diff --git "a/EgoQA-GeLM-7B/trainer_state.json" "b/EgoQA-GeLM-7B/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/EgoQA-GeLM-7B/trainer_state.json"
@@ -0,0 +1,9808 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 10.0,
+  "eval_steps": 500,
+  "global_step": 1630,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0816326530612243e-07,
+      "loss": 11.0156,
+      "step": 1
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.163265306122449e-07,
+      "loss": 10.5312,
+      "step": 2
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2244897959183673e-06,
+      "loss": 10.9531,
+      "step": 3
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6326530612244897e-06,
+      "loss": 10.4062,
+      "step": 4
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.0408163265306125e-06,
+      "loss": 10.5156,
+      "step": 5
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.4489795918367347e-06,
+      "loss": 10.5156,
+      "step": 6
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 2.8571428571428573e-06,
+      "loss": 9.8281,
+      "step": 7
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 3.2653061224489794e-06,
+      "loss": 8.9531,
+      "step": 8
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 3.6734693877551024e-06,
+      "loss": 9.0156,
+      "step": 9
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.081632653061225e-06,
+      "loss": 8.9062,
+      "step": 10
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.489795918367348e-06,
+      "loss": 8.0781,
+      "step": 11
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 4.897959183673469e-06,
+      "loss": 7.625,
+      "step": 12
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.306122448979593e-06,
+      "loss": 7.7188,
+      "step": 13
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 5.7142857142857145e-06,
+      "loss": 7.0391,
+      "step": 14
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.122448979591837e-06,
+      "loss": 6.6875,
+      "step": 15
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 6.530612244897959e-06,
+      "loss": 6.4922,
+      "step": 16
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 6.938775510204082e-06,
+      "loss": 6.1953,
+      "step": 17
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 7.346938775510205e-06,
+      "loss": 6.2578,
+      "step": 18
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 7.755102040816327e-06,
+      "loss": 5.8906,
+      "step": 19
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 8.16326530612245e-06,
+      "loss": 5.7656,
+      "step": 20
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.571428571428571e-06,
+      "loss": 5.4844,
+      "step": 21
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 8.979591836734695e-06,
+      "loss": 5.4922,
+      "step": 22
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.387755102040818e-06,
+      "loss": 5.3594,
+      "step": 23
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.795918367346939e-06,
+      "loss": 5.0625,
+      "step": 24
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.0204081632653063e-05,
+      "loss": 4.9219,
+      "step": 25
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.0612244897959186e-05,
+      "loss": 4.9609,
+      "step": 26
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1020408163265306e-05,
+      "loss": 4.6172,
+      "step": 27
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.1428571428571429e-05,
+      "loss": 4.7578,
+      "step": 28
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.1836734693877552e-05,
+      "loss": 4.4219,
+      "step": 29
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.2244897959183674e-05,
+      "loss": 4.4688,
+      "step": 30
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.2653061224489798e-05,
+      "loss": 4.8281,
+      "step": 31
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.3061224489795918e-05,
+      "loss": 4.6484,
+      "step": 32
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.3469387755102042e-05,
+      "loss": 4.6172,
+      "step": 33
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.3877551020408165e-05,
+      "loss": 4.5938,
+      "step": 34
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.4285714285714287e-05,
+      "loss": 4.5156,
+      "step": 35
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.469387755102041e-05,
+      "loss": 4.5938,
+      "step": 36
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.510204081632653e-05,
+      "loss": 4.2109,
+      "step": 37
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.5510204081632655e-05,
+      "loss": 4.1094,
+      "step": 38
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.5918367346938776e-05,
+      "loss": 4.0742,
+      "step": 39
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.63265306122449e-05,
+      "loss": 4.1406,
+      "step": 40
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.673469387755102e-05,
+      "loss": 3.9453,
+      "step": 41
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7142857142857142e-05,
+      "loss": 3.7773,
+      "step": 42
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7551020408163266e-05,
+      "loss": 4.0781,
+      "step": 43
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.795918367346939e-05,
+      "loss": 4.3906,
+      "step": 44
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.836734693877551e-05,
+      "loss": 4.2031,
+      "step": 45
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8775510204081636e-05,
+      "loss": 4.0469,
+      "step": 46
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9183673469387756e-05,
+      "loss": 4.3359,
+      "step": 47
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9591836734693877e-05,
+      "loss": 3.9688,
+      "step": 48
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 2e-05,
+      "loss": 4.1289,
+      "step": 49
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9999980257330883e-05,
+      "loss": 4.2734,
+      "step": 50
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9999921029401478e-05,
+      "loss": 4.2344,
+      "step": 51
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9999822316445652e-05,
+      "loss": 4.1797,
+      "step": 52
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9999684118853177e-05,
+      "loss": 4.0859,
+      "step": 53
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9999506437169723e-05,
+      "loss": 4.082,
+      "step": 54
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9999289272096886e-05,
