9975,
+      "step": 1192
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.883865850737552e-05,
+      "loss": 1.1425,
+      "step": 1193
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8836327169484122e-05,
+      "loss": 1.1915,
+      "step": 1194
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8833993638466694e-05,
+      "loss": 1.0505,
+      "step": 1195
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8831657914902394e-05,
+      "loss": 1.0817,
+      "step": 1196
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.882931999937094e-05,
+      "loss": 0.965,
+      "step": 1197
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8826979892452586e-05,
+      "loss": 0.8995,
+      "step": 1198
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8824637594728135e-05,
+      "loss": 1.0087,
+      "step": 1199
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8822293106778934e-05,
+      "loss": 0.9791,
+      "step": 1200
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.881994642918687e-05,
+      "loss": 1.2353,
+      "step": 1201
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.881759756253437e-05,
+      "loss": 0.9326,
+      "step": 1202
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.881524650740442e-05,
+      "loss": 1.0108,
+      "step": 1203
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.881289326438053e-05,
+      "loss": 1.1179,
+      "step": 1204
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8810537834046763e-05,
+      "loss": 1.0539,
+      "step": 1205
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8808180216987726e-05,
+      "loss": 1.034,
+      "step": 1206
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8805820413788565e-05,
+      "loss": 1.0861,
+      "step": 1207
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8803458425034965e-05,
+      "loss": 1.1736,
+      "step": 1208
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8801094251313167e-05,
+      "loss": 1.0633,
+      "step": 1209
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.879872789320994e-05,
+      "loss": 1.0155,
+      "step": 1210
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.87963593513126e-05,
+      "loss": 1.1206,
+      "step": 1211
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.879398862620901e-05,
+      "loss": 1.0301,
+      "step": 1212
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8791615718487565e-05,
+      "loss": 1.0873,
+      "step": 1213
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8789240628737208e-05,
+      "loss": 1.0283,
+      "step": 1214
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8786863357547423e-05,
+      "loss": 0.9739,
+      "step": 1215
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.878448390550824e-05,
+      "loss": 1.059,
+      "step": 1216
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8782102273210223e-05,
+      "loss": 1.0393,
+      "step": 1217
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8779718461244475e-05,
+      "loss": 1.086,
+      "step": 1218
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.877733247020265e-05,
+      "loss": 1.0966,
+      "step": 1219
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8774944300676936e-05,
+      "loss": 1.0614,
+      "step": 1220
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.877255395326006e-05,
+      "loss": 1.0675,
+      "step": 1221
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8770161428545297e-05,
+      "loss": 0.9688,
+      "step": 1222
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8767766727126455e-05,
+      "loss": 1.0434,
+      "step": 1223
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.876536984959789e-05,
+      "loss": 1.1013,
+      "step": 1224
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8762970796554484e-05,
+      "loss": 1.0637,
+      "step": 1225
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.876056956859168e-05,
+      "loss": 1.0011,
+      "step": 1226
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8758166166305436e-05,
+      "loss": 1.0294,
+      "step": 1227
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8755760590292275e-05,
+      "loss": 1.0021,
+      "step": 1228
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.875335284114924e-05,
+      "loss": 0.9689,
+      "step": 1229
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.875094291947392e-05,
+      "loss": 1.0504,
+      "step": 1230
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.874853082586445e-05,
+      "loss": 1.1079,
+      "step": 1231
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8746116560919492e-05,
+      "loss": 0.9676,
+      "step": 1232
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.874370012523826e-05,
+      "loss": 1.0932,
+      "step": 1233
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8741281519420493e-05,
+      "loss": 0.9579,
+      "step": 1234
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8738860744066475e-05,
+      "loss": 1.0063,
+      "step": 1235
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.873643779977703e-05,
+      "loss": 1.0613,
+      "step": 1236
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.873401268715352e-05,
+      "loss": 0.9698,
+      "step": 1237
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8731585406797843e-05,
+      "loss": 1.0035,
+      "step": 1238
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.872915595931244e-05,
+      "loss": 1.0602,
+      "step": 1239
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.872672434530028e-05,
+      "loss": 0.9768,
+      "step": 1240
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.872429056536488e-05,
+      "loss": 0.9326,
+      "step": 1241
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.872185462011029e-05,
+      "loss": 1.1173,
+      "step": 1242
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.871941651014109e-05,
+      "loss": 1.0718,
+      "step": 1243
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.871697623606242e-05,
+      "loss": 1.0508,
+      "step": 1244
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8714533798479926e-05,
+      "loss": 0.9914,
+      "step": 1245
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.871208919799982e-05,
+      "loss": 0.9698,
+      "step": 1246
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.870964243522883e-05,
+      "loss": 1.1307,
+      "step": 1247
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8707193510774234e-05,
+      "loss": 1.0218,
+      "step": 1248
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8704742425243837e-05,
+      "loss": 1.1711,
+      "step": 1249
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8702289179245987e-05,
+      "loss": 1.1832,
+      "step": 1250
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8699833773389562e-05,
+      "loss": 0.9909,
+      "step": 1251
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8697376208283985e-05,
+      "loss": 1.1857,
+      "step": 1252
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8694916484539204e-05,
+      "loss": 1.165,
+      "step": 1253
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8692454602765713e-05,
+      "loss": 1.0787,
+      "step": 1254
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8689990563574535e-05,
+      "loss": 1.1184,
+      "step": 1255
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.868752436757723e-05,
+      "loss": 1.1197,
+      "step": 1256
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8685056015385897e-05,
+      "loss": 0.9376,
+      "step": 1257
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8682585507613163e-05,
+      "loss": 1.0885,
+      "step": 1258
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8680112844872194e-05,
+      "loss": 1.0868,
+      "step": 1259
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8677638027776693e-05,
+      "loss": 1.1907,
+      "step": 1260
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8675161056940896e-05,
+      "loss": 1.139,
+      "step": 1261
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.867268193297957e-05,
+      "loss": 1.0938,
+      "step": 1262
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8670200656508024e-05,
+      "loss": 1.03,
+      "step": 1263
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.866771722814209e-05,
+      "loss": 1.0779,
+      "step": 1264
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.866523164849815e-05,
+      "loss": 1.2211,
+      "step": 1265
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.86627439181931e-05,
+      "loss": 1.0922,
+      "step": 1266
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.866025403784439e-05,
+      "loss": 0.9383,
+      "step": 1267
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8657762008069987e-05,
+      "loss": 0.9293,
+      "step": 1268
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8655267829488402e-05,
+      "loss": 0.9369,
+      "step": 1269
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8652771502718675e-05,
+      "loss": 1.1731,
+      "step": 1270
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8650273028380384e-05,
+      "loss": 1.0936,
+      "step": 1271
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.864777240709363e-05,
+      "loss": 1.1244,
+      "step": 1272
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8645269639479054e-05,
+      "loss": 1.079,
+      "step": 1273
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8642764726157834e-05,
+      "loss": 1.1028,
+      "step": 1274
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.864025766775167e-05,
+      "loss": 1.0632,
+      "step": 1275
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.86377484648828e-05,
+      "loss": 0.9421,
+      "step": 1276
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8635237118173993e-05,
+      "loss": 1.079,
+      "step": 1277
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8632723628248557e-05,
+      "loss": 1.1021,
+      "step": 1278
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8630207995730322e-05,
+      "loss": 1.2376,
+      "step": 1279
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8627690221243648e-05,
+      "loss": 0.9203,
+      "step": 1280
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8625170305413444e-05,
+      "loss": 1.0825,
+      "step": 1281
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.862264824886513e-05,
+      "loss": 1.0858,
+      "step": 1282
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8620124052224667e-05,
+      "loss": 0.9478,
+      "step": 1283
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8617597716118548e-05,
+      "loss": 0.9929,
+      "step": 1284
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8615069241173796e-05,
+      "loss": 1.1461,
+      "step": 1285
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.861253862801796e-05,
+      "loss": 0.9812,
+      "step": 1286
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.861000587727913e-05,
+      "loss": 1.0936,
+      "step": 1287
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8607470989585918e-05,
+      "loss": 1.1539,
+      "step": 1288
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8604933965567465e-05,
+      "loss": 1.0632,
+      "step": 1289
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.860239480585345e-05,
+      "loss": 1.0238,
+      "step": 1290
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8599853511074076e-05,
+      "loss": 1.093,
+      "step": 1291
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.859731008186008e-05,
+      "loss": 1.1021,
+      "step": 1292
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8594764518842722e-05,
+      "loss": 1.1901,
+      "step": 1293
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8592216822653803e-05,
+      "loss": 1.0521,
+      "step": 1294
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8589666993925638e-05,
+      "loss": 1.1542,
+      "step": 1295
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8587115033291086e-05,
+      "loss": 0.8345,
+      "step": 1296
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8584560941383533e-05,
+      "loss": 1.1897,
+      "step": 1297
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8582004718836883e-05,
+      "loss": 1.1564,
+      "step": 1298
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8579446366285574e-05,
+      "loss": 1.0463,
+      "step": 1299
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8576885884364583e-05,
+      "loss": 1.0317,
+      "step": 1300
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.85743232737094e-05,
+      "loss": 1.055,
+      "step": 1301
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8571758534956054e-05,
+      "loss": 1.1097,
+      "step": 1302
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8569191668741098e-05,
+      "loss": 0.9993,
+      "step": 1303
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8566622675701607e-05,
+      "loss": 1.1032,
+      "step": 1304
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8564051556475205e-05,
+      "loss": 0.9441,
+      "step": 1305
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8561478311700015e-05,
+      "loss": 1.0544,
+      "step": 1306
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.855890294201471e-05,
+      "loss": 1.0137,
+      "step": 1307
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.855632544805848e-05,
+      "loss": 1.0144,
+      "step": 1308
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.855374583047104e-05,
+      "loss": 0.9884,
+      "step": 1309
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8551164089892643e-05,
+      "loss": 0.8365,
+      "step": 1310
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.854858022696406e-05,
+      "loss": 0.9167,
+      "step": 1311
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8545994242326587e-05,
+      "loss": 1.109,
+      "step": 1312
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8543406136622053e-05,
+      "loss": 0.9133,
+      "step": 1313
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8540815910492816e-05,
+      "loss": 0.9706,
+      "step": 1314
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.853822356458175e-05,
+      "loss": 1.0367,
+      "step": 1315
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8535629099532262e-05,
+      "loss": 1.0904,
+      "step": 1316
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.853303251598828e-05,
+      "loss": 1.1468,
+      "step": 1317
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8530433814594267e-05,
+      "loss": 0.9566,
+      "step": 1318
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8527832995995207e-05,
+      "loss": 1.0,
+      "step": 1319
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8525230060836602e-05,
+      "loss": 0.9064,
+      "step": 1320
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8522625009764487e-05,
+      "loss": 0.9325,
+      "step": 1321
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8520017843425426e-05,
+      "loss": 1.0262,
+      "step": 1322
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8517408562466494e-05,
+      "loss": 0.967,
+      "step": 1323
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8514797167535307e-05,
+      "loss": 1.0531,
+      "step": 1324
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.851218365928e-05,
+      "loss": 0.948,
+      "step": 1325
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8509568038349222e-05,
+      "loss": 1.1809,
+      "step": 1326
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8506950305392163e-05,
+      "loss": 1.1456,
+      "step": 1327
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8504330461058523e-05,
+      "loss": 1.1595,
+      "step": 1328
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8501708505998538e-05,
+      "loss": 1.1007,
+      "step": 1329
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.849908444086296e-05,
+      "loss": 1.0457,
+      "step": 1330
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.849645826630307e-05,
+      "loss": 1.0237,
+      "step": 1331
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8493829982970665e-05,
+      "loss": 1.0804,
+      "step": 1332
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.849119959151807e-05,
+      "loss": 1.0699,
+      "step": 1333
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8488567092598137e-05,
+      "loss": 1.2075,
+      "step": 1334
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8485932486864234e-05,
+      "loss": 1.0954,
+      "step": 1335
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8483295774970257e-05,
+      "loss": 1.1002,
+      "step": 1336
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8480656957570622e-05,
+      "loss": 0.9454,
+      "step": 1337
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8478016035320266e-05,
+      "loss": 1.058,
+      "step": 1338
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8475373008874658e-05,
+      "loss": 0.9621,
+      "step": 1339
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8472727878889776e-05,
+      "loss": 1.0849,
+      "step": 1340
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.847008064602213e-05,
+      "loss": 0.944,
+      "step": 1341
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.846743131092874e-05,
+      "loss": 1.1205,
+      "step": 1342
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8464779874267166e-05,
+      "loss": 1.1322,
+      "step": 1343
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8462126336695473e-05,
+      "loss": 1.1476,
+      "step": 1344
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8459470698872258e-05,
+      "loss": 0.9257,
+      "step": 1345
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8456812961456635e-05,
+      "loss": 0.9236,
+      "step": 1346
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8454153125108232e-05,
+      "loss": 1.0115,
+      "step": 1347
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8451491190487213e-05,
+      "loss": 1.0947,
+      "step": 1348
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8448827158254258e-05,
+      "loss": 1.0949,
+      "step": 1349
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8446161029070553e-05,
+      "loss": 1.0774,
+      "step": 1350
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8443492803597825e-05,
+      "loss": 1.0748,
+      "step": 1351
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.844082248249831e-05,
+      "loss": 0.9093,
+      "step": 1352
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8438150066434765e-05,
+      "loss": 0.9431,
+      "step": 1353
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.843547555607047e-05,
+      "loss": 1.0566,
+      "step": 1354
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8432798952069222e-05,
+      "loss": 1.0688,
+      "step": 1355
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8430120255095342e-05,
+      "loss": 1.0081,
+      "step": 1356
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8427439465813666e-05,
+      "loss": 1.1734,
+      "step": 1357
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.842475658488955e-05,
+      "loss": 1.144,
+      "step": 1358
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8422071612988865e-05,
+      "loss": 0.9791,
+      "step": 1359
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8419384550778012e-05,
+      "loss": 0.8386,
+      "step": 1360
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8416695398923904e-05,
+      "loss": 1.1238,
+      "step": 1361
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.841400415809397e-05,
+      "loss": 1.1259,
+      "step": 1362
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8411310828956165e-05,
+      "loss": 1.0248,
+      "step": 1363
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8408615412178954e-05,
+      "loss": 0.9197,
+      "step": 1364
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.840591790843133e-05,
+      "loss": 1.1605,
+      "step": 1365
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8403218318382788e-05,
+      "loss": 1.0935,
+      "step": 1366
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8400516642703357e-05,
+      "loss": 0.991,
+      "step": 1367
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.839781288206358e-05,
+      "loss": 1.2362,
+      "step": 1368
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.839510703713451e-05,
+      "loss": 0.9849,
+      "step": 1369
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.839239910858772e-05,
+      "loss": 1.0473,
+      "step": 1370
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8389689097095313e-05,
+      "loss": 1.0172,
+      "step": 1371
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.838697700332989e-05,
+      "loss": 1.0066,
+      "step": 1372
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8384262827964578e-05,
+      "loss": 1.0912,
+      "step": 1373
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8381546571673025e-05,
+      "loss": 1.0495,
+      "step": 1374
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.837882823512938e-05,
+      "loss": 1.2483,
+      "step": 1375
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.837610781900833e-05,
+      "loss": 1.0739,
+      "step": 1376
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8373385323985062e-05,
+      "loss": 0.9809,
+      "step": 1377
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8370660750735283e-05,
+      "loss": 1.0494,
+      "step": 1378
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8367934099935218e-05,
+      "loss": 1.0219,
+      "step": 1379
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8365205372261604e-05,
+      "loss": 1.031,
+      "step": 1380
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8362474568391696e-05,
+      "loss": 1.0157,
+      "step": 1381
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8359741689003266e-05,
+      "loss": 1.0324,
+      "step": 1382
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8357006734774597e-05,
+      "loss": 0.9853,
+      "step": 1383
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8354269706384492e-05,
+      "loss": 0.9362,
+      "step": 1384
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8351530604512264e-05,
+      "loss": 1.1007,
+      "step": 1385
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.834878942983774e-05,
+      "loss": 0.9768,
+      "step": 1386
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.834604618304127e-05,
+      "loss": 1.0637,
+      "step": 1387
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8343300864803706e-05,
+      "loss": 1.0938,
+      "step": 1388
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8340553475806423e-05,
+      "loss": 1.0358,
+      "step": 1389
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.833780401673131e-05,
+      "loss": 1.2053,
+      "step": 1390
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8335052488260764e-05,
+      "loss": 0.959,
+      "step": 1391
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8332298891077697e-05,
+      "loss": 1.1023,
+      "step": 1392
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8329543225865536e-05,
+      "loss": 1.1131,
+      "step": 1393
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.832678549330823e-05,
+      "loss": 1.007,
+      "step": 1394
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.832402569409022e-05,
+      "loss": 0.9844,
+      "step": 1395
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8321263828896477e-05,
+      "loss": 0.9334,
+      "step": 1396
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8318499898412487e-05,
+      "loss": 1.1242,
+      "step": 1397
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8315733903324235e-05,
+      "loss": 1.0295,
+      "step": 1398
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8312965844318227e-05,
+      "loss": 1.0853,
+      "step": 1399
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8310195722081475e-05,
+      "loss": 0.9646,
+      "step": 1400
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8307423537301516e-05,
+      "loss": 1.0,
+      "step": 1401
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8304649290666384e-05,
+      "loss": 1.1103,
+      "step": 1402
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8301872982864637e-05,
+      "loss": 1.0355,
+      "step": 1403
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.829909461458533e-05,
+      "loss": 1.1799,
+      "step": 1404
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.829631418651805e-05,
+      "loss": 0.9495,
+      "step": 1405
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.829353169935287e-05,
+      "loss": 1.0702,
+      "step": 1406
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8290747153780396e-05,
+      "loss": 1.0245,
+      "step": 1407
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8287960550491736e-05,
+      "loss": 1.0289,
+      "step": 1408
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8285171890178505e-05,
+      "loss": 1.1624,
+      "step": 1409
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8282381173532837e-05,
+      "loss": 1.0751,
+      "step": 1410
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.827958840124737e-05,
+      "loss": 1.1887,
+      "step": 1411
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8276793574015253e-05,
+      "loss": 0.97,
+      "step": 1412
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8273996692530147e-05,
+      "loss": 1.1352,
+      "step": 1413
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8271197757486224e-05,
+      "loss": 0.9234,
+      "step": 1414
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8268396769578163e-05,
+      "loss": 1.0743,
+      "step": 1415
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.826559372950115e-05,
+      "loss": 1.0046,
+      "step": 1416
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.826278863795089e-05,
+      "loss": 1.0625,
+      "step": 1417
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8259981495623585e-05,
+      "loss": 1.0342,
+      "step": 1418
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.825717230321595e-05,
+      "loss": 1.0106,
+      "step": 1419
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.825436106142522e-05,
+      "loss": 1.0435,
+      "step": 1420
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8251547770949116e-05,
+      "loss": 1.0159,
+      "step": 1421
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8248732432485893e-05,
+      "loss": 1.1001,
+      "step": 1422
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8245915046734296e-05,
+      "loss": 1.1229,
+      "step": 1423
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.824309561439358e-05,
+      "loss": 0.9444,
+      "step": 1424
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8240274136163526e-05,
+      "loss": 0.937,
+      "step": 1425
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8237450612744392e-05,
+      "loss": 1.0347,
+      "step": 1426
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8234625044836972e-05,
+      "loss": 1.1392,
+      "step": 1427
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.823179743314255e-05,
+      "loss": 1.1109,
+      "step": 1428
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.822896777836293e-05,
+      "loss": 1.0769,
+      "step": 1429
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8226136081200407e-05,
+      "loss": 1.1769,
+      "step": 1430
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8223302342357796e-05,
+      "loss": 0.9775,
+      "step": 1431
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8220466562538416e-05,
+      "loss": 1.1149,
+      "step": 1432
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.821762874244609e-05,
+      "loss": 1.0954,
+      "step": 1433
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8214788882785152e-05,
+      "loss": 1.1527,
+      "step": 1434
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8211946984260437e-05,
+      "loss": 0.9797,
+      "step": 1435
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.820910304757728e-05,
+      "loss": 1.0859,
+      "step": 1436
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8206257073441546e-05,
+      "loss": 1.0384,
+      "step": 1437
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8203409062559577e-05,
+      "loss": 1.042,
+      "step": 1438
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.820055901563824e-05,
+      "loss": 1.0702,
+      "step": 1439
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.819770693338489e-05,
+      "loss": 1.1606,
+      "step": 1440
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8194852816507407e-05,
+      "loss": 1.0242,
+      "step": 1441
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8191996665714168e-05,
+      "loss": 1.0968,
+      "step": 1442
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.818913848171405e-05,
+      "loss": 1.1647,
+      "step": 1443
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8186278265216436e-05,
+      "loss": 0.9891,
+      "step": 1444
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8183416016931216e-05,
+      "loss": 1.1532,
+      "step": 1445
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8180551737568785e-05,
+      "loss": 1.0451,
+      "step": 1446
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.817768542784004e-05,
+      "loss": 1.0133,
+      "step": 1447
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8174817088456387e-05,
+      "loss": 0.9731,
+      "step": 1448
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8171946720129726e-05,
+      "loss": 0.9388,
+      "step": 1449
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.816907432357247e-05,
+      "loss": 0.9925,
+      "step": 1450
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8166199899497527e-05,
+      "loss": 1.1662,
+      "step": 1451
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8163323448618314e-05,
+      "loss": 1.0034,
+      "step": 1452
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8160444971648753e-05,
+      "loss": 1.0814,
+      "step": 1453
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8157564469303262e-05,
+      "loss": 1.1394,
+      "step": 1454
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8154681942296766e-05,
+      "loss": 1.0766,
+      "step": 1455
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8151797391344695e-05,
+      "loss": 0.9661,
+      "step": 1456
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8148910817162973e-05,
+      "loss": 0.9614,
+      "step": 1457
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8146022220468035e-05,
+      "loss": 1.1614,
+      "step": 1458
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8143131601976813e-05,
+      "loss": 1.0612,
+      "step": 1459
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8140238962406743e-05,
+      "loss": 1.0169,
+      "step": 1460
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.813734430247576e-05,
+      "loss": 1.0227,
+      "step": 1461
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8134447622902302e-05,
+      "loss": 1.0176,
+      "step": 1462
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.813154892440531e-05,
+      "loss": 0.8889,
+      "step": 1463
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.812864820770422e-05,
+      "loss": 1.1485,
+      "step": 1464
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.812574547351898e-05,
+      "loss": 0.9893,
+      "step": 1465
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8122840722570032e-05,
+      "loss": 1.1769,
+      "step": 1466
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.811993395557831e-05,
+      "loss": 1.126,
+      "step": 1467
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.811702517326527e-05,
+      "loss": 0.9582,
+      "step": 1468
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8114114376352844e-05,
+      "loss": 1.1461,
+      "step": 1469
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8111201565563475e-05,
+      "loss": 1.0633,
+      "step": 1470
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8108286741620113e-05,
+      "loss": 1.0718,
+      "step": 1471
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8105369905246198e-05,
+      "loss": 1.0271,
+      "step": 1472
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8102451057165672e-05,
+      "loss": 1.095,
+      "step": 1473
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.809953019810298e-05,
+      "loss": 1.0014,
+      "step": 1474
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8096607328783053e-05,
+      "loss": 0.9859,
+      "step": 1475
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8093682449931342e-05,
+      "loss": 1.0699,
+      "step": 1476
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8090755562273772e-05,
+      "loss": 1.0507,
+      "step": 1477
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8087826666536793e-05,
+      "loss": 1.0061,
+      "step": 1478
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8084895763447333e-05,
+      "loss": 1.1059,
+      "step": 1479
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8081962853732832e-05,
+      "loss": 1.1032,
+      "step": 1480
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8079027938121208e-05,
+      "loss": 1.1743,
+      "step": 1481
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8076091017340906e-05,
+      "loss": 1.0744,
+      "step": 1482
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8073152092120844e-05,
+      "loss": 1.0507,
+      "step": 1483
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8070211163190447e-05,
+      "loss": 1.082,
+      "step": 1484
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.806726823127964e-05,
+      "loss": 0.9934,
+      "step": 1485
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.806432329711884e-05,
+      "loss": 1.0665,
+      "step": 1486
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8061376361438964e-05,
+      "loss": 1.1174,
+      "step": 1487
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8058427424971422e-05,
+      "loss": 1.0959,
+      "step": 1488
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8055476488448123e-05,
+      "loss": 1.0805,
+      "step": 1489
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.805252355260148e-05,
+      "loss": 1.0338,
+      "step": 1490
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8049568618164387e-05,
+      "loss": 0.9513,
+      "step": 1491
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8046611685870243e-05,
+      "loss": 1.0875,
+      "step": 1492
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8043652756452945e-05,
+      "loss": 1.1753,
+      "step": 1493
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8040691830646882e-05,
+      "loss": 0.9974,
+      "step": 1494
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8037728909186932e-05,
+      "loss": 1.0144,
+      "step": 1495
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8034763992808484e-05,
+      "loss": 0.8835,
+      "step": 1496
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8031797082247408e-05,
+      "loss": 1.0443,
+      "step": 1497
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8028828178240082e-05,
+      "loss": 1.0091,
+      "step": 1498
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.802585728152336e-05,
+      "loss": 1.0554,
+      "step": 1499
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.802288439283461e-05,
+      "loss": 1.1278,
+      "step": 1500
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.801990951291168e-05,
+      "loss": 1.0526,
+      "step": 1501
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8016932642492923e-05,
+      "loss": 0.9737,
+      "step": 1502
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8013953782317176e-05,
+      "loss": 0.9278,
+      "step": 1503
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.801097293312378e-05,
+      "loss": 1.1038,
+      "step": 1504
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8007990095652565e-05,
+      "loss": 0.9224,
+      "step": 1505
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.800500527064385e-05,
+      "loss": 0.9794,
+      "step": 1506
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8002018458838456e-05,
+      "loss": 1.0691,
+      "step": 1507
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7999029660977692e-05,
+      "loss": 0.9376,
+      "step": 1508
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.799603887780336e-05,
+      "loss": 0.9596,
+      "step": 1509
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7993046110057753e-05,
+      "loss": 1.0638,
+      "step": 1510
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.799005135848366e-05,
+      "loss": 0.9693,
+      "step": 1511
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7987054623824363e-05,
+      "loss": 0.984,
+      "step": 1512
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7984055906823634e-05,
+      "loss": 0.973,
+      "step": 1513
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7981055208225737e-05,
+      "loss": 1.1249,
+      "step": 1514
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.797805252877543e-05,
+      "loss": 1.1263,
+      "step": 1515
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7975047869217955e-05,
+      "loss": 0.9645,
+      "step": 1516
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.797204123029906e-05,
+      "loss": 1.0919,
+      "step": 1517
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7969032612764963e-05,
+      "loss": 1.0074,
+      "step": 1518
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7966022017362397e-05,
+      "loss": 1.0573,
+      "step": 1519
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.796300944483858e-05,
+      "loss": 1.1499,
+      "step": 1520
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7959994895941197e-05,
+      "loss": 1.0668,
+      "step": 1521
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7956978371418453e-05,
+      "loss": 1.0644,
+      "step": 1522
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7953959872019036e-05,
+      "loss": 1.019,
+      "step": 1523
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.795093939849211e-05,
+      "loss": 1.0951,
+      "step": 1524
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7947916951587345e-05,
+      "loss": 1.0696,
+      "step": 1525
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7944892532054896e-05,
+      "loss": 1.034,
+      "step": 1526
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7941866140645405e-05,
+      "loss": 1.0275,
+      "step": 1527
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7938837778110005e-05,
+      "loss": 0.9791,
+      "step": 1528
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7935807445200318e-05,
+      "loss": 1.0677,
+      "step": 1529
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.793277514266846e-05,
+      "loss": 0.898,
+      "step": 1530
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7929740871267022e-05,
+      "loss": 1.0353,
+      "step": 1531
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7926704631749102e-05,
+      "loss": 0.9816,
+      "step": 1532
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.792366642486827e-05,
+      "loss": 0.9118,
+      "step": 1533
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.79206262513786e-05,
+      "loss": 0.993,
+      "step": 1534
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7917584112034634e-05,
+      "loss": 0.9035,
+      "step": 1535
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7914540007591426e-05,
+      "loss": 0.9684,
+      "step": 1536
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7911493938804495e-05,
+      "loss": 1.0267,
+      "step": 1537
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7908445906429868e-05,
+      "loss": 1.1483,
+      "step": 1538
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.790539591122404e-05,
+      "loss": 1.0096,
+      "step": 1539
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7902343953944007e-05,
+      "loss": 0.961,
+      "step": 1540
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.789929003534725e-05,
+      "loss": 0.8598,
+      "step": 1541
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.789623415619173e-05,
+      "loss": 0.9138,
+      "step": 1542
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.78931763172359e-05,
+      "loss": 0.9519,
+      "step": 1543
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7890116519238695e-05,
+      "loss": 0.9979,
+      "step": 1544
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7887054762959545e-05,
+      "loss": 1.2285,
+      "step": 1545
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7883991049158356e-05,
+      "loss": 0.9724,
+      "step": 1546
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.788092537859553e-05,
+      "loss": 1.051,
+      "step": 1547
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.787785775203194e-05,
+      "loss": 1.1234,
+      "step": 1548
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.787478817022896e-05,
+      "loss": 0.943,
+      "step": 1549
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7871716633948444e-05,
+      "loss": 1.1483,
+      "step": 1550
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7868643143952722e-05,
+      "loss": 0.9226,
+      "step": 1551
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7865567701004627e-05,
+      "loss": 1.2343,
+      "step": 1552
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.786249030586746e-05,
+      "loss": 0.9533,
+      "step": 1553
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7859410959305014e-05,
+      "loss": 1.0752,
+      "step": 1554
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7856329662081565e-05,
+      "loss": 0.9512,
+      "step": 1555
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.785324641496187e-05,
+      "loss": 1.0476,
+      "step": 1556
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7850161218711182e-05,
+      "loss": 1.0456,
+      "step": 1557
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7847074074095223e-05,
+      "loss": 0.9801,
+      "step": 1558
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7843984981880206e-05,
+      "loss": 1.0588,
+      "step": 1559
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7840893942832826e-05,
+      "loss": 1.0143,
+      "step": 1560
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7837800957720262e-05,
+      "loss": 1.0881,
+      "step": 1561
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7834706027310172e-05,
+      "loss": 0.9193,
+      "step": 1562
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7831609152370705e-05,
+      "loss": 1.0382,
+      "step": 1563
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7828510333670487e-05,
+      "loss": 1.1225,
+      "step": 1564
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.782540957197862e-05,
+      "loss": 0.9518,
+      "step": 1565
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7822306868064708e-05,
+      "loss": 1.0734,
+      "step": 1566
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7819202222698812e-05,
+      "loss": 1.0173,
+      "step": 1567
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7816095636651495e-05,
+      "loss": 0.9991,
+      "step": 1568
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.781298711069379e-05,
+      "loss": 0.931,
+      "step": 1569
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7809876645597218e-05,
+      "loss": 1.0675,
+      "step": 1570
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7806764242133775e-05,
+      "loss": 1.0344,
+      "step": 1571
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.780364990107595e-05,
+      "loss": 1.124,
+      "step": 1572
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7800533623196695e-05,
+      "loss": 0.9696,
+      "step": 1573
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.779741540926946e-05,
+      "loss": 1.005,
+      "step": 1574
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.779429526006816e-05,
+      "loss": 1.0076,
+      "step": 1575
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7791173176367208e-05,
+      "loss": 1.0768,
+      "step": 1576
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7788049158941478e-05,
+      "loss": 0.9776,
+      "step": 1577
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7784923208566338e-05,
+      "loss": 1.061,
+      "step": 1578
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.778179532601763e-05,
+      "loss": 1.0581,
+      "step": 1579
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.777866551207168e-05,
+      "loss": 1.1186,
+      "step": 1580
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7775533767505288e-05,
+      "loss": 1.0593,
+      "step": 1581
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.777240009309573e-05,
+      "loss": 1.0695,
+      "step": 1582
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7769264489620776e-05,
+      "loss": 0.9791,
+      "step": 1583
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7766126957858654e-05,
+      "loss": 0.9486,
+      "step": 1584
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7762987498588086e-05,
+      "loss": 0.9508,
+      "step": 1585
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.775984611258827e-05,
+      "loss": 1.0289,
+      "step": 1586
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7756702800638884e-05,
+      "loss": 1.0436,
+      "step": 1587
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.775355756352007e-05,
+      "loss": 1.021,
+      "step": 1588
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7750410402012465e-05,
+      "loss": 1.0255,
+      "step": 1589
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7747261316897172e-05,
+      "loss": 1.0803,
+      "step": 1590
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7744110308955778e-05,
+      "loss": 0.9836,
+      "step": 1591
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.774095737897035e-05,
+      "loss": 1.1388,
+      "step": 1592
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7737802527723424e-05,
+      "loss": 1.1598,
+      "step": 1593
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7734645755998008e-05,
+      "loss": 0.9558,
+      "step": 1594
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7731487064577605e-05,
+      "loss": 0.9935,
+      "step": 1595
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.772832645424618e-05,
+      "loss": 1.0311,
+      "step": 1596
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.772516392578818e-05,
+      "loss": 1.1139,
+      "step": 1597
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7721999479988524e-05,
+      "loss": 0.9878,
+      "step": 1598
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7718833117632616e-05,
+      "loss": 0.8651,
+      "step": 1599
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7715664839506323e-05,
+      "loss": 0.9543,
+      "step": 1600
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7712494646395993e-05,
+      "loss": 1.1811,
+      "step": 1601
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7709322539088455e-05,
+      "loss": 1.0429,
+      "step": 1602
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7706148518371e-05,
+      "loss": 1.107,
+      "step": 1603
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.770297258503141e-05,
+      "loss": 0.9741,
+      "step": 1604
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7699794739857926e-05,
+      "loss": 1.0841,
+      "step": 1605
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.769661498363928e-05,
+      "loss": 1.0843,
+      "step": 1606
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7693433317164666e-05,
+      "loss": 1.196,
+      "step": 1607
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.769024974122375e-05,
+      "loss": 1.0321,
+      "step": 1608
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.768706425660668e-05,
+      "loss": 1.1549,
+      "step": 1609
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7683876864104083e-05,
+      "loss": 0.9817,
+      "step": 1610
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.768068756450704e-05,
+      "loss": 1.0128,
+      "step": 1611
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7677496358607123e-05,
+      "loss": 0.9847,
+      "step": 1612
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.767430324719637e-05,
+      "loss": 1.0976,
+      "step": 1613
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7671108231067298e-05,
+      "loss": 1.0502,
+      "step": 1614
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.766791131101288e-05,
+      "loss": 1.0821,
+      "step": 1615
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7664712487826586e-05,
+      "loss": 0.9886,
+      "step": 1616
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.766151176230234e-05,
+      "loss": 1.0152,
+      "step": 1617
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7658309135234544e-05,
+      "loss": 1.0656,
+      "step": 1618
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.765510460741807e-05,
+      "loss": 1.0338,
+      "step": 1619
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7651898179648265e-05,
+      "loss": 0.9196,
+      "step": 1620
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.764868985272095e-05,
+      "loss": 1.0167,
+      "step": 1621
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7645479627432407e-05,
+      "loss": 0.9977,
+      "step": 1622
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.76422675045794e-05,
+      "loss": 1.2564,
+      "step": 1623
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7639053484959156e-05,
+      "loss": 0.987,
+      "step": 1624
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.763583756936938e-05,
+      "loss": 1.1253,
+      "step": 1625
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7632619758608244e-05,
+      "loss": 1.0418,
+      "step": 1626
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7629400053474387e-05,
+      "loss": 1.0161,
+      "step": 1627
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.762617845476692e-05,
+      "loss": 0.8886,
+      "step": 1628
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7622954963285436e-05,
+      "loss": 1.0453,
+      "step": 1629
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7619729579829973e-05,
+      "loss": 1.0294,
+      "step": 1630
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7616502305201064e-05,
+      "loss": 1.0789,
+      "step": 1631
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7613273140199693e-05,
+      "loss": 0.9205,
+      "step": 1632
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7610042085627324e-05,
+      "loss": 1.0134,
+      "step": 1633
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.760680914228589e-05,
+      "loss": 0.9511,
+      "step": 1634
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7603574310977776e-05,
+      "loss": 1.1561,
+      "step": 1635
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.760033759250586e-05,
+      "loss": 1.1372,
+      "step": 1636
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7597098987673477e-05,
+      "loss": 0.9992,
+      "step": 1637
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7593858497284427e-05,
+      "loss": 0.9162,
+      "step": 1638
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7590616122142982e-05,
+      "loss": 1.0704,
+      "step": 1639
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7587371863053882e-05,
+      "loss": 0.9554,
+      "step": 1640
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7584125720822326e-05,
+      "loss": 0.9892,
+      "step": 1641
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7580877696254e-05,
+      "loss": 0.9244,
+      "step": 1642
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.757762779015504e-05,
+      "loss": 1.0168,
+      "step": 1643
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7574376003332055e-05,
+      "loss": 1.2326,
+      "step": 1644
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7571122336592115e-05,
+      "loss": 1.0708,
+      "step": 1645
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7567866790742768e-05,
+      "loss": 1.1989,
+      "step": 1646
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7564609366592017e-05,
+      "loss": 1.0485,
+      "step": 1647
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.756135006494834e-05,
+      "loss": 1.118,
+      "step": 1648
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7558088886620673e-05,
+      "loss": 1.1721,
+      "step": 1649
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.755482583241843e-05,
+      "loss": 0.8714,
+      "step": 1650
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.755156090315147e-05,
+      "loss": 1.0072,
+      "step": 1651
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.754829409963014e-05,
+      "loss": 1.0478,
+      "step": 1652
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.754502542266524e-05,
+      "loss": 1.0369,
+      "step": 1653
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7541754873068032e-05,
+      "loss": 1.1204,
+      "step": 1654
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7538482451650255e-05,
+      "loss": 1.1894,
+      "step": 1655
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7535208159224098e-05,
+      "loss": 0.9408,
+      "step": 1656
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7531931996602227e-05,
+      "loss": 1.0413,
+      "step": 1657
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7528653964597765e-05,
+      "loss": 1.1764,
+      "step": 1658
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7525374064024302e-05,
+      "loss": 1.0955,
+      "step": 1659
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.752209229569589e-05,
+      "loss": 0.9395,
+      "step": 1660
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7518808660427046e-05,
+      "loss": 1.1806,
+      "step": 1661
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7515523159032745e-05,
+      "loss": 1.0791,
+      "step": 1662
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.751223579232844e-05,
+      "loss": 1.0875,
+      "step": 1663
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7508946561130024e-05,
+      "loss": 1.0617,
+      "step": 1664
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.750565546625387e-05,
+      "loss": 0.9304,
+      "step": 1665
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7502362508516816e-05,
+      "loss": 1.0472,
+      "step": 1666
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.749906768873615e-05,
+      "loss": 0.8698,
+      "step": 1667
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.749577100772962e-05,
+      "loss": 1.0557,
+      "step": 1668
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7492472466315455e-05,
+      "loss": 1.0743,
+      "step": 1669
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.748917206531233e-05,
+      "loss": 0.9907,
+      "step": 1670
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7485869805539385e-05,
+      "loss": 1.2578,
+      "step": 1671
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7482565687816224e-05,
+      "loss": 0.9641,
+      "step": 1672
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.747925971296291e-05,
+      "loss": 1.0431,
+      "step": 1673
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7475951881799963e-05,
+      "loss": 1.1413,
+      "step": 1674
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7472642195148372e-05,
+      "loss": 1.1624,
+      "step": 1675
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.746933065382958e-05,
+      "loss": 1.1292,
+      "step": 1676
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7466017258665497e-05,
+      "loss": 0.9652,
+      "step": 1677
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7462702010478482e-05,
+      "loss": 1.1101,
+      "step": 1678
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7459384910091362e-05,
+      "loss": 1.032,
+      "step": 1679
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7456065958327433e-05,
+      "loss": 1.001,
+      "step": 1680
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7452745156010428e-05,
+      "loss": 1.0592,
+      "step": 1681
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.744942250396455e-05,
+      "loss": 1.172,
+      "step": 1682
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7446098003014472e-05,
+      "loss": 0.9714,
+      "step": 1683
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7442771653985308e-05,
+      "loss": 0.9743,
+      "step": 1684
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7439443457702644e-05,
+      "loss": 0.8662,
+      "step": 1685
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.743611341499252e-05,
+      "loss": 0.9542,
+      "step": 1686
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.743278152668143e-05,
+      "loss": 1.1204,
+      "step": 1687
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7429447793596327e-05,
+      "loss": 0.9336,
+      "step": 1688
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7426112216564632e-05,
+      "loss": 0.9815,
+      "step": 1689
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7422774796414212e-05,
+      "loss": 0.9667,
+      "step": 1690
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7419435533973396e-05,
+      "loss": 1.074,
+      "step": 1691
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7416094430070968e-05,
+      "loss": 0.9961,
+      "step": 1692
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7412751485536178e-05,
+      "loss": 1.0313,
+      "step": 1693
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.740940670119872e-05,
+      "loss": 1.0581,
+      "step": 1694
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7406060077888745e-05,
+      "loss": 1.0358,
+      "step": 1695
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.740271161643688e-05,
+      "loss": 1.1239,
+      "step": 1696
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7399361317674182e-05,
+      "loss": 1.1759,
+      "step": 1697
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.739600918243218e-05,
+      "loss": 0.9677,
+      "step": 1698
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7392655211542854e-05,
+      "loss": 1.1268,
+      "step": 1699
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7389299405838646e-05,
+      "loss": 1.0462,
+      "step": 1700
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.738594176615244e-05,
+      "loss": 1.0635,
+      "step": 1701
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.738258229331759e-05,
+      "loss": 0.9739,
+      "step": 1702
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.737922098816789e-05,
+      "loss": 1.1004,
+      "step": 1703
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.73758578515376e-05,
+      "loss": 1.1009,
+      "step": 1704
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7372492884261436e-05,
+      "loss": 1.1065,
+      "step": 1705
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7369126087174558e-05,
+      "loss": 0.9784,
+      "step": 1706
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7365757461112595e-05,
+      "loss": 0.9644,
+      "step": 1707
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7362387006911605e-05,
+      "loss": 1.0716,
+      "step": 1708
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.735901472540813e-05,
+      "loss": 0.9792,
+      "step": 1709
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7355640617439144e-05,
+      "loss": 1.0527,
+      "step": 1710
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.735226468384208e-05,
+      "loss": 0.9323,
+      "step": 1711
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.734888692545483e-05,
+      "loss": 1.0719,
+      "step": 1712
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7345507343115735e-05,
+      "loss": 1.196,
+      "step": 1713
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.734212593766358e-05,
+      "loss": 1.1954,
+      "step": 1714
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7338742709937624e-05,
+      "loss": 1.1035,
+      "step": 1715
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7335357660777554e-05,
+      "loss": 1.0841,
+      "step": 1716
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.733197079102352e-05,
+      "loss": 0.9134,
+      "step": 1717
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.732858210151613e-05,
+      "loss": 1.1482,
+      "step": 1718
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7325191593096434e-05,
+      "loss": 1.0814,
+      "step": 1719
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7321799266605936e-05,
+      "loss": 1.0493,
+      "step": 1720
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7318405122886595e-05,
+      "loss": 1.0152,
+      "step": 1721
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7315009162780817e-05,
+      "loss": 1.1134,
+      "step": 1722
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7311611387131454e-05,
+      "loss": 0.893,
+      "step": 1723
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7308211796781825e-05,
+      "loss": 1.0313,
+      "step": 1724
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7304810392575683e-05,
+      "loss": 1.0205,
+      "step": 1725
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7301407175357236e-05,
+      "loss": 1.0115,
+      "step": 1726
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7298002145971144e-05,
+      "loss": 0.9689,
+      "step": 1727
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.729459530526252e-05,
+      "loss": 0.7768,
+      "step": 1728
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.729118665407692e-05,
+      "loss": 0.9464,
+      "step": 1729
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7287776193260347e-05,
+      "loss": 1.0047,
+      "step": 1730
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7284363923659263e-05,
+      "loss": 1.0403,
+      "step": 1731
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7280949846120577e-05,
+      "loss": 1.0885,
+      "step": 1732
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7277533961491637e-05,
+      "loss": 0.9231,
+      "step": 1733
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.727411627062025e-05,
+      "loss": 0.9846,
+      "step": 1734
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7270696774354664e-05,
+      "loss": 0.9829,
+      "step": 1735
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.726727547354358e-05,
+      "loss": 1.0301,
+      "step": 1736
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7263852369036146e-05,
+      "loss": 1.0214,
+      "step": 1737
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7260427461681958e-05,
+      "loss": 1.0158,
+      "step": 1738
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.725700075233106e-05,
+      "loss": 1.0227,
+      "step": 1739
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7253572241833934e-05,
+      "loss": 1.0098,
+      "step": 1740
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7250141931041524e-05,
+      "loss": 0.9149,
+      "step": 1741
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7246709820805212e-05,
+      "loss": 1.0033,
+      "step": 1742
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7243275911976826e-05,
+      "loss": 1.0205,
+      "step": 1743
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7239840205408647e-05,
+      "loss": 0.9692,
+      "step": 1744
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7236402701953393e-05,
+      "loss": 1.0908,
+      "step": 1745
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7232963402464237e-05,
+      "loss": 0.9333,
+      "step": 1746
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7229522307794787e-05,
+      "loss": 0.943,
+      "step": 1747
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.722607941879911e-05,
+      "loss": 1.2042,
+      "step": 1748
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.722263473633171e-05,
+      "loss": 1.0609,
+      "step": 1749
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7219188261247536e-05,
+      "loss": 0.9184,
+      "step": 1750
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.721573999440198e-05,
+      "loss": 1.129,
+      "step": 1751
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.721228993665089e-05,
+      "loss": 0.933,
+      "step": 1752
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.720883808885055e-05,
+      "loss": 0.9702,
+      "step": 1753
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7205384451857677e-05,
+      "loss": 0.9317,
+      "step": 1754
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.720192902652946e-05,
+      "loss": 1.0665,
+      "step": 1755
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7198471813723504e-05,
+      "loss": 1.0452,
+      "step": 1756
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.719501281429788e-05,
+      "loss": 0.878,
+      "step": 1757
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7191552029111078e-05,
+      "loss": 0.9375,
+      "step": 1758
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7188089459022057e-05,
+      "loss": 0.9561,
+      "step": 1759
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7184625104890204e-05,
+      "loss": 0.9861,
+      "step": 1760
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.718115896757535e-05,
+      "loss": 1.064,
+      "step": 1761
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.717769104793777e-05,
+      "loss": 0.9909,
+      "step": 1762
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.717422134683819e-05,
+      "loss": 1.1465,
+      "step": 1763
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.717074986513776e-05,
+      "loss": 0.9658,
+      "step": 1764
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7167276603698083e-05,
+      "loss": 0.947,
+      "step": 1765
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.716380156338121e-05,
+      "loss": 1.0804,
+      "step": 1766
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.716032474504962e-05,
+      "loss": 1.0412,
+      "step": 1767
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.715684614956624e-05,
+      "loss": 0.9062,
+      "step": 1768
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7153365777794436e-05,
+      "loss": 0.9145,
+      "step": 1769
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7149883630598025e-05,
+      "loss": 0.92,
+      "step": 1770
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.714639970884125e-05,
+      "loss": 0.9788,
+      "step": 1771
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7142914013388795e-05,
+      "loss": 1.0534,
+      "step": 1772
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.71394265451058e-05,
+      "loss": 0.8981,
+      "step": 1773
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7135937304857824e-05,
+      "loss": 0.921,
+      "step": 1774
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.713244629351089e-05,
+      "loss": 1.148,
+      "step": 1775
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7128953511931436e-05,
+      "loss": 1.0276,
+      "step": 1776
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7125458960986355e-05,
+      "loss": 1.1536,
+      "step": 1777
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7121962641542977e-05,
+      "loss": 0.8659,
+      "step": 1778
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7118464554469062e-05,
+      "loss": 1.0124,
+      "step": 1779
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.711496470063282e-05,
+      "loss": 0.945,
+      "step": 1780
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7111463080902894e-05,
+      "loss": 0.8665,
+      "step": 1781
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7107959696148364e-05,
+      "loss": 0.9994,
+      "step": 1782
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.710445454723875e-05,
+      "loss": 0.8864,
+      "step": 1783
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.710094763504402e-05,
+      "loss": 1.026,
+      "step": 1784
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7097438960434552e-05,
+      "loss": 1.0641,
+      "step": 1785
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7093928524281193e-05,
+      "loss": 1.0252,
+      "step": 1786
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7090416327455204e-05,
+      "loss": 0.9959,
+      "step": 1787
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7086902370828303e-05,
+      "loss": 1.02,
+      "step": 1788
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7083386655272628e-05,
+      "loss": 1.0742,
+      "step": 1789
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7079869181660755e-05,
+      "loss": 0.9733,
+      "step": 1790
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7076349950865707e-05,
+      "loss": 0.9484,
+      "step": 1791
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.707282896376094e-05,
+      "loss": 0.9963,
+      "step": 1792
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.706930622122033e-05,
+      "loss": 1.0271,
+      "step": 1793
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7065781724118214e-05,
+      "loss": 1.0011,
+      "step": 1794
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7062255473329344e-05,
+      "loss": 0.9644,
+      "step": 1795
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7058727469728925e-05,
+      "loss": 0.9476,
+      "step": 1796
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.705519771419258e-05,
+      "loss": 0.9351,
+      "step": 1797
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7051666207596376e-05,
+      "loss": 1.0373,
+      "step": 1798
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7048132950816812e-05,
+      "loss": 1.0807,
+      "step": 1799
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7044597944730823e-05,
+      "loss": 0.8762,
+      "step": 1800
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.704106119021578e-05,
+      "loss": 0.918,
+      "step": 1801
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.703752268814948e-05,
+      "loss": 1.0035,
+      "step": 1802
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.703398243941016e-05,
+      "loss": 0.9479,
+      "step": 1803
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.703044044487649e-05,
+      "loss": 0.96,
+      "step": 1804
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.702689670542758e-05,
+      "loss": 1.0229,
+      "step": 1805
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7023351221942957e-05,
+      "loss": 0.9802,
+      "step": 1806
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7019803995302593e-05,
+      "loss": 0.9521,
+      "step": 1807
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.701625502638689e-05,
+      "loss": 1.0211,
+      "step": 1808
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7012704316076678e-05,
+      "loss": 1.0005,
+      "step": 1809
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7009151865253228e-05,
+      "loss": 0.9531,
+      "step": 1810
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7005597674798234e-05,
+      "loss": 1.0776,
+      "step": 1811
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7002041745593833e-05,
+      "loss": 0.9932,
+      "step": 1812
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.699848407852258e-05,
+      "loss": 0.9732,
+      "step": 1813
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6994924674467465e-05,
+      "loss": 1.1675,
+      "step": 1814
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.699136353431192e-05,
+      "loss": 1.0335,
+      "step": 1815
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6987800658939795e-05,
+      "loss": 0.9217,
+      "step": 1816
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6984236049235373e-05,
+      "loss": 1.1743,
+      "step": 1817
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.698066970608338e-05,
+      "loss": 0.8949,
+      "step": 1818
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6977101630368946e-05,
+      "loss": 1.0119,
+      "step": 1819
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.697353182297766e-05,
+      "loss": 1.0094,
+      "step": 1820
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6969960284795522e-05,
+      "loss": 1.0132,
+      "step": 1821
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.696638701670897e-05,
+      "loss": 1.1,
+      "step": 1822
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6962812019604865e-05,
+      "loss": 1.0518,
+      "step": 1823
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.695923529437051e-05,
+      "loss": 1.1162,
+      "step": 1824
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6955656841893616e-05,
+      "loss": 1.0037,
+      "step": 1825
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6952076663062342e-05,
+      "loss": 1.0303,
+      "step": 1826
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6948494758765266e-05,
+      "loss": 1.0421,
+      "step": 1827
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6944911129891396e-05,
+      "loss": 1.2274,
+      "step": 1828
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6941325777330167e-05,
+      "loss": 0.8531,
+      "step": 1829
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.693773870197145e-05,
+      "loss": 0.9954,
+      "step": 1830
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6934149904705525e-05,
+      "loss": 1.0117,
+      "step": 1831
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6930559386423123e-05,
+      "loss": 0.9238,
+      "step": 1832
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.692696714801538e-05,
+      "loss": 1.1238,
+      "step": 1833
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6923373190373874e-05,
+      "loss": 0.9905,
+      "step": 1834
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6919777514390604e-05,
+      "loss": 1.0346,
+      "step": 1835
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6916180120958e-05,
+      "loss": 0.9858,
+      "step": 1836
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.691258101096891e-05,
+      "loss": 1.0317,
+      "step": 1837
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.690898018531661e-05,
+      "loss": 0.9863,
+      "step": 1838
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6905377644894812e-05,
+      "loss": 1.0902,
+      "step": 1839
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6901773390597644e-05,
+      "loss": 1.2134,
+      "step": 1840
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6898167423319657e-05,
+      "loss": 1.009,
+      "step": 1841
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.689455974395583e-05,
+      "loss": 0.97,
+      "step": 1842
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.689095035340158e-05,
+      "loss": 1.1021,
+      "step": 1843
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6887339252552726e-05,
+      "loss": 1.026,
+      "step": 1844
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6883726442305525e-05,
+      "loss": 0.9715,
+      "step": 1845
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6880111923556663e-05,
+      "loss": 0.9966,
+      "step": 1846
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.687649569720323e-05,
+      "loss": 1.0037,
+      "step": 1847
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.687287776414276e-05,
+      "loss": 1.0994,
+      "step": 1848
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6869258125273208e-05,
+      "loss": 0.8819,
+      "step": 1849
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.686563678149294e-05,
+      "loss": 0.9422,
+      "step": 1850
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6862013733700757e-05,
+      "loss": 1.1876,
+      "step": 1851
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.685838898279588e-05,
+      "loss": 1.0329,
+      "step": 1852
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6854762529677945e-05,
+      "loss": 1.2055,
+      "step": 1853
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6851134375247025e-05,
+      "loss": 1.0397,
+      "step": 1854
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.68475045204036e-05,
+      "loss": 1.1463,
+      "step": 1855
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6843872966048588e-05,
+      "loss": 1.0824,
+      "step": 1856