+      "loss": 4.1875,
+      "step": 55
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9999032624492144e-05,
+      "loss": 3.9062,
+      "step": 56
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.999873649536887e-05,
+      "loss": 4.0664,
+      "step": 57
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9998400885896355e-05,
+      "loss": 3.9922,
+      "step": 58
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9998025797399753e-05,
+      "loss": 4.0312,
+      "step": 59
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9997611231360117e-05,
+      "loss": 3.9297,
+      "step": 60
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9997157189414373e-05,
+      "loss": 4.0781,
+      "step": 61
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9996663673355326e-05,
+      "loss": 3.8125,
+      "step": 62
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9996130685131637e-05,
+      "loss": 3.9375,
+      "step": 63
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.999555822684783e-05,
+      "loss": 4.1602,
+      "step": 64
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9994946300764276e-05,
+      "loss": 3.9336,
+      "step": 65
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.999429490929718e-05,
+      "loss": 4.207,
+      "step": 66
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.999360405501859e-05,
+      "loss": 3.7422,
+      "step": 67
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9992873740656372e-05,
+      "loss": 3.9414,
+      "step": 68
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9992103969094182e-05,
+      "loss": 3.8711,
+      "step": 69
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.99912947433715e-05,
+      "loss": 4.25,
+      "step": 70
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.999044606668358e-05,
+      "loss": 3.8203,
+      "step": 71
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.998955794238145e-05,
+      "loss": 3.9688,
+      "step": 72
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9988630373971896e-05,
+      "loss": 3.9414,
+      "step": 73
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9987663365117456e-05,
+      "loss": 3.5312,
+      "step": 74
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.99866569196364e-05,
+      "loss": 3.8789,
+      "step": 75
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9985611041502704e-05,
+      "loss": 3.9062,
+      "step": 76
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9984525734846056e-05,
+      "loss": 3.75,
+      "step": 77
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.998340100395183e-05,
+      "loss": 3.8984,
+      "step": 78
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9982236853261067e-05,
+      "loss": 4.0781,
+      "step": 79
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9981033287370443e-05,
+      "loss": 3.8672,
+      "step": 80
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9979790311032288e-05,
+      "loss": 3.7461,
+      "step": 81
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9978507929154534e-05,
+      "loss": 3.6602,
+      "step": 82
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9977186146800707e-05,
+      "loss": 3.8555,
+      "step": 83
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9975824969189913e-05,
+      "loss": 3.8086,
+      "step": 84
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.997442440169681e-05,
+      "loss": 3.9727,
+      "step": 85
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.997298444985158e-05,
+      "loss": 3.6172,
+      "step": 86
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9971505119339923e-05,
+      "loss": 3.8359,
+      "step": 87
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9969986416003026e-05,
+      "loss": 3.8594,
+      "step": 88
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9968428345837542e-05,
+      "loss": 3.7227,
+      "step": 89
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.996683091499556e-05,
+      "loss": 4.125,
+      "step": 90
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9965194129784597e-05,
+      "loss": 3.832,
+      "step": 91
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9963517996667548e-05,
+      "loss": 4.0,
+      "step": 92
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9961802522262685e-05,
+      "loss": 4.0703,
+      "step": 93
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.996004771334361e-05,
+      "loss": 3.7461,
+      "step": 94
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9958253576839256e-05,
+      "loss": 3.9727,
+      "step": 95
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9956420119833826e-05,
+      "loss": 4.0664,
+      "step": 96
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9954547349566783e-05,
+      "loss": 4.2539,
+      "step": 97
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9952635273432835e-05,
+      "loss": 4.0156,
+      "step": 98
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9950683898981866e-05,
+      "loss": 4.1406,
+      "step": 99
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.994869323391895e-05,
+      "loss": 4.1523,
+      "step": 100
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9946663286104303e-05,
+      "loss": 3.9023,
+      "step": 101
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.994459406355323e-05,
+      "loss": 3.8086,
+      "step": 102
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.994248557443613e-05,
+      "loss": 4.0391,
+      "step": 103
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9940337827078448e-05,
+      "loss": 3.9453,
+      "step": 104
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9938150829960634e-05,
+      "loss": 4.0039,
+      "step": 105
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.993592459171812e-05,
+      "loss": 3.9883,
+      "step": 106
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9933659121141283e-05,
+      "loss": 3.6758,
+      "step": 107
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.993135442717541e-05,
+      "loss": 3.793,
+      "step": 108
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9929010518920667e-05,
+      "loss": 3.7383,
+      "step": 109
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9926627405632048e-05,
+      "loss": 3.7227,
+      "step": 110
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.992420509671936e-05,
+      "loss": 3.9023,
+      "step": 111
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.992174360174717e-05,