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.684023971308331e-05,
+      "loss": 1.0678,
+      "step": 1857
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6836604762409526e-05,
+      "loss": 0.935,
+      "step": 1858
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6832968114929407e-05,
+      "loss": 0.9682,
+      "step": 1859
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6829329771545548e-05,
+      "loss": 0.9675,
+      "step": 1860
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6825689733160966e-05,
+      "loss": 1.07,
+      "step": 1861
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6822048000679097e-05,
+      "loss": 1.0306,
+      "step": 1862
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6818404575003802e-05,
+      "loss": 1.1237,
+      "step": 1863
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6814759457039352e-05,
+      "loss": 0.9788,
+      "step": 1864
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.681111264769045e-05,
+      "loss": 0.9164,
+      "step": 1865
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.680746414786221e-05,
+      "loss": 1.0995,
+      "step": 1866
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.680381395846017e-05,
+      "loss": 1.0285,
+      "step": 1867
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.680016208039029e-05,
+      "loss": 1.0395,
+      "step": 1868
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6796508514558935e-05,
+      "loss": 0.9611,
+      "step": 1869
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.679285326187291e-05,
+      "loss": 1.1602,
+      "step": 1870
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6789196323239422e-05,
+      "loss": 0.9429,
+      "step": 1871
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6785537699566105e-05,
+      "loss": 1.1065,
+      "step": 1872
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.678187739176101e-05,
+      "loss": 1.1696,
+      "step": 1873
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6778215400732598e-05,
+      "loss": 1.0964,
+      "step": 1874
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.677455172738976e-05,
+      "loss": 1.0428,
+      "step": 1875
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6770886372641795e-05,
+      "loss": 0.9783,
+      "step": 1876
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6767219337398427e-05,
+      "loss": 0.8558,
+      "step": 1877
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6763550622569792e-05,
+      "loss": 1.0934,
+      "step": 1878
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6759880229066444e-05,
+      "loss": 1.143,
+      "step": 1879
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6756208157799354e-05,
+      "loss": 0.9462,
+      "step": 1880
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6752534409679908e-05,
+      "loss": 0.954,
+      "step": 1881
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6748858985619907e-05,
+      "loss": 0.9359,
+      "step": 1882
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.674518188653158e-05,
+      "loss": 1.0216,
+      "step": 1883
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6741503113327547e-05,
+      "loss": 0.9767,
+      "step": 1884
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6737822666920872e-05,
+      "loss": 1.0442,
+      "step": 1885
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6734140548225015e-05,
+      "loss": 0.9494,
+      "step": 1886
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6730456758153856e-05,
+      "loss": 1.0197,
+      "step": 1887
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6726771297621693e-05,
+      "loss": 0.9755,
+      "step": 1888
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6723084167543238e-05,
+      "loss": 0.9334,
+      "step": 1889
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.671939536883361e-05,
+      "loss": 1.0638,
+      "step": 1890
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.671570490240835e-05,
+      "loss": 1.1623,
+      "step": 1891
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6712012769183418e-05,
+      "loss": 1.0526,
+      "step": 1892
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.670831897007517e-05,
+      "loss": 1.184,
+      "step": 1893
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6704623506000392e-05,
+      "loss": 0.9754,
+      "step": 1894
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.670092637787627e-05,
+      "loss": 1.0249,
+      "step": 1895
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.669722758662042e-05,
+      "loss": 0.9588,
+      "step": 1896
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6693527133150854e-05,
+      "loss": 1.0746,
+      "step": 1897
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6689825018386008e-05,
+      "loss": 1.0306,
+      "step": 1898
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.668612124324472e-05,
+      "loss": 0.9509,
+      "step": 1899
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.668241580864625e-05,
+      "loss": 1.0106,
+      "step": 1900
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.667870871551027e-05,
+      "loss": 1.0525,
+      "step": 1901
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6674999964756852e-05,
+      "loss": 1.0335,
+      "step": 1902
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6671289557306484e-05,
+      "loss": 1.1511,
+      "step": 1903
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.666757749408008e-05,
+      "loss": 0.8989,
+      "step": 1904
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6663863775998942e-05,
+      "loss": 1.016,
+      "step": 1905
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.66601484039848e-05,
+      "loss": 0.9497,
+      "step": 1906
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6656431378959786e-05,
+      "loss": 1.1673,
+      "step": 1907
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6652712701846442e-05,
+      "loss": 1.0154,
+      "step": 1908
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6648992373567728e-05,
+      "loss": 0.9858,
+      "step": 1909
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6645270395047004e-05,
+      "loss": 0.9651,
+      "step": 1910
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6641546767208043e-05,
+      "loss": 1.0391,
+      "step": 1911
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6637821490975034e-05,
+      "loss": 0.9528,
+      "step": 1912
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.663409456727256e-05,
+      "loss": 1.0423,
+      "step": 1913
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6630365997025633e-05,
+      "loss": 0.9589,
+      "step": 1914
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6626635781159656e-05,
+      "loss": 1.1264,
+      "step": 1915
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.662290392060045e-05,
+      "loss": 1.121,
+      "step": 1916
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.661917041627424e-05,
+      "loss": 1.0798,
+      "step": 1917
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.661543526910766e-05,
+      "loss": 0.9722,
+      "step": 1918
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6611698480027754e-05,
+      "loss": 0.871,
+      "step": 1919
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6607960049961972e-05,
+      "loss": 1.0867,
+      "step": 1920
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6604219979838167e-05,
+      "loss": 1.206,
+      "step": 1921
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.660047827058461e-05,
+      "loss": 0.9014,
+      "step": 1922
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6596734923129963e-05,
+      "loss": 0.8642,
+      "step": 1923
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.659298993840331e-05,
+      "loss": 0.9908,
+      "step": 1924
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6589243317334136e-05,
+      "loss": 0.9603,
+      "step": 1925
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.658549506085233e-05,
+      "loss": 1.0095,
+      "step": 1926
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.658174516988818e-05,
+      "loss": 1.0736,
+      "step": 1927
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6577993645372397e-05,
+      "loss": 0.9685,
+      "step": 1928
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6574240488236083e-05,
+      "loss": 1.0151,
+      "step": 1929
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6570485699410753e-05,
+      "loss": 1.1087,
+      "step": 1930
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6566729279828326e-05,
+      "loss": 1.0809,
+      "step": 1931
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6562971230421118e-05,
+      "loss": 1.0769,
+      "step": 1932
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.655921155212186e-05,
+      "loss": 1.0758,
+      "step": 1933
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.655545024586368e-05,
+      "loss": 0.9239,
+      "step": 1934
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.655168731258012e-05,
+      "loss": 1.2618,
+      "step": 1935
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6547922753205103e-05,
+      "loss": 1.0518,
+      "step": 1936
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6544156568672987e-05,
+      "loss": 0.9783,
+      "step": 1937
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.654038875991851e-05,
+      "loss": 0.9338,
+      "step": 1938
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.653661932787682e-05,
+      "loss": 1.0397,
+      "step": 1939
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6532848273483472e-05,
+      "loss": 0.9969,
+      "step": 1940
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6529075597674422e-05,
+      "loss": 0.9436,
+      "step": 1941
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6525301301386017e-05,
+      "loss": 1.0582,
+      "step": 1942
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6521525385555024e-05,
+      "loss": 0.9414,
+      "step": 1943
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.651774785111861e-05,
+      "loss": 0.9549,
+      "step": 1944
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6513968699014323e-05,
+      "loss": 1.092,
+      "step": 1945
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.651018793018013e-05,
+      "loss": 0.9463,
+      "step": 1946
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6506405545554408e-05,
+      "loss": 1.0509,
+      "step": 1947
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.650262154607591e-05,
+      "loss": 1.1525,
+      "step": 1948
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.649883593268381e-05,
+      "loss": 0.92,
+      "step": 1949
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6495048706317673e-05,
+      "loss": 1.0741,
+      "step": 1950
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.649125986791747e-05,
+      "loss": 1.1691,
+      "step": 1951
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6487469418423562e-05,
+      "loss": 1.068,
+      "step": 1952
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6483677358776725e-05,
+      "loss": 1.0249,
+      "step": 1953
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6479883689918125e-05,
+      "loss": 0.9204,
+      "step": 1954
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6476088412789323e-05,
+      "loss": 1.0178,
+      "step": 1955
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6472291528332295e-05,
+      "loss": 1.1058,
+      "step": 1956
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6468493037489398e-05,
+      "loss": 1.1182,
+      "step": 1957
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.64646929412034e-05,
+      "loss": 1.0449,
+      "step": 1958
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6460891240417463e-05,
+      "loss": 1.0357,
+      "step": 1959
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6457087936075144e-05,
+      "loss": 1.1474,
+      "step": 1960
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.64532830291204e-05,
+      "loss": 0.99,
+      "step": 1961
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.64494765204976e-05,
+      "loss": 1.0271,
+      "step": 1962
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.644566841115148e-05,
+      "loss": 1.0996,
+      "step": 1963
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6441858702027206e-05,
+      "loss": 1.1393,
+      "step": 1964
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6438047394070315e-05,
+      "loss": 0.9757,
+      "step": 1965
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6434234488226757e-05,
+      "loss": 1.0671,
+      "step": 1966
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6430419985442877e-05,
+      "loss": 1.0099,
+      "step": 1967
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6426603886665407e-05,
+      "loss": 0.9294,
+      "step": 1968
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6422786192841484e-05,
+      "loss": 1.2129,
+      "step": 1969
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6418966904918634e-05,
+      "loss": 1.0452,
+      "step": 1970
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6415146023844788e-05,
+      "loss": 1.0097,
+      "step": 1971
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6411323550568264e-05,
+      "loss": 0.982,
+      "step": 1972
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.640749948603778e-05,
+      "loss": 1.0879,
+      "step": 1973
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6403673831202443e-05,
+      "loss": 0.9839,
+      "step": 1974
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.639984658701176e-05,
+      "loss": 1.0362,
+      "step": 1975
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6396017754415634e-05,
+      "loss": 0.8817,
+      "step": 1976
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.639218733436436e-05,
+      "loss": 0.9518,
+      "step": 1977
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.638835532780862e-05,
+      "loss": 1.1562,
+      "step": 1978
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6384521735699502e-05,
+      "loss": 1.084,
+      "step": 1979
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.638068655898848e-05,
+      "loss": 1.0497,
+      "step": 1980
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.637684979862742e-05,
+      "loss": 1.0446,
+      "step": 1981
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.637301145556859e-05,
+      "loss": 1.0662,
+      "step": 1982
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6369171530764644e-05,
+      "loss": 0.9861,
+      "step": 1983
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6365330025168625e-05,
+      "loss": 1.0874,
+      "step": 1984
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6361486939733975e-05,
+      "loss": 1.0245,
+      "step": 1985
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6357642275414527e-05,
+      "loss": 1.1084,
+      "step": 1986
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6353796033164503e-05,
+      "loss": 1.0104,
+      "step": 1987
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.634994821393852e-05,
+      "loss": 1.0237,
+      "step": 1988
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6346098818691584e-05,
+      "loss": 0.9584,
+      "step": 1989
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6342247848379092e-05,
+      "loss": 1.115,
+      "step": 1990
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6338395303956837e-05,
+      "loss": 1.015,
+      "step": 1991
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.633454118638099e-05,
+      "loss": 0.9737,
+      "step": 1992
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.633068549660813e-05,
+      "loss": 1.0493,
+      "step": 1993
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6326828235595212e-05,
+      "loss": 1.0586,
+      "step": 1994
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.632296940429959e-05,
+      "loss": 0.9976,
+      "step": 1995
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6319109003679002e-05,
+      "loss": 1.0176,
+      "step": 1996
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6315247034691574e-05,
+      "loss": 1.1289,
+      "step": 1997
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6311383498295833e-05,
+      "loss": 1.0711,
+      "step": 1998
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.630751839545068e-05,
+      "loss": 0.998,
+      "step": 1999
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.630365172711541e-05,
+      "loss": 0.9654,
+      "step": 2000
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6299783494249718e-05,
+      "loss": 0.9433,
+      "step": 2001
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6295913697813663e-05,
+      "loss": 1.1032,
+      "step": 2002
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6292042338767714e-05,
+      "loss": 1.0357,
+      "step": 2003
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6288169418072725e-05,
+      "loss": 1.0128,
+      "step": 2004
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6284294936689927e-05,
+      "loss": 0.8796,
+      "step": 2005
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6280418895580946e-05,
+      "loss": 1.1608,
+      "step": 2006
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.627654129570779e-05,
+      "loss": 1.0213,
+      "step": 2007
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.627266213803286e-05,
+      "loss": 0.9496,
+      "step": 2008
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.626878142351894e-05,
+      "loss": 0.9993,
+      "step": 2009
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6264899153129206e-05,
+      "loss": 1.0457,
+      "step": 2010
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.626101532782721e-05,
+      "loss": 1.2133,
+      "step": 2011
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6257129948576893e-05,
+      "loss": 1.0316,
+      "step": 2012
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.625324301634259e-05,
+      "loss": 1.0546,
+      "step": 2013
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6249354532089013e-05,
+      "loss": 1.066,
+      "step": 2014
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.624546449678126e-05,
+      "loss": 1.0107,
+      "step": 2015
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.624157291138482e-05,
+      "loss": 1.0583,
+      "step": 2016
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6237679776865552e-05,
+      "loss": 0.9929,
+      "step": 2017
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6233785094189724e-05,
+      "loss": 0.8966,
+      "step": 2018
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6229888864323965e-05,
+      "loss": 1.02,
+      "step": 2019
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6225991088235306e-05,
+      "loss": 1.1362,
+      "step": 2020
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6222091766891135e-05,
+      "loss": 1.0129,
+      "step": 2021
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6218190901259257e-05,
+      "loss": 0.9945,
+      "step": 2022
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6214288492307843e-05,
+      "loss": 0.9032,
+      "step": 2023
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6210384541005445e-05,
+      "loss": 1.0606,
+      "step": 2024
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6206479048321e-05,
+      "loss": 0.8197,
+      "step": 2025
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6202572015223835e-05,
+      "loss": 1.1201,
+      "step": 2026
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6198663442683643e-05,
+      "loss": 1.0728,
+      "step": 2027
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6194753331670517e-05,
+      "loss": 0.9549,
+      "step": 2028
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.619084168315493e-05,
+      "loss": 1.1443,
+      "step": 2029
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.618692849810772e-05,
+      "loss": 0.9875,
+      "step": 2030
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.618301377750012e-05,
+      "loss": 0.9789,
+      "step": 2031
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6179097522303745e-05,
+      "loss": 0.9797,
+      "step": 2032
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6175179733490583e-05,
+      "loss": 1.0453,
+      "step": 2033
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.617126041203301e-05,
+      "loss": 1.142,
+      "step": 2034
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6167339558903777e-05,
+      "loss": 0.9937,
+      "step": 2035
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6163417175076023e-05,
+      "loss": 1.0432,
+      "step": 2036
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.615949326152325e-05,
+      "loss": 0.9748,
+      "step": 2037
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.615556781921936e-05,
+      "loss": 0.9192,
+      "step": 2038
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6151640849138623e-05,
+      "loss": 0.9544,
+      "step": 2039
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.614771235225569e-05,
+      "loss": 1.1121,
+      "step": 2040
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6143782329545595e-05,
+      "loss": 0.9844,
+      "step": 2041
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6139850781983745e-05,
+      "loss": 1.0186,
+      "step": 2042
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6135917710545923e-05,
+      "loss": 0.9484,
+      "step": 2043
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6131983116208303e-05,
+      "loss": 0.9553,
+      "step": 2044
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6128046999947424e-05,
+      "loss": 0.9728,
+      "step": 2045
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6124109362740205e-05,
+      "loss": 1.0786,
+      "step": 2046
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6120170205563952e-05,
+      "loss": 0.9617,
+      "step": 2047
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6116229529396337e-05,
+      "loss": 1.0872,
+      "step": 2048
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6112287335215416e-05,
+      "loss": 1.0411,
+      "step": 2049
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6108343623999618e-05,
+      "loss": 1.1735,
+      "step": 2050
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6104398396727744e-05,
+      "loss": 0.9671,
+      "step": 2051
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6100451654378985e-05,
+      "loss": 1.1172,
+      "step": 2052
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.60965033979329e-05,
+      "loss": 1.0129,
+      "step": 2053
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6092553628369417e-05,
+      "loss": 0.9758,
+      "step": 2054
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.608860234666885e-05,
+      "loss": 0.9876,
+      "step": 2055
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6084649553811892e-05,
+      "loss": 0.9667,
+      "step": 2056
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6080695250779587e-05,
+      "loss": 1.1118,
+      "step": 2057
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6076739438553383e-05,
+      "loss": 0.9385,
+      "step": 2058
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.607278211811509e-05,
+      "loss": 0.8383,
+      "step": 2059
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6068823290446887e-05,
+      "loss": 0.9846,
+      "step": 2060
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.606486295653133e-05,
+      "loss": 1.0561,
+      "step": 2061
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6060901117351363e-05,
+      "loss": 1.0041,
+      "step": 2062
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6056937773890283e-05,
+      "loss": 0.9399,
+      "step": 2063
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6052972927131768e-05,
+      "loss": 0.9063,
+      "step": 2064
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.604900657805988e-05,
+      "loss": 1.0728,
+      "step": 2065
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.604503872765903e-05,
+      "loss": 1.0716,
+      "step": 2066
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.604106937691403e-05,
+      "loss": 1.0609,
+      "step": 2067
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.603709852681004e-05,
+      "loss": 0.9192,
+      "step": 2068
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6033126178332608e-05,
+      "loss": 1.0298,
+      "step": 2069
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6029152332467646e-05,
+      "loss": 1.0024,
+      "step": 2070
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.602517699020144e-05,
+      "loss": 0.9745,
+      "step": 2071
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6021200152520647e-05,
+      "loss": 1.0274,
+      "step": 2072
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6017221820412298e-05,
+      "loss": 0.932,
+      "step": 2073
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6013241994863785e-05,
+      "loss": 1.0411,
+      "step": 2074
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.600926067686289e-05,
+      "loss": 0.9408,
+      "step": 2075
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6005277867397745e-05,
+      "loss": 1.0506,
+      "step": 2076
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6001293567456865e-05,
+      "loss": 1.0407,
+      "step": 2077
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5997307778029123e-05,
+      "loss": 0.9422,
+      "step": 2078
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5993320500103776e-05,
+      "loss": 1.0793,
+      "step": 2079
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.598933173467044e-05,
+      "loss": 1.0241,
+      "step": 2080
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5985341482719113e-05,
+      "loss": 1.0685,
+      "step": 2081
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.598134974524014e-05,
+      "loss": 1.0742,
+      "step": 2082
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5977356523224252e-05,
+      "loss": 0.9107,
+      "step": 2083
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.597336181766255e-05,
+      "loss": 1.0721,
+      "step": 2084
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.596936562954649e-05,
+      "loss": 1.003,
+      "step": 2085
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5965367959867898e-05,
+      "loss": 1.0301,
+      "step": 2086
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5961368809618992e-05,
+      "loss": 0.9977,
+      "step": 2087
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5957368179792318e-05,
+      "loss": 1.0487,
+      "step": 2088
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5953366071380817e-05,
+      "loss": 1.0375,
+      "step": 2089
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5949362485377796e-05,
+      "loss": 0.891,
+      "step": 2090
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5945357422776916e-05,
+      "loss": 1.0286,
+      "step": 2091
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.594135088457221e-05,
+      "loss": 1.0012,
+      "step": 2092
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.593734287175808e-05,
+      "loss": 1.0554,
+      "step": 2093
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.593333338532929e-05,
+      "loss": 0.9914,
+      "step": 2094
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.592932242628097e-05,
+      "loss": 1.0459,
+      "step": 2095
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5925309995608628e-05,
+      "loss": 0.9919,
+      "step": 2096
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.592129609430811e-05,
+      "loss": 0.988,
+      "step": 2097
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5917280723375657e-05,
+      "loss": 0.9748,
+      "step": 2098
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5913263883807856e-05,
+      "loss": 0.9963,
+      "step": 2099
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5909245576601663e-05,
+      "loss": 1.0372,
+      "step": 2100
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5905225802754398e-05,
+      "loss": 1.0337,
+      "step": 2101
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5901204563263747e-05,
+      "loss": 1.0317,
+      "step": 2102
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5897181859127763e-05,
+      "loss": 1.0127,
+      "step": 2103
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.589315769134485e-05,
+      "loss": 0.9457,
+      "step": 2104
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5889132060913796e-05,
+      "loss": 0.9767,
+      "step": 2105
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5885104968833726e-05,
+      "loss": 1.0603,
+      "step": 2106
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5881076416104146e-05,
+      "loss": 1.1419,
+      "step": 2107
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5877046403724922e-05,
+      "loss": 1.0358,
+      "step": 2108
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.587301493269628e-05,
+      "loss": 1.0051,
+      "step": 2109
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5868982004018806e-05,
+      "loss": 1.0382,
+      "step": 2110
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.586494761869345e-05,
+      "loss": 0.9527,
+      "step": 2111
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5860911777721528e-05,
+      "loss": 1.1209,
+      "step": 2112
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.585687448210471e-05,
+      "loss": 1.0123,
+      "step": 2113
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5852835732845023e-05,
+      "loss": 1.0271,
+      "step": 2114
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.584879553094487e-05,
+      "loss": 1.0441,
+      "step": 2115
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5844753877407002e-05,
+      "loss": 0.9736,
+      "step": 2116
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5840710773234537e-05,
+      "loss": 0.9191,
+      "step": 2117
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.583666621943095e-05,
+      "loss": 1.0246,
+      "step": 2118
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5832620217000076e-05,
+      "loss": 0.9818,
+      "step": 2119
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5828572766946107e-05,
+      "loss": 1.2,
+      "step": 2120
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.58245238702736e-05,
+      "loss": 0.925,
+      "step": 2121
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.582047352798747e-05,
+      "loss": 1.089,
+      "step": 2122
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5816421741092982e-05,
+      "loss": 1.0427,
+      "step": 2123
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5812368510595774e-05,
+      "loss": 0.8557,
+      "step": 2124
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.580831383750183e-05,
+      "loss": 1.0111,
+      "step": 2125
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5804257722817496e-05,
+      "loss": 1.0926,
+      "step": 2126
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5800200167549485e-05,
+      "loss": 0.9211,
+      "step": 2127
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5796141172704846e-05,
+      "loss": 1.003,
+      "step": 2128
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5792080739291006e-05,
+      "loss": 1.2255,
+      "step": 2129
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5788018868315746e-05,
+      "loss": 1.1413,
+      "step": 2130
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.578395556078719e-05,
+      "loss": 1.027,
+      "step": 2131
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5779890817713827e-05,
+      "loss": 1.1323,
+      "step": 2132
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.577582464010451e-05,
+      "loss": 1.0497,
+      "step": 2133
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5771757028968445e-05,
+      "loss": 1.119,
+      "step": 2134
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5767687985315178e-05,
+      "loss": 1.0982,
+      "step": 2135
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5763617510154628e-05,
+      "loss": 1.0613,
+      "step": 2136
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5759545604497067e-05,
+      "loss": 0.9906,
+      "step": 2137
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5755472269353115e-05,
+      "loss": 0.9977,
+      "step": 2138
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5751397505733745e-05,
+      "loss": 1.0021,
+      "step": 2139
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5747321314650302e-05,
+      "loss": 0.9746,
+      "step": 2140
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5743243697114468e-05,
+      "loss": 1.0476,
+      "step": 2141
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.573916465413828e-05,
+      "loss": 1.049,
+      "step": 2142
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.573508418673414e-05,
+      "loss": 1.0726,
+      "step": 2143
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5731002295914793e-05,
+      "loss": 1.0915,
+      "step": 2144
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5726918982693344e-05,
+      "loss": 0.9512,
+      "step": 2145
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5722834248083242e-05,
+      "loss": 1.0624,
+      "step": 2146
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5718748093098303e-05,
+      "loss": 1.1176,
+      "step": 2147
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5714660518752682e-05,
+      "loss": 1.1434,
+      "step": 2148
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.571057152606089e-05,
+      "loss": 1.0406,
+      "step": 2149
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.57064811160378e-05,
+      "loss": 1.0249,
+      "step": 2150
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5702389289698618e-05,
+      "loss": 0.9099,
+      "step": 2151
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5698296048058922e-05,
+      "loss": 1.0486,
+      "step": 2152
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5694201392134624e-05,
+      "loss": 0.964,
+      "step": 2153
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5690105322941995e-05,
+      "loss": 0.9812,
+      "step": 2154
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.568600784149766e-05,
+      "loss": 1.0081,
+      "step": 2155
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5681908948818584e-05,
+      "loss": 1.0489,
+      "step": 2156
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.56778086459221e-05,
+      "loss": 0.9073,
+      "step": 2157
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5673706933825867e-05,
+      "loss": 1.0442,
+      "step": 2158
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5669603813547913e-05,
+      "loss": 0.9817,
+      "step": 2159
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.566549928610661e-05,
+      "loss": 0.9939,
+      "step": 2160
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.566139335252068e-05,
+      "loss": 1.0376,
+      "step": 2161
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.565728601380918e-05,
+      "loss": 1.0424,
+      "step": 2162
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5653177270991547e-05,
+      "loss": 1.0266,
+      "step": 2163
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.564906712508753e-05,
+      "loss": 0.9796,
+      "step": 2164
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5644955577117256e-05,
+      "loss": 1.027,
+      "step": 2165
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.564084262810118e-05,
+      "loss": 1.0176,
+      "step": 2166
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.563672827906012e-05,
+      "loss": 1.0435,
+      "step": 2167
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5632612531015224e-05,
+      "loss": 1.1543,
+      "step": 2168
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.562849538498801e-05,
+      "loss": 1.0298,
+      "step": 2169
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5624376842000315e-05,
+      "loss": 1.0423,
+      "step": 2170
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.562025690307435e-05,
+      "loss": 1.0461,
+      "step": 2171
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5616135569232656e-05,
+      "loss": 0.9795,
+      "step": 2172
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5612012841498124e-05,
+      "loss": 1.0135,
+      "step": 2173
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5607888720893994e-05,
+      "loss": 1.0518,
+      "step": 2174
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5603763208443848e-05,
+      "loss": 1.0112,
+      "step": 2175
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.559963630517161e-05,
+      "loss": 1.0193,
+      "step": 2176
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5595508012101565e-05,
+      "loss": 1.039,
+      "step": 2177
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.559137833025832e-05,
+      "loss": 1.0767,
+      "step": 2178
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5587247260666845e-05,
+      "loss": 1.0606,
+      "step": 2179
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5583114804352443e-05,
+      "loss": 1.1026,
+      "step": 2180
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5578980962340775e-05,
+      "loss": 1.0205,
+      "step": 2181
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.557484573565783e-05,
+      "loss": 0.8921,
+      "step": 2182
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5570709125329947e-05,
+      "loss": 1.0017,
+      "step": 2183
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5566571132383814e-05,
+      "loss": 0.9515,
+      "step": 2184
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.556243175784645e-05,
+      "loss": 1.1022,
+      "step": 2185
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5558291002745232e-05,
+      "loss": 1.1211,
+      "step": 2186
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.555414886810787e-05,
+      "loss": 0.9476,
+      "step": 2187
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5550005354962413e-05,
+      "loss": 1.0176,
+      "step": 2188
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5545860464337265e-05,
+      "loss": 0.9961,
+      "step": 2189
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.554171419726116e-05,
+      "loss": 1.0843,
+      "step": 2190
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5537566554763177e-05,
+      "loss": 0.9766,
+      "step": 2191
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5533417537872738e-05,
+      "loss": 1.0227,
+      "step": 2192
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5529267147619608e-05,
+      "loss": 0.9629,
+      "step": 2193
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5525115385033882e-05,
+      "loss": 1.1604,
+      "step": 2194
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5520962251146013e-05,
+      "loss": 1.114,
+      "step": 2195
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5516807746986785e-05,
+      "loss": 1.0321,
+      "step": 2196
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5512651873587316e-05,
+      "loss": 1.0109,
+      "step": 2197
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5508494631979074e-05,
+      "loss": 1.0644,
+      "step": 2198
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5504336023193865e-05,
+      "loss": 1.1513,
+      "step": 2199
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5500176048263824e-05,
+      "loss": 1.0047,
+      "step": 2200
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5496014708221442e-05,
+      "loss": 0.9713,
+      "step": 2201
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5491852004099533e-05,
+      "loss": 1.0384,
+      "step": 2202
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.548768793693126e-05,
+      "loss": 0.8687,
+      "step": 2203
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.548352250775012e-05,
+      "loss": 0.9893,
+      "step": 2204
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.547935571758995e-05,
+      "loss": 0.8818,
+      "step": 2205
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.547518756748492e-05,
+      "loss": 1.1213,
+      "step": 2206
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5471018058469547e-05,
+      "loss": 1.0937,
+      "step": 2207
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5466847191578674e-05,
+      "loss": 0.871,
+      "step": 2208
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.546267496784749e-05,
+      "loss": 1.0764,
+      "step": 2209
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5458501388311508e-05,
+      "loss": 0.9768,
+      "step": 2210
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5454326454006603e-05,
+      "loss": 0.9587,
+      "step": 2211
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5450150165968957e-05,
+      "loss": 0.8945,
+      "step": 2212
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5445972525235106e-05,
+      "loss": 1.0587,
+      "step": 2213
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5441793532841914e-05,
+      "loss": 0.9339,
+      "step": 2214
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5437613189826587e-05,
+      "loss": 0.9191,
+      "step": 2215
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5433431497226655e-05,
+      "loss": 0.9838,
+      "step": 2216
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.542924845608e-05,
+      "loss": 0.8402,
+      "step": 2217
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5425064067424823e-05,
+      "loss": 1.007,
+      "step": 2218
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5420878332299668e-05,
+      "loss": 0.8972,
+      "step": 2219
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5416691251743406e-05,
+      "loss": 0.9888,
+      "step": 2220
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5412502826795254e-05,
+      "loss": 1.0269,
+      "step": 2221
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5408313058494745e-05,
+      "loss": 0.9446,
+      "step": 2222
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5404121947881765e-05,
+      "loss": 1.1374,
+      "step": 2223
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.539992949599652e-05,
+      "loss": 1.0992,
+      "step": 2224
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5395735703879553e-05,
+      "loss": 0.9751,
+      "step": 2225
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5391540572571737e-05,
+      "loss": 1.0221,
+      "step": 2226
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.538734410311429e-05,
+      "loss": 0.915,
+      "step": 2227
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5383146296548733e-05,
+      "loss": 0.9285,
+      "step": 2228
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5378947153916953e-05,
+      "loss": 1.0427,
+      "step": 2229
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5374746676261147e-05,
+      "loss": 0.8968,
+      "step": 2230
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5370544864623855e-05,
+      "loss": 0.9012,
+      "step": 2231
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5366341720047934e-05,
+      "loss": 0.9298,
+      "step": 2232
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.536213724357659e-05,
+      "loss": 1.1206,
+      "step": 2233
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.535793143625334e-05,
+      "loss": 0.9037,
+      "step": 2234
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5353724299122053e-05,
+      "loss": 1.0954,
+      "step": 2235
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5349515833226907e-05,
+      "loss": 0.9698,
+      "step": 2236
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.534530603961242e-05,
+      "loss": 0.9414,
+      "step": 2237
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5341094919323447e-05,
+      "loss": 0.9726,
+      "step": 2238
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5336882473405155e-05,
+      "loss": 0.9073,
+      "step": 2239
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5332668702903056e-05,
+      "loss": 0.9594,
+      "step": 2240
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5328453608862975e-05,
+      "loss": 0.9909,
+      "step": 2241
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5324237192331085e-05,
+      "loss": 1.1005,
+      "step": 2242
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5320019454353868e-05,
+      "loss": 0.9875,
+      "step": 2243
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5315800395978145e-05,
+      "loss": 1.0517,
+      "step": 2244
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.531158001825107e-05,
+      "loss": 1.0156,
+      "step": 2245
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5307358322220102e-05,
+      "loss": 1.0314,
+      "step": 2246
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.530313530893305e-05,
+      "loss": 1.0422,
+      "step": 2247
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5298910979438045e-05,
+      "loss": 0.8951,
+      "step": 2248
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.529468533478354e-05,
+      "loss": 1.0641,
+      "step": 2249
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5290458376018305e-05,
+      "loss": 0.888,
+      "step": 2250
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.528623010419146e-05,
+      "loss": 0.9308,
+      "step": 2251
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5282000520352438e-05,
+      "loss": 0.9567,
+      "step": 2252
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5277769625550992e-05,
+      "loss": 0.9105,
+      "step": 2253
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5273537420837203e-05,
+      "loss": 1.0047,
+      "step": 2254
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5269303907261487e-05,
+      "loss": 1.0601,
+      "step": 2255
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5265069085874574e-05,
+      "loss": 1.2335,
+      "step": 2256
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5260832957727528e-05,
+      "loss": 1.0945,
+      "step": 2257
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5256595523871726e-05,
+      "loss": 0.9533,
+      "step": 2258
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5252356785358875e-05,
+      "loss": 0.9469,
+      "step": 2259
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.524811674324101e-05,
+      "loss": 0.935,
+      "step": 2260
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5243875398570484e-05,
+      "loss": 1.0144,
+      "step": 2261
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5239632752399968e-05,
+      "loss": 0.9413,
+      "step": 2262
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5235388805782472e-05,
+      "loss": 0.9939,
+      "step": 2263
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5231143559771315e-05,
+      "loss": 0.9556,
+      "step": 2264
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5226897015420143e-05,
+      "loss": 1.2351,
+      "step": 2265
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5222649173782924e-05,
+      "loss": 1.0737,
+      "step": 2266
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.521840003591395e-05,
+      "loss": 0.9851,
+      "step": 2267
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5214149602867827e-05,
+      "loss": 0.8587,
+      "step": 2268
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5209897875699495e-05,
+      "loss": 1.0218,
+      "step": 2269
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5205644855464207e-05,
+      "loss": 0.9167,
+      "step": 2270
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5201390543217537e-05,
+      "loss": 1.0379,
+      "step": 2271
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5197134940015381e-05,
+      "loss": 1.0969,
+      "step": 2272
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5192878046913956e-05,
+      "loss": 1.0249,
+      "step": 2273
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5188619864969796e-05,
+      "loss": 0.9692,
+      "step": 2274
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5184360395239761e-05,
+      "loss": 0.938,
+      "step": 2275
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5180099638781027e-05,
+      "loss": 0.9708,
+      "step": 2276
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5175837596651088e-05,
+      "loss": 0.9827,
+      "step": 2277
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5171574269907761e-05,
+      "loss": 1.1285,
+      "step": 2278
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5167309659609175e-05,
+      "loss": 1.0902,
+      "step": 2279
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5163043766813785e-05,
+      "loss": 1.0243,
+      "step": 2280
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.515877659258036e-05,
+      "loss": 1.0013,
+      "step": 2281
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.515450813796799e-05,
+      "loss": 1.0556,
+      "step": 2282
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.515023840403608e-05,
+      "loss": 0.9342,
+      "step": 2283
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5145967391844352e-05,
+      "loss": 1.0364,
+      "step": 2284
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.514169510245285e-05,
+      "loss": 1.0031,
+      "step": 2285
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.513742153692193e-05,
+      "loss": 0.9104,
+      "step": 2286
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5133146696312269e-05,
+      "loss": 0.982,
+      "step": 2287
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5128870581684853e-05,
+      "loss": 0.8462,
+      "step": 2288
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5124593194100996e-05,
+      "loss": 1.058,
+      "step": 2289
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5120314534622314e-05,
+      "loss": 1.1842,
+      "step": 2290
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5116034604310752e-05,
+      "loss": 1.0422,
+      "step": 2291
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5111753404228559e-05,
+      "loss": 0.9677,
+      "step": 2292
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5107470935438313e-05,
+      "loss": 1.06,
+      "step": 2293
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5103187199002889e-05,
+      "loss": 0.9859,
+      "step": 2294
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.509890219598549e-05,
+      "loss": 0.9413,
+      "step": 2295
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5094615927449626e-05,
+      "loss": 0.9169,
+      "step": 2296
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.509032839445913e-05,
+      "loss": 0.9494,
+      "step": 2297
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5086039598078142e-05,
+      "loss": 1.0869,
+      "step": 2298
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5081749539371115e-05,
+      "loss": 0.9558,
+      "step": 2299
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5077458219402817e-05,
+      "loss": 0.9892,
+      "step": 2300
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.507316563923833e-05,
+      "loss": 0.9282,
+      "step": 2301
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5068871799943043e-05,
+      "loss": 0.9354,
+      "step": 2302
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5064576702582673e-05,
+      "loss": 1.0461,
+      "step": 2303
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5060280348223231e-05,
+      "loss": 1.0571,
+      "step": 2304
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.505598273793105e-05,
+      "loss": 1.0207,
+      "step": 2305
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5051683872772771e-05,
+      "loss": 0.9,
+      "step": 2306
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.504738375381535e-05,
+      "loss": 0.9546,
+      "step": 2307
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5043082382126047e-05,
+      "loss": 1.0453,
+      "step": 2308
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.5038779758772444e-05,
+      "loss": 1.0566,
+      "step": 2309
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.5034475884822426e-05,
+      "loss": 0.9839,
+      "step": 2310
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.5030170761344187e-05,
+      "loss": 1.015,
+      "step": 2311
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.502586438940624e-05,
+      "loss": 0.9385,
+      "step": 2312
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.5021556770077395e-05,
+      "loss": 0.9248,
+      "step": 2313
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.501724790442678e-05,
+      "loss": 0.894,
+      "step": 2314
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.5012937793523836e-05,
+      "loss": 0.8905,
+      "step": 2315
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.5008626438438306e-05,
+      "loss": 1.121,
+      "step": 2316
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.500431384024024e-05,
+      "loss": 1.0856,
+      "step": 2317
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 0.7907,
+      "step": 2318
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4995684918788263e-05,
+      "loss": 1.0001,
+      "step": 2319
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4991368597676003e-05,
+      "loss": 0.9911,
+      "step": 2320
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4987051037734508e-05,
+      "loss": 1.0075,
+      "step": 2321
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4982732240035368e-05,
+      "loss": 0.9805,
+      "step": 2322
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4978412205650492e-05,
+      "loss": 1.0569,
+      "step": 2323
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.497409093565208e-05,
+      "loss": 1.0001,
+      "step": 2324
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4969768431112646e-05,
+      "loss": 1.0834,
+      "step": 2325
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4965444693105019e-05,
+      "loss": 0.9736,
+      "step": 2326
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4961119722702319e-05,
+      "loss": 1.1085,
+      "step": 2327
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4956793520977983e-05,
+      "loss": 1.0699,
+      "step": 2328
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4952466089005748e-05,
+      "loss": 0.9602,
+      "step": 2329
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4948137427859658e-05,
+      "loss": 0.9788,
+      "step": 2330
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.494380753861406e-05,
+      "loss": 0.8727,
+      "step": 2331
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4939476422343613e-05,
+      "loss": 0.8316,
+      "step": 2332
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.493514408012327e-05,
+      "loss": 0.8812,
+      "step": 2333
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4930810513028298e-05,
+      "loss": 0.9157,
+      "step": 2334
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.492647572213426e-05,
+      "loss": 1.0819,
+      "step": 2335
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4922139708517027e-05,
+      "loss": 1.0026,
+      "step": 2336
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4917802473252773e-05,
+      "loss": 1.0126,
+      "step": 2337
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4913464017417978e-05,
+      "loss": 0.9611,
+      "step": 2338
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4909124342089421e-05,
+      "loss": 1.1165,
+      "step": 2339
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4904783448344182e-05,
+      "loss": 0.989,
+      "step": 2340
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4900441337259649e-05,
+      "loss": 0.9806,
+      "step": 2341
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4896098009913507e-05,
+      "loss": 1.105,
+      "step": 2342
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4891753467383743e-05,
+      "loss": 1.0833,
+      "step": 2343
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4887407710748653e-05,
+      "loss": 0.9577,
+      "step": 2344
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.488306074108683e-05,
+      "loss": 1.0293,
+      "step": 2345
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4878712559477159e-05,
+      "loss": 0.9413,
+      "step": 2346
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4874363166998842e-05,
+      "loss": 0.9837,
+      "step": 2347
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4870012564731368e-05,
+      "loss": 1.0526,
+      "step": 2348
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4865660753754534e-05,
+      "loss": 0.9773,
+      "step": 2349
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4861307735148435e-05,
+      "loss": 0.9394,
+      "step": 2350
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4856953509993464e-05,
+      "loss": 1.1021,
+      "step": 2351
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4852598079370318e-05,
+      "loss": 0.8875,
+      "step": 2352
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4848241444359986e-05,
+      "loss": 1.0269,
+      "step": 2353
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4843883606043761e-05,
+      "loss": 1.1553,
+      "step": 2354
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4839524565503236e-05,
+      "loss": 0.9944,
+      "step": 2355
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4835164323820297e-05,
+      "loss": 1.0454,
+      "step": 2356
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4830802882077136e-05,
+      "loss": 1.2588,
+      "step": 2357
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.482644024135623e-05,
+      "loss": 1.0876,
+      "step": 2358
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4822076402740369e-05,
+      "loss": 1.0226,
+      "step": 2359
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4817711367312628e-05,
+      "loss": 1.0827,
+      "step": 2360
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4813345136156388e-05,
+      "loss": 1.0817,
+      "step": 2361
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4808977710355318e-05,
+      "loss": 1.0386,
+      "step": 2362
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4804609090993394e-05,
+      "loss": 0.9755,
+      "step": 2363
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4800239279154877e-05,
+      "loss": 0.821,
+      "step": 2364
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.479586827592433e-05,
+      "loss": 0.9933,
+      "step": 2365
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4791496082386615e-05,
+      "loss": 0.9864,
+      "step": 2366
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4787122699626883e-05,
+      "loss": 1.0768,
+      "step": 2367
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4782748128730585e-05,
+      "loss": 0.9414,
+      "step": 2368
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.477837237078346e-05,
+      "loss": 1.0553,
+      "step": 2369
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4773995426871548e-05,
+      "loss": 0.991,
+      "step": 2370
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.476961729808118e-05,
+      "loss": 1.0564,
+      "step": 2371
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.476523798549899e-05,
+      "loss": 1.0415,
+      "step": 2372
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.476085749021189e-05,
+      "loss": 0.9139,
+      "step": 2373
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4756475813307099e-05,
+      "loss": 1.0881,
+      "step": 2374
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4752092955872121e-05,
+      "loss": 0.9314,
+      "step": 2375
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4747708918994757e-05,
+      "loss": 0.8559,
+      "step": 2376
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4743323703763099e-05,
+      "loss": 0.9137,
+      "step": 2377
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4738937311265538e-05,
+      "loss": 0.9933,
+      "step": 2378
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4734549742590747e-05,
+      "loss": 0.8361,
+      "step": 2379
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4730160998827695e-05,
+      "loss": 1.0522,
+      "step": 2380
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.472577108106564e-05,
+      "loss": 0.9843,
+      "step": 2381
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4721379990394146e-05,
+      "loss": 1.0847,
+      "step": 2382
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4716987727903045e-05,
+      "loss": 0.98,
+      "step": 2383
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4712594294682478e-05,
+      "loss": 1.0445,
+      "step": 2384
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4708199691822869e-05,
+      "loss": 0.892,
+      "step": 2385
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4703803920414931e-05,
+      "loss": 0.9877,
+      "step": 2386
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.469940698154967e-05,
+      "loss": 0.9174,
+      "step": 2387
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4695008876318386e-05,
+      "loss": 0.9204,
+      "step": 2388
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4690609605812656e-05,
+      "loss": 1.0817,
+      "step": 2389
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.468620917112436e-05,
+      "loss": 0.9289,
+      "step": 2390
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4681807573345656e-05,
+      "loss": 0.9485,
+      "step": 2391
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4677404813568998e-05,
+      "loss": 0.9986,
+      "step": 2392
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4673000892887128e-05,
+      "loss": 0.9138,
+      "step": 2393
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.466859581239307e-05,
+      "loss": 1.0142,
+      "step": 2394
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4664189573180144e-05,
+      "loss": 1.1843,
+      "step": 2395
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4659782176341948e-05,
+      "loss": 0.9281,
+      "step": 2396
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4655373622972378e-05,
+      "loss": 1.0248,
+      "step": 2397
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.465096391416561e-05,
+      "loss": 1.0101,
+      "step": 2398
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4646553051016106e-05,
+      "loss": 1.1285,
+      "step": 2399
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4642141034618622e-05,
+      "loss": 1.0117,
+      "step": 2400
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.463772786606819e-05,
+      "loss": 1.0641,
+      "step": 2401
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4633313546460138e-05,
+      "loss": 1.006,
+      "step": 2402
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4628898076890074e-05,
+      "loss": 0.9549,
+      "step": 2403
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.462448145845389e-05,
+      "loss": 0.9336,
+      "step": 2404
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4620063692247764e-05,
+      "loss": 1.0604,
+      "step": 2405
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.461564477936817e-05,
+      "loss": 1.0933,
+      "step": 2406
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4611224720911843e-05,
+      "loss": 0.8231,
+      "step": 2407
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4606803517975826e-05,
+      "loss": 1.0955,
+      "step": 2408
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4602381171657438e-05,
+      "loss": 1.0019,
+      "step": 2409
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.459795768305427e-05,
+      "loss": 1.1151,
+      "step": 2410
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4593533053264217e-05,
+      "loss": 1.0552,
+      "step": 2411
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4589107283385442e-05,
+      "loss": 1.0706,
+      "step": 2412
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4584680374516395e-05,
+      "loss": 0.9611,
+      "step": 2413
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4580252327755813e-05,
+      "loss": 1.1545,
+      "step": 2414
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4575823144202707e-05,
+      "loss": 0.8825,
+      "step": 2415
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.457139282495638e-05,
+      "loss": 1.0356,
+      "step": 2416
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4566961371116408e-05,
+      "loss": 1.0765,
+      "step": 2417
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.456252878378266e-05,
+      "loss": 0.9893,
+      "step": 2418
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.455809506405527e-05,
+      "loss": 0.9298,
+      "step": 2419
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4553660213034666e-05,
+      "loss": 0.9856,
+      "step": 2420
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4549224231821552e-05,
+      "loss": 1.1342,
+      "step": 2421
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4544787121516915e-05,
+      "loss": 0.924,
+      "step": 2422
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4540348883222015e-05,
+      "loss": 1.0049,
+      "step": 2423
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4535909518038405e-05,
+      "loss": 0.9276,
+      "step": 2424
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4531469027067907e-05,
+      "loss": 0.8669,
+      "step": 2425
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4527027411412624e-05,
+      "loss": 0.9648,
+      "step": 2426
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.452258467217494e-05,
+      "loss": 1.0499,
+      "step": 2427
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4518140810457521e-05,
+      "loss": 0.9354,
+      "step": 2428
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4513695827363302e-05,
+      "loss": 0.9287,
+      "step": 2429
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4509249723995508e-05,
+      "loss": 0.9869,
+      "step": 2430
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4504802501457636e-05,
+      "loss": 1.0566,
+      "step": 2431
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4500354160853458e-05,
+      "loss": 0.943,
+      "step": 2432
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.449590470328703e-05,
+      "loss": 0.9647,
+      "step": 2433
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4491454129862681e-05,
+      "loss": 1.1038,
+      "step": 2434
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4487002441685014e-05,
+      "loss": 0.8132,
+      "step": 2435
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4482549639858919e-05,
+      "loss": 0.9369,
+      "step": 2436
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4478095725489555e-05,
+      "loss": 1.0705,
+      "step": 2437
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4473640699682351e-05,
+      "loss": 0.9358,
+      "step": 2438
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4469184563543029e-05,
+      "loss": 0.9987,
+      "step": 2439
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4464727318177572e-05,
+      "loss": 1.05,
+      "step": 2440
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.446026896469224e-05,
+      "loss": 1.108,
+      "step": 2441
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4455809504193576e-05,
+      "loss": 1.1097,
+      "step": 2442
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.445134893778839e-05,
+      "loss": 0.978,
+      "step": 2443
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4446887266583769e-05,
+      "loss": 1.0508,
+      "step": 2444
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4442424491687074e-05,
+      "loss": 1.0425,
+      "step": 2445
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4437960614205945e-05,
+      "loss": 0.8926,
+      "step": 2446
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4433495635248284e-05,
+      "loss": 0.9259,
+      "step": 2447
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4429029555922278e-05,
+      "loss": 0.9995,
+      "step": 2448
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4424562377336381e-05,
+      "loss": 1.0433,
+      "step": 2449
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4420094100599323e-05,
+      "loss": 1.0766,
+      "step": 2450
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4415624726820099e-05,
+      "loss": 0.9507,
+      "step": 2451
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.441115425710799e-05,
+      "loss": 1.1172,
+      "step": 2452
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4406682692572535e-05,
+      "loss": 1.1602,
+      "step": 2453
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4402210034323554e-05,
+      "loss": 0.9719,
+      "step": 2454
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4397736283471136e-05,
+      "loss": 0.9925,
+      "step": 2455
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4393261441125636e-05,
+      "loss": 0.9587,
+      "step": 2456
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4388785508397688e-05,
+      "loss": 0.9964,
+      "step": 2457
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4384308486398195e-05,
+      "loss": 0.9932,
+      "step": 2458
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4379830376238321e-05,
+      "loss": 0.9456,
+      "step": 2459
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4375351179029515e-05,
+      "loss": 0.9268,
+      "step": 2460
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4370870895883485e-05,
+      "loss": 1.1786,
+      "step": 2461
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4366389527912212e-05,
+      "loss": 1.0339,
+      "step": 2462
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4361907076227944e-05,
+      "loss": 1.0517,
+      "step": 2463
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4357423541943205e-05,
+      "loss": 0.9776,
+      "step": 2464
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4352938926170776e-05,
+      "loss": 0.9851,
+      "step": 2465
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.434845323002372e-05,
+      "loss": 0.9245,
+      "step": 2466
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4343966454615357e-05,
+      "loss": 1.0283,
+      "step": 2467
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.433947860105928e-05,
+      "loss": 0.8453,
+      "step": 2468
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4334989670469345e-05,
+      "loss": 1.0206,
+      "step": 2469
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.433049966395969e-05,
+      "loss": 0.9546,
+      "step": 2470
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4326008582644694e-05,
+      "loss": 1.1396,
+      "step": 2471
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.432151642763903e-05,
+      "loss": 0.8789,
+      "step": 2472
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4317023200057618e-05,
+      "loss": 0.9522,
+      "step": 2473
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4312528901015657e-05,
+      "loss": 0.9804,
+      "step": 2474
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.43080335316286e-05,
+      "loss": 0.9595,
+      "step": 2475
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4303537093012179e-05,
+      "loss": 1.0361,
+      "step": 2476
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4299039586282377e-05,
+      "loss": 1.0566,
+      "step": 2477
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4294541012555458e-05,
+      "loss": 1.0236,
+      "step": 2478
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4290041372947937e-05,
+      "loss": 0.9013,
+      "step": 2479
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4285540668576599e-05,
+      "loss": 1.0144,
+      "step": 2480
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4281038900558492e-05,
+      "loss": 0.9842,
+      "step": 2481
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4276536070010935e-05,
+      "loss": 1.0101,
+      "step": 2482
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4272032178051498e-05,
+      "loss": 1.1316,
+      "step": 2483
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4267527225798025e-05,
+      "loss": 0.9525,
+      "step": 2484
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4263021214368619e-05,
+      "loss": 0.9021,
+      "step": 2485
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4258514144881647e-05,
+      "loss": 0.932,
+      "step": 2486
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4254006018455737e-05,
+      "loss": 1.1381,
+      "step": 2487
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4249496836209783e-05,
+      "loss": 0.9751,
+      "step": 2488
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4244986599262929e-05,
+      "loss": 0.9064,
+      "step": 2489
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4240475308734602e-05,
+      "loss": 1.0367,
+      "step": 2490
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4235962965744472e-05,
+      "loss": 1.0245,
+      "step": 2491
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.423144957141248e-05,
+      "loss": 1.0094,
+      "step": 2492
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4226935126858818e-05,
+      "loss": 0.9948,
+      "step": 2493
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4222419633203954e-05,
+      "loss": 0.9129,
+      "step": 2494
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4217903091568605e-05,
+      "loss": 0.8996,
+      "step": 2495
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.421338550307375e-05,
+      "loss": 1.0159,
+      "step": 2496
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4208866868840624e-05,
+      "loss": 0.8693,
+      "step": 2497
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.420434718999074e-05,
+      "loss": 1.0098,
+      "step": 2498
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.419982646764584e-05,
+      "loss": 0.8227,
+      "step": 2499
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4195304702927952e-05,
+      "loss": 0.8598,
+      "step": 2500
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4190781896959349e-05,
+      "loss": 0.9552,
+      "step": 2501
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4186258050862567e-05,
+      "loss": 0.9501,
+      "step": 2502
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4181733165760396e-05,
+      "loss": 1.0013,
+      "step": 2503
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4177207242775892e-05,
+      "loss": 1.0519,
+      "step": 2504
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4172680283032354e-05,
+      "loss": 1.118,
+      "step": 2505
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4168152287653358e-05,
+      "loss": 1.059,
+      "step": 2506
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4163623257762721e-05,
+      "loss": 1.0704,
+      "step": 2507
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4159093194484523e-05,
+      "loss": 0.9674,
+      "step": 2508
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4154562098943099e-05,
+      "loss": 1.0765,
+      "step": 2509
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4150029972263046e-05,
+      "loss": 1.2137,
+      "step": 2510
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4145496815569205e-05,
+      "loss": 0.9391,
+      "step": 2511
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4140962629986685e-05,
+      "loss": 0.9549,
+      "step": 2512
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4136427416640842e-05,
+      "loss": 1.1108,
+      "step": 2513
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4131891176657293e-05,
+      "loss": 0.9317,
+      "step": 2514
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4127353911161903e-05,
+      "loss": 0.8014,
+      "step": 2515
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.41228156212808e-05,
+      "loss": 0.9258,
+      "step": 2516
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4118276308140355e-05,
+      "loss": 1.0423,
+      "step": 2517
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4113735972867209e-05,
+      "loss": 0.9638,
+      "step": 2518
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.410919461658824e-05,
+      "loss": 0.9606,
+      "step": 2519
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4104652240430588e-05,
+      "loss": 1.0383,
+      "step": 2520
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4100108845521645e-05,
+      "loss": 0.9923,
+      "step": 2521
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4095564432989063e-05,
+      "loss": 1.0019,
+      "step": 2522
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4091019003960724e-05,
+      "loss": 1.0672,
+      "step": 2523
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.408647255956479e-05,
+      "loss": 0.9706,
+      "step": 2524
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.408192510092966e-05,
+      "loss": 1.2206,
+      "step": 2525
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4077376629183987e-05,
+      "loss": 0.9889,
+      "step": 2526
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4072827145456671e-05,
+      "loss": 0.9462,
+      "step": 2527
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.406827665087688e-05,
+      "loss": 0.9141,
+      "step": 2528
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4063725146574006e-05,
+      "loss": 1.0151,
+      "step": 2529
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4059172633677714e-05,
+      "loss": 1.0529,
+      "step": 2530
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4054619113317914e-05,
+      "loss": 0.9101,
+      "step": 2531
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.405006458662476e-05,
+      "loss": 0.9952,
+      "step": 2532
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4045509054728655e-05,
+      "loss": 1.0077,
+      "step": 2533
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4040952518760267e-05,
+      "loss": 0.9366,
+      "step": 2534
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4036394979850494e-05,
+      "loss": 0.9417,
+      "step": 2535
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4031836439130495e-05,
+      "loss": 0.8675,
+      "step": 2536
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4027276897731674e-05,
+      "loss": 1.0185,
+      "step": 2537
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4022716356785682e-05,
+      "loss": 0.9567,
+      "step": 2538
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4018154817424414e-05,
+      "loss": 1.0593,
+      "step": 2539
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.401359228078003e-05,
+      "loss": 1.045,
+      "step": 2540
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4009028747984915e-05,
+      "loss": 1.1272,
+      "step": 2541
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4004464220171716e-05,
+      "loss": 0.9046,
+      "step": 2542
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3999898698473322e-05,
+      "loss": 0.9305,
+      "step": 2543
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3995332184022872e-05,
+      "loss": 1.0501,
+      "step": 2544
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3990764677953743e-05,
+      "loss": 1.1258,
+      "step": 2545
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3986196181399574e-05,
+      "loss": 1.1219,
+      "step": 2546
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3981626695494228e-05,
+      "loss": 0.9761,
+      "step": 2547
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3977056221371832e-05,
+      "loss": 1.0409,
+      "step": 2548
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.397248476016675e-05,
+      "loss": 1.0037,
+      "step": 2549
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3967912313013596e-05,
+      "loss": 0.9447,
+      "step": 2550
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.396333888104722e-05,
+      "loss": 0.9652,
+      "step": 2551
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3958764465402726e-05,
+      "loss": 0.94,
+      "step": 2552
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3954189067215453e-05,
+      "loss": 0.9674,
+      "step": 2553
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3949612687620993e-05,
+      "loss": 0.9939,
+      "step": 2554
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.394503532775518e-05,
+      "loss": 0.9182,
+      "step": 2555
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3940456988754078e-05,
+      "loss": 1.0894,
+      "step": 2556
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3935877671754014e-05,
+      "loss": 0.9536,
+      "step": 2557
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3931297377891549e-05,
+      "loss": 1.0286,
+      "step": 2558
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3926716108303478e-05,
+      "loss": 0.9764,
+      "step": 2559
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3922133864126855e-05,
+      "loss": 1.0571,
+      "step": 2560
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3917550646498961e-05,
+      "loss": 0.9887,
+      "step": 2561
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3912966456557324e-05,
+      "loss": 1.0385,
+      "step": 2562
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3908381295439717e-05,
+      "loss": 0.9162,
+      "step": 2563
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3903795164284148e-05,
+      "loss": 0.9961,
+      "step": 2564
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3899208064228873e-05,
+      "loss": 0.9649,
+      "step": 2565
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3894619996412382e-05,
+      "loss": 0.8778,
+      "step": 2566
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3890030961973402e-05,
+      "loss": 1.0155,
+      "step": 2567
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3885440962050916e-05,
+      "loss": 0.9689,
+      "step": 2568
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3880849997784126e-05,
+      "loss": 0.9743,
+      "step": 2569
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3876258070312488e-05,
+      "loss": 0.9908,
+      "step": 2570
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3871665180775692e-05,
+      "loss": 0.9646,
+      "step": 2571
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3867071330313661e-05,
+      "loss": 0.9972,
+      "step": 2572
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3862476520066572e-05,
+      "loss": 1.0049,
+      "step": 2573
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3857880751174825e-05,
+      "loss": 1.0519,
+      "step": 2574
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3853284024779066e-05,
+      "loss": 0.8886,
+      "step": 2575
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3848686342020174e-05,
+      "loss": 1.0081,
+      "step": 2576
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.384408770403927e-05,
+      "loss": 0.9983,
+      "step": 2577
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3839488111977707e-05,
+      "loss": 1.047,
+      "step": 2578
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.383488756697708e-05,
+      "loss": 0.928,
+      "step": 2579
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.383028607017922e-05,
+      "loss": 0.941,
+      "step": 2580
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3825683622726182e-05,
+      "loss": 0.9685,
+      "step": 2581
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3821080225760282e-05,
+      "loss": 0.9167,
+      "step": 2582
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3816475880424047e-05,
+      "loss": 0.9386,
+      "step": 2583
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3811870587860253e-05,
+      "loss": 1.0007,
+      "step": 2584
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3807264349211901e-05,
+      "loss": 0.9749,
+      "step": 2585
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3802657165622246e-05,
+      "loss": 0.8462,
+      "step": 2586
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.379804903823475e-05,
+      "loss": 0.9679,
+      "step": 2587
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3793439968193137e-05,
+      "loss": 1.0566,
+      "step": 2588
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3788829956641344e-05,
+      "loss": 0.9991,
+      "step": 2589
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.378421900472355e-05,
+      "loss": 0.9326,
+      "step": 2590
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3779607113584167e-05,
+      "loss": 1.0115,
+      "step": 2591
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3774994284367847e-05,
+      "loss": 0.8504,
+      "step": 2592
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3770380518219458e-05,
+      "loss": 1.0685,
+      "step": 2593
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3765765816284115e-05,
+      "loss": 1.0477,
+      "step": 2594
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3761150179707164e-05,
+      "loss": 1.053,
+      "step": 2595
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3756533609634176e-05,
+      "loss": 0.9748,
+      "step": 2596
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3751916107210954e-05,
+      "loss": 0.929,
+      "step": 2597
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3747297673583544e-05,
+      "loss": 1.1566,
+      "step": 2598
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3742678309898208e-05,
+      "loss": 0.9706,
+      "step": 2599
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3738058017301447e-05,
+      "loss": 1.0588,
+      "step": 2600
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3733436796939996e-05,
+      "loss": 1.0694,
+      "step": 2601
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.372881464996081e-05,
+      "loss": 1.0026,
+      "step": 2602
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3724191577511079e-05,
+      "loss": 1.0654,
+      "step": 2603
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3719567580738231e-05,
+      "loss": 1.0383,
+      "step": 2604
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3714942660789907e-05,
+      "loss": 1.0487,
+      "step": 2605
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3710316818813989e-05,
+      "loss": 1.0023,
+      "step": 2606
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3705690055958587e-05,
+      "loss": 0.972,
+      "step": 2607
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3701062373372034e-05,
+      "loss": 0.8728,
+      "step": 2608
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3696433772202892e-05,
+      "loss": 0.9835,
+      "step": 2609
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3691804253599963e-05,
+      "loss": 0.9726,
+      "step": 2610
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3687173818712258e-05,
+      "loss": 1.0116,
+      "step": 2611
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3682542468689028e-05,
+      "loss": 0.8231,
+      "step": 2612
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3677910204679747e-05,
+      "loss": 0.8553,
+      "step": 2613
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3673277027834118e-05,
+      "loss": 1.0219,
+      "step": 2614
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3668642939302065e-05,
+      "loss": 1.0296,
+      "step": 2615
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3664007940233748e-05,
+      "loss": 0.9746,
+      "step": 2616
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3659372031779543e-05,
+      "loss": 0.9749,
+      "step": 2617
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3654735215090059e-05,
+      "loss": 1.1461,
+      "step": 2618
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3650097491316123e-05,
+      "loss": 0.8739,
+      "step": 2619
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3645458861608799e-05,
+      "loss": 0.9783,
+      "step": 2620
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3640819327119359e-05,
+      "loss": 0.964,
+      "step": 2621
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3636178888999316e-05,
+      "loss": 0.9837,
+      "step": 2622
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.36315375484004e-05,
+      "loss": 0.9906,
+      "step": 2623
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3626895306474563e-05,
+      "loss": 1.1847,
+      "step": 2624
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3622252164373986e-05,
+      "loss": 0.9744,
+      "step": 2625
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3617608123251065e-05,
+      "loss": 1.1798,
+      "step": 2626
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3612963184258428e-05,
+      "loss": 1.011,
+      "step": 2627
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3608317348548923e-05,
+      "loss": 1.0155,
+      "step": 2628
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3603670617275617e-05,
+      "loss": 0.9895,
+      "step": 2629
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3599022991591806e-05,
+      "loss": 0.9992,
+      "step": 2630
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3594374472651004e-05,
+      "loss": 0.8785,
+      "step": 2631
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3589725061606944e-05,
+      "loss": 0.8923,
+      "step": 2632
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.358507475961358e-05,
+      "loss": 0.8884,
+      "step": 2633
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.35804235678251e-05,
+      "loss": 1.2066,
+      "step": 2634
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3575771487395899e-05,
+      "loss": 0.9745,
+      "step": 2635
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3571118519480596e-05,
+      "loss": 1.0412,
+      "step": 2636
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3566464665234026e-05,
+      "loss": 0.9806,
+      "step": 2637
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3561809925811261e-05,
+      "loss": 0.9842,
+      "step": 2638
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.355715430236757e-05,
+      "loss": 0.9787,
+      "step": 2639
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.355249779605846e-05,
+      "loss": 1.1148,
+      "step": 2640
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3547840408039644e-05,
+      "loss": 1.0429,
+      "step": 2641
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.354318213946706e-05,
+      "loss": 1.012,
+      "step": 2642
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3538522991496864e-05,
+      "loss": 0.9297,
+      "step": 2643
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3533862965285432e-05,
+      "loss": 1.0552,
+      "step": 2644
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3529202061989352e-05,
+      "loss": 1.0658,
+      "step": 2645
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.352454028276544e-05,
+      "loss": 0.977,
+      "step": 2646
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3519877628770717e-05,
+      "loss": 1.0143,
+      "step": 2647
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3515214101162431e-05,
+      "loss": 1.0202,
+      "step": 2648
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3510549701098037e-05,
+      "loss": 0.9726,
+      "step": 2649
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.350588442973522e-05,
+      "loss": 0.9033,
+      "step": 2650
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3501218288231873e-05,
+      "loss": 1.0431,
+      "step": 2651
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.34965512777461e-05,
+      "loss": 1.055,
+      "step": 2652
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.349188339943623e-05,
+      "loss": 1.1015,
+      "step": 2653
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3487214654460805e-05,
+      "loss": 0.942,
+      "step": 2654
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3482545043978577e-05,
+      "loss": 1.0129,
+      "step": 2655
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3477874569148521e-05,
+      "loss": 0.9738,
+      "step": 2656
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.347320323112982e-05,
+      "loss": 0.9606,
+      "step": 2657
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3468531031081874e-05,
+      "loss": 0.9803,
+      "step": 2658
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3463857970164296e-05,
+      "loss": 1.0222,
+      "step": 2659
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3459184049536914e-05,
+      "loss": 0.9146,
+      "step": 2660
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3454509270359763e-05,
+      "loss": 0.9658,
+      "step": 2661
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3449833633793106e-05,
+      "loss": 0.9946,
+      "step": 2662
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3445157140997405e-05,
+      "loss": 1.0204,
+      "step": 2663
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3440479793133336e-05,
+      "loss": 0.9646,
+      "step": 2664
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3435801591361794e-05,
+      "loss": 1.0627,
+      "step": 2665
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.343112253684388e-05,
+      "loss": 0.823,
+      "step": 2666
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3426442630740908e-05,
+      "loss": 1.0244,
+      "step": 2667
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3421761874214408e-05,
+      "loss": 0.9342,
+      "step": 2668
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3417080268426112e-05,
+      "loss": 1.0059,
+      "step": 2669
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3412397814537972e-05,
+      "loss": 0.9128,
+      "step": 2670
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3407714513712144e-05,
+      "loss": 0.9839,
+      "step": 2671
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3403030367110997e-05,
+      "loss": 0.8417,
+      "step": 2672
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3398345375897111e-05,
+      "loss": 0.9978,
+      "step": 2673
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.339365954123327e-05,
+      "loss": 0.8315,
+      "step": 2674
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3388972864282477e-05,
+      "loss": 1.1182,
+      "step": 2675
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3384285346207935e-05,
+      "loss": 0.9762,
+      "step": 2676
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.337959698817306e-05,
+      "loss": 1.0081,
+      "step": 2677
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3374907791341473e-05,
+      "loss": 1.0417,
+      "step": 2678
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3370217756877011e-05,
+      "loss": 0.9655,
+      "step": 2679
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3365526885943712e-05,
+      "loss": 1.0003,
+      "step": 2680
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3360835179705822e-05,
+      "loss": 0.9396,
+      "step": 2681
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3356142639327798e-05,
+      "loss": 1.023,
+      "step": 2682
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.33514492659743e-05,
+      "loss": 0.9986,
+      "step": 2683
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3346755060810195e-05,
+      "loss": 1.0863,
+      "step": 2684
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.334206002500056e-05,
+      "loss": 0.8503,
+      "step": 2685
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.333736415971068e-05,
+      "loss": 1.1151,
+      "step": 2686
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3332667466106032e-05,
+      "loss": 0.8796,
+      "step": 2687
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.332796994535232e-05,
+      "loss": 0.9655,
+      "step": 2688
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3323271598615435e-05,
+      "loss": 1.102,
+      "step": 2689
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3318572427061481e-05,
+      "loss": 0.9517,
+      "step": 2690
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3313872431856765e-05,
+      "loss": 1.1204,
+      "step": 2691
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.33091716141678e-05,
+      "loss": 0.9112,
+      "step": 2692
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3304469975161305e-05,
+      "loss": 1.0104,
+      "step": 2693
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3299767516004193e-05,
+      "loss": 0.9797,
+      "step": 2694
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3295064237863595e-05,
+      "loss": 0.9406,
+      "step": 2695
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3290360141906833e-05,
+      "loss": 0.9027,
+      "step": 2696
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3285655229301437e-05,
+      "loss": 0.926,
+      "step": 2697
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3280949501215144e-05,
+      "loss": 1.0233,
+      "step": 2698
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3276242958815884e-05,
+      "loss": 0.8985,
+      "step": 2699
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3271535603271794e-05,
+      "loss": 0.9643,
+      "step": 2700
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.326682743575122e-05,
+      "loss": 1.0542,
+      "step": 2701
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.326211845742269e-05,
+      "loss": 0.9315,
+      "step": 2702
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3257408669454956e-05,
+      "loss": 1.0338,
+      "step": 2703
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3252698073016958e-05,
+      "loss": 1.0632,
+      "step": 2704
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.324798666927784e-05,
+      "loss": 1.0213,
+      "step": 2705
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3243274459406936e-05,
+      "loss": 0.976,
+      "step": 2706
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3238561444573803e-05,
+      "loss": 0.8978,
+      "step": 2707
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.323384762594818e-05,
+      "loss": 1.1531,
+      "step": 2708
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3229133004700005e-05,
+      "loss": 1.0146,
+      "step": 2709
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3224417581999426e-05,
+      "loss": 0.9136,
+      "step": 2710
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3219701359016782e-05,
+      "loss": 1.0077,
+      "step": 2711
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3214984336922609e-05,
+      "loss": 0.9832,
+      "step": 2712
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3210266516887653e-05,
+      "loss": 0.9463,
+      "step": 2713
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3205547900082844e-05,
+      "loss": 1.056,
+      "step": 2714
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3200828487679318e-05,
+      "loss": 1.0036,
+      "step": 2715
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3196108280848406e-05,
+      "loss": 0.967,
+      "step": 2716
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3191387280761633e-05,
+      "loss": 0.9258,
+      "step": 2717
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3186665488590728e-05,
+      "loss": 1.055,
+      "step": 2718
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3181942905507614e-05,
+      "loss": 1.0284,
+      "step": 2719
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.317721953268441e-05,
+      "loss": 0.7995,
+      "step": 2720
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3172495371293424e-05,
+      "loss": 0.954,
+      "step": 2721
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3167770422507169e-05,
+      "loss": 1.0785,
+      "step": 2722
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3163044687498352e-05,
+      "loss": 0.9277,
+      "step": 2723
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3158318167439872e-05,
+      "loss": 0.9239,
+      "step": 2724
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3153590863504824e-05,
+      "loss": 0.8976,
+      "step": 2725
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3148862776866503e-05,
+      "loss": 1.0034,
+      "step": 2726
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3144133908698384e-05,
+      "loss": 0.9871,
+      "step": 2727
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3139404260174151e-05,
+      "loss": 1.05,
+      "step": 2728
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3134673832467676e-05,
+      "loss": 0.9449,
+      "step": 2729
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3129942626753023e-05,
+      "loss": 1.0246,
+      "step": 2730
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3125210644204449e-05,
+      "loss": 0.9387,
+      "step": 2731
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3120477885996408e-05,
+      "loss": 1.0837,
+      "step": 2732
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3115744353303542e-05,
+      "loss": 0.8693,
+      "step": 2733
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3111010047300682e-05,
+      "loss": 0.9774,
+      "step": 2734
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.310627496916287e-05,
+      "loss": 0.951,
+      "step": 2735
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3101539120065307e-05,
+      "loss": 0.9729,
+      "step": 2736
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3096802501183417e-05,
+      "loss": 1.0183,
+      "step": 2737
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3092065113692796e-05,
+      "loss": 0.9487,
+      "step": 2738
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.308732695876924e-05,
+      "loss": 0.9605,
+      "step": 2739
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3082588037588728e-05,
+      "loss": 1.1366,
+      "step": 2740
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3077848351327438e-05,
+      "loss": 1.0422,
+      "step": 2741
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3073107901161728e-05,
+      "loss": 0.9113,
+      "step": 2742
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3068366688268153e-05,
+      "loss": 0.9994,
+      "step": 2743
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3063624713823457e-05,
+      "loss": 1.0794,
+      "step": 2744
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.305888197900457e-05,
+      "loss": 1.1914,
+      "step": 2745
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3054138484988606e-05,
+      "loss": 0.9741,
+      "step": 2746
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3049394232952883e-05,
+      "loss": 1.0497,
+      "step": 2747
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.304464922407489e-05,
+      "loss": 0.9281,
+      "step": 2748
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3039903459532315e-05,
+      "loss": 0.9177,
+      "step": 2749
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.303515694050303e-05,
+      "loss": 1.0147,
+      "step": 2750
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3030409668165089e-05,
+      "loss": 0.7937,
+      "step": 2751
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3025661643696742e-05,
+      "loss": 1.008,
+      "step": 2752
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3020912868276422e-05,
+      "loss": 1.0116,
+      "step": 2753
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3016163343082746e-05,
+      "loss": 1.0348,
+      "step": 2754
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3011413069294521e-05,
+      "loss": 0.912,
+      "step": 2755
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3006662048090734e-05,
+      "loss": 1.0457,
+      "step": 2756
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3001910280650566e-05,
+      "loss": 0.9078,
+      "step": 2757
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2997157768153377e-05,
+      "loss": 1.1,
+      "step": 2758
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2992404511778715e-05,
+      "loss": 0.9975,
+      "step": 2759
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2987650512706306e-05,
+      "loss": 1.0135,
+      "step": 2760
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2982895772116072e-05,
+      "loss": 1.0061,
+      "step": 2761
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2978140291188107e-05,
+      "loss": 0.9323,
+      "step": 2762
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.29733840711027e-05,
+      "loss": 0.9249,
+      "step": 2763
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2968627113040308e-05,
+      "loss": 0.8884,
+      "step": 2764
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2963869418181596e-05,
+      "loss": 1.0563,
+      "step": 2765
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2959110987707381e-05,
+      "loss": 1.0429,
+      "step": 2766
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2954351822798688e-05,
+      "loss": 0.9601,
+      "step": 2767
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2949591924636713e-05,
+      "loss": 0.9564,
+      "step": 2768
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2944831294402833e-05,
+      "loss": 1.0612,
+      "step": 2769
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.294006993327861e-05,
+      "loss": 0.9663,
+      "step": 2770
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.293530784244579e-05,
+      "loss": 1.0102,
+      "step": 2771
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2930545023086292e-05,
+      "loss": 1.0129,
+      "step": 2772
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2925781476382225e-05,
+      "loss": 1.172,
+      "step": 2773
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2921017203515866e-05,
+      "loss": 0.8988,
+      "step": 2774
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2916252205669695e-05,
+      "loss": 0.9507,
+      "step": 2775
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2911486484026346e-05,
+      "loss": 0.9219,
+      "step": 2776
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2906720039768645e-05,
+      "loss": 0.9818,
+      "step": 2777
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.29019528740796e-05,
+      "loss": 1.0091,
+      "step": 2778
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.289718498814239e-05,
+      "loss": 0.8957,
+      "step": 2779
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2892416383140377e-05,
+      "loss": 0.977,
+      "step": 2780
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2887647060257113e-05,
+      "loss": 0.9793,
+      "step": 2781
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2882877020676302e-05,
+      "loss": 0.9016,
+      "step": 2782
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2878106265581845e-05,
+      "loss": 0.9912,
+      "step": 2783
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2873334796157821e-05,
+      "loss": 0.8854,
+      "step": 2784
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2868562613588478e-05,
+      "loss": 0.8494,
+      "step": 2785
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.286378971905824e-05,
+      "loss": 0.9583,
+      "step": 2786
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2859016113751723e-05,
+      "loss": 0.9793,
+      "step": 2787
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.28542417988537e-05,
+      "loss": 0.8992,
+      "step": 2788
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2849466775549131e-05,
+      "loss": 0.9101,
+      "step": 2789
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.284469104502315e-05,
+      "loss": 1.0248,
+      "step": 2790
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2839914608461066e-05,
+      "loss": 0.9035,
+      "step": 2791
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2835137467048361e-05,
+      "loss": 0.9588,
+      "step": 2792
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2830359621970702e-05,
+      "loss": 0.8863,
+      "step": 2793
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2825581074413914e-05,
+      "loss": 0.9557,
+      "step": 2794
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2820801825564008e-05,
+      "loss": 1.0901,
+      "step": 2795
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2816021876607168e-05,
+      "loss": 0.9563,
+      "step": 2796
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2811241228729752e-05,
+      "loss": 0.9147,
+      "step": 2797
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.280645988311828e-05,
+      "loss": 1.0232,
+      "step": 2798
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2801677840959467e-05,
+      "loss": 1.1271,
+      "step": 2799
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2796895103440181e-05,
+      "loss": 0.9335,
+      "step": 2800
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2792111671747474e-05,
+      "loss": 0.8819,
+      "step": 2801
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2787327547068563e-05,
+      "loss": 1.005,
+      "step": 2802
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2782542730590843e-05,
+      "loss": 0.8512,
+      "step": 2803
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2777757223501875e-05,
+      "loss": 1.0228,
+      "step": 2804
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.27729710269894e-05,
+      "loss": 0.9431,
+      "step": 2805
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.276818414224132e-05,
+      "loss": 0.9133,
+      "step": 2806
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2763396570445714e-05,
+      "loss": 0.986,
+      "step": 2807
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2758608312790833e-05,
+      "loss": 1.0376,
+      "step": 2808
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2753819370465088e-05,
+      "loss": 0.9435,
+      "step": 2809
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2749029744657073e-05,
+      "loss": 0.9371,
+      "step": 2810
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2744239436555547e-05,
+      "loss": 0.9615,
+      "step": 2811
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2739448447349431e-05,
+      "loss": 0.9598,
+      "step": 2812
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2734656778227827e-05,
+      "loss": 0.9909,
+      "step": 2813
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2729864430379997e-05,
+      "loss": 1.1534,
+      "step": 2814
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2725071404995377e-05,
+      "loss": 1.0758,
+      "step": 2815
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2720277703263565e-05,
+      "loss": 1.0764,
+      "step": 2816
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2715483326374333e-05,
+      "loss": 0.9139,
+      "step": 2817
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2710688275517616e-05,
+      "loss": 1.1071,
+      "step": 2818
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.270589255188352e-05,
+      "loss": 0.9509,
+      "step": 2819
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2701096156662317e-05,
+      "loss": 1.1478,
+      "step": 2820
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2696299091044444e-05,
+      "loss": 0.9479,
+      "step": 2821
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2691501356220502e-05,
+      "loss": 0.9933,
+      "step": 2822
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.268670295338127e-05,
+      "loss": 1.0703,
+      "step": 2823
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2681903883717673e-05,
+      "loss": 0.8743,
+      "step": 2824
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.267710414842082e-05,
+      "loss": 1.0156,
+      "step": 2825
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2672303748681977e-05,
+      "loss": 1.0874,
+      "step": 2826
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2667502685692577e-05,
+      "loss": 1.0755,
+      "step": 2827
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2662700960644213e-05,
+      "loss": 1.1892,
+      "step": 2828
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2657898574728649e-05,
+      "loss": 1.0046,
+      "step": 2829
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2653095529137805e-05,
+      "loss": 0.9832,
+      "step": 2830
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2648291825063776e-05,
+      "loss": 0.9022,
+      "step": 2831
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2643487463698812e-05,
+      "loss": 0.9736,
+      "step": 2832
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2638682446235324e-05,
+      "loss": 1.1061,
+      "step": 2833
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2633876773865894e-05,
+      "loss": 1.0646,
+      "step": 2834
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2629070447783263e-05,
+      "loss": 1.0998,
+      "step": 2835
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2624263469180329e-05,
+      "loss": 0.9358,
+      "step": 2836
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2619455839250161e-05,
+      "loss": 0.9905,
+      "step": 2837
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2614647559185986e-05,
+      "loss": 0.9666,
+      "step": 2838
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2609838630181188e-05,
+      "loss": 0.9857,
+      "step": 2839
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2605029053429318e-05,
+      "loss": 0.8794,
+      "step": 2840
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2600218830124087e-05,
+      "loss": 1.0962,
+      "step": 2841
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.259540796145936e-05,
+      "loss": 1.0633,
+      "step": 2842
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2590596448629173e-05,
+      "loss": 0.9341,
+      "step": 2843
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2585784292827711e-05,
+      "loss": 0.9847,
+      "step": 2844
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.258097149524933e-05,
+      "loss": 0.9805,
+      "step": 2845
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2576158057088532e-05,
+      "loss": 1.0087,
+      "step": 2846
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.257134397953999e-05,
+      "loss": 0.9261,
+      "step": 2847
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2566529263798528e-05,
+      "loss": 0.8675,
+      "step": 2848
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.256171391105913e-05,
+      "loss": 0.9963,
+      "step": 2849
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2556897922516939e-05,
+      "loss": 0.9057,
+      "step": 2850
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2552081299367266e-05,
+      "loss": 0.9399,
+      "step": 2851
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2547264042805551e-05,
+      "loss": 1.0598,
+      "step": 2852
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2542446154027423e-05,
+      "loss": 0.891,
+      "step": 2853
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2537627634228652e-05,
+      "loss": 0.845,
+      "step": 2854
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2532808484605167e-05,
+      "loss": 1.0197,
+      "step": 2855
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2527988706353051e-05,
+      "loss": 0.9217,
+      "step": 2856
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.252316830066855e-05,
+      "loss": 1.0229,
+      "step": 2857
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2518347268748054e-05,
+      "loss": 0.8937,
+      "step": 2858
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2513525611788125e-05,
+      "loss": 1.0,
+      "step": 2859
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2508703330985464e-05,
+      "loss": 0.8945,
+      "step": 2860
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2503880427536937e-05,
+      "loss": 0.9794,
+      "step": 2861
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2499056902639554e-05,
+      "loss": 0.9365,
+      "step": 2862
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2494232757490498e-05,
+      "loss": 1.0781,
+      "step": 2863
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2489407993287084e-05,
+      "loss": 0.906,
+      "step": 2864
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.24845826112268e-05,
+      "loss": 0.9921,
+      "step": 2865
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.247975661250727e-05,
+      "loss": 0.9426,
+      "step": 2866
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2474929998326283e-05,
+      "loss": 0.9906,
+      "step": 2867
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2470102769881778e-05,
+      "loss": 0.9628,
+      "step": 2868
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2465274928371849e-05,
+      "loss": 0.8764,
+      "step": 2869
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2460446474994728e-05,
+      "loss": 1.0113,
+      "step": 2870
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2455617410948818e-05,
+      "loss": 0.9746,
+      "step": 2871
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2450787737432663e-05,
+      "loss": 0.9705,
+      "step": 2872
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2445957455644963e-05,
+      "loss": 1.0283,
+      "step": 2873
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2441126566784561e-05,
+      "loss": 1.0713,
+      "step": 2874
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2436295072050463e-05,
+      "loss": 0.9444,
+      "step": 2875
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2431462972641812e-05,
+      "loss": 1.0315,
+      "step": 2876
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2426630269757911e-05,
+      "loss": 1.004,
+      "step": 2877
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2421796964598208e-05,
+      "loss": 1.0291,
+      "step": 2878
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2416963058362303e-05,
+      "loss": 0.9893,
+      "step": 2879
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.241212855224994e-05,
+      "loss": 1.0183,
+      "step": 2880
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2407293447461024e-05,
+      "loss": 1.0158,
+      "step": 2881
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2402457745195592e-05,
+      "loss": 1.0379,
+      "step": 2882
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2397621446653843e-05,
+      "loss": 0.9607,
+      "step": 2883
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2392784553036116e-05,
+      "loss": 0.9155,
+      "step": 2884
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2387947065542902e-05,
+      "loss": 0.9955,
+      "step": 2885
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2383108985374836e-05,
+      "loss": 1.0332,
+      "step": 2886
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2378270313732707e-05,
+      "loss": 0.8325,
+      "step": 2887
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.237343105181744e-05,
+      "loss": 0.9868,
+      "step": 2888
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2368591200830113e-05,
+      "loss": 0.8954,
+      "step": 2889
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.236375076197195e-05,
+      "loss": 0.9723,
+      "step": 2890
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2358909736444325e-05,
+      "loss": 1.1171,
+      "step": 2891
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2354068125448744e-05,
+      "loss": 1.0838,
+      "step": 2892
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2349225930186875e-05,
+      "loss": 0.9925,
+      "step": 2893
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2344383151860517e-05,
+      "loss": 1.0317,
+      "step": 2894
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2339539791671627e-05,
+      "loss": 0.9007,
+      "step": 2895
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2334695850822293e-05,
+      "loss": 0.8932,
+      "step": 2896
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.232985133051476e-05,
+      "loss": 0.8741,
+      "step": 2897
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2325006231951402e-05,
+      "loss": 0.9464,
+      "step": 2898
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2320160556334755e-05,
+      "loss": 0.9475,
+      "step": 2899
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2315314304867477e-05,
+      "loss": 1.0823,
+      "step": 2900
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2310467478752388e-05,
+      "loss": 1.104,
+      "step": 2901
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.230562007919244e-05,
+      "loss": 1.0155,
+      "step": 2902
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2300772107390731e-05,
+      "loss": 1.0142,
+      "step": 2903
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2295923564550497e-05,
+      "loss": 0.9652,
+      "step": 2904
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2291074451875122e-05,
+      "loss": 0.9839,
+      "step": 2905
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2286224770568125e-05,
+      "loss": 0.9612,
+      "step": 2906
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2281374521833173e-05,
+      "loss": 1.0401,
+      "step": 2907
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2276523706874068e-05,
+      "loss": 1.0332,
+      "step": 2908
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2271672326894754e-05,
+      "loss": 0.9891,
+      "step": 2909
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2266820383099316e-05,
+      "loss": 0.8968,
+      "step": 2910
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2261967876691982e-05,
+      "loss": 0.8983,
+      "step": 2911
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2257114808877111e-05,
+      "loss": 0.9078,
+      "step": 2912
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2252261180859213e-05,
+      "loss": 1.0145,
+      "step": 2913
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2247406993842922e-05,
+      "loss": 0.9395,
+      "step": 2914
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2242552249033034e-05,
+      "loss": 0.9039,
+      "step": 2915
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2237696947634453e-05,
+      "loss": 0.9366,
+      "step": 2916
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2232841090852244e-05,
+      "loss": 1.0233,
+      "step": 2917
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2227984679891605e-05,
+      "loss": 0.9483,
+      "step": 2918
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2223127715957867e-05,
+      "loss": 0.9155,
+      "step": 2919
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2218270200256501e-05,
+      "loss": 1.0104,
+      "step": 2920
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.221341213399312e-05,
+      "loss": 0.8296,
+      "step": 2921
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2208553518373456e-05,
+      "loss": 1.0938,
+      "step": 2922
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2203694354603401e-05,
+      "loss": 0.875,
+      "step": 2923
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2198834643888968e-05,
+      "loss": 1.1426,
+      "step": 2924
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.219397438743631e-05,
+      "loss": 1.0759,
+      "step": 2925
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2189113586451712e-05,
+      "loss": 0.8136,
+      "step": 2926
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2184252242141603e-05,
+      "loss": 0.9589,
+      "step": 2927
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2179390355712533e-05,
+      "loss": 1.1103,
+      "step": 2928
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2174527928371203e-05,
+      "loss": 0.8993,
+      "step": 2929
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2169664961324432e-05,
+      "loss": 1.0117,
+      "step": 2930
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2164801455779186e-05,
+      "loss": 0.9157,
+      "step": 2931
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2159937412942558e-05,
+      "loss": 0.9928,
+      "step": 2932
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2155072834021777e-05,
+      "loss": 1.0047,
+      "step": 2933
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2150207720224199e-05,
+      "loss": 0.8367,
+      "step": 2934
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2145342072757322e-05,
+      "loss": 0.9322,
+      "step": 2935
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2140475892828771e-05,
+      "loss": 1.0062,
+      "step": 2936
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2135609181646304e-05,
+      "loss": 1.0257,
+      "step": 2937
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2130741940417809e-05,
+      "loss": 1.0199,
+      "step": 2938
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2125874170351313e-05,
+      "loss": 0.9767,
+      "step": 2939
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2121005872654962e-05,
+      "loss": 1.0044,
+      "step": 2940
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2116137048537045e-05,
+      "loss": 1.1156,
+      "step": 2941
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2111267699205972e-05,
+      "loss": 0.9483,
+      "step": 2942
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2106397825870292e-05,
+      "loss": 0.9793,
+      "step": 2943
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2101527429738672e-05,
+      "loss": 0.9505,
+      "step": 2944
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2096656512019931e-05,
+      "loss": 1.0622,
+      "step": 2945
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2091785073922988e-05,
+      "loss": 0.9501,
+      "step": 2946
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2086913116656915e-05,
+      "loss": 0.9505,
+      "step": 2947
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2082040641430903e-05,
+      "loss": 0.93,
+      "step": 2948
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2077167649454269e-05,
+      "loss": 0.9417,
+      "step": 2949
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2072294141936462e-05,
+      "loss": 1.1213,
+      "step": 2950
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2067420120087068e-05,
+      "loss": 0.9726,
+      "step": 2951
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2062545585115777e-05,
+      "loss": 0.9766,
+      "step": 2952
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.205767053823243e-05,
+      "loss": 0.9004,
+      "step": 2953
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2052794980646985e-05,
+      "loss": 0.9542,
+      "step": 2954
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2047918913569524e-05,
+      "loss": 0.9486,
+      "step": 2955
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2043042338210262e-05,
+      "loss": 1.0275,
+      "step": 2956
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2038165255779542e-05,
+      "loss": 1.0995,
+      "step": 2957
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2033287667487815e-05,
+      "loss": 1.0774,
+      "step": 2958
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.2028409574545683e-05,
+      "loss": 0.9947,
+      "step": 2959
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.2023530978163856e-05,
+      "loss": 0.9607,
+      "step": 2960
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.2018651879553173e-05,
+      "loss": 0.8789,
+      "step": 2961
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.2013772279924597e-05,
+      "loss": 1.0249,
+      "step": 2962
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.2008892180489224e-05,
+      "loss": 1.0084,
+      "step": 2963
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.2004011582458261e-05,
+      "loss": 0.9565,
+      "step": 2964
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1999130487043046e-05,
+      "loss": 0.9627,
+      "step": 2965
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1994248895455037e-05,
+      "loss": 0.9443,
+      "step": 2966
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.198936680890582e-05,
+      "loss": 0.9006,
+      "step": 2967
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.19844842286071e-05,
+      "loss": 1.1238,
+      "step": 2968
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1979601155770706e-05,
+      "loss": 1.0797,
+      "step": 2969
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1974717591608589e-05,
+      "loss": 0.9816,
+      "step": 2970
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.196983353733282e-05,
+      "loss": 0.977,
+      "step": 2971
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1964948994155597e-05,
+      "loss": 0.9056,
+      "step": 2972
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1960063963289232e-05,
+      "loss": 0.9808,
+      "step": 2973
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.195517844594616e-05,
+      "loss": 1.0359,
+      "step": 2974
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1950292443338947e-05,
+      "loss": 1.0034,
+      "step": 2975
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1945405956680266e-05,
+      "loss": 0.9772,
+      "step": 2976
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1940518987182911e-05,
+      "loss": 1.0332,
+      "step": 2977
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1935631536059808e-05,
+      "loss": 0.995,
+      "step": 2978
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1930743604523989e-05,
+      "loss": 1.1052,
+      "step": 2979
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1925855193788611e-05,
+      "loss": 0.9999,
+      "step": 2980
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1920966305066953e-05,
+      "loss": 0.9464,
+      "step": 2981
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1916076939572407e-05,
+      "loss": 1.0055,
+      "step": 2982
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1911187098518487e-05,
+      "loss": 0.9019,
+      "step": 2983
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1906296783118818e-05,
+      "loss": 0.9246,
+      "step": 2984
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.190140599458716e-05,
+      "loss": 1.0942,
+      "step": 2985
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1896514734137366e-05,
+      "loss": 1.0634,
+      "step": 2986
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1891623002983427e-05,
+      "loss": 0.9371,
+      "step": 2987
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1886730802339441e-05,
+      "loss": 1.0069,
+      "step": 2988
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1881838133419625e-05,
+      "loss": 0.9903,
+      "step": 2989
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1876944997438306e-05,
+      "loss": 0.9654,
+      "step": 2990
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.187205139560994e-05,
+      "loss": 0.8685,
+      "step": 2991
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.186715732914909e-05,
+      "loss": 0.9669,
+      "step": 2992
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1862262799270432e-05,
+      "loss": 0.8659,
+      "step": 2993
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1857367807188762e-05,
+      "loss": 0.9342,
+      "step": 2994
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1852472354118987e-05,
+      "loss": 0.9876,
+      "step": 2995
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1847576441276131e-05,
+      "loss": 0.9637,
+      "step": 2996
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1842680069875339e-05,
+      "loss": 0.9886,
+      "step": 2997
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1837783241131853e-05,
+      "loss": 0.9053,
+      "step": 2998
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1832885956261041e-05,
+      "loss": 0.9998,
+      "step": 2999
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1827988216478385e-05,
+      "loss": 0.8669,
+      "step": 3000
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1823090022999473e-05,
+      "loss": 0.7194,
+      "step": 3001
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1818191377040005e-05,
+      "loss": 1.182,
+      "step": 3002
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1813292279815802e-05,
+      "loss": 1.0905,
+      "step": 3003
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.180839273254279e-05,
+      "loss": 0.8948,
+      "step": 3004
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1803492736437012e-05,
+      "loss": 0.9536,
+      "step": 3005
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1798592292714614e-05,
+      "loss": 1.0106,
+      "step": 3006
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1793691402591862e-05,
+      "loss": 1.032,
+      "step": 3007
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1788790067285125e-05,
+      "loss": 0.8928,
+      "step": 3008
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1783888288010892e-05,
+      "loss": 0.9532,
+      "step": 3009
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1778986065985754e-05,
+      "loss": 1.0151,
+      "step": 3010
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1774083402426415e-05,
+      "loss": 0.7592,
+      "step": 3011
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1769180298549686e-05,
+      "loss": 0.9448,
+      "step": 3012
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.176427675557249e-05,
+      "loss": 0.9391,
+      "step": 3013
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.175937277471186e-05,
+      "loss": 0.8152,
+      "step": 3014
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1754468357184935e-05,
+      "loss": 0.9791,
+      "step": 3015
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1749563504208965e-05,
+      "loss": 0.9692,
+      "step": 3016
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1744658217001306e-05,
+      "loss": 1.218,
+      "step": 3017
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1739752496779422e-05,
+      "loss": 1.0126,
+      "step": 3018
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1734846344760882e-05,
+      "loss": 0.8845,
+      "step": 3019
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1729939762163369e-05,
+      "loss": 0.9043,
+      "step": 3020
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1725032750204669e-05,
+      "loss": 0.9851,
+      "step": 3021
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1720125310102672e-05,
+      "loss": 0.8462,
+      "step": 3022
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1715217443075377e-05,
+      "loss": 1.0234,
+      "step": 3023
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.171030915034089e-05,
+      "loss": 0.8693,
+      "step": 3024
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1705400433117417e-05,
+      "loss": 0.9947,
+      "step": 3025
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1700491292623281e-05,
+      "loss": 0.9228,
+      "step": 3026
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1695581730076898e-05,
+      "loss": 1.1267,
+      "step": 3027
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1690671746696794e-05,
+      "loss": 0.8649,
+      "step": 3028
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1685761343701598e-05,
+      "loss": 1.0157,
+      "step": 3029
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1680850522310045e-05,
+      "loss": 1.0056,
+      "step": 3030
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1675939283740973e-05,
+      "loss": 0.9295,
+      "step": 3031
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1671027629213325e-05,
+      "loss": 0.9042,
+      "step": 3032
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1666115559946144e-05,
+      "loss": 0.9914,
+      "step": 3033
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1661203077158577e-05,
+      "loss": 0.9737,
+      "step": 3034
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1656290182069874e-05,
+      "loss": 1.0505,
+      "step": 3035
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1651376875899389e-05,
+      "loss": 0.9126,
+      "step": 3036
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1646463159866574e-05,
+      "loss": 1.1126,
+      "step": 3037
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1641549035190991e-05,
+      "loss": 0.8926,
+      "step": 3038
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1636634503092293e-05,
+      "loss": 0.8873,
+      "step": 3039
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1631719564790239e-05,
+      "loss": 1.0196,
+      "step": 3040
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1626804221504689e-05,
+      "loss": 0.8795,
+      "step": 3041
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1621888474455605e-05,
+      "loss": 1.028,
+      "step": 3042
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1616972324863045e-05,
+      "loss": 0.9771,
+      "step": 3043
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.161205577394717e-05,
+      "loss": 1.0881,
+      "step": 3044
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1607138822928243e-05,
+      "loss": 0.9328,
+      "step": 3045
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.160222147302662e-05,
+      "loss": 0.9416,
+      "step": 3046
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1597303725462758e-05,
+      "loss": 1.0511,
+      "step": 3047
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1592385581457217e-05,
+      "loss": 1.0366,
+      "step": 3048
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.158746704223065e-05,
+      "loss": 0.9864,
+      "step": 3049
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1582548109003812e-05,
+      "loss": 1.0379,
+      "step": 3050
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1577628782997554e-05,
+      "loss": 1.0067,
+      "step": 3051
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1572709065432826e-05,
+      "loss": 0.8887,
+      "step": 3052
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1567788957530667e-05,
+      "loss": 1.0742,
+      "step": 3053
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1562868460512228e-05,
+      "loss": 1.0906,
+      "step": 3054
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1557947575598747e-05,
+      "loss": 0.951,
+      "step": 3055
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1553026304011557e-05,
+      "loss": 1.0909,
+      "step": 3056
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1548104646972091e-05,
+      "loss": 0.9578,
+      "step": 3057
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1543182605701875e-05,
+      "loss": 0.9311,
+      "step": 3058
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.153826018142253e-05,
+      "loss": 1.0464,
+      "step": 3059
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1533337375355776e-05,
+      "loss": 1.0327,
+      "step": 3060
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1528414188723426e-05,
+      "loss": 0.9048,
+      "step": 3061
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1523490622747388e-05,
+      "loss": 0.9952,
+      "step": 3062
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.151856667864966e-05,
+      "loss": 1.0114,
+      "step": 3063
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1513642357652338e-05,
+      "loss": 0.7669,
+      "step": 3064
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1508717660977609e-05,
+      "loss": 0.8715,
+      "step": 3065
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1503792589847757e-05,
+      "loss": 0.9697,
+      "step": 3066
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1498867145485155e-05,
+      "loss": 0.9022,
+      "step": 3067
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1493941329112272e-05,
+      "loss": 0.803,
+      "step": 3068
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1489015141951667e-05,
+      "loss": 0.921,
+      "step": 3069
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.148408858522599e-05,
+      "loss": 0.8843,
+      "step": 3070
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1479161660157982e-05,
+      "loss": 1.1231,
+      "step": 3071
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1474234367970485e-05,
+      "loss": 0.8998,
+      "step": 3072
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1469306709886421e-05,
+      "loss": 1.0333,
+      "step": 3073
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1464378687128805e-05,
+      "loss": 0.9408,
+      "step": 3074
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1459450300920748e-05,
+      "loss": 0.9888,
+      "step": 3075
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.145452155248544e-05,
+      "loss": 0.9957,
+      "step": 3076
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1449592443046176e-05,
+      "loss": 0.9252,
+      "step": 3077
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.144466297382633e-05,
+      "loss": 0.9264,
+      "step": 3078
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1439733146049368e-05,
+      "loss": 1.0067,
+      "step": 3079
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1434802960938846e-05,
+      "loss": 1.0007,
+      "step": 3080
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1429872419718406e-05,
+      "loss": 1.0486,
+      "step": 3081
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1424941523611785e-05,
+      "loss": 0.9707,
+      "step": 3082
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1420010273842794e-05,
+      "loss": 0.9078,
+      "step": 3083
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1415078671635351e-05,
+      "loss": 1.0462,
+      "step": 3084
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1410146718213448e-05,
+      "loss": 0.917,
+      "step": 3085
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1405214414801167e-05,
+      "loss": 0.9143,
+      "step": 3086
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1400281762622673e-05,
+      "loss": 1.0224,
+      "step": 3087
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1395348762902234e-05,
+      "loss": 0.8888,
+      "step": 3088
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.139041541686418e-05,
+      "loss": 0.9526,
+      "step": 3089
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1385481725732945e-05,
+      "loss": 0.9729,
+      "step": 3090
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1380547690733047e-05,
+      "loss": 1.1105,
+      "step": 3091
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1375613313089079e-05,
+      "loss": 1.0582,
+      "step": 3092
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1370678594025723e-05,
+      "loss": 0.8121,
+      "step": 3093
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.136574353476776e-05,
+      "loss": 1.0114,
+      "step": 3094
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1360808136540029e-05,
+      "loss": 0.9884,
+      "step": 3095
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.135587240056748e-05,
+      "loss": 0.9629,
+      "step": 3096
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1350936328075129e-05,
+      "loss": 1.151,
+      "step": 3097
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.134599992028808e-05,
+      "loss": 0.9749,
+      "step": 3098
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1341063178431522e-05,
+      "loss": 0.9081,
+      "step": 3099
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1336126103730729e-05,
+      "loss": 1.0112,
+      "step": 3100
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1331188697411052e-05,
+      "loss": 0.9723,
+      "step": 3101
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1326250960697927e-05,
+      "loss": 0.9263,
+      "step": 3102
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1321312894816873e-05,
+      "loss": 0.9722,
+      "step": 3103
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1316374500993493e-05,
+      "loss": 0.9984,
+      "step": 3104
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1311435780453461e-05,
+      "loss": 0.9924,
+      "step": 3105
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1306496734422548e-05,
+      "loss": 0.9717,
+      "step": 3106
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1301557364126587e-05,
+      "loss": 1.0213,
+      "step": 3107
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.129661767079151e-05,
+      "loss": 1.047,
+      "step": 3108
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1291677655643316e-05,
+      "loss": 0.8792,
+      "step": 3109
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1286737319908092e-05,
+      "loss": 0.9315,
+      "step": 3110
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1281796664811996e-05,
+      "loss": 1.1819,
+      "step": 3111
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1276855691581279e-05,
+      "loss": 0.9657,
+      "step": 3112
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1271914401442252e-05,
+      "loss": 1.0519,
+      "step": 3113
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1266972795621323e-05,
+      "loss": 0.9651,
+      "step": 3114
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.126203087534497e-05,
+      "loss": 1.0167,
+      "step": 3115
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1257088641839744e-05,
+      "loss": 1.0372,
+      "step": 3116
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1252146096332282e-05,
+      "loss": 0.9833,
+      "step": 3117
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1247203240049298e-05,
+      "loss": 0.9389,
+      "step": 3118
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1242260074217577e-05,
+      "loss": 0.9276,
+      "step": 3119
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1237316600063985e-05,
+      "loss": 0.8803,
+      "step": 3120
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1232372818815469e-05,
+      "loss": 0.9541,
+      "step": 3121
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1227428731699042e-05,
+      "loss": 0.9923,
+      "step": 3122
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1222484339941799e-05,
+      "loss": 0.9195,
+      "step": 3123
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1217539644770911e-05,
+      "loss": 0.984,
+      "step": 3124
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1212594647413625e-05,
+      "loss": 0.9298,
+      "step": 3125
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1207649349097258e-05,
+      "loss": 0.9592,
+      "step": 3126
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1202703751049205e-05,
+      "loss": 1.0833,
+      "step": 3127
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1197757854496938e-05,
+      "loss": 0.8558,
+      "step": 3128
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1192811660667997e-05,
+      "loss": 1.0563,
+      "step": 3129
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.118786517079e-05,
+      "loss": 0.929,
+      "step": 3130
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.118291838609064e-05,
+      "loss": 0.7851,
+      "step": 3131
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.117797130779768e-05,
+      "loss": 0.8942,
+      "step": 3132
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1173023937138951e-05,
+      "loss": 1.0833,
+      "step": 3133
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1168076275342374e-05,
+      "loss": 0.9888,
+      "step": 3134
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1163128323635917e-05,
+      "loss": 0.9028,
+      "step": 3135
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1158180083247643e-05,
+      "loss": 0.9575,
+      "step": 3136
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1153231555405673e-05,
+      "loss": 0.8575,
+      "step": 3137
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1148282741338203e-05,
+      "loss": 0.9495,
+      "step": 3138
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1143333642273502e-05,
+      "loss": 0.9935,
+      "step": 3139
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1138384259439911e-05,
+      "loss": 0.9202,
+      "step": 3140
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1133434594065831e-05,
+      "loss": 0.952,
+      "step": 3141
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.112848464737975e-05,
+      "loss": 1.0275,
+      "step": 3142
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1123534420610206e-05,
+      "loss": 0.9165,
+      "step": 3143
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1118583914985827e-05,
+      "loss": 0.9704,
+      "step": 3144
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1113633131735294e-05,
+      "loss": 0.9499,
+      "step": 3145
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1108682072087369e-05,
+      "loss": 1.0807,
+      "step": 3146
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1103730737270868e-05,
+      "loss": 0.9839,
+      "step": 3147
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1098779128514692e-05,
+      "loss": 0.9252,
+      "step": 3148
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1093827247047797e-05,
+      "loss": 1.0274,
+      "step": 3149
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1088875094099216e-05,
+      "loss": 0.9961,
+      "step": 3150
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1083922670898039e-05,
+      "loss": 1.0327,
+      "step": 3151
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.107896997867344e-05,
+      "loss": 0.8911,
+      "step": 3152
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1074017018654638e-05,
+      "loss": 0.9704,
+      "step": 3153
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.1069063792070935e-05,
+      "loss": 0.9713,
+      "step": 3154
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.1064110300151692e-05,
+      "loss": 1.0005,
+      "step": 3155
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.1059156544126337e-05,
+      "loss": 0.9159,
+      "step": 3156
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.1054202525224365e-05,
+      "loss": 0.9243,
+      "step": 3157
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.1049248244675337e-05,
+      "loss": 1.0251,
+      "step": 3158
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.1044293703708873e-05,
+      "loss": 0.9639,
+      "step": 3159
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.1039338903554663e-05,
+      "loss": 0.9708,
+      "step": 3160
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.1034383845442464e-05,
+      "loss": 0.9507,
+      "step": 3161
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.102942853060209e-05,
+      "loss": 0.9486,
+      "step": 3162
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.102447296026342e-05,
+      "loss": 0.8778,
+      "step": 3163
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.1019517135656406e-05,
+      "loss": 1.056,
+      "step": 3164
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.1014561058011045e-05,
+      "loss": 0.8754,
+      "step": 3165
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.1009604728557412e-05,
+      "loss": 0.9178,
+      "step": 3166
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.1004648148525643e-05,
+      "loss": 0.8578,
+      "step": 3167
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0999691319145929e-05,
+      "loss": 0.9137,
+      "step": 3168
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0994734241648523e-05,
+      "loss": 0.8765,
+      "step": 3169
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0989776917263756e-05,
+      "loss": 0.9497,
+      "step": 3170
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0984819347221994e-05,
+      "loss": 0.9765,
+      "step": 3171
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0979861532753686e-05,
+      "loss": 1.0378,
+      "step": 3172
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0974903475089329e-05,
+      "loss": 0.9801,
+      "step": 3173
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0969945175459488e-05,
+      "loss": 0.9078,
+      "step": 3174
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0964986635094781e-05,
+      "loss": 1.0353,
+      "step": 3175
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0960027855225894e-05,
+      "loss": 1.1471,
+      "step": 3176
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0955068837083563e-05,
+      "loss": 0.8657,
+      "step": 3177
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0950109581898592e-05,
+      "loss": 0.9693,
+      "step": 3178
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0945150090901838e-05,
+      "loss": 1.0189,
+      "step": 3179
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.094019036532422e-05,
+      "loss": 1.0252,
+      "step": 3180
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0935230406396707e-05,
+      "loss": 0.9072,
+      "step": 3181
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0930270215350345e-05,
+      "loss": 0.9827,
+      "step": 3182
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0925309793416212e-05,
+      "loss": 0.9751,
+      "step": 3183
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0920349141825464e-05,
+      "loss": 0.9458,
+      "step": 3184