+      "loss": 4.0078,
+      "step": 112
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.991924293043478e-05,
+      "loss": 4.2109,
+      "step": 113
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.991670309265617e-05,
+      "loss": 3.7461,
+      "step": 114
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9914124098439976e-05,
+      "loss": 4.0039,
+      "step": 115
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9911505957969443e-05,
+      "loss": 3.8867,
+      "step": 116
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.990884868158239e-05,
+      "loss": 3.9883,
+      "step": 117
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9906152279771162e-05,
+      "loss": 3.8359,
+      "step": 118
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.990341676318259e-05,
+      "loss": 3.6719,
+      "step": 119
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9900642142617958e-05,
+      "loss": 3.5898,
+      "step": 120
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9897828429032946e-05,
+      "loss": 3.9922,
+      "step": 121
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.98949756335376e-05,
+      "loss": 3.8711,
+      "step": 122
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9892083767396274e-05,
+      "loss": 3.6797,
+      "step": 123
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9889152842027607e-05,
+      "loss": 4.0078,
+      "step": 124
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9886182869004447e-05,
+      "loss": 3.8164,
+      "step": 125
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9883173860053845e-05,
+      "loss": 3.6953,
+      "step": 126
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9880125827056967e-05,
+      "loss": 3.7344,
+      "step": 127
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9877038782049074e-05,
+      "loss": 3.6562,
+      "step": 128
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9873912737219468e-05,
+      "loss": 3.5625,
+      "step": 129
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.987074770491145e-05,
+      "loss": 4.0859,
+      "step": 130
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9867543697622248e-05,
+      "loss": 3.7344,
+      "step": 131
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9864300728002997e-05,
+      "loss": 3.9453,
+      "step": 132
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.986101880885867e-05,
+      "loss": 4.1211,
+      "step": 133
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.985769795314804e-05,
+      "loss": 3.7344,
+      "step": 134
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9854338173983615e-05,
+      "loss": 3.6875,
+      "step": 135
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9850939484631598e-05,
+      "loss": 3.8125,
+      "step": 136
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9847501898511824e-05,
+      "loss": 3.707,
+      "step": 137
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.9844025429197727e-05,
+      "loss": 4.0781,
+      "step": 138
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.984051009041626e-05,
+      "loss": 3.8281,
+      "step": 139
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.983695589604785e-05,
+      "loss": 4.0391,
+      "step": 140
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.9833362860126364e-05,
+      "loss": 3.6719,
+      "step": 141
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.982973099683902e-05,
+      "loss": 4.0898,
+      "step": 142
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.9826060320526355e-05,
+      "loss": 3.8281,
+      "step": 143
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.982235084568216e-05,
+      "loss": 3.9219,
+      "step": 144
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.9818602586953414e-05,
+      "loss": 3.9961,
+      "step": 145
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9814815559140258e-05,
+      "loss": 3.8125,
+      "step": 146
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.9810989777195884e-05,
+      "loss": 3.8164,
+      "step": 147
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9807125256226532e-05,
+      "loss": 4.1094,
+      "step": 148
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9803222011491385e-05,
+      "loss": 3.4805,
+      "step": 149
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.979928005840255e-05,
+      "loss": 3.7305,
+      "step": 150
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9795299412524948e-05,
+      "loss": 3.793,
+      "step": 151
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.9791280089576302e-05,
+      "loss": 4.0312,
+      "step": 152
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.978722210542704e-05,
+      "loss": 3.6953,
+      "step": 153
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9783125476100254e-05,
+      "loss": 3.7891,
+      "step": 154
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.977899021777162e-05,
+      "loss": 3.6523,
+      "step": 155
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.977481634676935e-05,
+      "loss": 3.9414,
+      "step": 156
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.9770603879574108e-05,
+      "loss": 3.9609,
+      "step": 157
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9766352832818972e-05,
+      "loss": 3.4336,
+      "step": 158
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9762063223289334e-05,
+      "loss": 3.6484,
+      "step": 159
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.975773506792287e-05,
+      "loss": 3.8281,
+      "step": 160
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9753368383809445e-05,
+      "loss": 3.7578,
+      "step": 161
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.974896318819106e-05,
+      "loss": 3.8555,
+      "step": 162
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.974451949846177e-05,
+      "loss": 3.7617,
+      "step": 163
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.974003733216765e-05,
+      "loss": 3.5039,
+      "step": 164
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.9735516707006676e-05,
+      "loss": 3.7344,
+      "step": 165
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.973095764082869e-05,
+      "loss": 3.6172,
+      "step": 166
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.972636015163532e-05,
+      "loss": 3.7734,
+      "step": 167
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.9721724257579907e-05,
+      "loss": 3.543,
+      "step": 168