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0915388261809305e-05,
+      "loss": 0.9214,
+      "step": 3185
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0910427154598999e-05,
+      "loss": 0.8485,
+      "step": 3186
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.090546582142586e-05,
+      "loss": 0.8898,
+      "step": 3187
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0900504263521268e-05,
+      "loss": 0.9781,
+      "step": 3188
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0895542482116648e-05,
+      "loss": 0.9318,
+      "step": 3189
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.089058047844349e-05,
+      "loss": 0.9251,
+      "step": 3190
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0885618253733333e-05,
+      "loss": 0.9075,
+      "step": 3191
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0880655809217772e-05,
+      "loss": 0.9805,
+      "step": 3192
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0875693146128456e-05,
+      "loss": 0.7785,
+      "step": 3193
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0870730265697098e-05,
+      "loss": 0.985,
+      "step": 3194
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.086576716915545e-05,
+      "loss": 1.0093,
+      "step": 3195
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0860803857735325e-05,
+      "loss": 0.8457,
+      "step": 3196
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0855840332668586e-05,
+      "loss": 0.9841,
+      "step": 3197
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.085087659518716e-05,
+      "loss": 1.0094,
+      "step": 3198
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0845912646523007e-05,
+      "loss": 0.9458,
+      "step": 3199
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.084094848790816e-05,
+      "loss": 0.8948,
+      "step": 3200
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0835984120574692e-05,
+      "loss": 1.0269,
+      "step": 3201
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0831019545754731e-05,
+      "loss": 1.0075,
+      "step": 3202
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0826054764680452e-05,
+      "loss": 0.8424,
+      "step": 3203
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0821089778584093e-05,
+      "loss": 0.9473,
+      "step": 3204
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0816124588697928e-05,
+      "loss": 0.9689,
+      "step": 3205
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0811159196254293e-05,
+      "loss": 1.1136,
+      "step": 3206
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0806193602485568e-05,
+      "loss": 0.9717,
+      "step": 3207
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0801227808624189e-05,
+      "loss": 0.9166,
+      "step": 3208
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0796261815902628e-05,
+      "loss": 1.0523,
+      "step": 3209
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0791295625553431e-05,
+      "loss": 0.9876,
+      "step": 3210
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0786329238809165e-05,
+      "loss": 1.0707,
+      "step": 3211
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0781362656902466e-05,
+      "loss": 0.8721,
+      "step": 3212
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0776395881066008e-05,
+      "loss": 1.0243,
+      "step": 3213
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.077142891253252e-05,
+      "loss": 0.9043,
+      "step": 3214
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.076646175253477e-05,
+      "loss": 0.8644,
+      "step": 3215
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0761494402305583e-05,
+      "loss": 0.9131,
+      "step": 3216
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0756526863077825e-05,
+      "loss": 0.9244,
+      "step": 3217
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0751559136084413e-05,
+      "loss": 0.8209,
+      "step": 3218
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0746591222558305e-05,
+      "loss": 0.9763,
+      "step": 3219
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0741623123732515e-05,
+      "loss": 0.9782,
+      "step": 3220
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0736654840840085e-05,
+      "loss": 0.9585,
+      "step": 3221
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0731686375114132e-05,
+      "loss": 0.9732,
+      "step": 3222
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0726717727787786e-05,
+      "loss": 0.8958,
+      "step": 3223
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0721748900094243e-05,
+      "loss": 1.0476,
+      "step": 3224
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0716779893266738e-05,
+      "loss": 0.817,
+      "step": 3225
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0711810708538548e-05,
+      "loss": 0.9669,
+      "step": 3226
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0706841347142995e-05,
+      "loss": 1.0107,
+      "step": 3227
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0701871810313453e-05,
+      "loss": 1.0457,
+      "step": 3228
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0696902099283323e-05,
+      "loss": 1.1749,
+      "step": 3229
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0691932215286069e-05,
+      "loss": 0.9579,
+      "step": 3230
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.068696215955518e-05,
+      "loss": 1.0379,
+      "step": 3231
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0681991933324203e-05,
+      "loss": 0.8846,
+      "step": 3232
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0677021537826708e-05,
+      "loss": 0.9456,
+      "step": 3233
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0672050974296335e-05,
+      "loss": 0.9282,
+      "step": 3234
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0667080243966734e-05,
+      "loss": 0.863,
+      "step": 3235
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.066210934807162e-05,
+      "loss": 1.0751,
+      "step": 3236
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0657138287844742e-05,
+      "loss": 1.1047,
+      "step": 3237
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0652167064519884e-05,
+      "loss": 0.9774,
+      "step": 3238
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.064719567933088e-05,
+      "loss": 0.9674,
+      "step": 3239
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0642224133511598e-05,
+      "loss": 0.9769,
+      "step": 3240
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0637252428295943e-05,
+      "loss": 0.9458,
+      "step": 3241
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.063228056491787e-05,
+      "loss": 0.974,
+      "step": 3242
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0627308544611366e-05,
+      "loss": 0.8862,
+      "step": 3243
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0622336368610455e-05,
+      "loss": 1.0281,
+      "step": 3244
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0617364038149203e-05,
+      "loss": 0.9256,
+      "step": 3245
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0612391554461718e-05,
+      "loss": 0.9588,
+      "step": 3246
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0607418918782137e-05,
+      "loss": 0.9143,
+      "step": 3247
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0602446132344645e-05,
+      "loss": 1.082,
+      "step": 3248
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0597473196383458e-05,
+      "loss": 0.9452,
+      "step": 3249
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.059250011213283e-05,
+      "loss": 0.9817,
+      "step": 3250
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0587526880827044e-05,
+      "loss": 0.9587,
+      "step": 3251
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0582553503700443e-05,
+      "loss": 1.0189,
+      "step": 3252
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0577579981987377e-05,
+      "loss": 0.9387,
+      "step": 3253
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0572606316922255e-05,
+      "loss": 0.8686,
+      "step": 3254
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.056763250973951e-05,
+      "loss": 0.9261,
+      "step": 3255
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0562658561673608e-05,
+      "loss": 0.9728,
+      "step": 3256
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.055768447395906e-05,
+      "loss": 0.9344,
+      "step": 3257
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0552710247830406e-05,
+      "loss": 1.0133,
+      "step": 3258
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0547735884522214e-05,
+      "loss": 0.9563,
+      "step": 3259
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0542761385269102e-05,
+      "loss": 0.9869,
+      "step": 3260
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0537786751305706e-05,
+      "loss": 0.9612,
+      "step": 3261
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0532811983866704e-05,
+      "loss": 1.1041,
+      "step": 3262
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0527837084186803e-05,
+      "loss": 0.9437,
+      "step": 3263
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0522862053500749e-05,
+      "loss": 1.0451,
+      "step": 3264
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0517886893043313e-05,
+      "loss": 1.0632,
+      "step": 3265
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.05129116040493e-05,
+      "loss": 1.0451,
+      "step": 3266
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0507936187753548e-05,
+      "loss": 1.0783,
+      "step": 3267
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0502960645390938e-05,
+      "loss": 0.945,
+      "step": 3268
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0497984978196356e-05,
+      "loss": 0.9578,
+      "step": 3269
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0493009187404741e-05,
+      "loss": 0.9368,
+      "step": 3270
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.048803327425106e-05,
+      "loss": 1.0574,
+      "step": 3271
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0483057239970297e-05,
+      "loss": 0.9058,
+      "step": 3272
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0478081085797482e-05,
+      "loss": 1.0746,
+      "step": 3273
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.047310481296767e-05,
+      "loss": 1.0834,
+      "step": 3274
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0468128422715935e-05,
+      "loss": 1.001,
+      "step": 3275
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0463151916277397e-05,
+      "loss": 1.0997,
+      "step": 3276
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0458175294887191e-05,
+      "loss": 1.0722,
+      "step": 3277
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0453198559780491e-05,
+      "loss": 0.9981,
+      "step": 3278
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0448221712192488e-05,
+      "loss": 1.1555,
+      "step": 3279
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0443244753358417e-05,
+      "loss": 0.8944,
+      "step": 3280
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.043826768451352e-05,
+      "loss": 0.9894,
+      "step": 3281
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.043329050689308e-05,
+      "loss": 0.9269,
+      "step": 3282
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.042831322173241e-05,
+      "loss": 1.0814,
+      "step": 3283
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0423335830266838e-05,
+      "loss": 1.0168,
+      "step": 3284
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0418358333731723e-05,
+      "loss": 0.8129,
+      "step": 3285
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0413380733362462e-05,
+      "loss": 0.9531,
+      "step": 3286
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0408403030394451e-05,
+      "loss": 0.9923,
+      "step": 3287
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.040342522606314e-05,
+      "loss": 0.9119,
+      "step": 3288
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0398447321603986e-05,
+      "loss": 0.8234,
+      "step": 3289
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0393469318252479e-05,
+      "loss": 0.8906,
+      "step": 3290
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0388491217244128e-05,
+      "loss": 0.8661,
+      "step": 3291
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0383513019814473e-05,
+      "loss": 0.9869,
+      "step": 3292
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.037853472719907e-05,
+      "loss": 0.9425,
+      "step": 3293
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0373556340633508e-05,
+      "loss": 1.0072,
+      "step": 3294
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.036857786135339e-05,
+      "loss": 1.0541,
+      "step": 3295
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.036359929059435e-05,
+      "loss": 1.0392,
+      "step": 3296
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0358620629592035e-05,
+      "loss": 0.944,
+      "step": 3297
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0353641879582129e-05,
+      "loss": 0.9381,
+      "step": 3298
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0348663041800317e-05,
+      "loss": 0.9139,
+      "step": 3299
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0343684117482332e-05,
+      "loss": 0.9685,
+      "step": 3300
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0338705107863905e-05,
+      "loss": 0.8782,
+      "step": 3301
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.03337260141808e-05,
+      "loss": 0.8843,
+      "step": 3302
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0328746837668799e-05,
+      "loss": 0.7584,
+      "step": 3303
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0323767579563713e-05,
+      "loss": 1.0256,
+      "step": 3304
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0318788241101351e-05,
+      "loss": 0.9311,
+      "step": 3305
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.031380882351757e-05,
+      "loss": 0.9382,
+      "step": 3306
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0308829328048227e-05,
+      "loss": 0.9477,
+      "step": 3307
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0303849755929206e-05,
+      "loss": 0.923,
+      "step": 3308
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0298870108396404e-05,
+      "loss": 0.9086,
+      "step": 3309
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.029389038668575e-05,
+      "loss": 0.9012,
+      "step": 3310
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.028891059203317e-05,
+      "loss": 1.0124,
+      "step": 3311
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0283930725674634e-05,
+      "loss": 1.0783,
+      "step": 3312
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0278950788846106e-05,
+      "loss": 1.0674,
+      "step": 3313
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0273970782783584e-05,
+      "loss": 0.9749,
+      "step": 3314
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0268990708723074e-05,
+      "loss": 1.0077,
+      "step": 3315
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0264010567900606e-05,
+      "loss": 0.9421,
+      "step": 3316
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0259030361552219e-05,
+      "loss": 1.1092,
+      "step": 3317
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0254050090913971e-05,
+      "loss": 0.9148,
+      "step": 3318
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.024906975722194e-05,
+      "loss": 1.0442,
+      "step": 3319
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0244089361712217e-05,
+      "loss": 0.9386,
+      "step": 3320
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.02391089056209e-05,
+      "loss": 0.9389,
+      "step": 3321
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0234128390184119e-05,
+      "loss": 0.94,
+      "step": 3322
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0229147816638007e-05,
+      "loss": 1.0177,
+      "step": 3323
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0224167186218712e-05,
+      "loss": 0.93,
+      "step": 3324
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0219186500162398e-05,
+      "loss": 1.0732,
+      "step": 3325
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0214205759705246e-05,
+      "loss": 1.051,
+      "step": 3326
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0209224966083444e-05,
+      "loss": 1.1628,
+      "step": 3327
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0204244120533198e-05,
+      "loss": 1.0283,
+      "step": 3328
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0199263224290727e-05,
+      "loss": 0.9572,
+      "step": 3329
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0194282278592258e-05,
+      "loss": 0.9109,
+      "step": 3330
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0189301284674035e-05,
+      "loss": 1.0784,
+      "step": 3331
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0184320243772314e-05,
+      "loss": 0.9909,
+      "step": 3332
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0179339157123356e-05,
+      "loss": 1.042,
+      "step": 3333
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0174358025963442e-05,
+      "loss": 1.0013,
+      "step": 3334
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.016937685152886e-05,
+      "loss": 0.9732,
+      "step": 3335
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.016439563505591e-05,
+      "loss": 1.1034,
+      "step": 3336
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0159414377780899e-05,
+      "loss": 1.0538,
+      "step": 3337
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0154433080940152e-05,
+      "loss": 0.9235,
+      "step": 3338
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.014945174576999e-05,
+      "loss": 1.0066,
+      "step": 3339
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0144470373506758e-05,
+      "loss": 1.0195,
+      "step": 3340
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0139488965386803e-05,
+      "loss": 1.0495,
+      "step": 3341
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0134507522646481e-05,
+      "loss": 0.9043,
+      "step": 3342
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0129526046522158e-05,
+      "loss": 0.8218,
+      "step": 3343
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.012454453825021e-05,
+      "loss": 0.8155,
+      "step": 3344
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0119562999067018e-05,
+      "loss": 0.8673,
+      "step": 3345
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.011458143020897e-05,
+      "loss": 0.9293,
+      "step": 3346
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0109599832912464e-05,
+      "loss": 0.8812,
+      "step": 3347
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0104618208413906e-05,
+      "loss": 1.001,
+      "step": 3348
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0099636557949702e-05,
+      "loss": 0.9652,
+      "step": 3349
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0094654882756275e-05,
+      "loss": 0.9159,
+      "step": 3350
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0089673184070043e-05,
+      "loss": 0.8238,
+      "step": 3351
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0084691463127441e-05,
+      "loss": 0.9834,
+      "step": 3352
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0079709721164899e-05,
+      "loss": 0.868,
+      "step": 3353
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0074727959418857e-05,
+      "loss": 1.1078,
+      "step": 3354
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0069746179125763e-05,
+      "loss": 0.9894,
+      "step": 3355
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0064764381522065e-05,
+      "loss": 1.0119,
+      "step": 3356
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0059782567844215e-05,
+      "loss": 0.993,
+      "step": 3357
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0054800739328677e-05,
+      "loss": 0.9808,
+      "step": 3358
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0049818897211903e-05,
+      "loss": 1.1198,
+      "step": 3359
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0044837042730365e-05,
+      "loss": 1.0442,
+      "step": 3360
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0039855177120526e-05,
+      "loss": 1.0281,
+      "step": 3361
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0034873301618862e-05,
+      "loss": 0.7898,
+      "step": 3362
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.002989141746184e-05,
+      "loss": 1.0507,
+      "step": 3363
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0024909525885944e-05,
+      "loss": 0.9973,
+      "step": 3364
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0019927628127642e-05,
+      "loss": 1.1292,
+      "step": 3365
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0014945725423417e-05,
+      "loss": 0.9455,
+      "step": 3366
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0009963819009748e-05,
+      "loss": 1.0218,
+      "step": 3367
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0004981910123115e-05,
+      "loss": 1.0019,
+      "step": 3368
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1e-05,
+      "loss": 1.0127,
+      "step": 3369
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.995018089876888e-06,
+      "loss": 0.9353,
+      "step": 3370
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.990036180990254e-06,
+      "loss": 1.0795,
+      "step": 3371
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.985054274576585e-06,
+      "loss": 1.0623,
+      "step": 3372
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.980072371872363e-06,
+      "loss": 0.8661,
+      "step": 3373
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.97509047411406e-06,
+      "loss": 1.0595,
+      "step": 3374
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.970108582538161e-06,
+      "loss": 1.0258,
+      "step": 3375
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.965126698381142e-06,
+      "loss": 0.9429,
+      "step": 3376
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.960144822879474e-06,
+      "loss": 1.0867,
+      "step": 3377
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.955162957269639e-06,
+      "loss": 0.8949,
+      "step": 3378
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.950181102788102e-06,
+      "loss": 0.9697,
+      "step": 3379
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.945199260671328e-06,
+      "loss": 0.9731,
+      "step": 3380
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.940217432155786e-06,
+      "loss": 1.0864,
+      "step": 3381
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.935235618477939e-06,
+      "loss": 0.9748,
+      "step": 3382
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.930253820874239e-06,
+      "loss": 1.0593,
+      "step": 3383
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.925272040581145e-06,
+      "loss": 1.0415,
+      "step": 3384
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.920290278835106e-06,
+      "loss": 0.9236,
+      "step": 3385
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.915308536872562e-06,
+      "loss": 0.8723,
+      "step": 3386
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.910326815929959e-06,
+      "loss": 0.8565,
+      "step": 3387
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.905345117243729e-06,
+      "loss": 1.0082,
+      "step": 3388
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.9003634420503e-06,
+      "loss": 0.9982,
+      "step": 3389
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.895381791586098e-06,
+      "loss": 1.0216,
+      "step": 3390
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.89040016708754e-06,
+      "loss": 0.9269,
+      "step": 3391
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.885418569791033e-06,
+      "loss": 1.0901,
+      "step": 3392
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.880437000932985e-06,
+      "loss": 0.9272,
+      "step": 3393
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.875455461749793e-06,
+      "loss": 0.9696,
+      "step": 3394
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.870473953477842e-06,
+      "loss": 1.0693,
+      "step": 3395
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.86549247735352e-06,
+      "loss": 0.9509,
+      "step": 3396
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.860511034613203e-06,
+      "loss": 1.0796,
+      "step": 3397
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.855529626493246e-06,
+      "loss": 0.9942,
+      "step": 3398
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.850548254230013e-06,
+      "loss": 1.0643,
+      "step": 3399
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.845566919059852e-06,
+      "loss": 1.0159,
+      "step": 3400
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.840585622219103e-06,
+      "loss": 0.9331,
+      "step": 3401
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.835604364944092e-06,
+      "loss": 1.012,
+      "step": 3402
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.830623148471144e-06,
+      "loss": 0.8705,
+      "step": 3403
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.825641974036561e-06,
+      "loss": 0.8804,
+      "step": 3404
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.820660842876648e-06,
+      "loss": 0.9851,
+      "step": 3405
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.815679756227689e-06,
+      "loss": 0.8149,
+      "step": 3406
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.810698715325967e-06,
+      "loss": 1.0055,
+      "step": 3407
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.805717721407744e-06,
+      "loss": 1.1304,
+      "step": 3408
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.80073677570928e-06,
+      "loss": 0.9587,
+      "step": 3409
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.795755879466803e-06,
+      "loss": 1.058,
+      "step": 3410
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.790775033916559e-06,
+      "loss": 0.8949,
+      "step": 3411
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.785794240294755e-06,
+      "loss": 0.9228,
+      "step": 3412
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.780813499837603e-06,
+      "loss": 0.9223,
+      "step": 3413
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.77583281378129e-06,
+      "loss": 0.9888,
+      "step": 3414
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.770852183361998e-06,
+      "loss": 0.8897,
+      "step": 3415
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.765871609815883e-06,
+      "loss": 0.9289,
+      "step": 3416
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.760891094379102e-06,
+      "loss": 0.9108,
+      "step": 3417
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.755910638287786e-06,
+      "loss": 0.9653,
+      "step": 3418
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.750930242778061e-06,
+      "loss": 1.0427,
+      "step": 3419
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.74594990908603e-06,
+      "loss": 1.0246,
+      "step": 3420
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.740969638447786e-06,
+      "loss": 0.966,
+      "step": 3421
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.735989432099397e-06,
+      "loss": 1.0098,
+      "step": 3422
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.731009291276928e-06,
+      "loss": 0.8131,
+      "step": 3423
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.726029217216418e-06,
+      "loss": 0.9321,
+      "step": 3424
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.721049211153895e-06,
+      "loss": 0.9474,
+      "step": 3425
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.716069274325371e-06,
+      "loss": 0.8535,
+      "step": 3426
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.711089407966833e-06,
+      "loss": 0.8944,
+      "step": 3427
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.706109613314256e-06,
+      "loss": 0.9193,
+      "step": 3428
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.7011298916036e-06,
+      "loss": 0.9064,
+      "step": 3429
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.696150244070797e-06,
+      "loss": 0.9823,
+      "step": 3430
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.691170671951775e-06,
+      "loss": 0.9257,
+      "step": 3431
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.686191176482433e-06,
+      "loss": 1.0452,
+      "step": 3432
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.681211758898654e-06,
+      "loss": 0.9196,
+      "step": 3433
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.676232420436294e-06,
+      "loss": 0.9543,
+      "step": 3434
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.671253162331203e-06,
+      "loss": 0.9512,
+      "step": 3435
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.666273985819203e-06,
+      "loss": 0.9104,
+      "step": 3436
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.6612948921361e-06,
+      "loss": 0.9152,
+      "step": 3437
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.656315882517674e-06,
+      "loss": 0.9878,
+      "step": 3438
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.651336958199688e-06,
+      "loss": 0.9789,
+      "step": 3439
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.646358120417876e-06,
+      "loss": 0.9299,
+      "step": 3440
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.641379370407968e-06,
+      "loss": 1.0741,
+      "step": 3441
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.636400709405653e-06,
+      "loss": 0.8736,
+      "step": 3442
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.63142213864661e-06,
+      "loss": 0.9779,
+      "step": 3443
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.626443659366495e-06,
+      "loss": 1.0176,
+      "step": 3444
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.62146527280093e-06,
+      "loss": 1.0789,
+      "step": 3445
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.61648698018553e-06,
+      "loss": 0.9112,
+      "step": 3446
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.611508782755875e-06,
+      "loss": 1.0265,
+      "step": 3447
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.606530681747524e-06,
+      "loss": 0.9767,
+      "step": 3448
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.601552678396016e-06,
+      "loss": 0.8601,
+      "step": 3449
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.596574773936863e-06,
+      "loss": 0.9434,
+      "step": 3450
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.591596969605549e-06,
+      "loss": 1.0084,
+      "step": 3451
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.586619266637545e-06,
+      "loss": 0.9556,
+      "step": 3452
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.581641666268278e-06,
+      "loss": 0.8962,
+      "step": 3453
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.576664169733166e-06,
+      "loss": 0.9888,
+      "step": 3454
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.571686778267592e-06,
+      "loss": 1.0487,
+      "step": 3455
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.56670949310692e-06,
+      "loss": 1.0198,
+      "step": 3456
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.561732315486482e-06,
+      "loss": 0.9241,
+      "step": 3457
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.55675524664159e-06,
+      "loss": 0.9544,
+      "step": 3458
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.551778287807513e-06,
+      "loss": 0.9617,
+      "step": 3459
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.546801440219512e-06,
+      "loss": 0.8604,
+      "step": 3460
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.54182470511281e-06,
+      "loss": 1.0183,
+      "step": 3461
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.536848083722605e-06,
+      "loss": 1.082,
+      "step": 3462
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.531871577284065e-06,
+      "loss": 1.0049,
+      "step": 3463
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.526895187032335e-06,
+      "loss": 0.9547,
+      "step": 3464
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.52191891420252e-06,
+      "loss": 1.0473,
+      "step": 3465
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.516942760029704e-06,
+      "loss": 0.8755,
+      "step": 3466
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.511966725748944e-06,
+      "loss": 1.0804,
+      "step": 3467
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.506990812595257e-06,
+      "loss": 0.9591,
+      "step": 3468
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.502015021803645e-06,
+      "loss": 0.8782,
+      "step": 3469
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.497039354609068e-06,
+      "loss": 0.9956,
+      "step": 3470
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.492063812246453e-06,
+      "loss": 0.9517,
+      "step": 3471
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.487088395950703e-06,
+      "loss": 1.0196,
+      "step": 3472
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.482113106956693e-06,
+      "loss": 0.9664,
+      "step": 3473
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.477137946499253e-06,
+      "loss": 0.9303,
+      "step": 3474
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.472162915813198e-06,
+      "loss": 0.973,
+      "step": 3475
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.467188016133301e-06,
+      "loss": 0.9875,
+      "step": 3476
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.462213248694299e-06,
+      "loss": 1.0087,
+      "step": 3477
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.457238614730901e-06,
+      "loss": 0.9717,
+      "step": 3478
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.452264115477788e-06,
+      "loss": 1.0569,
+      "step": 3479
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.447289752169597e-06,
+      "loss": 0.9473,
+      "step": 3480
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.442315526040942e-06,
+      "loss": 0.9467,
+      "step": 3481
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.437341438326397e-06,
+      "loss": 0.9994,
+      "step": 3482
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.432367490260496e-06,
+      "loss": 0.8566,
+      "step": 3483
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.427393683077748e-06,
+      "loss": 0.9565,
+      "step": 3484
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.422420018012624e-06,
+      "loss": 0.9328,
+      "step": 3485
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.41744649629956e-06,
+      "loss": 0.8735,
+      "step": 3486
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.412473119172956e-06,
+      "loss": 1.0656,
+      "step": 3487
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.407499887867177e-06,
+      "loss": 0.9863,
+      "step": 3488
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.402526803616547e-06,
+      "loss": 0.9626,
+      "step": 3489
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.397553867655356e-06,
+      "loss": 1.0321,
+      "step": 3490
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.392581081217864e-06,
+      "loss": 1.1173,
+      "step": 3491
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.387608445538285e-06,
+      "loss": 1.0419,
+      "step": 3492
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.382635961850799e-06,
+      "loss": 0.8928,
+      "step": 3493
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.37766363138955e-06,
+      "loss": 0.9409,
+      "step": 3494
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.37269145538864e-06,
+      "loss": 0.9364,
+      "step": 3495
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.367719435082131e-06,
+      "loss": 0.991,
+      "step": 3496
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.362747571704059e-06,
+      "loss": 1.0421,
+      "step": 3497
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.357775866488405e-06,
+      "loss": 0.9636,
+      "step": 3498
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.352804320669122e-06,
+      "loss": 0.9291,
+      "step": 3499
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.347832935480119e-06,
+      "loss": 1.0099,
+      "step": 3500
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.342861712155263e-06,
+      "loss": 1.0234,
+      "step": 3501
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.337890651928381e-06,
+      "loss": 0.8458,
+      "step": 3502
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.332919756033269e-06,
+      "loss": 0.878,
+      "step": 3503
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.327949025703669e-06,
+      "loss": 0.8016,
+      "step": 3504
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.32297846217329e-06,
+      "loss": 0.965,
+      "step": 3505
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.318008066675802e-06,
+      "loss": 0.9742,
+      "step": 3506
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.313037840444823e-06,
+      "loss": 1.1092,
+      "step": 3507
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.308067784713934e-06,
+      "loss": 0.9632,
+      "step": 3508
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.303097900716678e-06,
+      "loss": 1.0276,
+      "step": 3509
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.298128189686549e-06,
+      "loss": 1.01,
+      "step": 3510
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.293158652857007e-06,
+      "loss": 0.8558,
+      "step": 3511
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.288189291461456e-06,
+      "loss": 0.9905,
+      "step": 3512
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.283220106733269e-06,
+      "loss": 0.9443,
+      "step": 3513
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.278251099905759e-06,
+      "loss": 1.058,
+      "step": 3514
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.273282272212217e-06,
+      "loss": 1.0043,
+      "step": 3515
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.268313624885871e-06,
+      "loss": 0.9367,
+      "step": 3516
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.263345159159913e-06,
+      "loss": 1.0479,
+      "step": 3517
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.258376876267487e-06,
+      "loss": 0.9317,
+      "step": 3518
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.2534087774417e-06,
+      "loss": 0.905,
+      "step": 3519
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.248440863915592e-06,
+      "loss": 0.8969,
+      "step": 3520
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.243473136922178e-06,
+      "loss": 1.0857,
+      "step": 3521
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.23850559769442e-06,
+      "loss": 0.8966,
+      "step": 3522
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.233538247465232e-06,
+      "loss": 0.897,
+      "step": 3523
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.228571087467482e-06,
+      "loss": 0.9308,
+      "step": 3524
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.223604118933997e-06,
+      "loss": 1.0635,
+      "step": 3525
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.218637343097537e-06,
+      "loss": 0.9955,
+      "step": 3526
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.213670761190838e-06,
+      "loss": 0.9566,
+      "step": 3527
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.208704374446572e-06,
+      "loss": 0.9017,
+      "step": 3528
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.203738184097372e-06,
+      "loss": 0.9399,
+      "step": 3529
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.198772191375813e-06,
+      "loss": 0.8916,
+      "step": 3530
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.193806397514437e-06,
+      "loss": 0.9123,
+      "step": 3531
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.188840803745712e-06,
+      "loss": 1.0204,
+      "step": 3532
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.183875411302076e-06,
+      "loss": 1.0486,
+      "step": 3533
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.17891022141591e-06,
+      "loss": 0.9452,
+      "step": 3534
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.17394523531955e-06,
+      "loss": 0.9398,
+      "step": 3535
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.16898045424527e-06,
+      "loss": 0.9001,
+      "step": 3536
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.164015879425312e-06,
+      "loss": 0.9536,
+      "step": 3537
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.159051512091842e-06,
+      "loss": 1.026,
+      "step": 3538
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.154087353476995e-06,
+      "loss": 0.9947,
+      "step": 3539
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.149123404812842e-06,
+      "loss": 0.8266,
+      "step": 3540
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.144159667331415e-06,
+      "loss": 0.881,
+      "step": 3541
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.139196142264675e-06,
+      "loss": 1.008,
+      "step": 3542
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.134232830844556e-06,
+      "loss": 0.9826,
+      "step": 3543
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.129269734302905e-06,
+      "loss": 0.9456,
+      "step": 3544
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.124306853871545e-06,
+      "loss": 0.8561,
+      "step": 3545
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.119344190782231e-06,
+      "loss": 1.0234,
+      "step": 3546
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.114381746266672e-06,
+      "loss": 0.8084,
+      "step": 3547
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.109419521556512e-06,
+      "loss": 0.979,
+      "step": 3548
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.104457517883359e-06,
+      "loss": 0.9552,
+      "step": 3549
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.099495736478737e-06,
+      "loss": 0.9618,
+      "step": 3550
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.094534178574144e-06,
+      "loss": 0.9774,
+      "step": 3551
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.089572845401003e-06,
+      "loss": 1.0084,
+      "step": 3552
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.084611738190697e-06,
+      "loss": 0.8663,
+      "step": 3553
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.079650858174536e-06,
+      "loss": 0.8826,
+      "step": 3554
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.074690206583795e-06,
+      "loss": 1.0892,
+      "step": 3555
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.069729784649662e-06,
+      "loss": 1.0539,
+      "step": 3556
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.064769593603295e-06,
+      "loss": 0.9432,
+      "step": 3557
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.059809634675784e-06,
+      "loss": 0.8569,
+      "step": 3558
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.054849909098164e-06,
+      "loss": 0.9065,
+      "step": 3559
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.04989041810141e-06,
+      "loss": 0.9816,
+      "step": 3560
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.044931162916442e-06,
+      "loss": 0.8788,
+      "step": 3561
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.039972144774109e-06,
+      "loss": 1.079,
+      "step": 3562
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.035013364905222e-06,
+      "loss": 0.9425,
+      "step": 3563
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.030054824540514e-06,
+      "loss": 0.9768,
+      "step": 3564
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.025096524910673e-06,
+      "loss": 0.8798,
+      "step": 3565
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.020138467246314e-06,
+      "loss": 0.8491,
+      "step": 3566
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.015180652778011e-06,
+      "loss": 0.9825,
+      "step": 3567
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.010223082736249e-06,
+      "loss": 0.902,
+      "step": 3568
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.005265758351479e-06,
+      "loss": 0.7984,
+      "step": 3569
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.000308680854075e-06,
+      "loss": 0.9818,
+      "step": 3570
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.995351851474359e-06,
+      "loss": 0.946,
+      "step": 3571
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.990395271442588e-06,
+      "loss": 1.0072,
+      "step": 3572
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.98543894198896e-06,
+      "loss": 0.9087,
+      "step": 3573
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.980482864343601e-06,
+      "loss": 0.8447,
+      "step": 3574
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.975527039736582e-06,
+      "loss": 0.9907,
+      "step": 3575
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.970571469397913e-06,
+      "loss": 0.883,
+      "step": 3576
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.965616154557538e-06,
+      "loss": 0.901,
+      "step": 3577
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.960661096445337e-06,
+      "loss": 0.9329,
+      "step": 3578
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.95570629629113e-06,
+      "loss": 0.8804,
+      "step": 3579
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.950751755324668e-06,
+      "loss": 1.0903,
+      "step": 3580
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.945797474775638e-06,
+      "loss": 1.0333,
+      "step": 3581
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.940843455873664e-06,
+      "loss": 0.9822,
+      "step": 3582
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.935889699848311e-06,
+      "loss": 1.0756,
+      "step": 3583
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.930936207929066e-06,
+      "loss": 1.0066,
+      "step": 3584
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.925982981345364e-06,
+      "loss": 1.0132,
+      "step": 3585
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.921030021326564e-06,
+      "loss": 0.9857,
+      "step": 3586
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.916077329101963e-06,
+      "loss": 0.8126,
+      "step": 3587
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.911124905900787e-06,
+      "loss": 0.8973,
+      "step": 3588
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.906172752952206e-06,
+      "loss": 0.9393,
+      "step": 3589
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.90122087148531e-06,
+      "loss": 0.9057,
+      "step": 3590
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.896269262729134e-06,
+      "loss": 1.1602,
+      "step": 3591
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.891317927912638e-06,
+      "loss": 0.9154,
+      "step": 3592
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.88636686826471e-06,
+      "loss": 0.9721,
+      "step": 3593
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.881416085014174e-06,
+      "loss": 0.9295,
+      "step": 3594
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.876465579389795e-06,
+      "loss": 1.0089,
+      "step": 3595
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.871515352620252e-06,
+      "loss": 0.9167,
+      "step": 3596
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.866565405934169e-06,
+      "loss": 0.9872,
+      "step": 3597
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.861615740560094e-06,
+      "loss": 0.956,
+      "step": 3598
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.856666357726501e-06,
+      "loss": 0.9844,
+      "step": 3599
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.851717258661799e-06,
+      "loss": 0.9963,
+      "step": 3600
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.84676844459433e-06,
+      "loss": 0.9312,
+      "step": 3601
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.841819916752357e-06,
+      "loss": 1.0212,
+      "step": 3602
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.836871676364085e-06,
+      "loss": 0.8788,
+      "step": 3603
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.831923724657632e-06,
+      "loss": 0.9891,
+      "step": 3604
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.82697606286105e-06,
+      "loss": 0.9095,
+      "step": 3605
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.822028692202324e-06,
+      "loss": 0.9217,
+      "step": 3606
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.817081613909362e-06,
+      "loss": 0.8989,
+      "step": 3607
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.81213482921e-06,
+      "loss": 0.912,
+      "step": 3608
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.807188339332005e-06,
+      "loss": 0.9901,
+      "step": 3609
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.802242145503067e-06,
+      "loss": 0.9459,
+      "step": 3610
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.7972962489508e-06,
+      "loss": 0.8846,
+      "step": 3611
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.792350650902745e-06,
+      "loss": 0.9233,
+      "step": 3612
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.787405352586377e-06,
+      "loss": 0.8941,
+      "step": 3613
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.782460355229089e-06,
+      "loss": 0.9625,
+      "step": 3614
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.777515660058203e-06,
+      "loss": 0.9973,
+      "step": 3615
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.772571268300963e-06,
+      "loss": 0.9378,
+      "step": 3616
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.767627181184536e-06,
+      "loss": 0.9712,
+      "step": 3617
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.762683399936017e-06,
+      "loss": 0.8766,
+      "step": 3618
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.757739925782426e-06,
+      "loss": 0.932,
+      "step": 3619
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.752796759950704e-06,
+      "loss": 0.9483,
+      "step": 3620
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.74785390366772e-06,
+      "loss": 0.968,
+      "step": 3621
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.74291135816026e-06,
+      "loss": 0.8678,
+      "step": 3622
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.737969124655037e-06,
+      "loss": 0.846,
+      "step": 3623
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.733027204378679e-06,
+      "loss": 0.9661,
+      "step": 3624
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.72808559855775e-06,
+      "loss": 0.951,
+      "step": 3625
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.723144308418725e-06,
+      "loss": 0.9642,
+      "step": 3626
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.718203335188005e-06,
+      "loss": 0.9433,
+      "step": 3627
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.713262680091914e-06,
+      "loss": 0.9433,
+      "step": 3628
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.708322344356689e-06,
+      "loss": 0.9046,
+      "step": 3629
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.703382329208494e-06,
+      "loss": 1.0739,
+      "step": 3630
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.698442635873417e-06,
+      "loss": 0.941,
+      "step": 3631
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.693503265577457e-06,
+      "loss": 1.0087,
+      "step": 3632
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.68856421954654e-06,
+      "loss": 1.0109,
+      "step": 3633
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.683625499006512e-06,
+      "loss": 0.978,
+      "step": 3634
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.678687105183128e-06,
+      "loss": 0.8922,
+      "step": 3635
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.673749039302075e-06,
+      "loss": 0.8827,
+      "step": 3636
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.668811302588952e-06,
+      "loss": 1.036,
+      "step": 3637
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.663873896269273e-06,
+      "loss": 0.9063,
+      "step": 3638
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.65893682156848e-06,
+      "loss": 1.0719,
+      "step": 3639
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.654000079711925e-06,
+      "loss": 1.0575,
+      "step": 3640
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.649063671924876e-06,
+      "loss": 1.031,
+      "step": 3641
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.644127599432523e-06,
+      "loss": 0.9335,
+      "step": 3642
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.639191863459973e-06,
+      "loss": 0.8454,
+      "step": 3643
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.634256465232244e-06,
+      "loss": 0.9839,
+      "step": 3644
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.629321405974279e-06,
+      "loss": 0.8581,
+      "step": 3645
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.624386686910928e-06,
+      "loss": 0.9707,
+      "step": 3646
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.619452309266957e-06,
+      "loss": 1.106,
+      "step": 3647
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.614518274267057e-06,
+      "loss": 0.908,
+      "step": 3648
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.609584583135824e-06,
+      "loss": 0.9404,
+      "step": 3649
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.60465123709777e-06,
+      "loss": 1.1205,
+      "step": 3650
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.599718237377327e-06,
+      "loss": 0.9389,
+      "step": 3651
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.594785585198834e-06,
+      "loss": 0.9145,
+      "step": 3652
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.589853281786556e-06,
+      "loss": 0.9153,
+      "step": 3653
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.584921328364652e-06,
+      "loss": 1.055,
+      "step": 3654
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.579989726157208e-06,
+      "loss": 0.9457,
+      "step": 3655
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.575058476388218e-06,
+      "loss": 0.8378,
+      "step": 3656
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.570127580281596e-06,
+      "loss": 0.9704,
+      "step": 3657
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.565197039061154e-06,
+      "loss": 1.1149,
+      "step": 3658
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.560266853950633e-06,
+      "loss": 1.0897,
+      "step": 3659
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.555337026173674e-06,
+      "loss": 1.0162,
+      "step": 3660
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.550407556953829e-06,
+      "loss": 0.9688,
+      "step": 3661
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.545478447514562e-06,
+      "loss": 1.0618,
+      "step": 3662
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.540549699079257e-06,
+      "loss": 0.9396,
+      "step": 3663
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.535621312871197e-06,
+      "loss": 1.0387,
+      "step": 3664
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.530693290113582e-06,
+      "loss": 1.0101,
+      "step": 3665
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.525765632029517e-06,
+      "loss": 1.0885,
+      "step": 3666
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.52083833984202e-06,
+      "loss": 0.9329,
+      "step": 3667
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.515911414774012e-06,
+      "loss": 0.9648,
+      "step": 3668
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.510984858048336e-06,
+      "loss": 1.0608,
+      "step": 3669
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.506058670887728e-06,
+      "loss": 0.8715,
+      "step": 3670
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.501132854514847e-06,
+      "loss": 0.9554,
+      "step": 3671
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.496207410152246e-06,
+      "loss": 0.8784,
+      "step": 3672
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.491282339022395e-06,
+      "loss": 0.9582,
+      "step": 3673
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.486357642347664e-06,
+      "loss": 0.9933,
+      "step": 3674
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.481433321350343e-06,
+      "loss": 1.1241,
+      "step": 3675
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.476509377252614e-06,
+      "loss": 1.0095,
+      "step": 3676
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.471585811276576e-06,
+      "loss": 0.9662,
+      "step": 3677
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.466662624644226e-06,
+      "loss": 1.0473,
+      "step": 3678
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.461739818577474e-06,
+      "loss": 1.046,
+      "step": 3679
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.45681739429813e-06,
+      "loss": 0.9707,
+      "step": 3680
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.451895353027912e-06,
+      "loss": 0.9029,
+      "step": 3681
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.446973695988443e-06,
+      "loss": 1.0264,
+      "step": 3682
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.442052424401254e-06,
+      "loss": 0.8881,
+      "step": 3683
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.437131539487773e-06,
+      "loss": 0.9233,
+      "step": 3684
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.432211042469335e-06,
+      "loss": 1.0183,
+      "step": 3685
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.427290934567178e-06,
+      "loss": 0.8443,
+      "step": 3686
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.422371217002448e-06,
+      "loss": 0.8643,
+      "step": 3687
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.417451890996188e-06,
+      "loss": 1.1595,
+      "step": 3688
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.412532957769353e-06,
+      "loss": 0.9399,
+      "step": 3689
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.407614418542787e-06,
+      "loss": 0.9887,
+      "step": 3690
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.402696274537245e-06,
+      "loss": 1.0442,
+      "step": 3691
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.397778526973384e-06,
+      "loss": 0.919,
+      "step": 3692
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.39286117707176e-06,
+      "loss": 1.0544,
+      "step": 3693
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.38794422605283e-06,
+      "loss": 0.9218,
+      "step": 3694
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.383027675136958e-06,
+      "loss": 0.9727,
+      "step": 3695
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.3781115255444e-06,
+      "loss": 0.9515,
+      "step": 3696
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.373195778495313e-06,
+      "loss": 1.0006,
+      "step": 3697
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.368280435209765e-06,
+      "loss": 0.9451,
+      "step": 3698
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.363365496907712e-06,
+      "loss": 1.1389,
+      "step": 3699
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.35845096480901e-06,
+      "loss": 0.889,
+      "step": 3700
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.353536840133428e-06,
+      "loss": 0.9649,
+      "step": 3701
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.348623124100616e-06,
+      "loss": 0.9166,
+      "step": 3702
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.34370981793013e-06,
+      "loss": 0.845,
+      "step": 3703
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.338796922841427e-06,
+      "loss": 1.0348,
+      "step": 3704
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.33388444005386e-06,
+      "loss": 0.9456,
+      "step": 3705
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.328972370786676e-06,
+      "loss": 0.9793,
+      "step": 3706
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.32406071625903e-06,
+      "loss": 0.8863,
+      "step": 3707
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.319149477689959e-06,
+      "loss": 1.0423,
+      "step": 3708
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.314238656298405e-06,
+      "loss": 0.8481,
+      "step": 3709
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.309328253303209e-06,
+      "loss": 0.8906,
+      "step": 3710
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.304418269923106e-06,
+      "loss": 0.9169,
+      "step": 3711
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.29950870737672e-06,
+      "loss": 1.0755,
+      "step": 3712
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.294599566882586e-06,
+      "loss": 0.8609,
+      "step": 3713
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.289690849659116e-06,
+      "loss": 0.9791,
+      "step": 3714
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.284782556924625e-06,
+      "loss": 1.0892,
+      "step": 3715
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.279874689897332e-06,
+      "loss": 0.9119,
+      "step": 3716
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.274967249795333e-06,
+      "loss": 0.8952,
+      "step": 3717
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.27006023783663e-06,
+      "loss": 0.9844,
+      "step": 3718
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.265153655239121e-06,
+      "loss": 0.9714,
+      "step": 3719
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.260247503220583e-06,
+      "loss": 1.0426,
+      "step": 3720
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.255341782998696e-06,
+      "loss": 1.0568,
+      "step": 3721
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.250436495791037e-06,
+      "loss": 0.9779,
+      "step": 3722
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.245531642815068e-06,
+      "loss": 0.9472,
+      "step": 3723
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.24062722528814e-06,
+      "loss": 0.8778,
+      "step": 3724
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.235723244427511e-06,
+      "loss": 0.9156,
+      "step": 3725
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.23081970145032e-06,
+      "loss": 0.8978,
+      "step": 3726
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.225916597573588e-06,
+      "loss": 0.996,
+      "step": 3727
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.221013934014248e-06,
+      "loss": 0.9217,
+      "step": 3728
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.21611171198911e-06,
+      "loss": 0.9348,
+      "step": 3729
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.211209932714875e-06,
+      "loss": 0.9896,
+      "step": 3730
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.20630859740814e-06,
+      "loss": 1.0094,
+      "step": 3731
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.201407707285391e-06,
+      "loss": 0.9529,
+      "step": 3732
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.196507263562991e-06,
+      "loss": 0.6942,
+      "step": 3733
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.191607267457213e-06,
+      "loss": 0.8887,
+      "step": 3734
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.186707720184202e-06,
+      "loss": 0.9185,
+      "step": 3735
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.181808622959997e-06,
+      "loss": 0.9778,
+      "step": 3736
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.176909977000532e-06,
+      "loss": 0.948,
+      "step": 3737
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.17201178352162e-06,
+      "loss": 0.919,
+      "step": 3738
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.16711404373896e-06,
+      "loss": 0.9741,
+      "step": 3739
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.16221675886815e-06,
+      "loss": 1.005,
+      "step": 3740
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.157319930124665e-06,
+      "loss": 0.9397,
+      "step": 3741
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.152423558723867e-06,
+      "loss": 0.8958,
+      "step": 3742
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.147527645881015e-06,
+      "loss": 0.9548,
+      "step": 3743
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.142632192811245e-06,
+      "loss": 0.899,
+      "step": 3744
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.137737200729573e-06,
+      "loss": 0.9202,
+      "step": 3745
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.132842670850913e-06,
+      "loss": 0.9921,
+      "step": 3746
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.127948604390061e-06,
+      "loss": 0.9994,
+      "step": 3747
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.123055002561694e-06,
+      "loss": 0.9481,
+      "step": 3748
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.118161866580379e-06,
+      "loss": 0.9531,
+      "step": 3749
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.113269197660564e-06,
+      "loss": 1.0598,
+      "step": 3750
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.108376997016576e-06,
+      "loss": 1.0177,
+      "step": 3751
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.103485265862636e-06,
+      "loss": 0.9182,
+      "step": 3752
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.098594005412843e-06,
+      "loss": 1.1495,
+      "step": 3753
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.093703216881182e-06,
+      "loss": 0.9181,
+      "step": 3754
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.088812901481517e-06,
+      "loss": 0.9262,
+      "step": 3755
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.083923060427598e-06,
+      "loss": 0.904,
+      "step": 3756
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.079033694933049e-06,
+      "loss": 0.8921,
+      "step": 3757
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.074144806211392e-06,
+      "loss": 0.8352,
+      "step": 3758
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.069256395476012e-06,
+      "loss": 0.9815,
+      "step": 3759
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.064368463940194e-06,
+      "loss": 0.8353,
+      "step": 3760
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.05948101281709e-06,
+      "loss": 0.8489,
+      "step": 3761
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.05459404331974e-06,
+      "loss": 0.9214,
+      "step": 3762
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.049707556661057e-06,
+      "loss": 1.0024,
+      "step": 3763
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.044821554053841e-06,
+      "loss": 1.0242,
+      "step": 3764
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.039936036710771e-06,
+      "loss": 1.0023,
+      "step": 3765
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.035051005844408e-06,
+      "loss": 0.9407,
+      "step": 3766
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.030166462667182e-06,
+      "loss": 1.1121,
+      "step": 3767
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.025282408391418e-06,
+      "loss": 0.9204,
+      "step": 3768
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.020398844229299e-06,
+      "loss": 0.8563,
+      "step": 3769
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.015515771392902e-06,
+      "loss": 0.945,
+      "step": 3770
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.010633191094181e-06,
+      "loss": 0.8742,
+      "step": 3771
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.005751104544965e-06,
+      "loss": 0.9894,
+      "step": 3772
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.000869512956958e-06,
+      "loss": 0.9726,
+      "step": 3773
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.995988417541745e-06,
+      "loss": 1.0378,
+      "step": 3774
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.99110781951078e-06,
+      "loss": 1.0462,
+      "step": 3775
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.986227720075404e-06,
+      "loss": 0.9958,
+      "step": 3776
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.981348120446829e-06,
+      "loss": 1.085,
+      "step": 3777
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.976469021836146e-06,
+      "loss": 0.8323,
+      "step": 3778
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.971590425454318e-06,
+      "loss": 1.0258,
+      "step": 3779
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.966712332512188e-06,
+      "loss": 1.068,
+      "step": 3780
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.961834744220463e-06,
+      "loss": 0.9045,
+      "step": 3781
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.95695766178974e-06,
+      "loss": 0.8814,
+      "step": 3782
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.952081086430477e-06,
+      "loss": 0.9931,
+      "step": 3783
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.947205019353018e-06,
+      "loss": 0.9067,
+      "step": 3784
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.942329461767573e-06,
+      "loss": 0.9157,
+      "step": 3785
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.937454414884228e-06,
+      "loss": 0.9334,
+      "step": 3786
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.932579879912939e-06,
+      "loss": 0.979,
+      "step": 3787
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.92770585806354e-06,
+      "loss": 0.9496,
+      "step": 3788
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.922832350545733e-06,
+      "loss": 0.9662,
+      "step": 3789
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.9179593585691e-06,
+      "loss": 0.8924,
+      "step": 3790
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.913086883343087e-06,
+      "loss": 0.8875,
+      "step": 3791
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.908214926077011e-06,
+      "loss": 0.9793,
+      "step": 3792
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.903343487980072e-06,
+      "loss": 0.9586,
+      "step": 3793
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.89847257026133e-06,
+      "loss": 0.9786,
+      "step": 3794
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.893602174129712e-06,
+      "loss": 0.9224,
+      "step": 3795
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.888732300794031e-06,
+      "loss": 0.9857,
+      "step": 3796
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.88386295146296e-06,
+      "loss": 1.151,
+      "step": 3797
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.87899412734504e-06,
+      "loss": 0.9271,
+      "step": 3798
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.874125829648692e-06,
+      "loss": 0.8294,
+      "step": 3799
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.869258059582193e-06,
+      "loss": 0.9689,
+      "step": 3800
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.864390818353697e-06,
+      "loss": 1.013,
+      "step": 3801
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.859524107171232e-06,
+      "loss": 0.8779,
+      "step": 3802
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.85465792724268e-06,
+      "loss": 0.9631,
+      "step": 3803
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.849792279775801e-06,
+      "loss": 0.9993,
+      "step": 3804
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.844927165978228e-06,
+      "loss": 1.0277,
+      "step": 3805
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.840062587057445e-06,
+      "loss": 0.9685,
+      "step": 3806
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.835198544220815e-06,
+      "loss": 0.9834,
+      "step": 3807
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.830335038675571e-06,
+      "loss": 0.96,
+      "step": 3808
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.825472071628799e-06,
+      "loss": 0.8221,
+      "step": 3809
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.820609644287469e-06,
+      "loss": 0.9291,
+      "step": 3810
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.815747757858403e-06,
+      "loss": 0.9153,
+      "step": 3811
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.81088641354829e-06,
+      "loss": 0.9187,
+      "step": 3812
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.806025612563692e-06,
+      "loss": 1.0156,
+      "step": 3813
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.801165356111035e-06,
+      "loss": 0.9251,
+      "step": 3814
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.796305645396599e-06,
+      "loss": 1.0394,
+      "step": 3815
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.791446481626544e-06,
+      "loss": 0.9021,
+      "step": 3816
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.786587866006885e-06,
+      "loss": 0.9241,
+      "step": 3817
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.7817297997435e-06,
+      "loss": 0.9612,
+      "step": 3818
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.776872284042135e-06,
+      "loss": 0.85,
+      "step": 3819
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.772015320108396e-06,
+      "loss": 1.0698,
+      "step": 3820
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.767158909147756e-06,
+      "loss": 0.9758,
+      "step": 3821
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.762303052365549e-06,
+      "loss": 1.0268,
+      "step": 3822
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.757447750966973e-06,
+      "loss": 0.959,
+      "step": 3823
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.75259300615708e-06,
+      "loss": 1.0033,
+      "step": 3824
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.74773881914079e-06,
+      "loss": 0.9638,
+      "step": 3825
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.74288519112289e-06,
+      "loss": 0.8496,
+      "step": 3826
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.73803212330802e-06,
+      "loss": 1.0939,
+      "step": 3827
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.733179616900684e-06,
+      "loss": 0.8936,
+      "step": 3828
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.72832767310525e-06,
+      "loss": 0.9815,
+      "step": 3829
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.723476293125937e-06,
+      "loss": 0.8994,
+      "step": 3830
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.71862547816683e-06,
+      "loss": 0.9091,
+      "step": 3831
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.713775229431878e-06,
+      "loss": 0.9931,
+      "step": 3832
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.70892554812488e-06,
+      "loss": 1.1343,
+      "step": 3833
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.704076435449505e-06,
+      "loss": 0.9178,
+      "step": 3834
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.699227892609274e-06,
+      "loss": 1.0616,
+      "step": 3835
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.694379920807565e-06,
+      "loss": 0.8496,
+      "step": 3836
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.689532521247615e-06,
+      "loss": 0.9827,
+      "step": 3837
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.684685695132525e-06,
+      "loss": 0.8912,
+      "step": 3838
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.679839443665248e-06,
+      "loss": 0.9665,
+      "step": 3839
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.674993768048598e-06,
+      "loss": 0.8952,
+      "step": 3840
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.670148669485244e-06,
+      "loss": 0.8469,
+      "step": 3841
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.66530414917771e-06,
+      "loss": 0.8918,
+      "step": 3842
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.660460208328375e-06,
+      "loss": 0.9613,
+      "step": 3843
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.655616848139484e-06,
+      "loss": 0.9821,
+      "step": 3844
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.650774069813127e-06,
+      "loss": 0.8596,
+      "step": 3845
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.645931874551258e-06,
+      "loss": 0.9014,
+      "step": 3846
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.641090263555682e-06,
+      "loss": 0.8198,
+      "step": 3847
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.636249238028055e-06,
+      "loss": 0.9772,
+      "step": 3848
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.63140879916989e-06,
+      "loss": 0.9919,
+      "step": 3849
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.626568948182565e-06,
+      "loss": 0.8735,
+      "step": 3850
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.621729686267295e-06,
+      "loss": 0.9974,
+      "step": 3851
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.616891014625164e-06,
+      "loss": 0.9824,
+      "step": 3852
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.612052934457101e-06,
+      "loss": 0.961,
+      "step": 3853
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.607215446963888e-06,
+      "loss": 0.9748,
+      "step": 3854
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.60237855334616e-06,
+      "loss": 0.9575,
+      "step": 3855
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.597542254804411e-06,
+      "loss": 0.9231,
+      "step": 3856
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.592706552538978e-06,
+      "loss": 1.01,
+      "step": 3857
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5878714477500605e-06,
+      "loss": 1.0102,
+      "step": 3858
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5830369416377024e-06,
+      "loss": 1.0111,
+      "step": 3859
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5782030354017975e-06,
+      "loss": 0.9154,
+      "step": 3860
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5733697302420925e-06,
+      "loss": 0.9536,
+      "step": 3861
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.568537027358192e-06,
+      "loss": 1.1138,
+      "step": 3862
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.563704927949539e-06,
+      "loss": 0.8942,
+      "step": 3863
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.558873433215439e-06,
+      "loss": 0.9694,
+      "step": 3864
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.554042544355039e-06,
+      "loss": 1.0093,
+      "step": 3865
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5492122625673404e-06,
+      "loss": 0.9756,
+      "step": 3866
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.544382589051185e-06,
+      "loss": 1.1514,
+      "step": 3867
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.539553525005275e-06,
+      "loss": 0.9738,
+      "step": 3868
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5347250716281554e-06,
+      "loss": 1.009,
+      "step": 3869
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.529897230118223e-06,
+      "loss": 0.9346,
+      "step": 3870
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.525070001673716e-06,
+      "loss": 0.935,
+      "step": 3871
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.520243387492735e-06,
+      "loss": 0.7616,
+      "step": 3872
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.515417388773204e-06,
+      "loss": 0.8867,
+      "step": 3873
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.510592006712918e-06,
+      "loss": 0.9997,
+      "step": 3874
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.5057672425095045e-06,
+      "loss": 1.0128,
+      "step": 3875
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.500943097360447e-06,
+      "loss": 0.9672,
+      "step": 3876
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.496119572463066e-06,
+      "loss": 0.8859,
+      "step": 3877
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.4912966690145404e-06,
+      "loss": 1.0699,
+      "step": 3878
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.486474388211878e-06,
+      "loss": 0.886,
+      "step": 3879
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.481652731251947e-06,
+      "loss": 0.8546,
+      "step": 3880
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.4768316993314525e-06,
+      "loss": 0.9683,
+      "step": 3881
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.47201129364695e-06,
+      "loss": 1.0294,
+      "step": 3882
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.467191515394833e-06,
+      "loss": 0.9792,
+      "step": 3883
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.462372365771351e-06,
+      "loss": 0.9986,
+      "step": 3884
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.457553845972578e-06,
+      "loss": 1.0285,
+      "step": 3885
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.452735957194452e-06,
+      "loss": 0.8725,
+      "step": 3886
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.4479187006327395e-06,
+      "loss": 0.8671,
+      "step": 3887
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.443102077483061e-06,
+      "loss": 1.0478,
+      "step": 3888
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.438286088940871e-06,
+      "loss": 0.8884,
+      "step": 3889
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.433470736201478e-06,
+      "loss": 0.8484,
+      "step": 3890
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.428656020460014e-06,
+      "loss": 0.9598,
+      "step": 3891
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.423841942911472e-06,
+      "loss": 1.0483,
+      "step": 3892
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.419028504750672e-06,
+      "loss": 0.9028,
+      "step": 3893
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.4142157071722895e-06,
+      "loss": 1.0504,
+      "step": 3894
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.409403551370827e-06,
+      "loss": 0.9722,
+      "step": 3895
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.404592038540644e-06,
+      "loss": 0.9579,
+      "step": 3896
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.399781169875917e-06,
+      "loss": 1.0311,
+      "step": 3897
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.394970946570686e-06,
+      "loss": 0.8725,
+      "step": 3898
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.390161369818813e-06,
+      "loss": 0.9832,
+      "step": 3899
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.385352440814017e-06,
+      "loss": 0.9915,
+      "step": 3900
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3805441607498385e-06,
+      "loss": 0.9517,
+      "step": 3901
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.375736530819676e-06,
+      "loss": 1.0106,
+      "step": 3902
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.370929552216743e-06,
+      "loss": 0.9463,
+      "step": 3903
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.36612322613411e-06,
+      "loss": 0.917,
+      "step": 3904
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.361317553764678e-06,
+      "loss": 1.0752,
+      "step": 3905
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.356512536301193e-06,
+      "loss": 0.95,
+      "step": 3906
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.351708174936225e-06,
+      "loss": 0.8523,
+      "step": 3907
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.346904470862199e-06,
+      "loss": 0.924,
+      "step": 3908
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3421014252713554e-06,
+      "loss": 0.9624,
+      "step": 3909
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3372990393557916e-06,
+      "loss": 1.0633,
+      "step": 3910
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.332497314307425e-06,
+      "loss": 0.8874,
+      "step": 3911
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.327696251318025e-06,
+      "loss": 1.0089,
+      "step": 3912
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.32289585157918e-06,
+      "loss": 1.0028,
+      "step": 3913
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.318096116282334e-06,
+      "loss": 0.9127,
+      "step": 3914
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.313297046618737e-06,
+      "loss": 0.9145,
+      "step": 3915
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.308498643779501e-06,
+      "loss": 1.0962,
+      "step": 3916
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.30370090895556e-06,
+      "loss": 1.0164,
+      "step": 3917
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.298903843337686e-06,
+      "loss": 0.8371,
+      "step": 3918
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.294107448116481e-06,
+      "loss": 0.9236,
+      "step": 3919
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.28931172448239e-06,
+      "loss": 0.8847,
+      "step": 3920
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.284516673625672e-06,
+      "loss": 0.9136,
+      "step": 3921
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2797222967364404e-06,
+      "loss": 0.9053,
+      "step": 3922
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2749285950046266e-06,
+      "loss": 1.0835,
+      "step": 3923
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.270135569620005e-06,
+      "loss": 1.002,
+      "step": 3924
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.265343221772174e-06,
+      "loss": 1.0909,
+      "step": 3925
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.260551552650572e-06,
+      "loss": 0.7968,
+      "step": 3926
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.255760563444458e-06,
+      "loss": 0.7685,
+      "step": 3927
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2509702553429305e-06,
+      "loss": 0.974,
+      "step": 3928
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.246180629534913e-06,
+      "loss": 0.8521,
+      "step": 3929
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.241391687209171e-06,
+      "loss": 0.9776,
+      "step": 3930
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2366034295542864e-06,
+      "loss": 0.9606,
+      "step": 3931
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.231815857758682e-06,
+      "loss": 0.9937,
+      "step": 3932
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2270289730106036e-06,
+      "loss": 1.0602,
+      "step": 3933
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2222427764981285e-06,
+      "loss": 0.9886,
+      "step": 3934
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.217457269409161e-06,
+      "loss": 0.9115,
+      "step": 3935
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.21267245293144e-06,
+      "loss": 0.9185,
+      "step": 3936
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.207888328252528e-06,
+      "loss": 1.0208,
+      "step": 3937
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.203104896559819e-06,
+      "loss": 0.9131,
+      "step": 3938
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.198322159040536e-06,
+      "loss": 0.9088,
+      "step": 3939
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.193540116881722e-06,
+      "loss": 0.8912,
+      "step": 3940
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.188758771270252e-06,
+      "loss": 0.9684,
+      "step": 3941
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.183978123392834e-06,
+      "loss": 0.9088,
+      "step": 3942
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.179198174435993e-06,
+      "loss": 0.9574,
+      "step": 3943
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.174418925586088e-06,
+      "loss": 0.9648,
+      "step": 3944
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.169640378029304e-06,
+      "loss": 1.0619,
+      "step": 3945
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.1648625329516405e-06,
+      "loss": 0.9203,
+      "step": 3946
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.160085391538936e-06,
+      "loss": 0.9452,
+      "step": 3947
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.155308954976853e-06,
+      "loss": 0.9491,
+      "step": 3948
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.15053322445087e-06,
+      "loss": 0.9442,
+      "step": 3949
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.145758201146303e-06,
+      "loss": 0.8342,
+      "step": 3950
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.140983886248281e-06,
+      "loss": 1.0335,
+      "step": 3951
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.136210280941762e-06,
+      "loss": 1.0666,
+      "step": 3952
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.131437386411526e-06,
+      "loss": 1.0104,
+      "step": 3953
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.126665203842183e-06,
+      "loss": 0.9001,
+      "step": 3954
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.121893734418155e-06,
+      "loss": 0.9627,
+      "step": 3955
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.117122979323701e-06,
+      "loss": 1.0655,
+      "step": 3956
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.112352939742892e-06,
+      "loss": 0.831,
+      "step": 3957
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.1075836168596235e-06,
+      "loss": 1.0101,
+      "step": 3958
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.102815011857612e-06,
+      "loss": 0.9087,
+      "step": 3959
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.098047125920404e-06,
+      "loss": 1.0317,
+      "step": 3960
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0932799602313565e-06,
+      "loss": 0.8629,
+      "step": 3961
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.088513515973656e-06,
+      "loss": 1.0167,
+      "step": 3962
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.083747794330309e-06,
+      "loss": 1.0151,
+      "step": 3963
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.078982796484135e-06,
+      "loss": 0.9872,
+      "step": 3964
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.07421852361778e-06,
+      "loss": 1.0178,
+      "step": 3965
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.069454976913712e-06,
+      "loss": 0.9882,
+      "step": 3966
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0646921575542125e-06,
+      "loss": 0.9125,
+      "step": 3967
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.059930066721392e-06,
+      "loss": 0.9339,
+      "step": 3968
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.055168705597173e-06,
+      "loss": 0.8363,
+      "step": 3969
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.050408075363293e-06,
+      "loss": 1.0062,
+      "step": 3970
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.045648177201314e-06,
+      "loss": 0.8965,
+      "step": 3971
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.040889012292622e-06,
+      "loss": 1.0344,
+      "step": 3972
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.036130581818408e-06,
+      "loss": 1.1446,
+      "step": 3973
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.031372886959691e-06,
+      "loss": 0.9626,
+      "step": 3974
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.026615928897306e-06,
+      "loss": 0.8967,
+      "step": 3975
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.021859708811898e-06,
+      "loss": 1.0374,
+      "step": 3976
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.017104227883931e-06,
+      "loss": 0.9694,
+      "step": 3977
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.012349487293697e-06,
+      "loss": 0.9254,
+      "step": 3978
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.007595488221287e-06,
+      "loss": 0.935,
+      "step": 3979
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0028422318466235e-06,
+      "loss": 1.058,
+      "step": 3980
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.998089719349437e-06,
+      "loss": 0.9439,
+      "step": 3981
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.993337951909269e-06,
+      "loss": 0.8762,
+      "step": 3982
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.988586930705483e-06,
+      "loss": 1.0065,
+      "step": 3983
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.983836656917258e-06,
+      "loss": 0.8248,
+      "step": 3984
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.979087131723581e-06,
+      "loss": 0.8837,
+      "step": 3985
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.97433835630326e-06,
+      "loss": 0.9641,
+      "step": 3986
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.969590331834915e-06,
+      "loss": 0.9836,
+      "step": 3987
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.964843059496975e-06,
+      "loss": 1.0721,
+      "step": 3988
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.960096540467688e-06,
+      "loss": 0.9524,
+      "step": 3989
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.955350775925113e-06,
+      "loss": 0.9325,
+      "step": 3990
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.950605767047118e-06,
+      "loss": 0.8777,
+      "step": 3991
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.945861515011395e-06,
+      "loss": 1.0447,
+      "step": 3992
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9411180209954366e-06,
+      "loss": 0.7195,
+      "step": 3993
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.936375286176547e-06,
+      "loss": 0.9341,
+      "step": 3994
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.931633311731849e-06,
+      "loss": 0.9893,
+      "step": 3995
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.926892098838276e-06,
+      "loss": 1.046,
+      "step": 3996
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.922151648672565e-06,
+      "loss": 1.0458,
+      "step": 3997
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9174119624112735e-06,
+      "loss": 0.9944,
+      "step": 3998
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.912673041230764e-06,
+      "loss": 0.9386,
+      "step": 3999
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.907934886307206e-06,
+      "loss": 0.9817,
+      "step": 4000
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.903197498816586e-06,
+      "loss": 0.9637,
+      "step": 4001
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.898460879934695e-06,
+      "loss": 0.9301,
+      "step": 4002
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.893725030837134e-06,
+      "loss": 1.0455,
+      "step": 4003
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.888989952699317e-06,
+      "loss": 0.999,
+      "step": 4004
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.884255646696458e-06,
+      "loss": 0.9696,
+      "step": 4005
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.879522114003594e-06,
+      "loss": 0.9564,
+      "step": 4006
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.874789355795552e-06,
+      "loss": 0.9905,
+      "step": 4007
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.87005737324698e-06,
+      "loss": 1.039,
+      "step": 4008
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.8653261675323245e-06,
+      "loss": 1.1029,
+      "step": 4009
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.86059573982585e-06,
+      "loss": 0.9359,
+      "step": 4010
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.855866091301616e-06,
+      "loss": 0.9947,
+      "step": 4011
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.851137223133501e-06,
+      "loss": 0.8665,
+      "step": 4012
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.846409136495178e-06,
+      "loss": 1.0854,
+      "step": 4013
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.841681832560132e-06,
+      "loss": 0.8672,
+      "step": 4014
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.83695531250165e-06,
+      "loss": 0.8925,
+      "step": 4015
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.832229577492834e-06,
+      "loss": 1.0589,
+      "step": 4016
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.8275046287065785e-06,
+      "loss": 1.012,
+      "step": 4017
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.822780467315595e-06,
+      "loss": 0.9006,
+      "step": 4018
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.818057094492388e-06,
+      "loss": 0.9593,
+      "step": 4019
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.8133345114092745e-06,
+      "loss": 1.0131,
+      "step": 4020
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.808612719238369e-06,
+      "loss": 0.8829,
+      "step": 4021
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.803891719151599e-06,
+      "loss": 0.8945,
+      "step": 4022
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.799171512320683e-06,
+      "loss": 0.8718,
+      "step": 4023
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7944520999171596e-06,
+      "loss": 0.9242,
+      "step": 4024
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.78973348311235e-06,
+      "loss": 0.9795,
+      "step": 4025
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7850156630773926e-06,
+      "loss": 0.9863,
+      "step": 4026
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.780298640983221e-06,
+      "loss": 0.9643,
+      "step": 4027
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7755824180005755e-06,
+      "loss": 1.0691,
+      "step": 4028
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.770866995299994e-06,
+      "loss": 0.7846,
+      "step": 4029
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.766152374051824e-06,
+      "loss": 0.9304,
+      "step": 4030
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7614385554261986e-06,
+      "loss": 0.86,
+      "step": 4031
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.756725540593065e-06,
+      "loss": 0.9933,
+      "step": 4032
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.752013330722165e-06,
+      "loss": 0.9762,
+      "step": 4033
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.747301926983043e-06,
+      "loss": 0.9286,
+      "step": 4034
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.742591330545043e-06,
+      "loss": 0.9891,
+      "step": 4035
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.737881542577311e-06,
+      "loss": 0.9892,
+      "step": 4036
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.733172564248785e-06,
+      "loss": 1.0181,
+      "step": 4037
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.728464396728207e-06,
+      "loss": 1.0206,
+      "step": 4038
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.723757041184118e-06,
+      "loss": 1.0238,
+      "step": 4039
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.719050498784858e-06,
+      "loss": 0.8905,
+      "step": 4040
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.714344770698562e-06,
+      "loss": 0.8907,
+      "step": 4041
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.709639858093171e-06,
+      "loss": 0.8218,
+      "step": 4042
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.704935762136409e-06,
+      "loss": 0.9956,
+      "step": 4043
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7002324839958075e-06,
+      "loss": 0.8299,
+      "step": 4044
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.695530024838699e-06,
+      "loss": 0.9647,
+      "step": 4045
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.690828385832202e-06,
+      "loss": 1.0967,
+      "step": 4046
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.686127568143237e-06,
+      "loss": 0.9426,
+      "step": 4047
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.681427572938523e-06,
+      "loss": 0.98,
+      "step": 4048
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.676728401384569e-06,
+      "loss": 1.039,
+      "step": 4049
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.672030054647683e-06,
+      "loss": 0.9146,
+      "step": 4050
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.667332533893969e-06,
+      "loss": 0.8664,
+      "step": 4051
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.662635840289324e-06,
+      "loss": 0.9261,
+      "step": 4052
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.65793997499944e-06,
+      "loss": 1.0393,
+      "step": 4053
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.653244939189809e-06,
+      "loss": 0.8464,
+      "step": 4054
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.648550734025705e-06,
+      "loss": 0.9481,
+      "step": 4055
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.643857360672206e-06,
+      "loss": 0.7923,
+      "step": 4056
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.63916482029418e-06,
+      "loss": 0.8644,
+      "step": 4057
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.63447311405629e-06,
+      "loss": 0.9667,
+      "step": 4058
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.62978224312299e-06,
+      "loss": 0.9005,
+      "step": 4059
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.62509220865853e-06,
+      "loss": 0.8972,
+      "step": 4060
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.620403011826946e-06,
+      "loss": 0.9974,
+      "step": 4061
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.615714653792068e-06,
+      "loss": 0.9506,
+      "step": 4062
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.611027135717526e-06,
+      "loss": 0.9152,
+      "step": 4063
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.606340458766732e-06,
+      "loss": 0.9791,
+      "step": 4064
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.601654624102891e-06,
+      "loss": 0.8298,
+      "step": 4065
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.596969632889006e-06,
+      "loss": 0.8815,
+      "step": 4066
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.5922854862878595e-06,
+      "loss": 0.9679,
+      "step": 4067
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.587602185462029e-06,
+      "loss": 1.0335,
+      "step": 4068
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.582919731573889e-06,
+      "loss": 0.8992,
+      "step": 4069
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.578238125785594e-06,
+      "loss": 0.928,
+      "step": 4070
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.573557369259091e-06,
+      "loss": 0.9169,
+      "step": 4071
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.5688774631561226e-06,
+      "loss": 1.1127,
+      "step": 4072
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.5641984086382104e-06,
+      "loss": 0.9577,
+      "step": 4073
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.559520206866667e-06,
+      "loss": 0.8998,
+      "step": 4074
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.5548428590026e-06,
+      "loss": 1.0428,
+      "step": 4075
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.550166366206897e-06,
+      "loss": 0.8713,
+      "step": 4076
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.545490729640237e-06,
+      "loss": 0.9788,
+      "step": 4077
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.5408159504630904e-06,
+      "loss": 1.0077,
+      "step": 4078
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.5361420298357105e-06,
+      "loss": 0.9894,
+      "step": 4079
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.531468968918129e-06,
+      "loss": 1.0696,
+      "step": 4080
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.526796768870182e-06,
+      "loss": 0.8271,
+      "step": 4081
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.522125430851482e-06,
+      "loss": 1.1484,
+      "step": 4082
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.517454956021424e-06,
+      "loss": 0.9459,
+      "step": 4083
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.512785345539198e-06,
+      "loss": 0.8867,
+      "step": 4084
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.508116600563774e-06,
+      "loss": 1.01,
+      "step": 4085
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.503448722253902e-06,
+      "loss": 1.0121,
+      "step": 4086
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.498781711768131e-06,
+      "loss": 0.9873,
+      "step": 4087
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.494115570264781e-06,
+      "loss": 0.994,
+      "step": 4088
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.489450298901962e-06,
+      "loss": 0.9673,
+      "step": 4089
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.484785898837572e-06,
+      "loss": 0.9149,
+      "step": 4090
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.480122371229287e-06,
+      "loss": 0.7607,
+      "step": 4091
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.475459717234563e-06,
+      "loss": 0.9357,
+      "step": 4092
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.47079793801065e-06,
+      "loss": 1.0299,
+      "step": 4093
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.466137034714572e-06,
+      "loss": 0.9798,
+      "step": 4094
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.461477008503139e-06,
+      "loss": 1.0062,
+      "step": 4095
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.456817860532944e-06,
+      "loss": 0.8614,
+      "step": 4096
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.452159591960363e-06,
+      "loss": 0.9626,
+      "step": 4097
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.447502203941545e-06,
+      "loss": 0.8944,
+      "step": 4098
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4428456976324325e-06,
+      "loss": 1.0435,
+      "step": 4099
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.438190074188742e-06,
+      "loss": 0.9402,
+      "step": 4100
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.433535334765974e-06,
+      "loss": 0.9691,
+      "step": 4101
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.428881480519409e-06,
+      "loss": 0.9337,
+      "step": 4102
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.424228512604107e-06,
+      "loss": 1.0349,
+      "step": 4103
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4195764321749025e-06,
+      "loss": 0.9578,
+      "step": 4104
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.414925240386422e-06,
+      "loss": 0.9908,
+      "step": 4105
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4102749383930595e-06,
+      "loss": 0.9294,
+      "step": 4106
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.405625527349e-06,
+      "loss": 0.8372,
+      "step": 4107
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.400977008408196e-06,
+      "loss": 0.9465,
+      "step": 4108
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.396329382724387e-06,
+      "loss": 0.9628,
+      "step": 4109
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.391682651451081e-06,
+      "loss": 0.9133,
+      "step": 4110
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.387036815741575e-06,
+      "loss": 0.92,
+      "step": 4111
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.382391876748937e-06,
+      "loss": 1.0153,
+      "step": 4112
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.377747835626018e-06,
+      "loss": 0.9984,
+      "step": 4113
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.373104693525438e-06,
+      "loss": 0.9305,
+      "step": 4114
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.368462451599604e-06,
+      "loss": 0.9315,
+      "step": 4115
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.363821111000685e-06,
+      "loss": 0.875,
+      "step": 4116
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3591806728806434e-06,
+      "loss": 0.878,
+      "step": 4117
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3545411383912045e-06,
+      "loss": 0.831,
+      "step": 4118
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3499025086838785e-06,
+      "loss": 0.8305,
+      "step": 4119
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.345264784909946e-06,
+      "loss": 0.8934,
+      "step": 4120
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.340627968220463e-06,
+      "loss": 0.9973,
+      "step": 4121
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.335992059766256e-06,
+      "loss": 1.0299,
+      "step": 4122
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.331357060697938e-06,
+      "loss": 0.9332,
+      "step": 4123
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.326722972165885e-06,
+      "loss": 1.0298,
+      "step": 4124
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.322089795320254e-06,
+      "loss": 0.8876,
+      "step": 4125
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.317457531310974e-06,
+      "loss": 0.9167,
+      "step": 4126
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.312826181287746e-06,
+      "loss": 0.9845,
+      "step": 4127
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.30819574640004e-06,
+      "loss": 0.9745,
+      "step": 4128
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3035662277971085e-06,
+      "loss": 1.0353,
+      "step": 4129
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.298937626627968e-06,
+      "loss": 0.9005,
+      "step": 4130
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.294309944041415e-06,
+      "loss": 0.9075,
+      "step": 4131
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.289683181186013e-06,
+      "loss": 0.8313,
+      "step": 4132
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.285057339210099e-06,
+      "loss": 0.9995,
+      "step": 4133
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.280432419261774e-06,
+      "loss": 0.9798,
+      "step": 4134
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.275808422488924e-06,
+      "loss": 1.0609,
+      "step": 4135
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.271185350039193e-06,
+      "loss": 1.0398,
+      "step": 4136
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.266563203060008e-06,
+      "loss": 1.0541,
+      "step": 4137
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.261941982698556e-06,
+      "loss": 0.9883,
+      "step": 4138
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.257321690101799e-06,
+      "loss": 0.8989,
+      "step": 4139
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.252702326416461e-06,
+      "loss": 0.8972,
+      "step": 4140
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.24808389278905e-06,
+      "loss": 1.1542,
+      "step": 4141
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.2434663903658275e-06,
+      "loss": 0.9358,
+      "step": 4142
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.238849820292838e-06,
+      "loss": 1.0882,
+      "step": 4143
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.2342341837158855e-06,
+      "loss": 0.8946,
+      "step": 4144
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.229619481780544e-06,
+      "loss": 1.01,
+      "step": 4145
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.225005715632158e-06,
+      "loss": 0.9499,
+      "step": 4146
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.220392886415836e-06,
+      "loss": 0.9971,
+      "step": 4147
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.2157809952764525e-06,
+      "loss": 0.9612,
+      "step": 4148
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.211170043358661e-06,
+      "loss": 1.0434,
+      "step": 4149
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.206560031806868e-06,
+      "loss": 0.8855,
+      "step": 4150
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.20195096176525e-06,
+      "loss": 0.9626,
+      "step": 4151
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.19734283437776e-06,
+      "loss": 1.03,
+      "step": 4152
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.192735650788101e-06,
+      "loss": 0.8904,
+      "step": 4153
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.188129412139752e-06,
+      "loss": 0.8954,
+      "step": 4154
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.183524119575956e-06,
+      "loss": 0.927,
+      "step": 4155
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1789197742397215e-06,
+      "loss": 0.8772,
+      "step": 4156
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.174316377273817e-06,
+      "loss": 1.0405,
+      "step": 4157
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.169713929820786e-06,
+      "loss": 0.9343,
+      "step": 4158
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.165112433022924e-06,
+      "loss": 0.9767,
+      "step": 4159
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.160511888022294e-06,
+      "loss": 1.0688,
+      "step": 4160
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.155912295960733e-06,
+      "loss": 0.8661,
+      "step": 4161
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.151313657979827e-06,
+      "loss": 0.9451,
+      "step": 4162
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.146715975220936e-06,
+      "loss": 1.0329,
+      "step": 4163
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.142119248825178e-06,
+      "loss": 0.8463,
+      "step": 4164
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.137523479933431e-06,
+      "loss": 1.0442,
+      "step": 4165
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.13292866968634e-06,
+      "loss": 1.0285,
+      "step": 4166
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1283348192243134e-06,
+      "loss": 1.0208,
+      "step": 4167
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.123741929687514e-06,
+      "loss": 0.9122,
+      "step": 4168
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.119150002215876e-06,
+      "loss": 1.0846,
+      "step": 4169
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.114559037949088e-06,
+      "loss": 0.9791,
+      "step": 4170
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1099690380266e-06,
+      "loss": 0.9424,
+      "step": 4171
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.105380003587622e-06,
+      "loss": 0.9212,
+      "step": 4172
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.100791935771129e-06,
+      "loss": 0.9635,
+      "step": 4173
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.096204835715851e-06,
+      "loss": 1.0499,
+      "step": 4174
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.091618704560285e-06,
+      "loss": 0.977,
+      "step": 4175
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0870335434426795e-06,
+      "loss": 0.9211,
+      "step": 4176
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.082449353501046e-06,
+      "loss": 0.9238,
+      "step": 4177
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.077866135873149e-06,
+      "loss": 0.9974,
+      "step": 4178
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0732838916965246e-06,
+      "loss": 1.0571,
+      "step": 4179
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.068702622108454e-06,
+      "loss": 0.9906,
+      "step": 4180
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0641223282459875e-06,
+      "loss": 0.9531,
+      "step": 4181
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0595430112459255e-06,
+      "loss": 0.9242,
+      "step": 4182
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0549646722448275e-06,
+      "loss": 0.9546,
+      "step": 4183
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.050387312379009e-06,
+      "loss": 0.97,
+      "step": 4184
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.045810932784551e-06,
+      "loss": 0.9437,
+      "step": 4185
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.041235534597277e-06,
+      "loss": 0.9398,
+      "step": 4186
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.036661118952782e-06,
+      "loss": 0.9119,
+      "step": 4187
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0320876869864095e-06,
+      "loss": 1.0299,
+      "step": 4188
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.027515239833254e-06,
+      "loss": 1.0443,
+      "step": 4189
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.022943778628171e-06,
+      "loss": 0.904,
+      "step": 4190
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.018373304505776e-06,
+      "loss": 0.8659,
+      "step": 4191
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.01380381860043e-06,
+      "loss": 0.9751,
+      "step": 4192
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0092353220462565e-06,
+      "loss": 0.8937,
+      "step": 4193
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.004667815977132e-06,
+      "loss": 0.9078,
+      "step": 4194
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 6.000101301526682e-06,
+      "loss": 0.9374,
+      "step": 4195
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.995535779828286e-06,
+      "loss": 1.0569,
+      "step": 4196
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.990971252015088e-06,
+      "loss": 0.8743,
+      "step": 4197
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.986407719219973e-06,
+      "loss": 0.9208,
+      "step": 4198
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.981845182575585e-06,
+      "loss": 0.9609,
+      "step": 4199
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.977283643214324e-06,
+      "loss": 0.8778,
+      "step": 4200
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.9727231022683305e-06,
+      "loss": 0.8953,
+      "step": 4201
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.968163560869506e-06,
+      "loss": 0.8772,
+      "step": 4202
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.963605020149508e-06,
+      "loss": 0.9471,
+      "step": 4203
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.959047481239735e-06,
+      "loss": 1.0648,
+      "step": 4204
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.954490945271345e-06,
+      "loss": 0.9063,
+      "step": 4205
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.9499354133752475e-06,
+      "loss": 0.9877,
+      "step": 4206
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.945380886682093e-06,
+      "loss": 0.886,
+      "step": 4207
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.940827366322288e-06,
+      "loss": 0.997,
+      "step": 4208
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.936274853425999e-06,
+      "loss": 0.9284,
+      "step": 4209
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.931723349123124e-06,
+      "loss": 0.9233,
+      "step": 4210
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.927172854543328e-06,
+      "loss": 0.8587,
+      "step": 4211
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.922623370816015e-06,
+      "loss": 0.9664,
+      "step": 4212
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.9180748990703425e-06,
+      "loss": 0.9411,
+      "step": 4213
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.913527440435211e-06,
+      "loss": 0.9737,
+      "step": 4214
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.9089809960392775e-06,
+      "loss": 1.0381,
+      "step": 4215
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.9044355670109425e-06,
+      "loss": 1.0693,
+      "step": 4216
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.899891154478354e-06,
+      "loss": 1.1038,
+      "step": 4217
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.895347759569412e-06,
+      "loss": 0.9842,
+      "step": 4218
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.890805383411765e-06,
+      "loss": 1.0508,
+      "step": 4219
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.886264027132794e-06,
+      "loss": 0.8991,
+      "step": 4220
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8817236918596466e-06,
+      "loss": 0.8884,
+      "step": 4221
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.877184378719203e-06,
+      "loss": 0.9424,
+      "step": 4222
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.872646088838099e-06,
+      "loss": 0.938,
+      "step": 4223
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.868108823342707e-06,
+      "loss": 1.0212,
+      "step": 4224
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.863572583359162e-06,
+      "loss": 0.9387,
+      "step": 4225
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.859037370013317e-06,
+      "loss": 1.0731,
+      "step": 4226
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.854503184430798e-06,
+      "loss": 1.0185,
+      "step": 4227
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8499700277369565e-06,
+      "loss": 0.8208,
+      "step": 4228
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.845437901056902e-06,
+      "loss": 1.0556,
+      "step": 4229
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8409068055154774e-06,
+      "loss": 0.9715,
+      "step": 4230
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.836376742237284e-06,
+      "loss": 0.9634,
+      "step": 4231
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.831847712346646e-06,
+      "loss": 0.8474,
+      "step": 4232
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.827319716967645e-06,
+      "loss": 0.9478,
+      "step": 4233
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.822792757224114e-06,
+      "loss": 0.9186,
+      "step": 4234
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.818266834239606e-06,
+      "loss": 0.9331,
+      "step": 4235
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.813741949137434e-06,
+      "loss": 0.7891,
+      "step": 4236
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.809218103040653e-06,
+      "loss": 1.0925,
+      "step": 4237
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.804695297072053e-06,
+      "loss": 0.9186,
+      "step": 4238
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.800173532354163e-06,
+      "loss": 0.9618,
+      "step": 4239
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.795652810009263e-06,
+      "loss": 0.9485,
+      "step": 4240
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.791133131159372e-06,
+      "loss": 0.8719,
+      "step": 4241
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.786614496926252e-06,
+      "loss": 0.8136,
+      "step": 4242
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7820969084314e-06,
+      "loss": 1.0876,
+      "step": 4243
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.777580366796047e-06,
+      "loss": 0.962,
+      "step": 4244
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.773064873141182e-06,
+      "loss": 1.0359,
+      "step": 4245
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.768550428587526e-06,
+      "loss": 0.9671,
+      "step": 4246
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.764037034255531e-06,
+      "loss": 0.8469,
+      "step": 4247
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.759524691265399e-06,
+      "loss": 0.9657,
+      "step": 4248
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.755013400737073e-06,
+      "loss": 0.9809,
+      "step": 4249
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.750503163790225e-06,
+      "loss": 0.9879,
+      "step": 4250
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.745993981544267e-06,
+      "loss": 0.936,
+      "step": 4251
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.741485855118355e-06,
+      "loss": 0.8924,
+      "step": 4252
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.736978785631381e-06,
+      "loss": 0.9971,
+      "step": 4253
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.732472774201977e-06,
+      "loss": 0.8294,
+      "step": 4254
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.727967821948507e-06,
+      "loss": 0.9522,
+      "step": 4255
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7234639299890705e-06,
+      "loss": 0.9401,
+      "step": 4256
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.718961099441509e-06,
+      "loss": 0.9696,
+      "step": 4257
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.714459331423406e-06,
+      "loss": 0.905,
+      "step": 4258
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.709958627052066e-06,
+      "loss": 1.0044,
+      "step": 4259
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.7054589874445435e-06,
+      "loss": 0.9018,
+      "step": 4260
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.7009604137176245e-06,
+      "loss": 1.0084,
+      "step": 4261
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.696462906987827e-06,
+      "loss": 0.9779,
+      "step": 4262
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.691966468371403e-06,
+      "loss": 0.9833,
+      "step": 4263
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.6874710989843465e-06,
+      "loss": 0.9173,
+      "step": 4264
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.6829767999423825e-06,
+      "loss": 0.9617,
+      "step": 4265
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.678483572360973e-06,
+      "loss": 0.7661,
+      "step": 4266
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.673991417355312e-06,
+      "loss": 0.9578,
+      "step": 4267
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.669500336040316e-06,
+      "loss": 0.9293,
+      "step": 4268
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.665010329530656e-06,
+      "loss": 0.9729,
+      "step": 4269
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.660521398940726e-06,
+      "loss": 1.0164,
+      "step": 4270
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.656033545384646e-06,
+      "loss": 0.9316,
+      "step": 4271
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.6515467699762814e-06,
+      "loss": 0.9776,
+      "step": 4272
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.647061073829225e-06,
+      "loss": 1.0119,
+      "step": 4273
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.642576458056801e-06,
+      "loss": 0.8275,
+      "step": 4274
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.638092923772058e-06,
+      "loss": 1.101,
+      "step": 4275
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.63361047208779e-06,
+      "loss": 0.911,
+      "step": 4276
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.629129104116515e-06,
+      "loss": 0.9418,
+      "step": 4277
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.624648820970487e-06,
+      "loss": 1.0396,
+      "step": 4278
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.620169623761683e-06,
+      "loss": 0.9894,
+      "step": 4279
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.61569151360181e-06,
+      "loss": 0.8936,
+      "step": 4280
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.611214491602312e-06,
+      "loss": 0.969,
+      "step": 4281
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.606738558874367e-06,
+      "loss": 0.9747,
+      "step": 4282
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.602263716528868e-06,
+      "loss": 0.9737,
+      "step": 4283
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5977899656764455e-06,
+      "loss": 0.9558,
+      "step": 4284
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.593317307427464e-06,
+      "loss": 1.0399,
+      "step": 4285
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.588845742892015e-06,
+      "loss": 0.969,
+      "step": 4286
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.584375273179903e-06,
+      "loss": 0.9669,
+      "step": 4287
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5799058994006805e-06,
+      "loss": 0.949,
+      "step": 4288
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5754376226636185e-06,
+      "loss": 1.0497,
+      "step": 4289
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.570970444077723e-06,
+      "loss": 1.0516,
+      "step": 4290
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.566504364751717e-06,
+      "loss": 1.0027,
+      "step": 4291
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.562039385794059e-06,
+      "loss": 0.8409,
+      "step": 4292
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.557575508312927e-06,
+      "loss": 0.9564,
+      "step": 4293
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.553112733416235e-06,
+      "loss": 1.0574,
+      "step": 4294
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.548651062211613e-06,
+      "loss": 0.8769,
+      "step": 4295
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5441904958064255e-06,
+      "loss": 0.8885,
+      "step": 4296
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.539731035307759e-06,
+      "loss": 1.0803,
+      "step": 4297
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.535272681822435e-06,
+      "loss": 1.0065,
+      "step": 4298
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.530815436456975e-06,
+      "loss": 0.9598,
+      "step": 4299
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5263593003176495e-06,
+      "loss": 0.9209,
+      "step": 4300
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.521904274510448e-06,
+      "loss": 0.9468,
+      "step": 4301
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.517450360141083e-06,
+      "loss": 0.8848,
+      "step": 4302
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.512997558314987e-06,
+      "loss": 0.8426,
+      "step": 4303
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.508545870137325e-06,
+      "loss": 0.9561,
+      "step": 4304
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5040952967129724e-06,
+      "loss": 0.9973,
+      "step": 4305
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.499645839146546e-06,
+      "loss": 0.8409,
+      "step": 4306
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.495197498542367e-06,
+      "loss": 0.868,
+      "step": 4307
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4907502760044914e-06,
+      "loss": 0.9903,
+      "step": 4308
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.486304172636696e-06,
+      "loss": 0.954,
+      "step": 4309
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.481859189542482e-06,
+      "loss": 0.9664,
+      "step": 4310
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.477415327825064e-06,
+      "loss": 1.1091,
+      "step": 4311
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.472972588587379e-06,
+      "loss": 1.0577,
+      "step": 4312
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.468530972932095e-06,
+      "loss": 0.8125,
+      "step": 4313
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.464090481961597e-06,
+      "loss": 0.9237,
+      "step": 4314
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.459651116777985e-06,
+      "loss": 0.8953,
+      "step": 4315
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.45521287848309e-06,
+      "loss": 0.9273,
+      "step": 4316
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.450775768178451e-06,
+      "loss": 0.9408,
+      "step": 4317
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.446339786965339e-06,
+      "loss": 1.0056,
+      "step": 4318
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.441904935944733e-06,
+      "loss": 0.8268,
+      "step": 4319
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4374712162173425e-06,
+      "loss": 0.9933,
+      "step": 4320
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.433038628883589e-06,
+      "loss": 0.9819,
+      "step": 4321
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.428607175043621e-06,
+      "loss": 0.9264,
+      "step": 4322
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.424176855797296e-06,
+      "loss": 1.0086,
+      "step": 4323
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.419747672244191e-06,
+      "loss": 0.9933,
+      "step": 4324
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.415319625483606e-06,
+      "loss": 0.9907,
+      "step": 4325
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.410892716614562e-06,
+      "loss": 0.8943,
+      "step": 4326
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.4064669467357845e-06,
+      "loss": 0.9479,
+      "step": 4327
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.4020423169457325e-06,
+      "loss": 1.0355,
+      "step": 4328
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.3976188283425654e-06,
+      "loss": 0.9772,
+      "step": 4329
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.393196482024177e-06,
+      "loss": 0.8276,
+      "step": 4330
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.388775279088159e-06,
+      "loss": 0.9534,
+      "step": 4331
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.384355220631834e-06,
+      "loss": 0.8588,
+      "step": 4332
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.3799363077522335e-06,
+      "loss": 0.8568,
+      "step": 4333
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.375518541546113e-06,
+      "loss": 0.8735,
+      "step": 4334
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.371101923109932e-06,
+      "loss": 0.9171,
+      "step": 4335
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.366686453539865e-06,
+      "loss": 0.9283,
+      "step": 4336
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.36227213393181e-06,
+      "loss": 0.9936,
+      "step": 4337
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.357858965381382e-06,
+      "loss": 0.9763,
+      "step": 4338
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.353446948983896e-06,
+      "loss": 0.9277,
+      "step": 4339
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.349036085834395e-06,
+      "loss": 0.878,
+      "step": 4340
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.344626377027624e-06,
+      "loss": 0.9991,
+      "step": 4341
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.340217823658056e-06,
+      "loss": 0.9165,
+      "step": 4342
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.335810426819861e-06,
+      "loss": 0.8517,
+      "step": 4343
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.331404187606931e-06,
+      "loss": 0.8793,
+      "step": 4344
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.326999107112872e-06,
+      "loss": 1.0527,
+      "step": 4345
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.322595186431003e-06,
+      "loss": 0.9178,
+      "step": 4346
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.318192426654349e-06,
+      "loss": 0.8434,
+      "step": 4347
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.313790828875645e-06,
+      "loss": 0.978,
+      "step": 4348
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.309390394187346e-06,
+      "loss": 0.9245,
+      "step": 4349
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.3049911236816185e-06,
+      "loss": 0.8062,
+      "step": 4350
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.3005930184503306e-06,
+      "loss": 0.9443,
+      "step": 4351
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2961960795850694e-06,
+      "loss": 1.0155,
+      "step": 4352
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.291800308177134e-06,
+      "loss": 0.9884,
+      "step": 4353
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.287405705317526e-06,
+      "loss": 0.9322,
+      "step": 4354
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.283012272096957e-06,
+      "loss": 0.7963,
+      "step": 4355
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.278620009605856e-06,
+      "loss": 0.9596,
+      "step": 4356
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.274228918934357e-06,
+      "loss": 0.9826,
+      "step": 4357
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.269839001172309e-06,
+      "loss": 0.9397,
+      "step": 4358
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.265450257409259e-06,
+      "loss": 0.9845,
+      "step": 4359
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.261062688734465e-06,
+      "loss": 0.9644,
+      "step": 4360
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.256676296236901e-06,
+      "loss": 1.0413,
+      "step": 4361
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.252291081005247e-06,
+      "loss": 0.9866,
+      "step": 4362
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.247907044127882e-06,
+      "loss": 0.9127,
+      "step": 4363
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.243524186692904e-06,
+      "loss": 1.1344,
+      "step": 4364
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.239142509788114e-06,
+      "loss": 0.8368,
+      "step": 4365
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.234762014501011e-06,
+      "loss": 0.9948,
+      "step": 4366
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.230382701918821e-06,
+      "loss": 0.989,
+      "step": 4367
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.226004573128455e-06,
+      "loss": 0.9071,
+      "step": 4368
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.221627629216542e-06,
+      "loss": 1.0501,
+      "step": 4369
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.21725187126942e-06,
+      "loss": 0.8232,
+      "step": 4370
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2128773003731205e-06,
+      "loss": 0.8171,
+      "step": 4371
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.208503917613388e-06,
+      "loss": 0.9235,
+      "step": 4372
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.20413172407567e-06,
+      "loss": 1.0223,
+      "step": 4373
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1997607208451275e-06,
+      "loss": 1.0259,
+      "step": 4374
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.195390909006609e-06,
+      "loss": 0.8803,
+      "step": 4375
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.191022289644682e-06,
+      "loss": 0.9114,
+      "step": 4376
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.186654863843615e-06,
+      "loss": 0.8382,
+      "step": 4377
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.182288632687374e-06,
+      "loss": 1.0307,
+      "step": 4378
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.177923597259635e-06,
+      "loss": 0.9938,
+      "step": 4379
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.173559758643771e-06,
+      "loss": 1.1337,
+      "step": 4380
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.169197117922866e-06,
+      "loss": 0.9538,
+      "step": 4381
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.164835676179704e-06,
+      "loss": 1.0061,
+      "step": 4382
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.160475434496769e-06,
+      "loss": 0.9673,
+      "step": 4383
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.156116393956242e-06,
+      "loss": 0.9338,
+      "step": 4384
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1517585556400165e-06,
+      "loss": 0.9961,
+      "step": 4385
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.147401920629686e-06,
+      "loss": 0.8709,
+      "step": 4386
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.143046490006538e-06,
+      "loss": 0.9205,
+      "step": 4387
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.138692264851566e-06,
+      "loss": 0.9729,
+      "step": 4388
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.134339246245469e-06,
+      "loss": 0.9577,
+      "step": 4389
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.129987435268634e-06,
+      "loss": 1.0509,
+      "step": 4390
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.125636833001163e-06,
+      "loss": 0.9045,
+      "step": 4391
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.121287440522843e-06,
+      "loss": 0.898,
+      "step": 4392
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.116939258913172e-06,
+      "loss": 1.0934,
+      "step": 4393
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.112592289251348e-06,
+      "loss": 0.9636,
+      "step": 4394
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.108246532616261e-06,
+      "loss": 0.9437,
+      "step": 4395
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.103901990086498e-06,
+      "loss": 0.9794,
+      "step": 4396
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.099558662740354e-06,
+      "loss": 0.9925,
+      "step": 4397
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.095216551655822e-06,
+      "loss": 0.9149,
+      "step": 4398
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.090875657910582e-06,
+      "loss": 0.9617,
+      "step": 4399
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0865359825820215e-06,
+      "loss": 1.0206,
+      "step": 4400
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.08219752674723e-06,
+      "loss": 0.9705,
+      "step": 4401
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.077860291482976e-06,
+      "loss": 0.8336,
+      "step": 4402
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.073524277865745e-06,
+      "loss": 0.9268,
+      "step": 4403
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.069189486971706e-06,
+      "loss": 0.9335,
+      "step": 4404
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.064855919876731e-06,
+      "loss": 0.9693,
+      "step": 4405
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.060523577656391e-06,
+      "loss": 0.9031,
+      "step": 4406
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.056192461385945e-06,
+      "loss": 1.0771,
+      "step": 4407
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.051862572140347e-06,
+      "loss": 0.8793,
+      "step": 4408
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.047533910994255e-06,
+      "loss": 0.8762,
+      "step": 4409
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.043206479022022e-06,
+      "loss": 1.0422,
+      "step": 4410
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0388802772976835e-06,
+      "loss": 0.9324,
+      "step": 4411
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.034555306894981e-06,
+      "loss": 0.9538,
+      "step": 4412
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.030231568887356e-06,
+      "loss": 0.8654,
+      "step": 4413
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.025909064347923e-06,
+      "loss": 0.9073,
+      "step": 4414
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.021587794349514e-06,
+      "loss": 0.9494,
+      "step": 4415
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.017267759964632e-06,
+      "loss": 0.9739,
+      "step": 4416
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0129489622654934e-06,
+      "loss": 0.8609,
+      "step": 4417
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.008631402324e-06,
+      "loss": 0.8686,
+      "step": 4418
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.004315081211742e-06,
+      "loss": 1.0798,
+      "step": 4419
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.000000000000003e-06,
+      "loss": 0.9766,
+      "step": 4420
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.995686159759764e-06,
+      "loss": 0.9549,
+      "step": 4421
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.991373561561697e-06,
+      "loss": 0.926,
+      "step": 4422
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.987062206476167e-06,
+      "loss": 1.0884,
+      "step": 4423
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.98275209557322e-06,
+      "loss": 0.8334,
+      "step": 4424
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.978443229922606e-06,
+      "loss": 0.9599,
+      "step": 4425
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.974135610593765e-06,
+      "loss": 0.9217,
+      "step": 4426
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.969829238655817e-06,
+      "loss": 0.9096,
+      "step": 4427
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9655241151775776e-06,
+      "loss": 0.9601,
+      "step": 4428
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.961220241227556e-06,
+      "loss": 1.0551,
+      "step": 4429
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9569176178739565e-06,
+      "loss": 0.9386,
+      "step": 4430
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.952616246184654e-06,
+      "loss": 0.9724,
+      "step": 4431
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.948316127227234e-06,
+      "loss": 1.0269,
+      "step": 4432
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.944017262068953e-06,
+      "loss": 0.9761,
+      "step": 4433
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.93971965177677e-06,
+      "loss": 0.9042,
+      "step": 4434
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.93542329741733e-06,
+      "loss": 0.8314,
+      "step": 4435
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.931128200056956e-06,
+      "loss": 1.0118,
+      "step": 4436
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.926834360761672e-06,
+      "loss": 0.9441,
+      "step": 4437
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.922541780597186e-06,
+      "loss": 0.8917,
+      "step": 4438
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.918250460628891e-06,
+      "loss": 0.9308,
+      "step": 4439
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.913960401921861e-06,
+      "loss": 0.9009,
+      "step": 4440
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.90967160554087e-06,
+      "loss": 1.0261,
+      "step": 4441
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.905384072550377e-06,
+      "loss": 1.1984,
+      "step": 4442
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.901097804014513e-06,
+      "loss": 0.8336,
+      "step": 4443
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.896812800997116e-06,
+      "loss": 0.9633,
+      "step": 4444
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.892529064561691e-06,
+      "loss": 0.9126,
+      "step": 4445
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.88824659577144e-06,
+      "loss": 0.9255,
+      "step": 4446
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.883965395689251e-06,
+      "loss": 0.9593,
+      "step": 4447
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.879685465377687e-06,
+      "loss": 0.8824,
+      "step": 4448
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.875406805899004e-06,
+      "loss": 0.9967,
+      "step": 4449
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.871129418315148e-06,
+      "loss": 1.0059,
+      "step": 4450
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.866853303687736e-06,
+      "loss": 1.0399,
+      "step": 4451
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.862578463078073e-06,
+      "loss": 0.9372,
+      "step": 4452
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.858304897547151e-06,
+      "loss": 1.1341,
+      "step": 4453
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.854032608155651e-06,
+      "loss": 0.9862,
+      "step": 4454
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.849761595963923e-06,
+      "loss": 0.8341,
+      "step": 4455
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.845491862032014e-06,
+      "loss": 0.8658,
+      "step": 4456
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.841223407419643e-06,
+      "loss": 1.0121,
+      "step": 4457
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.836956233186217e-06,
+      "loss": 0.9806,
+      "step": 4458
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.832690340390829e-06,
+      "loss": 0.9502,
+      "step": 4459
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.828425730092242e-06,
+      "loss": 0.9529,
+      "step": 4460
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.824162403348912e-06,
+      "loss": 0.8746,
+      "step": 4461
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.819900361218975e-06,
+      "loss": 0.9152,
+      "step": 4462
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.815639604760243e-06,
+      "loss": 0.878,
+      "step": 4463
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.811380135030207e-06,
+      "loss": 1.0235,
+      "step": 4464
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.807121953086047e-06,
+      "loss": 0.8397,
+      "step": 4465
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.802865059984624e-06,
+      "loss": 0.9028,
+      "step": 4466
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.798609456782466e-06,
+      "loss": 0.905,
+      "step": 4467
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.794355144535798e-06,
+      "loss": 0.8419,
+      "step": 4468
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.790102124300507e-06,
+      "loss": 0.8572,
+      "step": 4469
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.785850397132173e-06,
+      "loss": 0.8896,
+      "step": 4470
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7815999640860555e-06,
+      "loss": 0.8758,
+      "step": 4471
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.777350826217079e-06,
+      "loss": 0.8363,
+      "step": 4472
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.773102984579858e-06,
+      "loss": 0.9149,
+      "step": 4473
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.768856440228688e-06,
+      "loss": 0.9108,
+      "step": 4474
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.764611194217532e-06,
+      "loss": 1.0356,
+      "step": 4475
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.760367247600034e-06,
+      "loss": 0.9375,
+      "step": 4476
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.756124601429519e-06,
+      "loss": 0.9397,
+      "step": 4477
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.751883256758994e-06,
+      "loss": 0.9164,
+      "step": 4478
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.747643214641126e-06,
+      "loss": 0.8185,
+      "step": 4479
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.74340447612828e-06,
+      "loss": 0.9098,
+      "step": 4480
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.739167042272474e-06,
+      "loss": 0.8249,
+      "step": 4481
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7349309141254256e-06,
+      "loss": 0.9768,
+      "step": 4482
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.730696092738516e-06,
+      "loss": 0.8414,
+      "step": 4483
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.726462579162799e-06,
+      "loss": 1.0901,
+      "step": 4484
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.722230374449012e-06,
+      "loss": 0.8307,
+      "step": 4485
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.717999479647566e-06,
+      "loss": 1.0546,
+      "step": 4486
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.713769895808543e-06,
+      "loss": 0.9355,
+      "step": 4487
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.709541623981697e-06,
+      "loss": 0.9738,
+      "step": 4488
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.705314665216466e-06,
+      "loss": 0.8368,
+      "step": 4489
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7010890205619566e-06,
+      "loss": 0.955,
+      "step": 4490
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.696864691066951e-06,
+      "loss": 0.9455,
+      "step": 4491
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6926416777799e-06,
+      "loss": 0.9133,
+      "step": 4492
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.688419981748937e-06,
+      "loss": 0.8948,
+      "step": 4493
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.684199604021855e-06,
+      "loss": 0.8947,
+      "step": 4494
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6799805456461355e-06,
+      "loss": 0.9932,
+      "step": 4495
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.675762807668918e-06,
+      "loss": 0.9557,
+      "step": 4496
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.671546391137024e-06,
+      "loss": 0.9369,
+      "step": 4497
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6673312970969485e-06,
+      "loss": 1.0119,
+      "step": 4498
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.663117526594849e-06,
+      "loss": 1.025,
+      "step": 4499
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.658905080676557e-06,
+      "loss": 0.9411,
+      "step": 4500
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.654693960387579e-06,
+      "loss": 1.0863,
+      "step": 4501
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.650484166773095e-06,
+      "loss": 0.9233,
+      "step": 4502
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.646275700877952e-06,
+      "loss": 0.9192,
+      "step": 4503
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.64206856374666e-06,
+      "loss": 0.9505,
+      "step": 4504
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.637862756423416e-06,
+      "loss": 1.0383,
+      "step": 4505
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.633658279952068e-06,
+      "loss": 0.8173,
+      "step": 4506
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.629455135376152e-06,
+      "loss": 1.0666,
+      "step": 4507
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.625253323738856e-06,
+      "loss": 1.0034,
+      "step": 4508
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.621052846083048e-06,
+      "loss": 0.9338,
+      "step": 4509
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6168537034512704e-06,
+      "loss": 0.903,
+      "step": 4510
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.61265589688572e-06,
+      "loss": 0.8691,
+      "step": 4511
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.608459427428266e-06,
+      "loss": 0.8371,
+      "step": 4512
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.60426429612045e-06,
+      "loss": 0.9188,
+      "step": 4513
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6000705040034796e-06,
+      "loss": 0.985,
+      "step": 4514
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.595878052118236e-06,
+      "loss": 0.9907,
+      "step": 4515
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.591686941505255e-06,
+      "loss": 0.9736,
+      "step": 4516
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.587497173204751e-06,
+      "loss": 0.8686,
+      "step": 4517
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.583308748256594e-06,
+      "loss": 0.9127,
+      "step": 4518
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.579121667700336e-06,
+      "loss": 0.9131,
+      "step": 4519
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.574935932575177e-06,
+      "loss": 0.9946,
+      "step": 4520
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.570751543919999e-06,
+      "loss": 0.8814,
+      "step": 4521
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.566568502773345e-06,
+      "loss": 0.8873,
+      "step": 4522
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.562386810173418e-06,
+      "loss": 1.0086,
+      "step": 4523
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.558206467158088e-06,
+      "loss": 0.996,
+      "step": 4524
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.5540274747648964e-06,
+      "loss": 0.9237,
+      "step": 4525
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.549849834031044e-06,
+      "loss": 0.9665,
+      "step": 4526
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.5456735459934e-06,
+      "loss": 1.0881,
+      "step": 4527
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.541498611688491e-06,
+      "loss": 0.8773,
+      "step": 4528
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.537325032152517e-06,
+      "loss": 0.8665,
+      "step": 4529
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.533152808421329e-06,
+      "loss": 1.0204,
+      "step": 4530
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.528981941530458e-06,
+      "loss": 1.1213,
+      "step": 4531
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.524812432515082e-06,
+      "loss": 1.0523,
+      "step": 4532
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.520644282410052e-06,
+      "loss": 0.9389,
+      "step": 4533
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.516477492249883e-06,
+      "loss": 0.7883,
+      "step": 4534
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.512312063068745e-06,
+      "loss": 0.9133,
+      "step": 4535
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.508147995900471e-06,
+      "loss": 0.9873,
+      "step": 4536
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.503985291778562e-06,
+      "loss": 0.9191,
+      "step": 4537
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4998239517361765e-06,
+      "loss": 0.9632,
+      "step": 4538
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.49566397680614e-06,
+      "loss": 0.9434,
+      "step": 4539
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.491505368020926e-06,
+      "loss": 1.0004,
+      "step": 4540
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.487348126412686e-06,
+      "loss": 0.9986,
+      "step": 4541
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4831922530132165e-06,
+      "loss": 0.9952,
+      "step": 4542
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4790377488539896e-06,
+      "loss": 1.02,
+      "step": 4543
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4748846149661194e-06,
+      "loss": 0.9426,
+      "step": 4544
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.470732852380396e-06,
+      "loss": 0.9666,
+      "step": 4545
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.466582462127262e-06,
+      "loss": 0.8945,
+      "step": 4546
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.462433445236829e-06,
+      "loss": 1.0529,
+      "step": 4547
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.458285802738844e-06,
+      "loss": 0.9965,
+      "step": 4548
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4541395356627374e-06,
+      "loss": 1.0182,
+      "step": 4549
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.449994645037585e-06,
+      "loss": 0.8828,
+      "step": 4550
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.445851131892132e-06,
+      "loss": 1.0057,
+      "step": 4551
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.441708997254768e-06,
+      "loss": 1.0063,
+      "step": 4552
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.437568242153551e-06,
+      "loss": 0.909,
+      "step": 4553
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.43342886761619e-06,
+      "loss": 0.903,
+      "step": 4554
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.429290874670057e-06,
+      "loss": 0.9342,
+      "step": 4555
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.425154264342174e-06,
+      "loss": 0.8898,
+      "step": 4556
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.421019037659226e-06,
+      "loss": 0.9683,
+      "step": 4557
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.416885195647555e-06,
+      "loss": 0.9502,
+      "step": 4558
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.412752739333162e-06,
+      "loss": 1.1017,
+      "step": 4559
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.408621669741685e-06,
+      "loss": 1.0941,
+      "step": 4560
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.404491987898439e-06,
+      "loss": 0.9677,
+      "step": 4561
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.400363694828389e-06,
+      "loss": 0.9717,
+      "step": 4562
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3962367915561564e-06,
+      "loss": 0.8838,
+      "step": 4563
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.392111279106007e-06,
+      "loss": 0.9229,
+      "step": 4564
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.387987158501875e-06,
+      "loss": 1.0194,
+      "step": 4565
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.383864430767346e-06,
+      "loss": 0.9095,
+      "step": 4566
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.379743096925655e-06,
+      "loss": 0.8143,
+      "step": 4567
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.375623157999688e-06,
+      "loss": 0.9648,
+      "step": 4568
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.371504615011995e-06,
+      "loss": 0.9564,
+      "step": 4569
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.367387468984775e-06,
+      "loss": 0.9213,
+      "step": 4570
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.363271720939885e-06,
+      "loss": 0.8325,
+      "step": 4571
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.359157371898823e-06,
+      "loss": 1.0062,
+      "step": 4572
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3550444228827475e-06,
+      "loss": 0.8751,
+      "step": 4573
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.35093287491247e-06,
+      "loss": 0.9312,
+      "step": 4574
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.346822729008459e-06,
+      "loss": 0.9843,
+      "step": 4575
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.342713986190819e-06,
+      "loss": 1.0613,
+      "step": 4576
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.338606647479323e-06,
+      "loss": 0.9032,
+      "step": 4577
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.334500713893391e-06,
+      "loss": 0.9013,
+      "step": 4578
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.330396186452091e-06,
+      "loss": 0.9467,
+      "step": 4579
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.326293066174135e-06,
+      "loss": 0.9082,
+      "step": 4580
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.322191354077903e-06,
+      "loss": 0.7809,
+      "step": 4581
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.318091051181413e-06,
+      "loss": 0.9967,
+      "step": 4582
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3139921585023425e-06,
+      "loss": 0.8935,
+      "step": 4583
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.30989467705801e-06,
+      "loss": 0.9496,
+      "step": 4584
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.305798607865381e-06,
+      "loss": 0.9395,
+      "step": 4585
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.301703951941081e-06,
+      "loss": 1.0036,
+      "step": 4586
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.297610710301384e-06,
+      "loss": 0.9281,
+      "step": 4587
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.293518883962204e-06,
+      "loss": 0.9614,
+      "step": 4588
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.289428473939109e-06,
+      "loss": 0.8749,
+      "step": 4589
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.285339481247322e-06,
+      "loss": 0.9661,
+      "step": 4590
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.281251906901702e-06,
+      "loss": 1.0514,
+      "step": 4591
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.277165751916761e-06,
+      "loss": 0.9179,
+      "step": 4592
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.27308101730666e-06,
+      "loss": 0.8385,
+      "step": 4593
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.268997704085207e-06,
+      "loss": 0.9035,
+      "step": 4594
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.264915813265863e-06,
+      "loss": 0.9381,
+      "step": 4595
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2608353458617244e-06,
+      "loss": 0.9105,
+      "step": 4596
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.256756302885537e-06,
+      "loss": 0.8929,
+      "step": 4597
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2526786853497006e-06,
+      "loss": 0.8537,
+      "step": 4598
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.248602494266257e-06,
+      "loss": 0.9513,
+      "step": 4599
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.24452773064689e-06,
+      "loss": 0.9026,
+      "step": 4600
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.240454395502935e-06,
+      "loss": 0.9216,
+      "step": 4601
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2363824898453746e-06,
+      "loss": 0.9401,
+      "step": 4602
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.232312014684828e-06,
+      "loss": 0.9249,
+      "step": 4603
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.22824297103156e-06,
+      "loss": 0.9811,
+      "step": 4604
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.224175359895489e-06,
+      "loss": 0.9856,
+      "step": 4605
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.220109182286172e-06,
+      "loss": 1.0736,
+      "step": 4606
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.216044439212815e-06,
+      "loss": 1.0297,
+      "step": 4607
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.211981131684262e-06,
+      "loss": 0.8617,
+      "step": 4608
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.207919260708996e-06,
+      "loss": 0.9867,
+      "step": 4609
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.203858827295157e-06,
+      "loss": 0.8911,
+      "step": 4610
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.199799832450522e-06,
+      "loss": 0.9659,
+      "step": 4611
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.195742277182504e-06,
+      "loss": 0.8814,
+      "step": 4612
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.191686162498171e-06,
+      "loss": 0.8913,
+      "step": 4613
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.187631489404229e-06,
+      "loss": 0.793,
+      "step": 4614
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.183578258907022e-06,
+      "loss": 0.9352,
+      "step": 4615
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.179526472012535e-06,
+      "loss": 0.9042,
+      "step": 4616
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.175476129726401e-06,
+      "loss": 0.8838,
+      "step": 4617
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.171427233053893e-06,
+      "loss": 1.0115,
+      "step": 4618
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.167379782999927e-06,
+      "loss": 0.8873,
+      "step": 4619
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.163333780569055e-06,
+      "loss": 0.9799,
+      "step": 4620
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.159289226765466e-06,
+      "loss": 0.9912,
+      "step": 4621
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.155246122593e-06,
+      "loss": 0.8459,
+      "step": 4622
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.151204469055135e-06,
+      "loss": 1.0186,
+      "step": 4623
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1471642671549816e-06,
+      "loss": 0.9228,
+      "step": 4624
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1431255178952955e-06,
+      "loss": 0.9409,
+      "step": 4625
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.139088222278477e-06,
+      "loss": 0.9027,
+      "step": 4626
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1350523813065535e-06,
+      "loss": 0.7928,
+      "step": 4627
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.131017995981198e-06,
+      "loss": 0.838,
+      "step": 4628
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.126985067303723e-06,
+      "loss": 0.9933,
+      "step": 4629
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.122953596275079e-06,
+      "loss": 0.8333,
+      "step": 4630
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.118923583895858e-06,
+      "loss": 0.9668,
+      "step": 4631
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.114895031166277e-06,
+      "loss": 0.9377,
+      "step": 4632
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.110867939086211e-06,
+      "loss": 0.8703,
+      "step": 4633
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.10684230865515e-06,
+      "loss": 1.031,
+      "step": 4634
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1028181408722416e-06,
+      "loss": 1.0426,
+      "step": 4635
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.098795436736255e-06,
+      "loss": 0.8893,
+      "step": 4636
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.094774197245604e-06,
+      "loss": 0.8587,
+      "step": 4637
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.090754423398339e-06,
+      "loss": 0.9665,
+      "step": 4638
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.086736116192151e-06,
+      "loss": 1.0031,
+      "step": 4639
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.082719276624347e-06,
+      "loss": 0.966,
+      "step": 4640
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.078703905691891e-06,
+      "loss": 1.0274,
+      "step": 4641
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.074690004391375e-06,
+      "loss": 0.9408,
+      "step": 4642
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.070677573719031e-06,
+      "loss": 0.7912,
+      "step": 4643
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.066666614670713e-06,
+      "loss": 1.1289,
+      "step": 4644
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.062657128241925e-06,
+      "loss": 0.9496,
+      "step": 4645
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.058649115427793e-06,
+      "loss": 0.9348,
+      "step": 4646
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.054642577223089e-06,
+      "loss": 0.9259,
+      "step": 4647
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.050637514622205e-06,
+      "loss": 1.0543,
+      "step": 4648
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.046633928619182e-06,
+      "loss": 0.8593,
+      "step": 4649
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.042631820207682e-06,
+      "loss": 0.989,
+      "step": 4650
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.038631190381016e-06,
+      "loss": 0.9119,
+      "step": 4651
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.034632040132104e-06,
+      "loss": 1.0107,
+      "step": 4652
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.030634370453516e-06,
+      "loss": 1.0481,
+      "step": 4653
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.026638182337453e-06,
+      "loss": 0.9813,
+      "step": 4654
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.022643476775751e-06,
+      "loss": 0.8978,
+      "step": 4655
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.018650254759863e-06,
+      "loss": 1.0096,
+      "step": 4656
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.014658517280893e-06,
+      "loss": 0.9518,
+      "step": 4657
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.010668265329559e-06,
+      "loss": 1.0101,
+      "step": 4658
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.006679499896227e-06,
+      "loss": 1.112,
+      "step": 4659
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 4.00269222197088e-06,
+      "loss": 0.9895,
+      "step": 4660
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.998706432543139e-06,
+      "loss": 0.9892,
+      "step": 4661
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.994722132602254e-06,
+      "loss": 0.9358,
+      "step": 4662
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.990739323137112e-06,
+      "loss": 0.9208,
+      "step": 4663
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.986758005136216e-06,
+      "loss": 0.9155,
+      "step": 4664
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.982778179587706e-06,
+      "loss": 0.9432,
+      "step": 4665
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9787998474793545e-06,
+      "loss": 0.9423,
+      "step": 4666
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.974823009798564e-06,
+      "loss": 0.9959,
+      "step": 4667
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.970847667532357e-06,
+      "loss": 0.928,
+      "step": 4668
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9668738216673975e-06,
+      "loss": 0.8433,
+      "step": 4669
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.962901473189963e-06,
+      "loss": 0.8634,
+      "step": 4670
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.958930623085976e-06,
+      "loss": 0.9913,
+      "step": 4671
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.954961272340973e-06,
+      "loss": 1.1366,
+      "step": 4672
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.950993421940125e-06,
+      "loss": 0.877,
+      "step": 4673
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.94702707286823e-06,
+      "loss": 0.9209,
+      "step": 4674
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.943062226109721e-06,
+      "loss": 0.9525,
+      "step": 4675
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.939098882648642e-06,
+      "loss": 0.9343,
+      "step": 4676
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9351370434686706e-06,
+      "loss": 0.9674,
+      "step": 4677
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.931176709553116e-06,
+      "loss": 0.9179,
+      "step": 4678
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.927217881884915e-06,
+      "loss": 0.9224,
+      "step": 4679
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.923260561446618e-06,
+      "loss": 0.9343,
+      "step": 4680
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.919304749220416e-06,
+      "loss": 1.09,
+      "step": 4681
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.915350446188113e-06,
+      "loss": 0.9658,
+      "step": 4682
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.911397653331151e-06,
+      "loss": 1.0116,
+      "step": 4683
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9074463716305845e-06,
+      "loss": 0.9136,
+      "step": 4684
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.903496602067102e-06,
+      "loss": 1.0294,
+      "step": 4685
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.899548345621014e-06,
+      "loss": 0.8123,
+      "step": 4686
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.895601603272256e-06,
+      "loss": 0.8961,
+      "step": 4687
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.891656376000389e-06,
+      "loss": 1.0069,
+      "step": 4688
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.887712664784587e-06,
+      "loss": 0.7854,
+      "step": 4689
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.883770470603664e-06,
+      "loss": 0.94,
+      "step": 4690
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.879829794436052e-06,
+      "loss": 0.8415,
+      "step": 4691
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.875890637259796e-06,
+      "loss": 0.9623,
+      "step": 4692
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.871953000052581e-06,
+      "loss": 1.0017,
+      "step": 4693
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.868016883791701e-06,
+      "loss": 0.8749,
+      "step": 4694
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.86408228945408e-06,
+      "loss": 0.9379,
+      "step": 4695
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.860149218016259e-06,
+      "loss": 0.8785,
+      "step": 4696
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8562176704544065e-06,
+      "loss": 0.8589,
+      "step": 4697
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.852287647744308e-06,
+      "loss": 0.8707,
+      "step": 4698
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.848359150861378e-06,
+      "loss": 0.8411,
+      "step": 4699
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.844432180780644e-06,
+      "loss": 0.9104,
+      "step": 4700
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.840506738476754e-06,
+      "loss": 0.9413,
+      "step": 4701
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.836582824923981e-06,
+      "loss": 1.0005,
+      "step": 4702
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.832660441096225e-06,
+      "loss": 0.8716,
+      "step": 4703
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.828739587966992e-06,
+      "loss": 0.9903,
+      "step": 4704
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.824820266509417e-06,
+      "loss": 0.9704,
+      "step": 4705
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.820902477696257e-06,
+      "loss": 1.0783,
+      "step": 4706
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.816986222499884e-06,
+      "loss": 0.9722,
+      "step": 4707
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8130715018922847e-06,
+      "loss": 0.8613,
+      "step": 4708
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8091583168450728e-06,
+      "loss": 0.8489,
+      "step": 4709
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8052466683294796e-06,
+      "loss": 1.061,
+      "step": 4710
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8013365573163586e-06,
+      "loss": 0.8895,
+      "step": 4711
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.797427984776173e-06,
+      "loss": 1.0698,
+      "step": 4712
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.793520951679003e-06,
+      "loss": 0.8258,
+      "step": 4713
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7896154589945578e-06,
+      "loss": 0.9433,
+      "step": 4714
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.78571150769216e-06,
+      "loss": 0.8252,
+      "step": 4715
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.781809098740743e-06,
+      "loss": 0.9403,
+      "step": 4716
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7779082331088645e-06,
+      "loss": 1.0073,
+      "step": 4717
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7740089117647005e-06,
+      "loss": 1.0703,
+      "step": 4718
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7701111356760343e-06,
+      "loss": 0.9911,
+      "step": 4719
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7662149058102783e-06,
+      "loss": 1.0062,
+      "step": 4720
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.762320223134448e-06,
+      "loss": 0.9216,
+      "step": 4721
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.758427088615183e-06,
+      "loss": 0.8888,
+      "step": 4722
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7545355032187437e-06,
+      "loss": 0.9707,
+      "step": 4723
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7506454679109937e-06,
+      "loss": 1.1334,
+      "step": 4724
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7467569836574147e-06,
+      "loss": 0.9946,
+      "step": 4725
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.74287005142311e-06,
+      "loss": 0.8983,
+      "step": 4726
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7389846721727964e-06,
+      "loss": 0.9984,
+      "step": 4727
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7351008468707972e-06,
+      "loss": 0.9644,
+      "step": 4728
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7312185764810592e-06,
+      "loss": 0.7871,
+      "step": 4729
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.727337861967143e-06,
+      "loss": 0.8498,
+      "step": 4730
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.723458704292212e-06,
+      "loss": 0.8113,
+      "step": 4731
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.719581104419059e-06,
+      "loss": 0.9221,
+      "step": 4732
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7157050633100756e-06,
+      "loss": 1.0481,
+      "step": 4733
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.711830581927275e-06,
+      "loss": 0.899,
+      "step": 4734
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7079576612322865e-06,
+      "loss": 1.0048,
+      "step": 4735
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7040863021863425e-06,
+      "loss": 0.9519,
+      "step": 4736
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7002165057502893e-06,
+      "loss": 1.0043,
+      "step": 4737
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.696348272884591e-06,
+      "loss": 1.0402,
+      "step": 4738
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.692481604549325e-06,
+      "loss": 1.0218,
+      "step": 4739
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6886165017041696e-06,
+      "loss": 1.047,
+      "step": 4740
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.684752965308425e-06,
+      "loss": 1.002,
+      "step": 4741
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6808909963210015e-06,
+      "loss": 0.8192,
+      "step": 4742
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6770305957004105e-06,
+      "loss": 1.0227,
+      "step": 4743
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6731717644047906e-06,
+      "loss": 0.939,
+      "step": 4744
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.669314503391872e-06,
+      "loss": 0.9278,
+      "step": 4745
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6654588136190107e-06,
+      "loss": 0.9607,
+      "step": 4746
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6616046960431685e-06,
+      "loss": 0.9289,
+      "step": 4747
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6577521516209134e-06,
+      "loss": 0.8418,
+      "step": 4748
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.65390118130842e-06,
+      "loss": 1.0081,
+      "step": 4749
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6500517860614825e-06,
+      "loss": 1.1475,
+      "step": 4750
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.646203966835502e-06,
+      "loss": 0.9408,
+      "step": 4751
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.642357724585477e-06,
+      "loss": 0.9628,
+      "step": 4752
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6385130602660267e-06,
+      "loss": 1.0198,
+      "step": 4753
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6346699748313796e-06,
+      "loss": 0.9769,
+      "step": 4754
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.630828469235359e-06,
+      "loss": 1.0364,
+      "step": 4755
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6269885444314135e-06,
+      "loss": 0.9248,
+      "step": 4756
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.623150201372582e-06,
+      "loss": 1.0547,
+      "step": 4757
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6193134410115228e-06,
+      "loss": 0.939,
+      "step": 4758
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6154782643005026e-06,
+      "loss": 0.9868,
+      "step": 4759
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.611644672191387e-06,
+      "loss": 0.9503,
+      "step": 4760
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.607812665635647e-06,
+      "loss": 0.8863,
+      "step": 4761
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.603982245584369e-06,
+      "loss": 0.9299,
+      "step": 4762
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.600153412988244e-06,
+      "loss": 1.0446,
+      "step": 4763
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5963261687975614e-06,
+      "loss": 0.8332,
+      "step": 4764
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5925005139622226e-06,
+      "loss": 1.0495,
+      "step": 4765
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5886764494317395e-06,
+      "loss": 1.0454,
+      "step": 4766
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5848539761552125e-06,
+      "loss": 0.9775,
+      "step": 4767
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.581033095081369e-06,
+      "loss": 0.8685,
+      "step": 4768
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5772138071585193e-06,
+      "loss": 1.0114,
+      "step": 4769
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.573396113334594e-06,
+      "loss": 0.9078,
+      "step": 4770
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5695800145571256e-06,
+      "loss": 0.9203,
+      "step": 4771
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5657655117732426e-06,
+      "loss": 1.0248,
+      "step": 4772
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5619526059296884e-06,
+      "loss": 0.8289,
+      "step": 4773
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.558141297972798e-06,
+      "loss": 0.8762,
+      "step": 4774
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5543315888485197e-06,
+      "loss": 1.0059,
+      "step": 4775
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5505234795024048e-06,
+      "loss": 0.8284,
+      "step": 4776
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.546716970879599e-06,
+      "loss": 0.9141,
+      "step": 4777
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5429120639248573e-06,
+      "loss": 0.8385,
+      "step": 4778
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5391087595825404e-06,
+      "loss": 0.9413,
+      "step": 4779
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.5353070587966034e-06,
+      "loss": 1.0498,
+      "step": 4780
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.5315069625106034e-06,
+      "loss": 0.9318,
+      "step": 4781
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.5277084716677055e-06,
+      "loss": 0.9781,
+      "step": 4782
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.523911587210678e-06,
+      "loss": 0.804,
+      "step": 4783
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.520116310081877e-06,
+      "loss": 0.9398,
+      "step": 4784
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.5163226412232774e-06,
+      "loss": 0.7637,
+      "step": 4785
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.51253058157644e-06,
+      "loss": 1.0719,
+      "step": 4786
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.508740132082533e-06,
+      "loss": 0.9214,
+      "step": 4787
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.5049512936823305e-06,
+      "loss": 1.056,
+      "step": 4788
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.5011640673161917e-06,
+      "loss": 0.9501,
+      "step": 4789
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4973784539240895e-06,
+      "loss": 1.0404,
+      "step": 4790
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4935944544455947e-06,
+      "loss": 0.9348,
+      "step": 4791
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4898120698198712e-06,
+      "loss": 0.8541,
+      "step": 4792
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4860313009856806e-06,
+      "loss": 0.905,
+      "step": 4793
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.482252148881393e-06,
+      "loss": 1.0613,
+      "step": 4794
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.478474614444974e-06,
+      "loss": 0.9522,
+      "step": 4795
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.474698698613982e-06,
+      "loss": 1.0138,
+      "step": 4796
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.470924402325583e-06,
+      "loss": 0.8741,
+      "step": 4797
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4671517265165288e-06,
+      "loss": 0.8419,
+      "step": 4798
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.46338067212318e-06,
+      "loss": 0.8884,
+      "step": 4799
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.459611240081494e-06,
+      "loss": 0.8593,
+      "step": 4800
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.455843431327015e-06,
+      "loss": 0.8193,
+      "step": 4801
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.452077246794896e-06,
+      "loss": 0.9947,
+      "step": 4802
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4483126874198857e-06,
+      "loss": 0.9722,
+      "step": 4803
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.444549754136323e-06,
+      "loss": 1.0294,
+      "step": 4804
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4407884478781426e-06,
+      "loss": 0.9212,
+      "step": 4805
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.437028769578883e-06,
+      "loss": 1.0345,
+      "step": 4806
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4332707201716774e-06,
+      "loss": 1.0759,
+      "step": 4807
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.429514300589247e-06,
+      "loss": 0.9275,
+      "step": 4808
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4257595117639196e-06,
+      "loss": 1.0831,
+      "step": 4809
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.422006354627606e-06,
+      "loss": 0.9873,
+      "step": 4810
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4182548301118213e-06,
+      "loss": 0.8626,
+      "step": 4811
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.414504939147677e-06,
+      "loss": 0.9764,
+      "step": 4812
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4107566826658655e-06,
+      "loss": 0.9439,
+      "step": 4813
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4070100615966895e-06,
+      "loss": 0.9781,
+      "step": 4814
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4032650768700392e-06,
+      "loss": 1.0825,
+      "step": 4815
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.399521729415397e-06,
+      "loss": 0.8782,
+      "step": 4816
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3957800201618363e-06,
+      "loss": 0.8457,
+      "step": 4817
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3920399500380307e-06,
+      "loss": 0.992,
+      "step": 4818
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.388301519972249e-06,
+      "loss": 0.9797,
+      "step": 4819
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.384564730892341e-06,
+      "loss": 0.8695,
+      "step": 4820
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3808295837257642e-06,
+      "loss": 0.9271,
+      "step": 4821
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3770960793995523e-06,
+      "loss": 0.9769,
+      "step": 4822
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3733642188403437e-06,
+      "loss": 1.0243,
+      "step": 4823
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.369634002974369e-06,
+      "loss": 0.9444,
+      "step": 4824
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3659054327274387e-06,
+      "loss": 0.8634,
+      "step": 4825
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.362178509024967e-06,
+      "loss": 0.9446,
+      "step": 4826
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.358453232791957e-06,
+      "loss": 0.8934,
+      "step": 4827
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.354729604953001e-06,
+      "loss": 0.8737,
+      "step": 4828
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3510076264322745e-06,
+      "loss": 0.9834,
+      "step": 4829
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3472872981535586e-06,
+      "loss": 0.8239,
+      "step": 4830
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.343568621040214e-06,
+      "loss": 0.8867,
+      "step": 4831
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3398515960152023e-06,
+      "loss": 0.9477,
+      "step": 4832
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3361362240010607e-06,
+      "loss": 0.9842,
+      "step": 4833
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3324225059199234e-06,
+      "loss": 1.0208,
+      "step": 4834
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.328710442693516e-06,
+      "loss": 1.0106,
+      "step": 4835
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3250000352431533e-06,
+      "loss": 1.0054,
+      "step": 4836
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.321291284489733e-06,
+      "loss": 0.9283,
+      "step": 4837
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.317584191353748e-06,
+      "loss": 0.9815,
+      "step": 4838
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3138787567552812e-06,
+      "loss": 0.8265,
+      "step": 4839
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.310174981613997e-06,
+      "loss": 0.9072,
+      "step": 4840
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3064728668491485e-06,
+      "loss": 0.901,
+      "step": 4841
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3027724133795814e-06,
+      "loss": 1.0021,
+      "step": 4842
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2990736221237283e-06,
+      "loss": 1.0465,
+      "step": 4843
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2953764939996124e-06,
+      "loss": 0.9966,
+      "step": 4844
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2916810299248315e-06,
+      "loss": 0.8789,
+      "step": 4845
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.287987230816585e-06,
+      "loss": 1.0353,
+      "step": 4846
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.284295097591648e-06,
+      "loss": 0.8698,
+      "step": 4847
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.280604631166392e-06,
+      "loss": 1.005,
+      "step": 4848
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.276915832456764e-06,
+      "loss": 0.8935,
+      "step": 4849
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.273228702378305e-06,
+      "loss": 0.8665,
+      "step": 4850
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2695432418461447e-06,
+      "loss": 0.8783,
+      "step": 4851
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2658594517749885e-06,
+      "loss": 0.9221,
+      "step": 4852
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.262177333079131e-06,
+      "loss": 1.034,
+      "step": 4853
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2584968866724533e-06,
+      "loss": 0.8476,
+      "step": 4854
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2548181134684233e-06,
+      "loss": 0.9334,
+      "step": 4855
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2511410143800938e-06,
+      "loss": 1.0236,
+      "step": 4856
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2474655903200935e-06,
+      "loss": 0.9219,
+      "step": 4857
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.24379184220065e-06,
+      "loss": 0.8825,
+      "step": 4858
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2401197709335574e-06,
+      "loss": 0.9838,
+      "step": 4859
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.236449377430211e-06,
+      "loss": 0.8774,
+      "step": 4860
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2327806626015744e-06,
+      "loss": 0.9493,
+      "step": 4861
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.229113627358205e-06,
+      "loss": 0.94,
+      "step": 4862
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2254482726102443e-06,
+      "loss": 0.9837,
+      "step": 4863
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.221784599267408e-06,
+      "loss": 0.9922,
+      "step": 4864
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.218122608238995e-06,
+      "loss": 0.9713,
+      "step": 4865
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.214462300433896e-06,
+      "loss": 0.9372,
+      "step": 4866
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2108036767605776e-06,
+      "loss": 0.9044,
+      "step": 4867
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2071467381270927e-06,
+      "loss": 0.9837,
+      "step": 4868
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2034914854410647e-06,
+      "loss": 0.7748,
+      "step": 4869
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.199837919609715e-06,
+      "loss": 1.0146,
+      "step": 4870
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1961860415398303e-06,
+      "loss": 0.9087,
+      "step": 4871
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1925358521377936e-06,
+      "loss": 0.9711,
+      "step": 4872
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.188887352309552e-06,
+      "loss": 1.14,
+      "step": 4873
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.185240542960648e-06,
+      "loss": 0.8937,
+      "step": 4874
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.181595424996201e-06,
+      "loss": 0.9702,
+      "step": 4875
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1779519993209064e-06,
+      "loss": 0.9761,
+      "step": 4876
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.174310266839037e-06,
+      "loss": 0.9255,
+      "step": 4877
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1706702284544543e-06,
+      "loss": 0.9958,
+      "step": 4878
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.167031885070595e-06,
+      "loss": 1.0124,
+      "step": 4879
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1633952375904773e-06,
+      "loss": 0.9722,
+      "step": 4880
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.159760286916691e-06,
+      "loss": 0.9595,
+      "step": 4881
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1561270339514173e-06,
+      "loss": 0.9826,
+      "step": 4882
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1524954795964002e-06,
+      "loss": 0.8955,
+      "step": 4883
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.148865624752979e-06,
+      "loss": 0.8915,
+      "step": 4884
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1452374703220556e-06,
+      "loss": 0.9056,
+      "step": 4885
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1416110172041216e-06,
+      "loss": 0.9703,
+      "step": 4886
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1379862662992443e-06,
+      "loss": 0.9477,
+      "step": 4887
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1343632185070638e-06,
+      "loss": 0.978,
+      "step": 4888
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1307418747267958e-06,
+      "loss": 0.944,
+      "step": 4889
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1271222358572405e-06,
+      "loss": 1.0727,
+      "step": 4890
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1235043027967717e-06,
+      "loss": 0.8418,
+      "step": 4891
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1198880764433437e-06,
+      "loss": 0.9441,
+      "step": 4892
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1162735576944758e-06,
+      "loss": 1.0141,
+      "step": 4893
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.112660747447278e-06,
+      "loss": 0.9494,
+      "step": 4894
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1090496465984234e-06,
+      "loss": 1.0006,
+      "step": 4895
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.105440256044171e-06,
+      "loss": 0.8957,
+      "step": 4896
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.101832576680347e-06,
+      "loss": 0.945,
+      "step": 4897
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.098226609402358e-06,
+      "loss": 0.8372,
+      "step": 4898
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0946223551051857e-06,
+      "loss": 0.8755,
+      "step": 4899
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0910198146833923e-06,
+      "loss": 1.0944,
+      "step": 4900
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0874189890310933e-06,
+      "loss": 1.0239,
+      "step": 4901
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0838198790420016e-06,
+      "loss": 0.9272,
+      "step": 4902
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0802224856093944e-06,
+      "loss": 0.9229,
+      "step": 4903
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0766268096261277e-06,
+      "loss": 0.8154,
+      "step": 4904
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.073032851984621e-06,
+      "loss": 0.9189,
+      "step": 4905
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0694406135768816e-06,
+      "loss": 0.9128,
+      "step": 4906
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0658500952944757e-06,
+      "loss": 0.9506,
+      "step": 4907
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.062261298028555e-06,
+      "loss": 0.8906,
+      "step": 4908
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.058674222669833e-06,
+      "loss": 0.8786,
+      "step": 4909
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.0550888701086047e-06,
+      "loss": 0.8121,
+      "step": 4910
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.0515052412347334e-06,
+      "loss": 0.8234,
+      "step": 4911
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.047923336937659e-06,
+      "loss": 0.9458,
+      "step": 4912
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.0443431581063876e-06,
+      "loss": 0.9556,
+      "step": 4913
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.040764705629494e-06,
+      "loss": 0.9287,
+      "step": 4914
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.0371879803951344e-06,
+      "loss": 0.9058,
+      "step": 4915
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.033612983291033e-06,
+      "loss": 0.8856,
+      "step": 4916
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.03003971520448e-06,
+      "loss": 0.9445,
+      "step": 4917
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.0264681770223416e-06,
+      "loss": 0.94,
+      "step": 4918
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.0228983696310567e-06,
+      "loss": 0.9423,
+      "step": 4919
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.019330293916628e-06,
+      "loss": 0.8832,
+      "step": 4920
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.0157639507646284e-06,
+      "loss": 1.0091,
+      "step": 4921
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.012199341060207e-06,
+      "loss": 0.8918,
+      "step": 4922
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.0086364656880795e-06,
+      "loss": 1.2037,
+      "step": 4923
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.0050753255325347e-06,
+      "loss": 0.8494,
+      "step": 4924
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 3.0015159214774246e-06,
+      "loss": 0.9514,
+      "step": 4925
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.997958254406169e-06,
+      "loss": 1.0429,
+      "step": 4926
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.994402325201764e-06,
+      "loss": 0.9614,
+      "step": 4927
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9908481347467745e-06,
+      "loss": 0.9222,
+      "step": 4928
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9872956839233234e-06,
+      "loss": 0.8904,
+      "step": 4929
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.983744973613112e-06,
+      "loss": 0.9103,
+      "step": 4930
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.98019600469741e-06,
+      "loss": 0.9938,
+      "step": 4931
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9766487780570474e-06,
+      "loss": 0.9606,
+      "step": 4932
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9731032945724234e-06,
+      "loss": 0.8488,
+      "step": 4933
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.969559555123509e-06,
+      "loss": 0.9033,
+      "step": 4934
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.96601756058984e-06,
+      "loss": 0.9465,
+      "step": 4935
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9624773118505244e-06,
+      "loss": 0.8405,
+      "step": 4936
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9589388097842265e-06,
+      "loss": 0.9552,
+      "step": 4937
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.95540205526918e-06,
+      "loss": 1.0055,
+      "step": 4938
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.951867049183189e-06,
+      "loss": 0.8727,
+      "step": 4939
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.948333792403627e-06,
+      "loss": 0.9096,
+      "step": 4940
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.944802285807422e-06,
+      "loss": 0.8931,
+      "step": 4941
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9412725302710753e-06,
+      "loss": 0.9546,
+      "step": 4942
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9377445266706563e-06,
+      "loss": 0.8766,
+      "step": 4943
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9342182758817915e-06,
+      "loss": 0.911,
+      "step": 4944
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.930693778779674e-06,
+      "loss": 0.8582,
+      "step": 4945
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.927171036239066e-06,
+      "loss": 0.9462,
+      "step": 4946
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9236500491342933e-06,
+      "loss": 0.8621,
+      "step": 4947
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9201308183392473e-06,
+      "loss": 0.9892,
+      "step": 4948
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.916613344727378e-06,
+      "loss": 0.903,
+      "step": 4949
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9130976291716994e-06,
+      "loss": 0.9782,
+      "step": 4950
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9095836725447948e-06,
+      "loss": 0.9629,
+      "step": 4951
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.906071475718811e-06,
+      "loss": 0.8275,
+      "step": 4952
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.902561039565449e-06,
+      "loss": 0.9926,
+      "step": 4953
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.899052364955984e-06,
+      "loss": 1.0524,
+      "step": 4954
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8955454527612503e-06,
+      "loss": 1.0119,
+      "step": 4955
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.892040303851641e-06,
+      "loss": 0.998,
+      "step": 4956
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8885369190971104e-06,
+      "loss": 1.0741,
+      "step": 4957
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8850352993671826e-06,
+      "loss": 0.9727,
+      "step": 4958
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8815354455309385e-06,
+      "loss": 0.8983,
+      "step": 4959
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8780373584570264e-06,
+      "loss": 0.9314,
+      "step": 4960
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8745410390136475e-06,
+      "loss": 0.983,
+      "step": 4961
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8710464880685662e-06,
+      "loss": 0.9006,
+      "step": 4962
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8675537064891123e-06,
+      "loss": 1.0545,
+      "step": 4963
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8640626951421767e-06,
+      "loss": 1.069,
+      "step": 4964
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8605734548942045e-06,
+      "loss": 0.824,
+      "step": 4965
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8570859866112066e-06,
+      "loss": 0.9586,
+      "step": 4966
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.853600291158757e-06,
+      "loss": 0.963,
+      "step": 4967
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8501163694019807e-06,
+      "loss": 1.0683,
+      "step": 4968
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8466342222055654e-06,
+      "loss": 0.9107,
+      "step": 4969
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.843153850433764e-06,
+      "loss": 0.9706,
+      "step": 4970
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8396752549503827e-06,
+      "loss": 0.8872,
+      "step": 4971
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8361984366187943e-06,
+      "loss": 0.8335,
+      "step": 4972
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.832723396301921e-06,
+      "loss": 0.9551,
+      "step": 4973
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8292501348622447e-06,
+      "loss": 0.8328,
+      "step": 4974
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.825778653161814e-06,
+      "loss": 0.9363,
+      "step": 4975
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8223089520622314e-06,
+      "loss": 0.9311,
+      "step": 4976
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8188410324246517e-06,
+      "loss": 0.8161,
+      "step": 4977
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.815374895109798e-06,
+      "loss": 0.9149,
+      "step": 4978
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8119105409779455e-06,
+      "loss": 0.9689,
+      "step": 4979
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8084479708889266e-06,
+      "loss": 0.8137,
+      "step": 4980
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.804987185702127e-06,
+      "loss": 0.9057,
+      "step": 4981
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.8015281862764976e-06,
+      "loss": 0.7131,
+      "step": 4982
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7980709734705425e-06,
+      "loss": 1.0627,
+      "step": 4983
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7946155481423244e-06,
+      "loss": 0.9861,
+      "step": 4984
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.791161911149455e-06,
+      "loss": 1.0008,
+      "step": 4985
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.787710063349113e-06,
+      "loss": 0.9889,
+      "step": 4986
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7842600055980208e-06,
+      "loss": 1.049,
+      "step": 4987
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.780811738752469e-06,
+      "loss": 1.0442,
+      "step": 4988
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7773652636682935e-06,
+      "loss": 1.0161,
+      "step": 4989
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7739205812008906e-06,
+      "loss": 0.938,
+      "step": 4990
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7704776922052123e-06,
+      "loss": 1.0207,
+      "step": 4991
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7670365975357704e-06,
+      "loss": 0.8808,
+      "step": 4992
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7635972980466108e-06,
+      "loss": 1.0141,
+      "step": 4993
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.760159794591356e-06,
+      "loss": 0.9155,
+      "step": 4994
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7567240880231748e-06,
+      "loss": 1.0221,
+      "step": 4995
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7532901791947918e-06,
+      "loss": 1.1221,
+      "step": 4996
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7498580689584776e-06,
+      "loss": 1.0094,
+      "step": 4997
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7464277581660703e-06,
+      "loss": 0.897,
+      "step": 4998
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.742999247668945e-06,
+      "loss": 0.9951,
+      "step": 4999
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.739572538318046e-06,
+      "loss": 0.9377,
+      "step": 5000
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7361476309638567e-06,
+      "loss": 0.8793,
+      "step": 5001
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7327245264564215e-06,
+      "loss": 0.8689,
+      "step": 5002
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.729303225645338e-06,
+      "loss": 1.0135,
+      "step": 5003
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7258837293797534e-06,
+      "loss": 1.0031,
+      "step": 5004
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.722466038508368e-06,
+      "loss": 0.9977,
+      "step": 5005
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7190501538794255e-06,
+      "loss": 1.0052,
+      "step": 5006
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7156360763407363e-06,
+      "loss": 1.0422,
+      "step": 5007
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7122238067396557e-06,
+      "loss": 0.9439,
+      "step": 5008
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7088133459230836e-06,
+      "loss": 1.0399,
+      "step": 5009
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7054046947374835e-06,
+      "loss": 0.9828,
+      "step": 5010
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7019978540288573e-06,
+      "loss": 0.8187,
+      "step": 5011
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6985928246427685e-06,
+      "loss": 0.8264,
+      "step": 5012
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6951896074243213e-06,
+      "loss": 0.9671,
+      "step": 5013
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6917882032181774e-06,
+      "loss": 0.8692,
+      "step": 5014
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6883886128685456e-06,
+      "loss": 1.0283,
+      "step": 5015
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.684990837219187e-06,
+      "loss": 1.0441,
+      "step": 5016
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.681594877113409e-06,
+      "loss": 0.928,
+      "step": 5017
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6782007333940665e-06,
+      "loss": 1.0104,
+      "step": 5018
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6748084069035673e-06,
+      "loss": 0.8979,
+      "step": 5019
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6714178984838725e-06,
+      "loss": 0.9942,
+      "step": 5020
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6680292089764805e-06,
+      "loss": 0.7257,
+      "step": 5021
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6646423392224506e-06,
+      "loss": 0.9675,
+      "step": 5022
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.661257290062379e-06,
+      "loss": 0.9438,
+      "step": 5023
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.657874062336421e-06,
+      "loss": 1.0405,
+      "step": 5024
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6544926568842687e-06,
+      "loss": 0.9867,
+      "step": 5025
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.651113074545171e-06,
+      "loss": 0.9277,
+      "step": 5026
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.64773531615792e-06,
+      "loss": 0.9611,
+      "step": 5027
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6443593825608606e-06,
+      "loss": 0.8518,
+      "step": 5028
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6409852745918752e-06,
+      "loss": 0.9067,
+      "step": 5029
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6376129930883976e-06,
+      "loss": 0.8232,
+      "step": 5030
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6342425388874094e-06,
+      "loss": 0.8755,
+      "step": 5031
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6308739128254437e-06,
+      "loss": 0.9879,
+      "step": 5032
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6275071157385656e-06,
+      "loss": 1.0546,
+      "step": 5033
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.624142148462403e-06,
+      "loss": 0.9368,
+      "step": 5034
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6207790118321143e-06,
+      "loss": 0.7848,
+      "step": 5035
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6174177066824182e-06,
+      "loss": 0.8958,
+      "step": 5036
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.614058233847564e-06,
+      "loss": 1.0068,
+      "step": 5037
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6107005941613573e-06,
+      "loss": 1.0463,
+      "step": 5038
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6073447884571456e-06,
+      "loss": 0.8158,
+      "step": 5039
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.603990817567823e-06,
+      "loss": 1.0757,
+      "step": 5040
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.6006386823258244e-06,
+      "loss": 1.0369,
+      "step": 5041
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5972883835631257e-06,
+      "loss": 0.9615,
+      "step": 5042
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.593939922111256e-06,
+      "loss": 1.0302,
+      "step": 5043
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.590593298801287e-06,
+      "loss": 0.8331,
+      "step": 5044
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.587248514463826e-06,
+      "loss": 0.8101,
+      "step": 5045
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5839055699290348e-06,
+      "loss": 0.9839,
+      "step": 5046
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.580564466026607e-06,
+      "loss": 0.9876,
+      "step": 5047
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5772252035857924e-06,
+      "loss": 0.9745,
+      "step": 5048
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5738877834353714e-06,
+      "loss": 0.8152,
+      "step": 5049
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5705522064036748e-06,
+      "loss": 0.8656,
+      "step": 5050
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5672184733185722e-06,
+      "loss": 0.9637,
+      "step": 5051
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5638865850074836e-06,
+      "loss": 0.9693,
+      "step": 5052
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5605565422973588e-06,
+      "loss": 0.913,
+      "step": 5053
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.557228346014694e-06,
+      "loss": 1.0376,
+      "step": 5054
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5539019969855307e-06,
+      "loss": 0.8494,
+      "step": 5055
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5505774960354533e-06,
+      "loss": 0.9414,
+      "step": 5056
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.547254843989577e-06,
+      "loss": 1.0113,
+      "step": 5057
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.54393404167257e-06,
+      "loss": 1.0356,
+      "step": 5058
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.540615089908638e-06,
+      "loss": 0.9175,
+      "step": 5059
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5372979895215233e-06,
+      "loss": 0.9455,
+      "step": 5060
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5339827413345075e-06,
+      "loss": 0.9132,
+      "step": 5061
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.530669346170421e-06,
+      "loss": 1.0902,
+      "step": 5062
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.527357804851629e-06,
+      "loss": 0.9928,
+      "step": 5063
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.524048118200039e-06,
+      "loss": 0.9947,
+      "step": 5064
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5207402870370945e-06,
+      "loss": 0.8634,
+      "step": 5065
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5174343121837775e-06,
+      "loss": 1.057,
+      "step": 5066
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5141301944606147e-06,
+      "loss": 0.8016,
+      "step": 5067
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.510827934687672e-06,
+      "loss": 0.8754,
+      "step": 5068
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.507527533684545e-06,
+      "loss": 0.8754,
+      "step": 5069
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5042289922703787e-06,
+      "loss": 0.9116,
+      "step": 5070
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.500932311263855e-06,
+      "loss": 0.8912,
+      "step": 5071
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.497637491483186e-06,
+      "loss": 1.0338,
+      "step": 5072
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4943445337461302e-06,
+      "loss": 0.8197,
+      "step": 5073
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4910534388699783e-06,
+      "loss": 0.8958,
+      "step": 5074
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4877642076715637e-06,
+      "loss": 0.9509,
+      "step": 5075
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.484476840967256e-06,
+      "loss": 0.9608,
+      "step": 5076
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.481191339572958e-06,
+      "loss": 0.8708,
+      "step": 5077
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4779077043041124e-06,
+      "loss": 0.9852,
+      "step": 5078
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4746259359756987e-06,
+      "loss": 0.9355,
+      "step": 5079
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.471346035402238e-06,
+      "loss": 1.0766,
+      "step": 5080
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4680680033977754e-06,
+      "loss": 0.9598,
+      "step": 5081
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.464791840775903e-06,
+      "loss": 1.0131,
+      "step": 5082
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.46151754834975e-06,
+      "loss": 1.0044,
+      "step": 5083
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4582451269319696e-06,
+      "loss": 0.8797,
+      "step": 5084
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4549745773347654e-06,
+      "loss": 0.9851,
+      "step": 5085
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.451705900369863e-06,
+      "loss": 0.9142,
+      "step": 5086
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4484390968485306e-06,
+      "loss": 0.9752,
+      "step": 5087
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4451741675815753e-06,
+      "loss": 0.9712,
+      "step": 5088
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4419111133793304e-06,
+      "loss": 1.117,
+      "step": 5089
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4386499350516646e-06,
+      "loss": 0.8778,
+      "step": 5090
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.435390633407986e-06,
+      "loss": 0.9438,
+      "step": 5091
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4321332092572368e-06,
+      "loss": 0.7817,
+      "step": 5092
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.428877663407888e-06,
+      "loss": 0.9359,
+      "step": 5093
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4256239966679485e-06,
+      "loss": 0.965,
+      "step": 5094
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.422372209844962e-06,
+      "loss": 0.8882,
+      "step": 5095
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4191223037459997e-06,
+      "loss": 1.0313,
+      "step": 5096
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4158742791776745e-06,
+      "loss": 0.9289,
+      "step": 5097
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4126281369461222e-06,
+      "loss": 0.8872,
+      "step": 5098
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.409383877857019e-06,
+      "loss": 0.9442,
+      "step": 5099
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4061415027155753e-06,
+      "loss": 0.8548,
+      "step": 5100
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4029010123265263e-06,
+      "loss": 0.989,
+      "step": 5101
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.399662407494141e-06,
+      "loss": 0.8747,
+      "step": 5102
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.396425689022226e-06,
+      "loss": 0.968,
+      "step": 5103
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3931908577141174e-06,
+      "loss": 0.8956,
+      "step": 5104
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.389957914372678e-06,
+      "loss": 0.9019,
+      "step": 5105
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.386726859800308e-06,
+      "loss": 0.9118,
+      "step": 5106
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3834976947989397e-06,
+      "loss": 1.0237,
+      "step": 5107
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3802704201700276e-06,
+      "loss": 0.9111,
+      "step": 5108
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.377045036714569e-06,
+      "loss": 0.9573,
+      "step": 5109
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3738215452330795e-06,
+      "loss": 0.8305,
+      "step": 5110
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3705999465256147e-06,
+      "loss": 1.0263,
+      "step": 5111
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.367380241391759e-06,
+      "loss": 0.9888,
+      "step": 5112
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3641624306306233e-06,
+      "loss": 0.9326,
+      "step": 5113
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.360946515040846e-06,
+      "loss": 1.015,
+      "step": 5114
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.357732495420603e-06,
+      "loss": 0.9273,
+      "step": 5115
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.354520372567597e-06,
+      "loss": 0.9355,
+      "step": 5116
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.351310147279053e-06,
+      "loss": 0.8294,
+      "step": 5117
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.348101820351735e-06,
+      "loss": 1.0389,
+      "step": 5118
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.344895392581933e-06,
+      "loss": 0.893,
+      "step": 5119
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.341690864765458e-06,
+      "loss": 0.9226,
+      "step": 5120
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3384882376976626e-06,
+      "loss": 0.924,
+      "step": 5121
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3352875121734142e-06,
+      "loss": 0.9888,
+      "step": 5122
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3320886889871175e-06,
+      "loss": 0.8722,
+      "step": 5123
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.328891768932705e-06,
+      "loss": 0.8186,
+      "step": 5124
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.325696752803629e-06,
+      "loss": 0.9059,
+      "step": 5125
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.322503641392879e-06,
+      "loss": 0.8148,
+      "step": 5126
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.319312435492962e-06,
+      "loss": 0.9807,
+      "step": 5127
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.316123135895919e-06,
+      "loss": 0.8678,
+      "step": 5128
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3129357433933207e-06,
+      "loss": 0.8938,
+      "step": 5129
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.309750258776252e-06,
+      "loss": 0.8634,
+      "step": 5130
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.306566682835336e-06,
+      "loss": 0.938,
+      "step": 5131
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.303385016360721e-06,
+      "loss": 0.8661,
+      "step": 5132
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3002052601420754e-06,
+      "loss": 1.0135,
+      "step": 5133
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2970274149685935e-06,
+      "loss": 0.8945,
+      "step": 5134
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2938514816290014e-06,
+      "loss": 1.0132,
+      "step": 5135
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.29067746091155e-06,
+      "loss": 0.8375,
+      "step": 5136
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2875053536040094e-06,
+      "loss": 0.9052,
+      "step": 5137
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2843351604936812e-06,
+      "loss": 0.9167,
+      "step": 5138
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.281166882367386e-06,
+      "loss": 0.993,
+      "step": 5139
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.278000520011474e-06,
+      "loss": 0.9854,
+      "step": 5140
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2748360742118226e-06,
+      "loss": 0.9615,
+      "step": 5141
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2716735457538206e-06,
+      "loss": 0.8997,
+      "step": 5142
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2685129354223954e-06,
+      "loss": 0.9944,
+      "step": 5143
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2653542440019947e-06,
+      "loss": 0.8074,
+      "step": 5144
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.262197472276583e-06,
+      "loss": 0.8936,
+      "step": 5145
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2590426210296523e-06,
+      "loss": 0.9379,
+      "step": 5146
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.255889691044221e-06,
+      "loss": 0.8584,
+      "step": 5147
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2527386831028297e-06,
+      "loss": 0.958,
+      "step": 5148
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.249589597987537e-06,
+      "loss": 0.9694,
+      "step": 5149
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.246442436479933e-06,
+      "loss": 1.0443,
+      "step": 5150
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2432971993611184e-06,
+      "loss": 0.9862,
+      "step": 5151
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2401538874117278e-06,
+      "loss": 1.0318,
+      "step": 5152
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.237012501411915e-06,
+      "loss": 1.0032,
+      "step": 5153
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.233873042141348e-06,
+      "loss": 1.0199,
+      "step": 5154
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.230735510379227e-06,
+      "loss": 1.0105,
+      "step": 5155
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2275999069042707e-06,
+      "loss": 1.0669,
+      "step": 5156
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2244662324947163e-06,
+      "loss": 0.9042,
+      "step": 5157
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2213344879283215e-06,
+      "loss": 0.8883,
+      "step": 5158
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2182046739823694e-06,
+      "loss": 0.9428,
+      "step": 5159
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2150767914336635e-06,
+      "loss": 0.8898,
+      "step": 5160
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2119508410585232e-06,
+      "loss": 0.8609,
+      "step": 5161
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.208826823632797e-06,
+      "loss": 0.7971,
+      "step": 5162
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.205704739931841e-06,
+      "loss": 0.9936,
+      "step": 5163
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2025845907305423e-06,
+      "loss": 1.0692,
+      "step": 5164
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1994663768033064e-06,
+      "loss": 1.0386,
+      "step": 5165
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1963500989240514e-06,
+      "loss": 1.0054,
+      "step": 5166
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.193235757866222e-06,
+      "loss": 0.9024,
+      "step": 5167
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1901233544027843e-06,
+      "loss": 0.9675,
+      "step": 5168
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.187012889306214e-06,
+      "loss": 0.9825,
+      "step": 5169
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.183904363348508e-06,
+      "loss": 0.9247,
+      "step": 5170
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.180797777301189e-06,
+      "loss": 0.9059,
+      "step": 5171
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.177693131935297e-06,
+      "loss": 0.8836,
+      "step": 5172
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1745904280213803e-06,
+      "loss": 0.8653,
+      "step": 5173
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.171489666329518e-06,
+      "loss": 1.0275,
+      "step": 5174
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.168390847629297e-06,
+      "loss": 0.8274,
+      "step": 5175
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1652939726898283e-06,
+      "loss": 0.8895,
+      "step": 5176
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1621990422797423e-06,
+      "loss": 0.8845,
+      "step": 5177
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1591060571671764e-06,
+      "loss": 0.8873,
+      "step": 5178
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.156015018119795e-06,
+      "loss": 0.8232,
+      "step": 5179
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1529259259047795e-06,
+      "loss": 0.9679,
+      "step": 5180
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1498387812888223e-06,
+      "loss": 0.8467,
+      "step": 5181
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1467535850381317e-06,
+      "loss": 0.9893,
+      "step": 5182
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.143670337918439e-06,
+      "loss": 0.9092,
+      "step": 5183
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.140589040694988e-06,
+      "loss": 0.9435,
+      "step": 5184
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1375096941325434e-06,
+      "loss": 0.9251,
+      "step": 5185
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.134432298995377e-06,
+      "loss": 0.952,
+      "step": 5186
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1313568560472787e-06,
+      "loss": 1.1201,
+      "step": 5187
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.128283366051559e-06,
+      "loss": 0.9297,
+      "step": 5188
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1252118297710423e-06,
+      "loss": 1.0832,
+      "step": 5189
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.122142247968062e-06,
+      "loss": 0.9465,
+      "step": 5190
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1190746214044734e-06,
+      "loss": 0.9663,
+      "step": 5191
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.116008950841646e-06,
+      "loss": 0.9441,
+      "step": 5192
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.11294523704046e-06,
+      "loss": 0.8441,
+      "step": 5193
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.109883480761308e-06,
+      "loss": 0.8315,
+      "step": 5194
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.106823682764103e-06,
+      "loss": 0.8985,
+      "step": 5195
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.103765843808271e-06,
+      "loss": 1.0296,
+      "step": 5196
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1007099646527518e-06,
+      "loss": 0.7732,
+      "step": 5197
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0976560460559913e-06,
+      "loss": 0.9145,
+      "step": 5198
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.094604088775961e-06,
+      "loss": 1.0365,
+      "step": 5199
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.091554093570134e-06,
+      "loss": 1.0205,
+      "step": 5200
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0885060611955056e-06,
+      "loss": 0.9107,
+      "step": 5201
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0854599924085773e-06,
+      "loss": 1.1132,
+      "step": 5202
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0824158879653655e-06,
+      "loss": 0.9535,
+      "step": 5203
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.079373748621405e-06,
+      "loss": 0.794,
+      "step": 5204
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.076333575131734e-06,
+      "loss": 0.9079,
+      "step": 5205
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0732953682509016e-06,
+      "loss": 1.0109,
+      "step": 5206
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0702591287329786e-06,
+      "loss": 0.8779,
+      "step": 5207
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0672248573315424e-06,
+      "loss": 0.9565,
+      "step": 5208
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0641925547996833e-06,
+      "loss": 1.0158,
+      "step": 5209
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0611622218899952e-06,
+      "loss": 1.0181,
+      "step": 5210
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0581338593545976e-06,
+      "loss": 0.9655,
+      "step": 5211
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0551074679451055e-06,
+      "loss": 0.9696,
+      "step": 5212
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0520830484126577e-06,
+      "loss": 0.9782,
+      "step": 5213
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.049060601507893e-06,
+      "loss": 0.8588,
+      "step": 5214
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.046040127980967e-06,
+      "loss": 0.902,
+      "step": 5215
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0430216285815485e-06,
+      "loss": 0.9915,
+      "step": 5216
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0400051040588076e-06,
+      "loss": 0.8872,
+      "step": 5217
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0369905551614266e-06,
+      "loss": 0.8291,
+      "step": 5218
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0339779826376014e-06,
+      "loss": 0.9375,
+      "step": 5219
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0309673872350365e-06,
+      "loss": 0.9146,
+      "step": 5220
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.027958769700946e-06,
+      "loss": 0.9307,
+      "step": 5221
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0249521307820463e-06,
+      "loss": 1.0258,
+      "step": 5222
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0219474712245747e-06,
+      "loss": 0.887,
+      "step": 5223
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0189447917742646e-06,
+      "loss": 1.0488,
+      "step": 5224
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0159440931763673e-06,
+      "loss": 0.9747,
+      "step": 5225
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0129453761756366e-06,
+      "loss": 0.9538,
+      "step": 5226
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0099486415163396e-06,
+      "loss": 0.8999,
+      "step": 5227
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.006953889942248e-06,
+      "loss": 0.8739,
+      "step": 5228
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0039611221966438e-06,
+      "loss": 0.9295,
+      "step": 5229
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.000970339022309e-06,
+      "loss": 1.001,
+      "step": 5230
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.997981541161543e-06,
+      "loss": 0.92,
+      "step": 5231
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.994994729356148e-06,
+      "loss": 0.7468,
+      "step": 5232
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9920099043474373e-06,
+      "loss": 0.9312,
+      "step": 5233
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.98902706687622e-06,
+      "loss": 0.9679,
+      "step": 5234
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9860462176828275e-06,
+      "loss": 0.9049,
+      "step": 5235
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9830673575070815e-06,
+      "loss": 0.9188,
+      "step": 5236
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9800904870883254e-06,
+      "loss": 1.0674,
+      "step": 5237
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.977115607165395e-06,
+      "loss": 0.8513,
+      "step": 5238
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.974142718476643e-06,
+      "loss": 0.7524,
+      "step": 5239
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9711718217599194e-06,
+      "loss": 0.9328,
+      "step": 5240
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9682029177525942e-06,
+      "loss": 0.9118,
+      "step": 5241
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.965236007191519e-06,
+      "loss": 1.035,
+      "step": 5242
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9622710908130692e-06,
+      "loss": 1.0273,
+      "step": 5243
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.959308169353121e-06,
+      "loss": 0.8397,
+      "step": 5244
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9563472435470577e-06,
+      "loss": 0.9436,
+      "step": 5245
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.953388314129757e-06,
+      "loss": 0.7526,
+      "step": 5246
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9504313818356156e-06,
+      "loss": 0.8392,
+      "step": 5247
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.947476447398522e-06,
+      "loss": 0.8749,
+      "step": 5248
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9445235115518767e-06,
+      "loss": 0.9195,
+      "step": 5249
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9415725750285796e-06,
+      "loss": 0.8932,
+      "step": 5250
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.938623638561038e-06,
+      "loss": 1.0005,
+      "step": 5251
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9356767028811594e-06,
+      "loss": 1.0051,
+      "step": 5252
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.932731768720364e-06,
+      "loss": 1.0262,
+      "step": 5253
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.929788836809555e-06,
+      "loss": 0.975,
+      "step": 5254
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.926847907879158e-06,
+      "loss": 0.9163,
+      "step": 5255
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.923908982659095e-06,
+      "loss": 0.9094,
+      "step": 5256
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.920972061878792e-06,
+      "loss": 0.9346,
+      "step": 5257
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.918037146267172e-06,
+      "loss": 0.9259,
+      "step": 5258
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9151042365526685e-06,
+      "loss": 0.8402,
+      "step": 5259
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9121733334632087e-06,
+      "loss": 0.9625,
+      "step": 5260
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.90924443772623e-06,
+      "loss": 0.9649,
+      "step": 5261
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9063175500686637e-06,
+      "loss": 1.0562,
+      "step": 5262
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.903392671216948e-06,
+      "loss": 0.9448,
+      "step": 5263
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9004698018970224e-06,
+      "loss": 0.9062,
+      "step": 5264
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8975489428343286e-06,
+      "loss": 0.8903,
+      "step": 5265
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8946300947538044e-06,
+      "loss": 0.9028,
+      "step": 5266
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.891713258379888e-06,
+      "loss": 0.902,
+      "step": 5267
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.888798434436525e-06,
+      "loss": 1.0032,
+      "step": 5268
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8858856236471612e-06,
+      "loss": 0.9171,
+      "step": 5269
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8829748267347337e-06,
+      "loss": 1.0168,
+      "step": 5270
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8800660444216878e-06,
+      "loss": 0.9419,
+      "step": 5271
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8771592774299696e-06,
+      "loss": 0.8914,
+      "step": 5272
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.874254526481021e-06,
+      "loss": 0.9604,
+      "step": 5273
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8713517922957803e-06,
+      "loss": 0.816,
+      "step": 5274
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8684510755946917e-06,
+      "loss": 0.9352,
+      "step": 5275
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.865552377097698e-06,
+      "loss": 0.9478,
+      "step": 5276
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8626556975242428e-06,
+      "loss": 0.9487,
+      "step": 5277
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8597610375932606e-06,
+      "loss": 0.9927,
+      "step": 5278
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8568683980231883e-06,
+      "loss": 0.9185,
+      "step": 5279
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8539777795319648e-06,
+      "loss": 0.9226,
+      "step": 5280
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.851089182837028e-06,
+      "loss": 0.9668,
+      "step": 5281
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8482026086553063e-06,
+      "loss": 1.0218,
+      "step": 5282
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8453180577032325e-06,
+      "loss": 0.8634,
+      "step": 5283
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8424355306967401e-06,
+      "loss": 0.9134,
+      "step": 5284
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8395550283512509e-06,
+      "loss": 1.1454,
+      "step": 5285
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8366765513816875e-06,
+      "loss": 1.0282,
+      "step": 5286
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8338001005024753e-06,
+      "loss": 0.8769,
+      "step": 5287
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.830925676427532e-06,
+      "loss": 0.9264,
+      "step": 5288
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8280532798702755e-06,
+      "loss": 0.9074,
+      "step": 5289
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8251829115436159e-06,
+      "loss": 0.8914,
+      "step": 5290
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.82231457215996e-06,
+      "loss": 0.779,
+      "step": 5291
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8194482624312148e-06,
+      "loss": 0.9175,
+      "step": 5292
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.816583983068787e-06,
+      "loss": 0.9483,
+      "step": 5293
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8137217347835666e-06,
+      "loss": 1.0365,
+      "step": 5294
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8108615182859525e-06,
+      "loss": 0.8883,
+      "step": 5295
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.808003334285834e-06,
+      "loss": 1.0402,
+      "step": 5296
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8051471834925948e-06,
+      "loss": 0.9236,
+      "step": 5297
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.802293066615113e-06,
+      "loss": 0.9987,
+      "step": 5298
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7994409843617666e-06,
+      "loss": 0.8519,
+      "step": 5299
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7965909374404255e-06,
+      "loss": 0.9784,
+      "step": 5300
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7937429265584572e-06,
+      "loss": 0.8475,
+      "step": 5301
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7908969524227216e-06,
+      "loss": 0.9248,
+      "step": 5302
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7880530157395682e-06,
+      "loss": 1.0655,
+      "step": 5303
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.78521111721485e-06,
+      "loss": 0.9058,
+      "step": 5304
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.782371257553912e-06,
+      "loss": 0.9747,
+      "step": 5305
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7795334374615859e-06,
+      "loss": 0.909,
+      "step": 5306
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7766976576422046e-06,
+      "loss": 0.8684,
+      "step": 5307
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7738639187995966e-06,
+      "loss": 0.7939,
+      "step": 5308
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7710322216370757e-06,
+      "loss": 1.0353,
+      "step": 5309
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7682025668574509e-06,
+      "loss": 0.9337,
+      "step": 5310
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7653749551630294e-06,
+      "loss": 0.8599,
+      "step": 5311
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7625493872556076e-06,
+      "loss": 0.8757,
+      "step": 5312
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7597258638364778e-06,
+      "loss": 1.0577,
+      "step": 5313
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7569043856064205e-06,
+      "loss": 0.9875,
+      "step": 5314
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.754084953265708e-06,
+      "loss": 0.8587,
+      "step": 5315
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7512675675141088e-06,
+      "loss": 0.9605,
+      "step": 5316
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7484522290508865e-06,
+      "loss": 0.9342,
+      "step": 5317
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7456389385747852e-06,
+      "loss": 1.0846,
+      "step": 5318
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7428276967840507e-06,
+      "loss": 0.7728,
+      "step": 5319
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7400185043764196e-06,
+      "loss": 0.9402,
+      "step": 5320
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.737211362049115e-06,
+      "loss": 1.0883,
+      "step": 5321
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7344062704988517e-06,
+      "loss": 1.078,
+      "step": 5322
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7316032304218389e-06,
+      "loss": 1.0167,
+      "step": 5323
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7288022425137763e-06,
+      "loss": 0.865,
+      "step": 5324
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7260033074698546e-06,
+      "loss": 0.9151,
+      "step": 5325
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7232064259847503e-06,
+      "loss": 0.9596,
+      "step": 5326
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7204115987526337e-06,
+      "loss": 1.0082,
+      "step": 5327
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7176188264671655e-06,
+      "loss": 1.0124,
+      "step": 5328
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.714828109821498e-06,
+      "loss": 0.9227,
+      "step": 5329
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7120394495082682e-06,
+      "loss": 0.9181,
+      "step": 5330
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7092528462196057e-06,
+      "loss": 0.8795,
+      "step": 5331
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7064683006471305e-06,
+      "loss": 0.971,
+      "step": 5332
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.703685813481958e-06,
+      "loss": 0.9836,
+      "step": 5333
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7009053854146718e-06,
+      "loss": 1.1025,
+      "step": 5334
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6981270171353658e-06,
+      "loss": 1.0442,
+      "step": 5335
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6953507093336153e-06,
+      "loss": 0.7967,
+      "step": 5336
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6925764626984865e-06,
+      "loss": 0.9947,
+      "step": 5337
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6898042779185253e-06,
+      "loss": 0.9004,
+      "step": 5338
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.687034155681777e-06,
+      "loss": 0.9094,
+      "step": 5339
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6842660966757673e-06,
+      "loss": 0.9854,
+      "step": 5340
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6815001015875166e-06,
+      "loss": 0.9271,
+      "step": 5341
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6787361711035232e-06,
+      "loss": 0.9312,
+      "step": 5342
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6759743059097822e-06,
+      "loss": 1.0175,
+      "step": 5343
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6732145066917737e-06,
+      "loss": 0.8884,
+      "step": 5344
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6704567741344667e-06,
+      "loss": 0.9457,
+      "step": 5345
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6677011089223072e-06,
+      "loss": 0.9544,
+      "step": 5346
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.66494751173924e-06,
+      "loss": 0.8959,
+      "step": 5347
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6621959832686919e-06,
+      "loss": 1.0843,
+      "step": 5348
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6594465241935775e-06,
+      "loss": 0.8324,
+      "step": 5349
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6566991351962947e-06,
+      "loss": 0.8764,
+      "step": 5350
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6539538169587332e-06,
+      "loss": 0.861,
+      "step": 5351
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6512105701622606e-06,
+      "loss": 0.8131,
+      "step": 5352
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6484693954877396e-06,
+      "loss": 0.8664,
+      "step": 5353
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6457302936155095e-06,
+      "loss": 0.9015,
+      "step": 5354
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6429932652254033e-06,
+      "loss": 0.9045,
+      "step": 5355
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6402583109967341e-06,
+      "loss": 0.935,
+      "step": 5356
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6375254316083056e-06,
+      "loss": 1.019,
+      "step": 5357
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6347946277384009e-06,
+      "loss": 0.8787,
+      "step": 5358
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6320659000647866e-06,
+      "loss": 0.9929,
+      "step": 5359
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6293392492647187e-06,
+      "loss": 0.8992,
+      "step": 5360
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6266146760149403e-06,
+      "loss": 0.895,
+      "step": 5361
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6238921809916708e-06,
+      "loss": 0.8921,
+      "step": 5362
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6211717648706216e-06,
+      "loss": 0.8843,
+      "step": 5363
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6184534283269794e-06,
+      "loss": 0.8489,
+      "step": 5364
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6157371720354252e-06,
+      "loss": 0.9173,
+      "step": 5365
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6130229966701139e-06,
+      "loss": 0.9174,
+      "step": 5366
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6103109029046892e-06,
+      "loss": 0.8682,
+      "step": 5367
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6076008914122798e-06,
+      "loss": 1.0307,
+      "step": 5368
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6048929628654953e-06,
+      "loss": 1.1724,
+      "step": 5369
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6021871179364267e-06,
+      "loss": 0.8206,
+      "step": 5370
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5994833572966462e-06,
+      "loss": 0.9352,
+      "step": 5371
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5967816816172154e-06,
+      "loss": 0.9949,
+      "step": 5372
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5940820915686762e-06,
+      "loss": 0.9402,
+      "step": 5373
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5913845878210477e-06,
+      "loss": 1.0072,
+      "step": 5374
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.588689171043838e-06,
+      "loss": 1.0744,
+      "step": 5375
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5859958419060307e-06,
+      "loss": 0.7621,
+      "step": 5376
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5833046010761e-06,
+      "loss": 1.0083,
+      "step": 5377
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5806154492219906e-06,
+      "loss": 0.9899,
+      "step": 5378
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5779283870111373e-06,
+      "loss": 0.8048,
+      "step": 5379
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5752434151104533e-06,
+      "loss": 1.0283,
+      "step": 5380
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.572560534186337e-06,
+      "loss": 0.8255,
+      "step": 5381
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.569879744904661e-06,
+      "loss": 0.8831,
+      "step": 5382
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5672010479307786e-06,
+      "loss": 1.132,
+      "step": 5383
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5645244439295315e-06,
+      "loss": 0.9323,
+      "step": 5384
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5618499335652382e-06,
+      "loss": 0.8729,
+      "step": 5385
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5591775175016933e-06,
+      "loss": 0.8829,
+      "step": 5386
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5565071964021783e-06,
+      "loss": 0.9259,
+      "step": 5387
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5538389709294488e-06,
+      "loss": 0.9489,
+      "step": 5388
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5511728417457472e-06,
+      "loss": 1.0106,
+      "step": 5389
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5485088095127876e-06,
+      "loss": 0.9209,
+      "step": 5390
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5458468748917689e-06,
+      "loss": 1.1079,
+      "step": 5391
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.543187038543368e-06,
+      "loss": 0.9046,
+      "step": 5392
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5405293011277434e-06,
+      "loss": 0.9315,
+      "step": 5393
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5378736633045289e-06,
+      "loss": 0.9972,
+      "step": 5394
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.535220125732837e-06,
+      "loss": 0.9497,
+      "step": 5395
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.532568689071261e-06,
+      "loss": 1.0542,
+      "step": 5396
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5299193539778756e-06,
+      "loss": 0.926,
+      "step": 5397
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5272721211102258e-06,
+      "loss": 0.9451,
+      "step": 5398
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5246269911253454e-06,
+      "loss": 0.9159,
+      "step": 5399
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.521983964679734e-06,
+      "loss": 0.852,
+      "step": 5400
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.519343042429382e-06,
+      "loss": 0.951,
+      "step": 5401
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5167042250297459e-06,
+      "loss": 0.9345,
+      "step": 5402
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5140675131357684e-06,
+      "loss": 0.9075,
+      "step": 5403
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5114329074018651e-06,
+      "loss": 0.9806,
+      "step": 5404
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.508800408481933e-06,
+      "loss": 0.972,
+      "step": 5405
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5061700170293402e-06,
+      "loss": 0.8361,
+      "step": 5406
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5035417336969338e-06,
+      "loss": 0.837,
+      "step": 5407
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5009155591370405e-06,
+      "loss": 1.0219,
+      "step": 5408
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4982914940014647e-06,
+      "loss": 0.9219,
+      "step": 5409
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4956695389414788e-06,
+      "loss": 0.9236,
+      "step": 5410
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.49304969460784e-06,
+      "loss": 1.0069,
+      "step": 5411
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.490431961650781e-06,
+      "loss": 0.9969,
+      "step": 5412
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.487816340720003e-06,
+      "loss": 0.9896,
+      "step": 5413
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4852028324646938e-06,
+      "loss": 0.9752,
+      "step": 5414
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.482591437533507e-06,
+      "loss": 0.9265,
+      "step": 5415
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4799821565745764e-06,
+      "loss": 0.9694,
+      "step": 5416
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4773749902355139e-06,
+      "loss": 0.8792,
+      "step": 5417
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.474769939163402e-06,
+      "loss": 1.0183,
+      "step": 5418
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4721670040047953e-06,
+      "loss": 0.961,
+      "step": 5419
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4695661854057319e-06,
+      "loss": 0.9326,
+      "step": 5420
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.466967484011721e-06,
+      "loss": 0.8092,
+      "step": 5421
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.464370900467741e-06,
+      "loss": 0.8415,
+      "step": 5422
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4617764354182518e-06,
+      "loss": 0.9509,
+      "step": 5423
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4591840895071863e-06,
+      "loss": 0.9362,
+      "step": 5424
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4565938633779474e-06,
+      "loss": 1.0481,
+      "step": 5425
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4540057576734168e-06,
+      "loss": 0.883,
+      "step": 5426
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4514197730359447e-06,
+      "loss": 1.0656,
+      "step": 5427
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.448835910107359e-06,
+      "loss": 0.9306,
+      "step": 5428
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4462541695289612e-06,
+      "loss": 1.0028,
+      "step": 5429
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4436745519415251e-06,
+      "loss": 0.8673,
+      "step": 5430
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4410970579852923e-06,
+      "loss": 0.968,
+      "step": 5431
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4385216882999854e-06,
+      "loss": 0.9788,
+      "step": 5432
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4359484435247983e-06,
+      "loss": 1.1167,
+      "step": 5433
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4333773242983917e-06,
+      "loss": 0.9694,
+      "step": 5434
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.430808331258905e-06,
+      "loss": 0.8715,
+      "step": 5435
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.428241465043949e-06,
+      "loss": 1.0308,
+      "step": 5436
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4256767262906024e-06,
+      "loss": 0.9649,
+      "step": 5437
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.423114115635421e-06,
+      "loss": 1.0701,
+      "step": 5438
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4205536337144277e-06,
+      "loss": 0.8333,
+      "step": 5439
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4179952811631203e-06,
+      "loss": 0.835,
+      "step": 5440
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4154390586164702e-06,
+      "loss": 1.0381,
+      "step": 5441
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4128849667089161e-06,
+      "loss": 0.8982,
+      "step": 5442
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4103330060743648e-06,
+      "loss": 0.8966,
+      "step": 5443
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4077831773462014e-06,
+      "loss": 0.934,
+      "step": 5444
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4052354811572821e-06,
+      "loss": 1.0828,
+      "step": 5445
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4026899181399245e-06,
+      "loss": 0.8215,
+      "step": 5446
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4001464889259265e-06,
+      "loss": 0.9774,
+      "step": 5447
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.397605194146553e-06,
+      "loss": 1.0356,
+      "step": 5448
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3950660344325374e-06,
+      "loss": 0.9056,
+      "step": 5449
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3925290104140865e-06,
+      "loss": 0.9514,
+      "step": 5450
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3899941227208713e-06,
+      "loss": 0.8852,
+      "step": 5451
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3874613719820396e-06,
+      "loss": 0.9159,
+      "step": 5452
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3849307588262072e-06,
+      "loss": 0.8617,
+      "step": 5453
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3824022838814566e-06,
+      "loss": 0.9543,
+      "step": 5454
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3798759477753365e-06,
+      "loss": 1.0271,
+      "step": 5455
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3773517511348745e-06,
+      "loss": 0.8656,
+      "step": 5456
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3748296945865602e-06,
+      "loss": 0.8212,
+      "step": 5457
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3723097787563533e-06,
+      "loss": 0.8723,
+      "step": 5458
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.369792004269681e-06,
+      "loss": 0.9932,
+      "step": 5459
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3672763717514459e-06,
+      "loss": 1.0273,
+      "step": 5460
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3647628818260072e-06,
+      "loss": 0.7786,
+      "step": 5461
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3622515351172039e-06,
+      "loss": 0.8811,
+      "step": 5462
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3597423322483339e-06,
+      "loss": 0.9622,
+      "step": 5463
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3572352738421679e-06,
+      "loss": 0.8197,
+      "step": 5464
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3547303605209483e-06,
+      "loss": 0.9115,
+      "step": 5465
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3522275929063755e-06,
+      "loss": 0.9402,
+      "step": 5466
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3497269716196204e-06,
+      "loss": 0.8778,
+      "step": 5467
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3472284972813265e-06,
+      "loss": 0.9059,
+      "step": 5468
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3447321705116012e-06,
+      "loss": 0.8999,
+      "step": 5469
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3422379919300167e-06,
+      "loss": 1.0301,
+      "step": 5470
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.339745962155613e-06,
+      "loss": 0.8862,
+      "step": 5471
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3372560818069002e-06,
+      "loss": 1.0342,
+      "step": 5472
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3347683515018539e-06,
+      "loss": 0.9421,
+      "step": 5473
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3322827718579122e-06,
+      "loss": 0.9651,
+      "step": 5474
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3297993434919798e-06,
+      "loss": 1.0531,
+      "step": 5475
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.327318067020431e-06,
+      "loss": 0.868,
+      "step": 5476
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.324838943059107e-06,
+      "loss": 0.927,
+      "step": 5477
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3223619722233083e-06,
+      "loss": 1.003,
+      "step": 5478
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3198871551278092e-06,
+      "loss": 0.8902,
+      "step": 5479
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3174144923868403e-06,
+      "loss": 0.998,
+      "step": 5480
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3149439846141054e-06,
+      "loss": 0.8993,
+      "step": 5481
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3124756324227716e-06,
+      "loss": 0.8916,
+      "step": 5482
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3100094364254666e-06,
+      "loss": 0.8243,
+      "step": 5483
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3075453972342878e-06,
+      "loss": 0.9005,
+      "step": 5484
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3050835154607976e-06,
+      "loss": 0.9738,
+      "step": 5485
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.302623791716019e-06,
+      "loss": 0.9086,
+      "step": 5486
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3001662266104399e-06,
+      "loss": 0.7879,
+      "step": 5487
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2977108207540145e-06,
+      "loss": 0.9261,
+      "step": 5488
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2952575747561647e-06,
+      "loss": 0.9139,
+      "step": 5489
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2928064892257674e-06,
+      "loss": 1.0013,
+      "step": 5490
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2903575647711707e-06,
+      "loss": 0.9024,
+      "step": 5491
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.287910802000182e-06,
+      "loss": 1.0109,
+      "step": 5492
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2854662015200737e-06,
+      "loss": 0.8381,
+      "step": 5493
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2830237639375841e-06,
+      "loss": 0.9677,
+      "step": 5494
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2805834898589098e-06,
+      "loss": 0.9914,
+      "step": 5495
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2781453798897137e-06,
+      "loss": 0.9202,
+      "step": 5496
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.275709434635123e-06,
+      "loss": 0.942,
+      "step": 5497
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.273275654699724e-06,
+      "loss": 0.8777,
+      "step": 5498
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2708440406875634e-06,
+      "loss": 0.8848,
+      "step": 5499
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2684145932021574e-06,
+      "loss": 0.9458,
+      "step": 5500
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2659873128464828e-06,
+      "loss": 1.0541,
+      "step": 5501
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2635622002229731e-06,
+      "loss": 0.9533,
+      "step": 5502
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2611392559335301e-06,
+      "loss": 0.9888,
+      "step": 5503
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2587184805795116e-06,
+      "loss": 0.874,
+      "step": 5504
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2562998747617417e-06,
+      "loss": 0.7549,
+      "step": 5505
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2538834390805087e-06,
+      "loss": 1.0336,
+      "step": 5506
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.251469174135551e-06,
+      "loss": 0.9578,
+      "step": 5507
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2490570805260792e-06,
+      "loss": 0.9558,
+      "step": 5508
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2466471588507633e-06,
+      "loss": 0.9516,
+      "step": 5509
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2442394097077303e-06,
+      "loss": 0.9369,
+      "step": 5510
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2418338336945667e-06,
+      "loss": 0.9001,
+      "step": 5511
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.239430431408325e-06,
+      "loss": 1.111,
+      "step": 5512
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.237029203445519e-06,
+      "loss": 1.0825,
+      "step": 5513
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2346301504021142e-06,
+      "loss": 0.7917,
+      "step": 5514
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2322332728735476e-06,
+      "loss": 0.9231,
+      "step": 5515
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2298385714547056e-06,
+      "loss": 0.9492,
+      "step": 5516
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2274460467399407e-06,
+      "loss": 0.8608,
+      "step": 5517
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2250556993230677e-06,
+      "loss": 0.8848,
+      "step": 5518
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2226675297973524e-06,
+      "loss": 0.9306,
+      "step": 5519
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2202815387555256e-06,
+      "loss": 0.9949,
+      "step": 5520
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2178977267897807e-06,
+      "loss": 1.0045,
+      "step": 5521
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2155160944917632e-06,
+      "loss": 1.0834,
+      "step": 5522
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2131366424525782e-06,
+      "loss": 0.9337,
+      "step": 5523
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2107593712627952e-06,
+      "loss": 0.8428,
+      "step": 5524
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2083842815124403e-06,
+      "loss": 0.9538,
+      "step": 5525
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.206011373790995e-06,
+      "loss": 0.7577,
+      "step": 5526
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.203640648687403e-06,
+      "loss": 0.8585,
+      "step": 5527
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2012721067900624e-06,
+      "loss": 0.8802,
+      "step": 5528
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1989057486868338e-06,
+      "loss": 0.9668,
+      "step": 5529
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1965415749650356e-06,
+      "loss": 0.8449,
+      "step": 5530
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.194179586211438e-06,
+      "loss": 0.8934,
+      "step": 5531
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1918197830122746e-06,
+      "loss": 0.9277,
+      "step": 5532
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.189462165953239e-06,
+      "loss": 0.9885,
+      "step": 5533
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1871067356194744e-06,
+      "loss": 0.8606,
+      "step": 5534
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1847534925955829e-06,
+      "loss": 1.0412,
+      "step": 5535
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1824024374656296e-06,
+      "loss": 0.9251,
+      "step": 5536
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1800535708131322e-06,
+      "loss": 1.0471,
+      "step": 5537
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.177706893221069e-06,
+      "loss": 0.8975,
+      "step": 5538
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1753624052718682e-06,
+      "loss": 0.9663,
+      "step": 5539
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.173020107547418e-06,
+      "loss": 0.9185,
+      "step": 5540
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1706800006290642e-06,
+      "loss": 0.8712,
+      "step": 5541
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1683420850976102e-06,
+      "loss": 0.7826,
+      "step": 5542
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1660063615333096e-06,
+      "loss": 0.8516,
+      "step": 5543
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1636728305158772e-06,
+      "loss": 0.9,
+      "step": 5544
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.161341492624486e-06,
+      "loss": 0.8465,
+      "step": 5545
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1590123484377559e-06,
+      "loss": 0.8546,
+      "step": 5546
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1566853985337668e-06,
+      "loss": 0.9134,
+      "step": 5547
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1543606434900568e-06,
+      "loss": 0.9815,
+      "step": 5548
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1520380838836154e-06,
+      "loss": 1.0613,
+      "step": 5549
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1497177202908926e-06,
+      "loss": 0.8499,
+      "step": 5550
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1473995532877835e-06,
+      "loss": 0.9447,
+      "step": 5551
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1450835834496488e-06,
+      "loss": 0.8851,
+      "step": 5552
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1427698113512952e-06,
+      "loss": 0.9232,
+      "step": 5553
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1404582375669915e-06,
+      "loss": 0.825,
+      "step": 5554
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1381488626704528e-06,
+      "loss": 1.0158,
+      "step": 5555
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1358416872348565e-06,
+      "loss": 0.902,
+      "step": 5556
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.13353671183283e-06,
+      "loss": 0.8672,
+      "step": 5557
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1312339370364534e-06,
+      "loss": 0.9953,
+      "step": 5558
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1289333634172617e-06,
+      "loss": 0.8812,
+      "step": 5559
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.126634991546246e-06,
+      "loss": 1.0329,
+      "step": 5560
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1243388219938478e-06,
+      "loss": 0.7741,
+      "step": 5561
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1220448553299657e-06,
+      "loss": 0.9831,
+      "step": 5562
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1197530921239463e-06,
+      "loss": 0.9593,
+      "step": 5563
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1174635329445948e-06,
+      "loss": 0.8173,
+      "step": 5564
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1151761783601634e-06,
+      "loss": 0.9338,
+      "step": 5565
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1128910289383643e-06,
+      "loss": 1.054,
+      "step": 5566
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1106080852463553e-06,
+      "loss": 1.0351,
+      "step": 5567
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1083273478507506e-06,
+      "loss": 0.7922,
+      "step": 5568
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1060488173176188e-06,
+      "loss": 0.8374,
+      "step": 5569
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.1037724942124772e-06,
+      "loss": 1.057,
+      "step": 5570
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.101498379100293e-06,
+      "loss": 0.9628,
+      "step": 5571
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0992264725454904e-06,
+      "loss": 0.9149,
+      "step": 5572
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0969567751119448e-06,
+      "loss": 0.8895,
+      "step": 5573
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0946892873629833e-06,
+      "loss": 0.8998,
+      "step": 5574
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0924240098613792e-06,
+      "loss": 0.8209,
+      "step": 5575
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0901609431693673e-06,
+      "loss": 0.9487,
+      "step": 5576
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0879000878486211e-06,
+      "loss": 0.8596,
+      "step": 5577
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0856414444602781e-06,
+      "loss": 1.1264,
+      "step": 5578
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.083385013564917e-06,
+      "loss": 0.9158,
+      "step": 5579
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0811307957225713e-06,
+      "loss": 0.9749,
+      "step": 5580
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.078878791492728e-06,
+      "loss": 0.8701,
+      "step": 5581
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0766290014343183e-06,
+      "loss": 0.9652,
+      "step": 5582
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0743814261057283e-06,
+      "loss": 0.9957,
+      "step": 5583
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0721360660647928e-06,
+      "loss": 0.9277,
+      "step": 5584
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.069892921868798e-06,
+      "loss": 0.9999,
+      "step": 5585
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0676519940744822e-06,
+      "loss": 0.9454,
+      "step": 5586
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.065413283238027e-06,
+      "loss": 0.9095,
+      "step": 5587
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0631767899150703e-06,
+      "loss": 0.9139,
+      "step": 5588
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0609425146606933e-06,
+      "loss": 0.8251,
+      "step": 5589
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.058710458029436e-06,
+      "loss": 0.9428,
+      "step": 5590
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0564806205752764e-06,
+      "loss": 0.868,
+      "step": 5591
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0542530028516495e-06,
+      "loss": 0.9962,
+      "step": 5592
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0520276054114388e-06,
+      "loss": 0.8728,
+      "step": 5593
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0498044288069775e-06,
+      "loss": 0.9801,
+      "step": 5594
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0475834735900391e-06,
+      "loss": 1.0515,
+      "step": 5595
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.045364740311855e-06,
+      "loss": 0.8814,
+      "step": 5596
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.043148229523101e-06,
+      "loss": 0.8916,
+      "step": 5597
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0409339417739074e-06,
+      "loss": 1.0841,
+      "step": 5598
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0387218776138407e-06,
+      "loss": 0.8349,
+      "step": 5599
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0365120375919292e-06,
+      "loss": 0.9102,
+      "step": 5600
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0343044222566367e-06,
+      "loss": 0.9291,
+      "step": 5601
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0320990321558855e-06,
+      "loss": 0.8665,
+      "step": 5602
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0298958678370363e-06,
+      "loss": 0.7588,
+      "step": 5603
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0276949298469053e-06,
+      "loss": 0.8942,
+      "step": 5604
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0254962187317518e-06,
+      "loss": 0.8396,
+      "step": 5605
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0232997350372875e-06,
+      "loss": 0.9428,
+      "step": 5606
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0211054793086583e-06,
+      "loss": 1.0037,
+      "step": 5607
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.018913452090472e-06,
+      "loss": 0.8734,
+      "step": 5608
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0167236539267744e-06,
+      "loss": 1.0142,
+      "step": 5609
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0145360853610663e-06,
+      "loss": 1.0044,
+      "step": 5610
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0123507469362826e-06,
+      "loss": 1.0126,
+      "step": 5611
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0101676391948156e-06,
+      "loss": 1.0248,
+      "step": 5612
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0079867626785022e-06,
+      "loss": 1.0555,
+      "step": 5613
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0058081179286216e-06,
+      "loss": 0.9483,
+      "step": 5614
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0036317054858978e-06,
+      "loss": 0.9245,
+      "step": 5615
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0014575258905067e-06,
+      "loss": 1.0666,
+      "step": 5616
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.992855796820678e-07,
+      "loss": 0.873,
+      "step": 5617
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.971158673996473e-07,
+      "loss": 0.7777,
+      "step": 5618
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.949483895817525e-07,
+      "loss": 0.9289,
+      "step": 5619
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.927831467663396e-07,
+      "loss": 0.9307,
+      "step": 5620
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.906201394908088e-07,
+      "loss": 0.9223,
+      "step": 5621
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.884593682920086e-07,
+      "loss": 0.7824,
+      "step": 5622
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.863008337062274e-07,
+      "loss": 0.9819,
+      "step": 5623
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.841445362692025e-07,
+      "loss": 0.7758,
+      "step": 5624
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.81990476516117e-07,
+      "loss": 1.0652,
+      "step": 5625
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.79838654981594e-07,
+      "loss": 1.0198,
+      "step": 5626
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.776890721997e-07,
+      "loss": 0.8849,
+      "step": 5627
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.755417287039536e-07,
+      "loss": 0.8995,
+      "step": 5628
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.73396625027312e-07,
+      "loss": 0.9825,
+      "step": 5629
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.7125376170218e-07,
+      "loss": 0.9529,
+      "step": 5630
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.691131392604013e-07,
+      "loss": 0.8587,
+      "step": 5631
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.669747582332645e-07,
+      "loss": 0.8823,
+      "step": 5632
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.648386191515057e-07,
+      "loss": 0.9675,
+      "step": 5633
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.627047225453057e-07,
+      "loss": 0.8893,
+      "step": 5634
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.605730689442794e-07,
+      "loss": 0.8862,
+      "step": 5635
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.584436588774937e-07,
+      "loss": 1.0541,
+      "step": 5636
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.563164928734593e-07,
+      "loss": 1.0118,
+      "step": 5637
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.541915714601224e-07,
+      "loss": 0.8757,
+      "step": 5638
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.520688951648771e-07,
+      "loss": 0.9944,
+      "step": 5639
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.499484645145595e-07,
+      "loss": 0.9621,
+      "step": 5640
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.478302800354489e-07,
+      "loss": 0.937,
+      "step": 5641
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.457143422532688e-07,
+      "loss": 0.9792,
+      "step": 5642
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.436006516931817e-07,
+      "loss": 0.9897,
+      "step": 5643
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.4148920887979e-07,
+      "loss": 0.893,
+      "step": 5644
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.393800143371445e-07,
+      "loss": 0.8715,
+      "step": 5645
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.372730685887365e-07,
+      "loss": 0.8617,
+      "step": 5646
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.351683721574956e-07,
+      "loss": 0.9241,
+      "step": 5647
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.33065925565797e-07,
+      "loss": 0.799,
+      "step": 5648
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.309657293354569e-07,
+      "loss": 0.9679,
+      "step": 5649
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.288677839877302e-07,
+      "loss": 1.0642,
+      "step": 5650
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.267720900433141e-07,
+      "loss": 1.005,
+      "step": 5651
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.24678648022348e-07,
+      "loss": 1.0149,
+      "step": 5652
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.22587458444415e-07,
+      "loss": 0.919,
+      "step": 5653
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.204985218285356e-07,
+      "loss": 1.0099,
+      "step": 5654
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.184118386931728e-07,
+      "loss": 0.8813,
+      "step": 5655
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.163274095562246e-07,
+      "loss": 0.8994,
+      "step": 5656
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.142452349350383e-07,
+      "loss": 0.9456,
+      "step": 5657
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.121653153463994e-07,
+      "loss": 0.9919,
+      "step": 5658
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.100876513065271e-07,
+      "loss": 0.9031,
+      "step": 5659
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.08012243331089e-07,
+      "loss": 1.0079,
+      "step": 5660
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.059390919351918e-07,
+      "loss": 0.9669,
+      "step": 5661
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.038681976333763e-07,
+      "loss": 1.0569,
+      "step": 5662
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.017995609396257e-07,
+      "loss": 1.0137,
+      "step": 5663
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.99733182367366e-07,
+      "loss": 0.847,
+      "step": 5664
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.976690624294604e-07,
+      "loss": 0.9633,
+      "step": 5665
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.956072016382133e-07,
+      "loss": 1.0126,
+      "step": 5666
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.935476005053645e-07,
+      "loss": 0.9431,
+      "step": 5667
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.914902595420949e-07,
+      "loss": 0.9236,
+      "step": 5668
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.89435179259025e-07,
+      "loss": 0.8345,
+      "step": 5669
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.873823601662168e-07,
+      "loss": 0.8582,
+      "step": 5670
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.85331802773165e-07,
+      "loss": 1.07,
+      "step": 5671
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.832835075888069e-07,
+      "loss": 0.8252,
+      "step": 5672
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.812374751215202e-07,
+      "loss": 0.922,
+      "step": 5673
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.791937058791167e-07,
+      "loss": 0.8595,
+      "step": 5674
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.77152200368846e-07,
+      "loss": 0.8565,
+      "step": 5675
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.751129590974005e-07,
+      "loss": 0.9514,
+      "step": 5676
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.730759825709079e-07,
+      "loss": 1.0142,
+      "step": 5677
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.710412712949356e-07,
+      "loss": 1.0319,
+      "step": 5678
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.690088257744867e-07,
+      "loss": 0.8536,
+      "step": 5679
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.669786465140006e-07,
+      "loss": 0.8283,
+      "step": 5680
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.649507340173569e-07,
+      "loss": 0.9346,
+      "step": 5681
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.629250887878749e-07,
+      "loss": 0.9061,
+      "step": 5682
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.609017113283036e-07,
+      "loss": 0.9372,
+      "step": 5683
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.588806021408358e-07,
+      "loss": 0.9176,
+      "step": 5684
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.568617617271003e-07,
+      "loss": 0.8947,
+      "step": 5685
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.548451905881639e-07,
+      "loss": 0.8146,
+      "step": 5686
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.528308892245218e-07,
+      "loss": 1.0323,
+      "step": 5687
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.508188581361154e-07,
+      "loss": 0.9324,
+      "step": 5688
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.488090978223196e-07,
+      "loss": 0.956,
+      "step": 5689
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.468016087819475e-07,
+      "loss": 0.9672,
+      "step": 5690
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.447963915132418e-07,
+      "loss": 1.0358,
+      "step": 5691
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.427934465138909e-07,
+      "loss": 1.0234,
+      "step": 5692
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.407927742810095e-07,
+      "loss": 0.854,
+      "step": 5693
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.387943753111572e-07,
+      "loss": 1.0174,
+      "step": 5694
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.367982501003213e-07,
+      "loss": 1.096,
+      "step": 5695
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.348043991439314e-07,
+      "loss": 0.9069,
+      "step": 5696
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.328128229368493e-07,
+      "loss": 0.9525,
+      "step": 5697
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.308235219733762e-07,
+      "loss": 0.8799,
+      "step": 5698
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.288364967472384e-07,
+      "loss": 0.9714,
+      "step": 5699
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.268517477516081e-07,
+      "loss": 0.8578,
+      "step": 5700
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.248692754790888e-07,
+      "loss": 0.8613,
+      "step": 5701
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.228890804217194e-07,
+      "loss": 0.8984,
+      "step": 5702
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.20911163070971e-07,
+      "loss": 0.8304,
+      "step": 5703
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.189355239177543e-07,
+      "loss": 1.0309,
+      "step": 5704
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.169621634524072e-07,
+      "loss": 0.9248,
+      "step": 5705
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.149910821647111e-07,
+      "loss": 0.8749,
+      "step": 5706
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.130222805438748e-07,
+      "loss": 1.0398,
+      "step": 5707
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.110557590785428e-07,
+      "loss": 1.0387,
+      "step": 5708
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.09091518256796e-07,
+      "loss": 0.8573,
+      "step": 5709
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.071295585661487e-07,
+      "loss": 0.9799,
+      "step": 5710
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.051698804935481e-07,
+      "loss": 1.062,
+      "step": 5711
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.032124845253719e-07,
+      "loss": 1.0543,
+      "step": 5712
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.012573711474358e-07,
+      "loss": 0.8506,
+      "step": 5713
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.993045408449918e-07,
+      "loss": 1.0125,
+      "step": 5714
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.973539941027153e-07,
+      "loss": 0.9353,
+      "step": 5715
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.95405731404727e-07,
+      "loss": 0.9635,
+      "step": 5716
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.934597532345678e-07,
+      "loss": 0.8812,
+      "step": 5717
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.915160600752248e-07,
+      "loss": 0.9568,
+      "step": 5718
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.895746524091063e-07,
+      "loss": 0.8838,
+      "step": 5719
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.876355307180616e-07,
+      "loss": 0.8316,
+      "step": 5720
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.85698695483369e-07,
+      "loss": 0.8322,
+      "step": 5721
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.837641471857416e-07,
+      "loss": 0.864,
+      "step": 5722
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.81831886305322e-07,
+      "loss": 1.0721,
+      "step": 5723
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.799019133216834e-07,
+      "loss": 1.0299,
+      "step": 5724
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.779742287138381e-07,
+      "loss": 0.83,
+      "step": 5725
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.760488329602256e-07,
+      "loss": 1.0294,
+      "step": 5726
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.741257265387159e-07,
+      "loss": 0.8591,
+      "step": 5727
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.722049099266171e-07,
+      "loss": 1.0646,
+      "step": 5728
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.702863836006602e-07,
+      "loss": 0.9807,
+      "step": 5729
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.683701480370176e-07,
+      "loss": 0.9503,
+      "step": 5730
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.664562037112822e-07,
+      "loss": 0.8412,
+      "step": 5731
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.645445510984883e-07,
+      "loss": 0.8791,
+      "step": 5732
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.626351906730956e-07,
+      "loss": 0.9645,
+      "step": 5733
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.607281229089992e-07,
+      "loss": 0.811,
+      "step": 5734
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.588233482795204e-07,
+      "loss": 0.8738,
+      "step": 5735
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.569208672574102e-07,
+      "loss": 0.9408,
+      "step": 5736
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.550206803148574e-07,
+      "loss": 0.9064,
+      "step": 5737
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.531227879234781e-07,
+      "loss": 0.8879,
+      "step": 5738
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.512271905543156e-07,
+      "loss": 0.8501,
+      "step": 5739
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.4933388867785e-07,
+      "loss": 0.9209,
+      "step": 5740
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.474428827639834e-07,
+      "loss": 0.9709,
+      "step": 5741
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.45554173282057e-07,
+      "loss": 1.1224,
+      "step": 5742
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.436677607008347e-07,
+      "loss": 0.9929,
+      "step": 5743
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.41783645488513e-07,
+      "loss": 0.9178,
+      "step": 5744
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.399018281127213e-07,
+      "loss": 0.9092,
+      "step": 5745
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.380223090405158e-07,
+      "loss": 0.8943,
+      "step": 5746
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.361450887383814e-07,
+      "loss": 0.9362,
+      "step": 5747
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.342701676722319e-07,
+      "loss": 0.8876,
+      "step": 5748
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.323975463074129e-07,
+      "loss": 0.861,
+      "step": 5749
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.305272251087014e-07,
+      "loss": 0.9509,
+      "step": 5750
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.286592045402963e-07,
+      "loss": 0.9255,
+      "step": 5751
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.267934850658309e-07,
+      "loss": 0.8926,
+      "step": 5752
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.249300671483684e-07,
+      "loss": 0.9086,
+      "step": 5753
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.230689512503963e-07,
+      "loss": 0.9817,
+      "step": 5754
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.21210137833832e-07,
+      "loss": 0.9471,
+      "step": 5755
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.193536273600243e-07,
+      "loss": 0.9761,
+      "step": 5756
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.174994202897479e-07,
+      "loss": 0.8904,
+      "step": 5757
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.156475170832078e-07,
+      "loss": 0.9533,
+      "step": 5758
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.137979182000343e-07,
+      "loss": 1.0608,
+      "step": 5759
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.119506240992858e-07,
+      "loss": 1.0538,
+      "step": 5760
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.10105635239452e-07,
+      "loss": 1.0572,
+      "step": 5761
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.082629520784501e-07,
+      "loss": 0.8794,
+      "step": 5762
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.064225750736209e-07,
+      "loss": 1.0598,
+      "step": 5763
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.045845046817368e-07,
+      "loss": 1.0206,
+      "step": 5764
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.027487413589973e-07,
+      "loss": 1.0317,
+      "step": 5765
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.009152855610247e-07,
+      "loss": 0.9023,
+      "step": 5766
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.99084137742877e-07,
+      "loss": 0.9463,
+      "step": 5767
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.97255298359032e-07,
+      "loss": 0.787,
+      "step": 5768
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.954287678633965e-07,
+      "loss": 0.8548,
+      "step": 5769
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.936045467093078e-07,
+      "loss": 0.8446,
+      "step": 5770
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.91782635349526e-07,
+      "loss": 0.9059,
+      "step": 5771
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.899630342362362e-07,
+      "loss": 0.8609,
+      "step": 5772
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.88145743821057e-07,
+      "loss": 0.8544,
+      "step": 5773
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.863307645550277e-07,
+      "loss": 0.8777,
+      "step": 5774
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.845180968886156e-07,
+      "loss": 0.9123,
+      "step": 5775
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.827077412717153e-07,
+      "loss": 1.0312,
+      "step": 5776
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.808996981536475e-07,
+      "loss": 0.9912,
+      "step": 5777
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.790939679831554e-07,
+      "loss": 0.8503,
+      "step": 5778
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.772905512084149e-07,
+      "loss": 0.8526,
+      "step": 5779
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.754894482770191e-07,
+      "loss": 0.959,
+      "step": 5780
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.736906596359949e-07,
+      "loss": 0.9676,
+      "step": 5781
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.718941857317907e-07,
+      "loss": 0.8099,
+      "step": 5782
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.70100027010282e-07,
+      "loss": 0.9051,
+      "step": 5783
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.683081839167649e-07,
+      "loss": 0.9988,
+      "step": 5784
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.665186568959681e-07,
+      "loss": 0.9915,
+      "step": 5785
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.647314463920418e-07,
+      "loss": 0.7947,
+      "step": 5786
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.629465528485602e-07,
+      "loss": 0.8471,
+      "step": 5787
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.611639767085243e-07,
+      "loss": 0.8876,
+      "step": 5788
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.593837184143614e-07,
+      "loss": 0.9731,
+      "step": 5789
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.57605778407917e-07,
+      "loss": 0.8705,
+      "step": 5790
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.558301571304693e-07,
+      "loss": 0.9437,
+      "step": 5791
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.540568550227155e-07,
+      "loss": 0.9491,
+      "step": 5792
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.522858725247793e-07,
+      "loss": 0.8835,
+      "step": 5793
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.505172100762114e-07,
+      "loss": 0.905,
+      "step": 5794
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.487508681159804e-07,
+      "loss": 0.9401,
+      "step": 5795
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.469868470824825e-07,
+      "loss": 1.073,
+      "step": 5796
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.452251474135374e-07,
+      "loss": 0.8792,
+      "step": 5797
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.434657695463919e-07,
+      "loss": 0.7968,
+      "step": 5798
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.417087139177103e-07,
+      "loss": 0.7926,
+      "step": 5799
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.399539809635846e-07,
+      "loss": 1.0651,
+      "step": 5800
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.38201571119531e-07,
+      "loss": 0.8365,
+      "step": 5801
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.364514848204861e-07,
+      "loss": 0.8149,
+      "step": 5802
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.347037225008123e-07,
+      "loss": 0.8885,
+      "step": 5803
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.329582845942939e-07,
+      "loss": 0.95,
+      "step": 5804
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.312151715341364e-07,
+      "loss": 0.8039,
+      "step": 5805
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.29474383752976e-07,
+      "loss": 0.8043,
+      "step": 5806
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.277359216828616e-07,
+      "loss": 0.9323,
+      "step": 5807
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.259997857552702e-07,
+      "loss": 0.9139,
+      "step": 5808
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.242659764011017e-07,
+      "loss": 0.8405,
+      "step": 5809
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.225344940506784e-07,
+      "loss": 0.9575,
+      "step": 5810
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.208053391337421e-07,
+      "loss": 0.9368,
+      "step": 5811
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.190785120794607e-07,
+      "loss": 0.9506,
+      "step": 5812
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.173540133164247e-07,
+      "loss": 1.0592,
+      "step": 5813
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.156318432726405e-07,
+      "loss": 0.8577,
+      "step": 5814
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.139120023755452e-07,
+      "loss": 0.97,
+      "step": 5815
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.121944910519905e-07,
+      "loss": 0.7938,
+      "step": 5816
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.104793097282535e-07,
+      "loss": 0.8753,
+      "step": 5817
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.087664588300346e-07,
+      "loss": 1.0498,
+      "step": 5818
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.070559387824526e-07,
+      "loss": 0.9115,
+      "step": 5819
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.053477500100458e-07,
+      "loss": 0.9909,
+      "step": 5820
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.036418929367793e-07,
+      "loss": 0.8974,
+      "step": 5821
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.01938367986038e-07,
+      "loss": 0.9546,
+      "step": 5822
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.002371755806268e-07,
+      "loss": 0.9196,
+      "step": 5823
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.985383161427705e-07,
+      "loss": 1.0297,
+      "step": 5824
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.96841790094117e-07,
+      "loss": 0.9492,
+      "step": 5825
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.951475978557364e-07,
+      "loss": 0.9623,
+      "step": 5826
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.934557398481145e-07,
+      "loss": 0.8343,
+      "step": 5827
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.91766216491162e-07,
+      "loss": 0.9408,
+      "step": 5828
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.900790282042079e-07,
+      "loss": 0.9023,
+      "step": 5829
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.883941754060063e-07,
+      "loss": 0.8977,
+      "step": 5830
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.867116585147237e-07,
+      "loss": 0.7474,
+      "step": 5831
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.850314779479549e-07,
+      "loss": 0.9937,
+      "step": 5832
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.833536341227075e-07,
+      "loss": 0.9778,
+      "step": 5833
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.81678127455414e-07,
+      "loss": 0.8817,
+      "step": 5834
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.800049583619294e-07,
+      "loss": 0.8963,
+      "step": 5835
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.783341272575183e-07,
+      "loss": 1.1353,
+      "step": 5836
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.766656345568755e-07,
+      "loss": 0.968,
+      "step": 5837
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.749994806741121e-07,
+      "loss": 0.88,
+      "step": 5838
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.733356660227563e-07,
+      "loss": 0.9672,
+      "step": 5839
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.716741910157563e-07,
+      "loss": 0.8526,
+      "step": 5840
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.700150560654827e-07,
+      "loss": 0.8079,
+      "step": 5841
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.683582615837224e-07,
+      "loss": 0.9325,
+      "step": 5842
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.667038079816822e-07,
+      "loss": 0.9777,
+      "step": 5843
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.650516956699903e-07,
+      "loss": 0.8721,
+      "step": 5844
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.634019250586875e-07,
+      "loss": 0.9141,
+      "step": 5845
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.617544965572408e-07,
+      "loss": 0.9777,
+      "step": 5846
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.60109410574532e-07,
+      "loss": 0.9489,
+      "step": 5847
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.584666675188599e-07,
+      "loss": 0.9692,
+      "step": 5848
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.568262677979463e-07,
+      "loss": 0.998,
+      "step": 5849
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.551882118189289e-07,
+      "loss": 0.9663,
+      "step": 5850
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.535524999883646e-07,
+      "loss": 0.9414,
+      "step": 5851
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.519191327122242e-07,
+      "loss": 0.8954,
+      "step": 5852
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.502881103959024e-07,
+      "loss": 0.9934,
+      "step": 5853
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.486594334442108e-07,
+      "loss": 1.0279,
+      "step": 5854
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.470331022613751e-07,
+      "loss": 1.0609,
+      "step": 5855
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.454091172510445e-07,
+      "loss": 1.0331,
+      "step": 5856
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.4378747881628e-07,
+      "loss": 0.9142,
+      "step": 5857
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.421681873595641e-07,
+      "loss": 0.8565,
+      "step": 5858
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.405512432827975e-07,
+      "loss": 0.9701,
+      "step": 5859
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.389366469872937e-07,
+      "loss": 0.8683,
+      "step": 5860
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.373243988737864e-07,
+      "loss": 0.8624,
+      "step": 5861
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.357144993424301e-07,
+      "loss": 1.1008,
+      "step": 5862
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.341069487927897e-07,
+      "loss": 0.7482,
+      "step": 5863
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.325017476238481e-07,
+      "loss": 0.9631,
+      "step": 5864
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.3089889623401e-07,
+      "loss": 0.9665,
+      "step": 5865
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.29298395021095e-07,
+      "loss": 0.9445,
+      "step": 5866
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.277002443823354e-07,
+      "loss": 0.8254,
+      "step": 5867
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.26104444714386e-07,
+      "loss": 0.8992,
+      "step": 5868
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.245109964133111e-07,
+      "loss": 0.7851,
+      "step": 5869
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.229198998745988e-07,
+      "loss": 0.6748,
+      "step": 5870
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.213311554931511e-07,
+      "loss": 0.834,
+      "step": 5871
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.197447636632824e-07,
+      "loss": 0.8369,
+      "step": 5872
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.181607247787257e-07,
+      "loss": 1.078,
+      "step": 5873
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.165790392326353e-07,
+      "loss": 0.8292,
+      "step": 5874
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.149997074175728e-07,
+      "loss": 0.9619,
+      "step": 5875
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.134227297255179e-07,
+      "loss": 0.977,
+      "step": 5876
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.118481065478687e-07,
+      "loss": 1.0243,
+      "step": 5877
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.102758382754414e-07,
+      "loss": 0.8871,
+      "step": 5878
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.087059252984583e-07,
+      "loss": 0.8877,
+      "step": 5879
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.071383680065678e-07,
+      "loss": 0.849,
+      "step": 5880
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.055731667888253e-07,
+      "loss": 0.8732,
+      "step": 5881
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.040103220337056e-07,
+      "loss": 0.9822,
+      "step": 5882
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.024498341290996e-07,
+      "loss": 0.9806,
+      "step": 5883
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.008917034623084e-07,
+      "loss": 0.8159,
+      "step": 5884
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.993359304200529e-07,
+      "loss": 0.9321,
+      "step": 5885
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.977825153884686e-07,
+      "loss": 0.8681,
+      "step": 5886
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.962314587531025e-07,
+      "loss": 0.8856,
+      "step": 5887
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.94682760898918e-07,
+      "loss": 0.9606,
+      "step": 5888
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.931364222102919e-07,
+      "loss": 1.0091,
+      "step": 5889
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.915924430710195e-07,
+      "loss": 1.0018,
+      "step": 5890
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.900508238643065e-07,
+      "loss": 0.9515,
+      "step": 5891
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.885115649727723e-07,
+      "loss": 1.0087,
+      "step": 5892
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.869746667784558e-07,
+      "loss": 1.0078,
+      "step": 5893
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.854401296628031e-07,
+      "loss": 1.0705,
+      "step": 5894
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.839079540066782e-07,
+      "loss": 1.0203,
+      "step": 5895
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.823781401903594e-07,
+      "loss": 0.8781,
+      "step": 5896
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.80850688593536e-07,
+      "loss": 0.9014,
+      "step": 5897
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.79325599595315e-07,
+      "loss": 0.9331,
+      "step": 5898
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.778028735742146e-07,
+      "loss": 0.9363,
+      "step": 5899
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.7628251090816477e-07,
+      "loss": 0.9392,
+      "step": 5900
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.7476451197451146e-07,
+      "loss": 1.0504,
+      "step": 5901
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.732488771500132e-07,
+      "loss": 0.8837,
+      "step": 5902
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.7173560681084474e-07,
+      "loss": 0.7946,
+      "step": 5903
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.702247013325878e-07,
+      "loss": 0.7788,
+      "step": 5904
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6871616109024244e-07,
+      "loss": 0.9303,
+      "step": 5905
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.67209986458218e-07,
+      "loss": 0.8386,
+      "step": 5906
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6570617781034e-07,
+      "loss": 0.9316,
+      "step": 5907
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6420473551984425e-07,
+      "loss": 0.8443,
+      "step": 5908
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6270565995938045e-07,
+      "loss": 0.8853,
+      "step": 5909
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.61208951501011e-07,
+      "loss": 0.74,
+      "step": 5910
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.597146105162109e-07,
+      "loss": 0.9481,
+      "step": 5911
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.582226373758647e-07,
+      "loss": 1.023,
+      "step": 5912
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.567330324502739e-07,
+      "loss": 1.085,
+      "step": 5913
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.552457961091483e-07,
+      "loss": 0.9427,
+      "step": 5914
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5376092872161494e-07,
+      "loss": 0.9129,
+      "step": 5915
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5227843065620446e-07,
+      "loss": 1.0122,
+      "step": 5916
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.507983022808682e-07,
+      "loss": 0.9129,
+      "step": 5917
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4932054396296334e-07,
+      "loss": 0.897,
+      "step": 5918
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4784515606926314e-07,
+      "loss": 0.8466,
+      "step": 5919
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4637213896594813e-07,
+      "loss": 0.8876,
+      "step": 5920
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4490149301861353e-07,
+      "loss": 0.9249,
+      "step": 5921
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.434332185922685e-07,
+      "loss": 0.9163,
+      "step": 5922
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4196731605132605e-07,
+      "loss": 0.9097,
+      "step": 5923
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4050378575961504e-07,
+      "loss": 0.8093,
+      "step": 5924
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3904262808037723e-07,
+      "loss": 0.8678,
+      "step": 5925
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3758384337626357e-07,
+      "loss": 0.9023,
+      "step": 5926
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.361274320093367e-07,
+      "loss": 0.8607,
+      "step": 5927
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.346733943410664e-07,
+      "loss": 0.9988,
+      "step": 5928
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3322173073234054e-07,
+      "loss": 0.9137,
+      "step": 5929
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.31772441543451e-07,
+      "loss": 1.0283,
+      "step": 5930
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3032552713410447e-07,
+      "loss": 0.8916,
+      "step": 5931
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2888098786341593e-07,
+      "loss": 0.878,
+      "step": 5932
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.274388240899119e-07,
+      "loss": 0.7539,
+      "step": 5933
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.259990361715305e-07,
+      "loss": 0.877,
+      "step": 5934
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.245616244656181e-07,
+      "loss": 0.9388,
+      "step": 5935
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.231265893289316e-07,
+      "loss": 0.8313,
+      "step": 5936
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2169393111763713e-07,
+      "loss": 0.931,
+      "step": 5937
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2026365018731476e-07,
+      "loss": 0.8734,
+      "step": 5938
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.188357468929538e-07,
+      "loss": 0.8638,
+      "step": 5939
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.174102215889464e-07,
+      "loss": 0.8986,
+      "step": 5940
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1598707462910395e-07,
+      "loss": 1.0421,
+      "step": 5941
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1456630636664163e-07,
+      "loss": 0.9462,
+      "step": 5942
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1314791715418745e-07,
+      "loss": 0.8684,
+      "step": 5943
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1173190734377534e-07,
+      "loss": 0.9959,
+      "step": 5944
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1031827728685304e-07,
+      "loss": 0.9361,
+      "step": 5945
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0890702733427323e-07,
+      "loss": 1.0606,
+      "step": 5946
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0749815783630464e-07,
+      "loss": 0.8796,
+      "step": 5947
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.060916691426142e-07,
+      "loss": 1.0204,
+      "step": 5948
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.046875616022883e-07,
+      "loss": 0.8928,
+      "step": 5949
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0328583556381693e-07,
+      "loss": 0.9297,
+      "step": 5950
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.018864913751042e-07,
+      "loss": 1.0321,
+      "step": 5951
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 4.004895293834532e-07,
+      "loss": 1.0009,
+      "step": 5952
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.990949499355878e-07,
+      "loss": 0.9851,
+      "step": 5953
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.977027533776312e-07,
+      "loss": 0.9658,
+      "step": 5954
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.963129400551191e-07,
+      "loss": 0.9491,
+      "step": 5955
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.9492551031299564e-07,
+      "loss": 0.9904,
+      "step": 5956
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.935404644956131e-07,
+      "loss": 0.8821,
+      "step": 5957
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.921578029467321e-07,
+      "loss": 0.9331,
+      "step": 5958
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.907775260095226e-07,
+      "loss": 1.0196,
+      "step": 5959
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.893996340265582e-07,
+      "loss": 0.754,
+      "step": 5960
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.8802412733982553e-07,
+      "loss": 0.8655,
+      "step": 5961
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.8665100629071694e-07,
+      "loss": 0.9173,
+      "step": 5962
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.8528027122003655e-07,
+      "loss": 0.8221,
+      "step": 5963
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.839119224679877e-07,
+      "loss": 0.9428,
+      "step": 5964
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.825459603741899e-07,
+      "loss": 1.0093,
+      "step": 5965
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.811823852776686e-07,
+      "loss": 0.8835,
+      "step": 5966
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7982119751685307e-07,
+      "loss": 0.8031,
+      "step": 5967
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.784623974295809e-07,
+      "loss": 1.0394,
+      "step": 5968
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7710598535310116e-07,
+      "loss": 0.9381,
+      "step": 5969
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.757519616240657e-07,
+      "loss": 0.8545,
+      "step": 5970
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.74400326578539e-07,
+      "loss": 1.0322,
+      "step": 5971
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7305108055198604e-07,
+      "loss": 1.1264,
+      "step": 5972
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.717042238792823e-07,
+      "loss": 1.0482,
+      "step": 5973
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7035975689470926e-07,
+      "loss": 0.9922,
+      "step": 5974
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.690176799319589e-07,
+      "loss": 0.8952,
+      "step": 5975
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6767799332412366e-07,
+      "loss": 0.9199,
+      "step": 5976
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6634069740370757e-07,
+      "loss": 1.0046,
+      "step": 5977
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.650057925026218e-07,
+      "loss": 0.8966,
+      "step": 5978
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6367327895217907e-07,
+      "loss": 0.8977,
+      "step": 5979
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6234315708310263e-07,
+      "loss": 0.76,
+      "step": 5980
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6101542722552175e-07,
+      "loss": 0.9316,
+      "step": 5981
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5969008970897057e-07,
+      "loss": 0.9468,
+      "step": 5982
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.583671448623915e-07,
+      "loss": 0.8984,
+      "step": 5983
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.57046593014132e-07,
+      "loss": 0.914,
+      "step": 5984
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.557284344919432e-07,
+      "loss": 0.8743,
+      "step": 5985
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.544126696229855e-07,
+      "loss": 1.0303,
+      "step": 5986
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.530992987338266e-07,
+      "loss": 0.9316,
+      "step": 5987
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.517883221504348e-07,
+      "loss": 0.8917,
+      "step": 5988
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.504797401981874e-07,
+      "loss": 1.0535,
+      "step": 5989
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.491735532018692e-07,
+      "loss": 0.8674,
+      "step": 5990
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4786976148566745e-07,
+      "loss": 0.8043,
+      "step": 5991
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.465683653731733e-07,
+      "loss": 0.8953,
+      "step": 5992
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.452693651873873e-07,
+      "loss": 0.8856,
+      "step": 5993
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4397276125071486e-07,
+      "loss": 1.0112,
+      "step": 5994
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.426785538849664e-07,
+      "loss": 0.9743,
+      "step": 5995
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4138674341135494e-07,
+      "loss": 0.9076,
+      "step": 5996
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4009733015050064e-07,
+      "loss": 1.0512,
+      "step": 5997
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3881031442242865e-07,
+      "loss": 0.9118,
+      "step": 5998
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.375256965465712e-07,
+      "loss": 0.9061,
+      "step": 5999
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3624347684175885e-07,
+      "loss": 0.9662,
+      "step": 6000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.349636556262348e-07,
+      "loss": 0.8613,
+      "step": 6001
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.336862332176416e-07,
+      "loss": 1.0272,
+      "step": 6002
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.324112099330301e-07,
+      "loss": 0.8908,
+      "step": 6003
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3113858608885163e-07,
+      "loss": 0.897,
+      "step": 6004
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.298683620009635e-07,
+      "loss": 0.9963,
+      "step": 6005
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2860053798463134e-07,
+      "loss": 0.9368,
+      "step": 6006
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2733511435452006e-07,
+      "loss": 0.9419,
+      "step": 6007
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2607209142470065e-07,
+      "loss": 0.894,
+      "step": 6008
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2481146950864793e-07,
+      "loss": 1.026,
+      "step": 6009
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.235532489192406e-07,
+      "loss": 0.9163,
+      "step": 6010
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2229742996876313e-07,
+      "loss": 0.9141,
+      "step": 6011
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2104401296890184e-07,
+      "loss": 1.0571,
+      "step": 6012
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.197929982307479e-07,
+      "loss": 0.7911,
+      "step": 6013
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.185443860647974e-07,
+      "loss": 0.8647,
+      "step": 6014
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1729817678094687e-07,
+      "loss": 1.2141,
+      "step": 6015
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.1605437068849776e-07,
+      "loss": 0.9548,
+      "step": 6016
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.148129680961576e-07,
+      "loss": 0.956,
+      "step": 6017
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.135739693120332e-07,
+      "loss": 1.0309,
+      "step": 6018
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.123373746436409e-07,
+      "loss": 0.9846,
+      "step": 6019
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.1110318439789286e-07,
+      "loss": 1.006,
+      "step": 6020
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.098713988811086e-07,
+      "loss": 0.8818,
+      "step": 6021
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.086420183990113e-07,
+      "loss": 1.0712,
+      "step": 6022
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.074150432567258e-07,
+      "loss": 1.0043,
+      "step": 6023
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.061904737587806e-07,
+      "loss": 0.9258,
+      "step": 6024
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.04968310209105e-07,
+      "loss": 0.9814,
+      "step": 6025
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.0374855291103623e-07,
+      "loss": 0.9674,
+      "step": 6026
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.0253120216730993e-07,
+      "loss": 0.8662,
+      "step": 6027
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.0131625828006327e-07,
+      "loss": 0.9474,
+      "step": 6028
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.0010372155084066e-07,
+      "loss": 0.9298,
+      "step": 6029
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.988935922805869e-07,
+      "loss": 0.7874,
+      "step": 6030
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.976858707696484e-07,
+      "loss": 0.9454,
+      "step": 6031
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9648055731777534e-07,
+      "loss": 0.9151,
+      "step": 6032
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.952776522241196e-07,
+      "loss": 0.9589,
+      "step": 6033
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.940771557872346e-07,
+      "loss": 0.9305,
+      "step": 6034
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.928790683050775e-07,
+      "loss": 0.924,
+      "step": 6035
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9168339007500603e-07,
+      "loss": 0.9225,
+      "step": 6036
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.904901213937805e-07,
+      "loss": 0.942,
+      "step": 6037
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8929926255756415e-07,
+      "loss": 0.8861,
+      "step": 6038
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.881108138619237e-07,
+      "loss": 0.8416,
+      "step": 6039
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.869247756018212e-07,
+      "loss": 0.8814,
+      "step": 6040
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.857411480716266e-07,
+      "loss": 0.8644,
+      "step": 6041
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.845599315651082e-07,
+      "loss": 0.9684,
+      "step": 6042
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8338112637543936e-07,
+      "loss": 0.9691,
+      "step": 6043
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8220473279519155e-07,
+      "loss": 0.967,
+      "step": 6044
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.81030751116339e-07,
+      "loss": 0.9754,
+      "step": 6045
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.798591816302565e-07,
+      "loss": 0.9306,
+      "step": 6046
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7869002462772357e-07,
+      "loss": 0.8721,
+      "step": 6047
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.775232803989147e-07,
+      "loss": 1.0011,
+      "step": 6048
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7635894923341064e-07,
+      "loss": 1.0437,
+      "step": 6049
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.751970314201935e-07,
+      "loss": 0.8149,
+      "step": 6050
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7403752724764365e-07,
+      "loss": 0.9748,
+      "step": 6051
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7288043700354207e-07,
+      "loss": 0.8112,
+      "step": 6052
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.717257609750723e-07,
+      "loss": 0.8858,
+      "step": 6053
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7057349944881963e-07,
+      "loss": 0.9142,
+      "step": 6054
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6942365271076853e-07,
+      "loss": 1.0566,
+      "step": 6055
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.68276221046303e-07,
+      "loss": 0.9382,
+      "step": 6056
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6713120474021083e-07,
+      "loss": 0.9224,
+      "step": 6057
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.659886040766768e-07,
+      "loss": 0.8384,
+      "step": 6058
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6484841933929083e-07,
+      "loss": 0.8879,
+      "step": 6059
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.637106508110354e-07,
+      "loss": 0.8628,
+      "step": 6060
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.625752987743024e-07,
+      "loss": 1.0328,
+      "step": 6061
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.614423635108787e-07,
+      "loss": 0.9658,
+      "step": 6062
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.603118453019515e-07,
+      "loss": 0.8897,
+      "step": 6063
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.591837444281109e-07,
+      "loss": 0.9167,
+      "step": 6064
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.580580611693417e-07,
+      "loss": 0.992,
+      "step": 6065
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.56934795805035e-07,
+      "loss": 0.854,
+      "step": 6066
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.558139486139788e-07,
+      "loss": 1.0227,
+      "step": 6067
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5469551987435836e-07,
+      "loss": 0.9711,
+      "step": 6068
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.53579509863765e-07,
+      "loss": 0.9211,
+      "step": 6069
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.524659188591827e-07,
+      "loss": 0.7691,
+      "step": 6070
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.513547471370004e-07,
+      "loss": 0.9198,
+      "step": 6071
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5024599497300404e-07,
+      "loss": 0.9964,
+      "step": 6072
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.491396626423792e-07,
+      "loss": 0.8205,
+      "step": 6073
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4803575041971063e-07,
+      "loss": 0.9803,
+      "step": 6074
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4693425857898576e-07,
+      "loss": 0.9136,
+      "step": 6075
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4583518739358714e-07,
+      "loss": 0.953,
+      "step": 6076
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.447385371362965e-07,
+      "loss": 0.8988,
+      "step": 6077
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.436443080792983e-07,
+      "loss": 0.7999,
+      "step": 6078
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4255250049417313e-07,
+      "loss": 0.9802,
+      "step": 6079
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4146311465189976e-07,
+      "loss": 0.9781,
+      "step": 6080
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.403761508228608e-07,
+      "loss": 0.9581,
+      "step": 6081
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.392916092768327e-07,
+      "loss": 0.896,
+      "step": 6082
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.3820949028299235e-07,
+      "loss": 0.9416,
+      "step": 6083
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.3712979410991489e-07,
+      "loss": 0.9461,
+      "step": 6084
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.3605252102557596e-07,
+      "loss": 0.8238,
+      "step": 6085
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.3497767129734837e-07,
+      "loss": 0.8154,
+      "step": 6086
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.339052451920043e-07,
+      "loss": 0.893,
+      "step": 6087
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.3283524297571304e-07,
+      "loss": 0.867,
+      "step": 6088
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.3176766491404213e-07,
+      "loss": 0.916,
+      "step": 6089
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.3070251127195853e-07,
+      "loss": 0.9145,
+      "step": 6090
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2963978231382966e-07,
+      "loss": 0.8329,
+      "step": 6091
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2857947830341677e-07,
+      "loss": 0.9357,
+      "step": 6092
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2752159950388263e-07,
+      "loss": 0.8891,
+      "step": 6093
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2646614617778394e-07,
+      "loss": 0.9017,
+      "step": 6094
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2541311858708225e-07,
+      "loss": 1.0291,
+      "step": 6095
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.243625169931285e-07,
+      "loss": 0.8719,
+      "step": 6096
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2331434165667963e-07,
+      "loss": 1.0846,
+      "step": 6097
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2226859283788538e-07,
+      "loss": 0.9526,
+      "step": 6098
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2122527079629474e-07,
+      "loss": 1.108,
+      "step": 6099
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2018437579085504e-07,
+      "loss": 1.0028,
+      "step": 6100
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1914590807990854e-07,
+      "loss": 0.9414,
+      "step": 6101
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1810986792119791e-07,
+      "loss": 0.8727,
+      "step": 6102
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1707625557186419e-07,
+      "loss": 0.8384,
+      "step": 6103
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.160450712884421e-07,
+      "loss": 1.0583,
+      "step": 6104
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1501631532686583e-07,
+      "loss": 0.9283,
+      "step": 6105
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1398998794246784e-07,
+      "loss": 0.8617,
+      "step": 6106
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1296608938997654e-07,
+      "loss": 1.0523,
+      "step": 6107
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1194461992351534e-07,
+      "loss": 0.9567,
+      "step": 6108
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.109255797966092e-07,
+      "loss": 0.9784,
+      "step": 6109
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.099089692621792e-07,
+      "loss": 0.9824,
+      "step": 6110
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0889478857254008e-07,
+      "loss": 0.8273,
+      "step": 6111
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0788303797940835e-07,
+      "loss": 0.9296,
+      "step": 6112
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0687371773389087e-07,
+      "loss": 0.8785,
+      "step": 6113
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0586682808649727e-07,
+      "loss": 0.7669,
+      "step": 6114
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0486236928713322e-07,
+      "loss": 0.9056,
+      "step": 6115
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0386034158509705e-07,
+      "loss": 0.924,
+      "step": 6116
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.028607452290876e-07,
+      "loss": 0.9843,
+      "step": 6117
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.018635804671998e-07,
+      "loss": 0.9303,
+      "step": 6118
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0086884754692226e-07,
+      "loss": 0.9818,
+      "step": 6119
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9987654671514534e-07,
+      "loss": 0.8458,
+      "step": 6120
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9888667821814756e-07,
+      "loss": 0.9958,
+      "step": 6121
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.978992423016124e-07,
+      "loss": 0.9248,
+      "step": 6122
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9691423921061602e-07,
+      "loss": 0.8838,
+      "step": 6123
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9593166918962847e-07,
+      "loss": 0.9109,
+      "step": 6124
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9495153248251796e-07,
+      "loss": 0.9066,
+      "step": 6125
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9397382933255104e-07,
+      "loss": 0.8555,
+      "step": 6126
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9299855998238691e-07,
+      "loss": 0.8939,
+      "step": 6127
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9202572467407976e-07,
+      "loss": 0.8922,
+      "step": 6128
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.910553236490853e-07,
+      "loss": 1.03,
+      "step": 6129
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9008735714824977e-07,
+      "loss": 1.0103,
+      "step": 6130
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8912182541181656e-07,
+      "loss": 0.9344,
+      "step": 6131
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.881587286794273e-07,
+      "loss": 0.8409,
+      "step": 6132
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8719806719011412e-07,
+      "loss": 0.9211,
+      "step": 6133
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8623984118230964e-07,
+      "loss": 0.9563,
+      "step": 6134
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8528405089383917e-07,
+      "loss": 0.8227,
+      "step": 6135
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8433069656192626e-07,
+      "loss": 0.9529,
+      "step": 6136
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.83379778423185e-07,
+      "loss": 0.9341,
+      "step": 6137
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.824312967136299e-07,
+      "loss": 1.0258,
+      "step": 6138
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8148525166866827e-07,
+      "loss": 0.9074,
+      "step": 6139
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8054164352310333e-07,
+      "loss": 0.9158,
+      "step": 6140
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7960047251113222e-07,
+      "loss": 0.8392,
+      "step": 6141
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.786617388663503e-07,
+      "loss": 0.9664,
+      "step": 6142
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7772544282174343e-07,
+      "loss": 0.8327,
+      "step": 6143
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7679158460969682e-07,
+      "loss": 0.9115,
+      "step": 6144
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.758601644619884e-07,
+      "loss": 0.9203,
+      "step": 6145
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.7493118260978992e-07,
+      "loss": 0.8131,
+      "step": 6146
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.740046392836714e-07,
+      "loss": 0.8172,
+      "step": 6147
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.7308053471359444e-07,
+      "loss": 0.9044,
+      "step": 6148
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.7215886912891667e-07,
+      "loss": 0.8826,
+      "step": 6149
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.7123964275838956e-07,
+      "loss": 0.845,
+      "step": 6150
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.7032285583016284e-07,
+      "loss": 0.9525,
+      "step": 6151
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6940850857177339e-07,
+      "loss": 0.8646,
+      "step": 6152
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6849660121015966e-07,
+      "loss": 0.8485,
+      "step": 6153
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6758713397165284e-07,
+      "loss": 0.9861,
+      "step": 6154
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6668010708197458e-07,
+      "loss": 0.9417,
+      "step": 6155
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6577552076624593e-07,
+      "loss": 0.9092,
+      "step": 6156
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.648733752489784e-07,
+      "loss": 1.0249,
+      "step": 6157
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6397367075408067e-07,
+      "loss": 1.0328,
+      "step": 6158
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.630764075048552e-07,
+      "loss": 0.8682,
+      "step": 6159
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.621815857239961e-07,
+      "loss": 0.9278,
+      "step": 6160
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.612892056335924e-07,
+      "loss": 0.8707,
+      "step": 6161
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6039926745512911e-07,
+      "loss": 0.8477,
+      "step": 6162
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5951177140948293e-07,
+      "loss": 0.8695,
+      "step": 6163
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.586267177169265e-07,
+      "loss": 0.9965,
+      "step": 6164
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5774410659712302e-07,
+      "loss": 0.8538,
+      "step": 6165
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5686393826913504e-07,
+      "loss": 0.9949,
+      "step": 6166
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5598621295141113e-07,
+      "loss": 1.1173,
+      "step": 6167
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5511093086180153e-07,
+      "loss": 0.9401,
+      "step": 6168
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.542380922175435e-07,
+      "loss": 1.0446,
+      "step": 6169
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5336769723527046e-07,
+      "loss": 0.8369,
+      "step": 6170
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5249974613101292e-07,
+      "loss": 1.0905,
+      "step": 6171
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5163423912018748e-07,
+      "loss": 0.915,
+      "step": 6172
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5077117641761008e-07,
+      "loss": 0.9365,
+      "step": 6173
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4991055823748822e-07,
+      "loss": 0.9861,
+      "step": 6174
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4905238479342e-07,
+      "loss": 0.9011,
+      "step": 6175
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4819665629840273e-07,
+      "loss": 1.1134,
+      "step": 6176
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4734337296482103e-07,
+      "loss": 0.826,
+      "step": 6177
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4649253500445438e-07,
+      "loss": 0.9482,
+      "step": 6178
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4564414262847938e-07,
+      "loss": 0.8122,
+      "step": 6179
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.447981960474587e-07,
+      "loss": 0.8426,
+      "step": 6180
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.439546954713522e-07,
+      "loss": 0.8926,
+      "step": 6181
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4311364110951132e-07,
+      "loss": 1.0008,
+      "step": 6182
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4227503317068347e-07,
+      "loss": 0.9473,
+      "step": 6183
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4143887186300442e-07,
+      "loss": 0.8744,
+      "step": 6184
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.406051573940048e-07,
+      "loss": 1.0312,
+      "step": 6185
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3977388997060803e-07,
+      "loss": 0.8615,
+      "step": 6186
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3894506979913013e-07,
+      "loss": 0.9727,
+      "step": 6187
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3811869708527992e-07,
+      "loss": 1.0302,
+      "step": 6188
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3729477203415664e-07,
+      "loss": 0.9443,
+      "step": 6189
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3647329485025452e-07,
+      "loss": 0.8891,
+      "step": 6190
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3565426573746155e-07,
+      "loss": 0.9473,
+      "step": 6191
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3483768489905514e-07,
+      "loss": 0.9751,
+      "step": 6192
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3402355253770316e-07,
+      "loss": 0.8946,
+      "step": 6193
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3321186885547176e-07,
+      "loss": 0.9224,
+      "step": 6194
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3240263405381647e-07,
+      "loss": 0.9546,
+      "step": 6195
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3159584833358107e-07,
+      "loss": 0.9705,
+      "step": 6196
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3079151189500982e-07,
+      "loss": 0.9083,
+      "step": 6197
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2998962493773194e-07,
+      "loss": 1.0982,
+      "step": 6198
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.291901876607704e-07,
+      "loss": 0.8285,
+      "step": 6199
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.283932002625432e-07,
+      "loss": 1.0651,
+      "step": 6200
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2759866294085653e-07,
+      "loss": 0.9807,
+      "step": 6201
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2680657589291045e-07,
+      "loss": 0.9885,
+      "step": 6202
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2601693931529658e-07,
+      "loss": 0.9345,
+      "step": 6203
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2522975340399924e-07,
+      "loss": 0.9589,
+      "step": 6204
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2444501835439105e-07,
+      "loss": 0.9423,
+      "step": 6205
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2366273436123954e-07,
+      "loss": 0.982,
+      "step": 6206
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2288290161870499e-07,
+      "loss": 0.9836,
+      "step": 6207
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2210552032033584e-07,
+      "loss": 1.0165,
+      "step": 6208
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.213305906590745e-07,
+      "loss": 1.0095,
+      "step": 6209
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.205581128272526e-07,
+      "loss": 0.9734,
+      "step": 6210
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1978808701659572e-07,
+      "loss": 0.9483,
+      "step": 6211
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1902051341822097e-07,
+      "loss": 0.8062,
+      "step": 6212
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1825539222263372e-07,
+      "loss": 0.9631,
+      "step": 6213
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1749272361973318e-07,
+      "loss": 0.9575,
+      "step": 6214
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1673250779881128e-07,
+      "loss": 0.9615,
+      "step": 6215
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1597474494854711e-07,
+      "loss": 0.937,
+      "step": 6216
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1521943525701351e-07,
+      "loss": 0.9737,
+      "step": 6217
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.144665789116739e-07,
+      "loss": 0.887,
+      "step": 6218
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1371617609938435e-07,
+      "loss": 0.8015,
+      "step": 6219
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1296822700638921e-07,
+      "loss": 0.9209,
+      "step": 6220
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1222273181832555e-07,
+      "loss": 0.8621,
+      "step": 6221
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1147969072022091e-07,
+      "loss": 0.997,
+      "step": 6222
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1073910389649334e-07,
+      "loss": 0.9407,
+      "step": 6223
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.1000097153095357e-07,
+      "loss": 0.9165,
+      "step": 6224
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0926529380680063e-07,
+      "loss": 0.8718,
+      "step": 6225
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0853207090662621e-07,
+      "loss": 0.9615,
+      "step": 6226
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0780130301241254e-07,
+      "loss": 0.9326,
+      "step": 6227
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.070729903055312e-07,
+      "loss": 0.8894,
+      "step": 6228
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0634713296674537e-07,
+      "loss": 0.797,
+      "step": 6229
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0562373117620761e-07,
+      "loss": 0.8783,
+      "step": 6230
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.049027851134643e-07,
+      "loss": 0.9678,
+      "step": 6231
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0418429495745008e-07,
+      "loss": 0.9552,
+      "step": 6232
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0346826088648787e-07,
+      "loss": 0.9467,
+      "step": 6233
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0275468307829439e-07,
+      "loss": 1.0577,
+      "step": 6234
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0204356170997576e-07,
+      "loss": 1.0674,
+      "step": 6235
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0133489695802967e-07,
+      "loss": 0.9243,
+      "step": 6236
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0062868899833989e-07,
+      "loss": 1.0075,
+      "step": 6237
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.992493800618397e-08,
+      "loss": 1.0707,
+      "step": 6238
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.922364415623109e-08,
+      "loss": 0.9059,
+      "step": 6239
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.852480762253646e-08,
+      "loss": 0.8646,
+      "step": 6240
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.782842857854691e-08,
+      "loss": 0.9365,
+      "step": 6241
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.713450719710082e-08,
+      "loss": 0.9455,
+      "step": 6242
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.64430436504249e-08,
+      "loss": 0.906,
+      "step": 6243
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.57540381101374e-08,
+      "loss": 0.9519,
+      "step": 6244
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.506749074724375e-08,
+      "loss": 0.8858,
+      "step": 6245
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.438340173214322e-08,
+      "loss": 0.9285,
+      "step": 6246
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.37017712346211e-08,
+      "loss": 1.0004,
+      "step": 6247
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.302259942385649e-08,
+      "loss": 1.0335,
+      "step": 6248
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.234588646841237e-08,
+      "loss": 0.8344,
+      "step": 6249
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.167163253624767e-08,
+      "loss": 0.8432,
+      "step": 6250
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.099983779470856e-08,
+      "loss": 0.9351,
+      "step": 6251
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.033050241053054e-08,
+      "loss": 1.017,
+      "step": 6252
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.966362654983629e-08,
+      "loss": 1.0021,
+      "step": 6253
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.89992103781434e-08,
+      "loss": 0.8549,
+      "step": 6254
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.833725406035443e-08,
+      "loss": 1.0302,
+      "step": 6255
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.76777577607657e-08,
+      "loss": 0.9467,
+      "step": 6256
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.702072164305741e-08,
+      "loss": 1.0422,
+      "step": 6257
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.636614587030357e-08,
+      "loss": 0.8476,
+      "step": 6258
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.571403060496641e-08,
+      "loss": 1.0739,
+      "step": 6259
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.50643760088965e-08,
+      "loss": 0.8678,
+      "step": 6260
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.441718224333484e-08,
+      "loss": 1.0343,
+      "step": 6261
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.377244946891183e-08,
+      "loss": 0.912,
+      "step": 6262
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.313017784564503e-08,
+      "loss": 0.8726,
+      "step": 6263
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.24903675329447e-08,
+      "loss": 0.9629,
+      "step": 6264
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.185301868960494e-08,
+      "loss": 1.0174,
+      "step": 6265
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.121813147381474e-08,
+      "loss": 0.8563,
+      "step": 6266
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.058570604314919e-08,
+      "loss": 0.8218,
+      "step": 6267
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.995574255457162e-08,
+      "loss": 0.8287,
+      "step": 6268
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.932824116443694e-08,
+      "loss": 0.7947,
+      "step": 6269
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.870320202848502e-08,
+      "loss": 0.9611,
+      "step": 6270
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.808062530184956e-08,
+      "loss": 0.9785,
+      "step": 6271
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.746051113904917e-08,
+      "loss": 0.8699,
+      "step": 6272
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.684285969399297e-08,
+      "loss": 0.8426,
+      "step": 6273
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.62276711199772e-08,
+      "loss": 0.9584,
+      "step": 6274
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.561494556969085e-08,
+      "loss": 0.9321,
+      "step": 6275
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.500468319520781e-08,
+      "loss": 0.9398,
+      "step": 6276
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.439688414799029e-08,
+      "loss": 1.0271,
+      "step": 6277
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.379154857889203e-08,
+      "loss": 1.0245,
+      "step": 6278
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.318867663815287e-08,
+      "loss": 1.1185,
+      "step": 6279
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.258826847540424e-08,
+      "loss": 0.9746,
+      "step": 6280
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.199032423966135e-08,
+      "loss": 0.8676,
+      "step": 6281
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.139484407933328e-08,
+      "loss": 0.9186,
+      "step": 6282
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.080182814221181e-08,
+      "loss": 1.0022,
+      "step": 6283
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 7.021127657548255e-08,
+      "loss": 0.9646,
+      "step": 6284
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.96231895257149e-08,
+      "loss": 0.8879,
+      "step": 6285
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.9037567138871e-08,
+      "loss": 0.9141,
+      "step": 6286
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.845440956029792e-08,
+      "loss": 0.9535,
+      "step": 6287
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.787371693473211e-08,
+      "loss": 1.0484,
+      "step": 6288
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.729548940629715e-08,
+      "loss": 0.9379,
+      "step": 6289
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.671972711850827e-08,
+      "loss": 0.9716,
+      "step": 6290
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.614643021426336e-08,
+      "loss": 0.789,
+      "step": 6291
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.557559883585418e-08,
+      "loss": 0.8879,
+      "step": 6292
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.500723312495517e-08,
+      "loss": 0.8896,
+      "step": 6293
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.444133322263457e-08,
+      "loss": 0.7886,
+      "step": 6294
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.38778992693434e-08,
+      "loss": 0.9047,
+      "step": 6295
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.331693140492312e-08,
+      "loss": 1.018,
+      "step": 6296
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.275842976860236e-08,
+      "loss": 1.0065,
+      "step": 6297
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.2202394498998e-08,
+      "loss": 1.0405,
+      "step": 6298
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.164882573411635e-08,
+      "loss": 1.0209,
+      "step": 6299
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.109772361134969e-08,
+      "loss": 0.7808,
+      "step": 6300
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.05490882674764e-08,
+      "loss": 0.9176,
+      "step": 6301
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.000291983866647e-08,
+      "loss": 0.8979,
+      "step": 6302
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.945921846047476e-08,
+      "loss": 0.9806,
+      "step": 6303
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.8917984267846674e-08,
+      "loss": 1.0004,
+      "step": 6304
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.837921739511143e-08,
+      "loss": 1.036,
+      "step": 6305
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.784291797598873e-08,
+      "loss": 0.8538,
+      "step": 6306
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.7309086143585433e-08,
+      "loss": 0.9898,
+      "step": 6307
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.6777722030394445e-08,
+      "loss": 1.0936,
+      "step": 6308
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.624882576829804e-08,
+      "loss": 0.9066,
+      "step": 6309
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.572239748856456e-08,
+      "loss": 1.0214,
+      "step": 6310
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.519843732185171e-08,
+      "loss": 0.9293,
+      "step": 6311
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.467694539820323e-08,
+      "loss": 0.9398,
+      "step": 6312
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.4157921847050046e-08,
+      "loss": 0.919,
+      "step": 6313
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.364136679721021e-08,
+      "loss": 0.8614,
+      "step": 6314
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.312728037689119e-08,
+      "loss": 0.9231,
+      "step": 6315
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.261566271368534e-08,
+      "loss": 0.8466,
+      "step": 6316
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.2106513934572226e-08,
+      "loss": 0.8932,
+      "step": 6317
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.159983416592185e-08,
+      "loss": 0.8036,
+      "step": 6318
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.109562353348807e-08,
+      "loss": 1.0221,
+      "step": 6319
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.059388216241301e-08,
+      "loss": 0.8967,
+      "step": 6320
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.0094610177227054e-08,
+      "loss": 0.8631,
+      "step": 6321
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.959780770184441e-08,
+      "loss": 0.9532,
+      "step": 6322
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.91034748595709e-08,
+      "loss": 1.2415,
+      "step": 6323
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.8611611773096147e-08,
+      "loss": 0.8126,
+      "step": 6324
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.812221856449917e-08,
+      "loss": 0.9329,
+      "step": 6325
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.7635295355241695e-08,
+      "loss": 0.8497,
+      "step": 6326
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.715084226617705e-08,
+      "loss": 0.9192,
+      "step": 6327
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.6668859417544624e-08,
+      "loss": 1.0003,
+      "step": 6328
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.618934692896759e-08,
+      "loss": 0.9533,
+      "step": 6329
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.571230491945966e-08,
+      "loss": 0.8645,
+      "step": 6330
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.5237733507420555e-08,
+      "loss": 1.0442,
+      "step": 6331
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.476563281063495e-08,
+      "loss": 0.9062,
+      "step": 6332
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.429600294627467e-08,
+      "loss": 0.9856,
+      "step": 6333
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.382884403090204e-08,
+      "loss": 1.0109,
+      "step": 6334
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.336415618046097e-08,
+      "loss": 0.8415,
+      "step": 6335
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.290193951028476e-08,
+      "loss": 0.8253,
+      "step": 6336
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.244219413509387e-08,
+      "loss": 0.8598,
+      "step": 6337
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.1984920168992585e-08,
+      "loss": 0.9313,
+      "step": 6338
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.153011772547566e-08,
+      "loss": 0.9063,
+      "step": 6339
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.10777869174217e-08,
+      "loss": 0.776,
+      "step": 6340
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.062792785709646e-08,
+      "loss": 0.9658,
+      "step": 6341
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 4.018054065615284e-08,
+      "loss": 0.9797,
+      "step": 6342
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.973562542562981e-08,
+      "loss": 0.8949,
+      "step": 6343
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.9293182275952354e-08,
+      "loss": 0.9501,
+      "step": 6344
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.8853211316932645e-08,
+      "loss": 0.8876,
+      "step": 6345
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.841571265776778e-08,
+      "loss": 0.9989,
+      "step": 6346
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.798068640704422e-08,
+      "loss": 0.9752,
+      "step": 6347
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.7548132672732276e-08,
+      "loss": 0.8573,
+      "step": 6348
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.7118051562189394e-08,
+      "loss": 0.9181,
+      "step": 6349
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.6690443182159086e-08,
+      "loss": 0.928,
+      "step": 6350
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.6265307638770895e-08,
+      "loss": 0.8909,
+      "step": 6351
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.5842645037542646e-08,
+      "loss": 0.8647,
+      "step": 6352
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.542245548337486e-08,
+      "loss": 0.8916,
+      "step": 6353
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.500473908055635e-08,
+      "loss": 0.9689,
+      "step": 6354
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.458949593276417e-08,
+      "loss": 0.9187,
+      "step": 6355
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.4176726143057003e-08,
+      "loss": 0.951,
+      "step": 6356
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.37664298138829e-08,
+      "loss": 0.8714,
+      "step": 6357
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.335860704707372e-08,
+      "loss": 0.9997,
+      "step": 6358
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.2953257943850735e-08,
+      "loss": 1.0158,
+      "step": 6359
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.255038260481902e-08,
+      "loss": 0.8631,
+      "step": 6360
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.214998112996859e-08,
+      "loss": 0.9768,
+      "step": 6361
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.175205361867773e-08,
+      "loss": 0.8933,
+      "step": 6362
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.135660016970965e-08,
+      "loss": 0.8032,
+      "step": 6363
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.096362088121363e-08,
+      "loss": 0.9267,
+      "step": 6364
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.057311585072498e-08,
+      "loss": 0.8182,
+      "step": 6365
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.018508517516505e-08,
+      "loss": 0.8763,
+      "step": 6366
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.979952895084015e-08,
+      "loss": 1.0071,
+      "step": 6367
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9416447273443726e-08,
+      "loss": 0.8833,
+      "step": 6368
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.903584023805306e-08,
+      "loss": 0.79,
+      "step": 6369
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8657707939134804e-08,
+      "loss": 0.9357,
+      "step": 6370
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8282050470538337e-08,
+      "loss": 1.0361,
+      "step": 6371
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.790886792549907e-08,
+      "loss": 0.9329,
+      "step": 6372
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.75381603966407e-08,
+      "loss": 0.8699,
+      "step": 6373
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7169927975967403e-08,
+      "loss": 1.0151,
+      "step": 6374
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6804170754876068e-08,
+      "loss": 1.0774,
+      "step": 6375
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6440888824142973e-08,
+      "loss": 0.8628,
+      "step": 6376
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.608008227393266e-08,
+      "loss": 1.0119,
+      "step": 6377
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5721751193796827e-08,
+      "loss": 0.9189,
+      "step": 6378
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5365895672671003e-08,
+      "loss": 0.9464,
+      "step": 6379
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5012515798875648e-08,
+      "loss": 0.9054,
+      "step": 6380
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.466161166011838e-08,
+      "loss": 0.9696,
+      "step": 6381
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.431318334348953e-08,
+      "loss": 0.822,
+      "step": 6382
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3967230935469933e-08,
+      "loss": 0.9075,
+      "step": 6383
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3623754521922003e-08,
+      "loss": 0.96,
+      "step": 6384
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3282754188093114e-08,
+      "loss": 0.9797,
+      "step": 6385
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2944230018620005e-08,
+      "loss": 1.0044,
+      "step": 6386
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2608182097519914e-08,
+      "loss": 0.9948,
+      "step": 6387
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.227461050820057e-08,
+      "loss": 0.9863,
+      "step": 6388
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1943515333451294e-08,
+      "loss": 0.8357,
+      "step": 6389
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.161489665544747e-08,
+      "loss": 0.9589,
+      "step": 6390
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1288754555750523e-08,
+      "loss": 1.06,
+      "step": 6391
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0965089115309035e-08,
+      "loss": 0.7979,
+      "step": 6392
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0643900414452077e-08,
+      "loss": 0.8185,
+      "step": 6393
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.032518853289922e-08,
+      "loss": 0.9655,
+      "step": 6394
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.000895354975052e-08,
+      "loss": 0.97,
+      "step": 6395
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9695195543496525e-08,
+      "loss": 0.879,
+      "step": 6396
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.938391459200828e-08,
+      "loss": 0.998,
+      "step": 6397
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9075110772543982e-08,
+      "loss": 0.9521,
+      "step": 6398
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8768784161747878e-08,
+      "loss": 0.9867,
+      "step": 6399
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.846493483564693e-08,
+      "loss": 1.0525,
+      "step": 6400
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.816356286965748e-08,
+      "loss": 0.9524,
+      "step": 6401
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7864668338575254e-08,
+      "loss": 1.103,
+      "step": 6402
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.756825131658646e-08,
+      "loss": 0.9856,
+      "step": 6403
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.727431187726003e-08,
+      "loss": 0.8838,
+      "step": 6404
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6982850093547608e-08,
+      "loss": 0.9987,
+      "step": 6405
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.669386603779133e-08,
+      "loss": 1.0171,
+      "step": 6406
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6407359781713817e-08,
+      "loss": 0.9847,
+      "step": 6407
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6123331396424857e-08,
+      "loss": 0.8747,
+      "step": 6408
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5841780952418063e-08,
+      "loss": 1.0115,
+      "step": 6409
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5562708519573088e-08,
+      "loss": 1.077,
+      "step": 6410
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.528611416715342e-08,
+      "loss": 0.9758,
+      "step": 6411
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.50119979638097e-08,
+      "loss": 0.9471,
+      "step": 6412
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.474035997757528e-08,
+      "loss": 0.9733,
+      "step": 6413
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4471200275868458e-08,
+      "loss": 0.7609,
+      "step": 6414
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4204518925493572e-08,
+      "loss": 1.0538,
+      "step": 6415
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3940315992639897e-08,
+      "loss": 0.9617,
+      "step": 6416
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3678591542880537e-08,
+      "loss": 1.0994,
+      "step": 6417
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.341934564117464e-08,
+      "loss": 0.9133,
+      "step": 6418
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3162578351864075e-08,
+      "loss": 1.1389,
+      "step": 6419
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2908289738678969e-08,
+      "loss": 0.8725,
+      "step": 6420
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2656479864731064e-08,
+      "loss": 0.9774,
+      "step": 6421
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2407148792519252e-08,
+      "loss": 1.0819,
+      "step": 6422
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2160296583925146e-08,
+      "loss": 1.0327,
+      "step": 6423
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1915923300215293e-08,
+      "loss": 1.0444,
+      "step": 6424
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1674029002043396e-08,
+      "loss": 0.9371,
+      "step": 6425
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1434613749444768e-08,
+      "loss": 0.9582,
+      "step": 6426
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1197677601841872e-08,
+      "loss": 0.9274,
+      "step": 6427
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0963220618041004e-08,
+      "loss": 0.9921,
+      "step": 6428
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0731242856233392e-08,
+      "loss": 1.0065,
+      "step": 6429
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0501744373992983e-08,
+      "loss": 1.1466,
+      "step": 6430
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.027472522828088e-08,
+      "loss": 0.8365,
+      "step": 6431
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0050185475443119e-08,
+      "loss": 0.9652,
+      "step": 6432
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.828125171207347e-09,
+      "loss": 0.92,
+      "step": 6433
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.608544370688366e-09,
+      "loss": 0.7573,
+      "step": 6434
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.391443128384802e-09,
+      "loss": 0.8696,
+      "step": 6435
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.176821498180222e-09,
+      "loss": 0.9467,
+      "step": 6436
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.964679533342013e-09,
+      "loss": 0.8789,
+      "step": 6437
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.755017286523615e-09,
+      "loss": 0.8738,
+      "step": 6438
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.54783480976118e-09,
+      "loss": 0.9999,
+      "step": 6439
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.343132154475797e-09,
+      "loss": 1.0927,
+      "step": 6440
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.140909371474603e-09,
+      "loss": 0.9057,
+      "step": 6441
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.94116651094745e-09,
+      "loss": 0.9808,
+      "step": 6442
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.74390362246913e-09,
+      "loss": 0.8965,
+      "step": 6443
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.549120755000472e-09,
+      "loss": 0.903,
+      "step": 6444
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.356817956882811e-09,
+      "loss": 0.9828,
+      "step": 6445
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.1669952758479655e-09,
+      "loss": 0.9431,
+      "step": 6446
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.979652759007139e-09,
+      "loss": 0.9392,
+      "step": 6447
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.79479045285647e-09,
+      "loss": 0.959,
+      "step": 6448
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.612408403280368e-09,
+      "loss": 0.9182,
+      "step": 6449
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.432506655542625e-09,
+      "loss": 0.9257,
+      "step": 6450
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.255085254295301e-09,
+      "loss": 0.9607,
+      "step": 6451
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.080144243573172e-09,
+      "loss": 0.9535,
+      "step": 6452
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.9076836667959494e-09,
+      "loss": 0.9613,
+      "step": 6453
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.737703566766062e-09,
+      "loss": 1.0515,
+      "step": 6454
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.5702039856719845e-09,
+      "loss": 0.8323,
+      "step": 6455
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.405184965087129e-09,
+      "loss": 0.9714,
+      "step": 6456
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.242646545967623e-09,
+      "loss": 0.8557,
+      "step": 6457
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.082588768654529e-09,
+      "loss": 1.0632,
+      "step": 6458
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.925011672872737e-09,
+      "loss": 0.9218,
+      "step": 6459
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.769915297733185e-09,
+      "loss": 0.8755,
+      "step": 6460
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.6172996817295255e-09,
+      "loss": 1.0468,
+      "step": 6461
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.467164862739237e-09,
+      "loss": 0.8862,
+      "step": 6462
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.319510878025846e-09,
+      "loss": 0.9207,
+      "step": 6463
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.174337764236702e-09,
+      "loss": 0.9835,
+      "step": 6464
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.031645557400765e-09,
+      "loss": 0.932,
+      "step": 6465
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.891434292936369e-09,
+      "loss": 0.9657,
+      "step": 6466
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.753704005640124e-09,
+      "loss": 0.9536,
+      "step": 6467
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.6184547296991278e-09,
+      "loss": 0.8576,
+      "step": 6468
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.485686498678753e-09,
+      "loss": 0.8569,
+      "step": 6469
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3553993455315294e-09,
+      "loss": 1.0756,
+      "step": 6470
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.2275933025960326e-09,
+      "loss": 0.9881,
+      "step": 6471
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.1022684015913352e-09,
+      "loss": 0.8453,
+      "step": 6472
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9794246736236652e-09,
+      "loss": 0.9181,
+      "step": 6473
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.8590621491797476e-09,
+      "loss": 0.915,
+      "step": 6474
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7411808581356837e-09,
+      "loss": 0.8898,
+      "step": 6475
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.62578082974696e-09,
+      "loss": 0.8299,
+      "step": 6476
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.5128620926551107e-09,
+      "loss": 0.9955,
+      "step": 6477
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.4024246748877155e-09,
+      "loss": 0.9197,
+      "step": 6478
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.294468603853961e-09,
+      "loss": 0.8397,
+      "step": 6479
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.188993906346859e-09,
+      "loss": 0.813,
+      "step": 6480
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0860006085454687e-09,
+      "loss": 1.0804,
+      "step": 6481
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9854887360126752e-09,
+      "loss": 0.9416,
+      "step": 6482
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8874583136940796e-09,
+      "loss": 1.0466,
+      "step": 6483
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7919093659202192e-09,
+      "loss": 0.9479,
+      "step": 6484
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.6988419164065684e-09,
+      "loss": 0.951,
+      "step": 6485
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.6082559882524274e-09,
+      "loss": 0.9937,
+      "step": 6486
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.520151603938702e-09,
+      "loss": 0.92,
+      "step": 6487
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4345287853334556e-09,
+      "loss": 1.0934,
+      "step": 6488
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3513875536874666e-09,
+      "loss": 0.8994,
+      "step": 6489
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2707279296375608e-09,
+      "loss": 0.9769,
+      "step": 6490
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1925499331999491e-09,
+      "loss": 0.8002,
+      "step": 6491
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.11685358378133e-09,
+      "loss": 0.9705,
+      "step": 6492
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0436389001666769e-09,
+      "loss": 0.9994,
+      "step": 6493
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.729059005281205e-10,
+      "loss": 0.9589,
+      "step": 6494
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.046546024216174e-10,
+      "loss": 0.8892,
+      "step": 6495
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.388850227869505e-10,
+      "loss": 0.9413,
+      "step": 6496
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.755971779466187e-10,
+      "loss": 0.9153,
+      "step": 6497
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.147910836091676e-10,
+      "loss": 0.9049,
+      "step": 6498
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.56466754866969e-10,
+      "loss": 0.9226,
+      "step": 6499
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.006242061940004e-10,
+      "loss": 0.9959,
+      "step": 6500
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.472634514513963e-10,
+      "loss": 0.9275,
+      "step": 6501
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.963845038818971e-10,
+      "loss": 0.7794,
+      "step": 6502
+    },
+    {
+      "epoch": 1.0,
+      "step": 6502,
+      "total_flos": 1901173095284736.0,
+      "train_loss": 1.0183198589869404,
+      "train_runtime": 56361.7935,
+      "train_samples_per_second": 11.076,
+      "train_steps_per_second": 0.115
+    }
+  ],
+  "max_steps": 6502,
+  "num_train_epochs": 1,
+  "total_flos": 1901173095284736.0,
+  "trial_name": null,
+  "trial_params": null
+}