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.9717049976967437e-05,
+      "loss": 3.7031,
+      "step": 169
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.971233732825446e-05,
+      "loss": 3.543,
+      "step": 170
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.9707586330049037e-05,
+      "loss": 3.6836,
+      "step": 171
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9702797001110642e-05,
+      "loss": 3.2969,
+      "step": 172
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.9697969360350098e-05,
+      "loss": 3.3789,
+      "step": 173
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.969310342682951e-05,
+      "loss": 3.5625,
+      "step": 174
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.9688199219762183e-05,
+      "loss": 3.9297,
+      "step": 175
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.9683256758512544e-05,
+      "loss": 3.6094,
+      "step": 176
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.967827606259607e-05,
+      "loss": 3.5547,
+      "step": 177
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.96732571516792e-05,
+      "loss": 3.5742,
+      "step": 178
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9668200045579283e-05,
+      "loss": 3.3047,
+      "step": 179
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.9663104764264468e-05,
+      "loss": 3.5117,
+      "step": 180
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.9657971327853644e-05,
+      "loss": 3.4805,
+      "step": 181
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.9652799756616364e-05,
+      "loss": 3.4453,
+      "step": 182
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.964759007097275e-05,
+      "loss": 3.5195,
+      "step": 183
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.964234229149342e-05,
+      "loss": 3.375,
+      "step": 184
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.963705643889941e-05,
+      "loss": 3.4648,
+      "step": 185
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.9631732534062088e-05,
+      "loss": 3.6719,
+      "step": 186
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.962637059800307e-05,
+      "loss": 3.582,
+      "step": 187
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.9620970651894146e-05,
+      "loss": 3.8086,
+      "step": 188
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.9615532717057185e-05,
+      "loss": 3.5234,
+      "step": 189
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9610056814964053e-05,
+      "loss": 3.6016,
+      "step": 190
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.9604542967236535e-05,
+      "loss": 3.6172,
+      "step": 191
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.9598991195646252e-05,
+      "loss": 3.3477,
+      "step": 192
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.959340152211455e-05,
+      "loss": 3.3125,
+      "step": 193
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.9587773968712458e-05,
+      "loss": 3.7891,
+      "step": 194
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.958210855766055e-05,
+      "loss": 3.8008,
+      "step": 195
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.95764053113289e-05,
+      "loss": 3.5156,
+      "step": 196
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.9570664252236966e-05,
+      "loss": 3.9531,
+      "step": 197
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.956488540305351e-05,
+      "loss": 3.3164,
+      "step": 198
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.9559068786596526e-05,
+      "loss": 3.6797,
+      "step": 199
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9553214425833108e-05,
+      "loss": 3.4844,
+      "step": 200
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.9547322343879397e-05,
+      "loss": 3.6641,
+      "step": 201
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.954139256400049e-05,
+      "loss": 3.5195,
+      "step": 202
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9535425109610317e-05,
+      "loss": 3.7773,
+      "step": 203
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.9529420004271568e-05,
+      "loss": 3.3711,
+      "step": 204
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.952337727169561e-05,
+      "loss": 3.8828,
+      "step": 205
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.951729693574238e-05,
+      "loss": 3.5781,
+      "step": 206
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.9511179020420284e-05,
+      "loss": 3.457,
+      "step": 207
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.950502354988612e-05,
+      "loss": 3.5312,
+      "step": 208
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.9498830548444972e-05,
+      "loss": 3.6367,
+      "step": 209
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.9492600040550114e-05,
+      "loss": 3.5625,
+      "step": 210
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.948633205080292e-05,
+      "loss": 3.5703,
+      "step": 211
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.948002660395276e-05,
+      "loss": 3.7461,
+      "step": 212
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9473683724896898e-05,
+      "loss": 3.7148,
+      "step": 213
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.9467303438680414e-05,
+      "loss": 3.5039,
+      "step": 214
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.946088577049608e-05,
+      "loss": 3.5273,
+      "step": 215
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.9454430745684276e-05,
+      "loss": 3.7188,
+      "step": 216
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.944793838973289e-05,
+      "loss": 3.5586,
+      "step": 217
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.94414087282772e-05,
+      "loss": 3.6602,
+      "step": 218
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.9434841787099804e-05,
+      "loss": 3.3633,
+      "step": 219
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.9428237592130487e-05,
+      "loss": 3.2969,
+      "step": 220
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.9421596169446135e-05,
+      "loss": 3.7031,
+      "step": 221
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.941491754527064e-05,
+      "loss": 3.4375,
+      "step": 222
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.940820174597476e-05,
+      "loss": 3.6016,
+      "step": 223
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.9401448798076064e-05,
+      "loss": 3.6406,
+      "step": 224
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.9394658728238797e-05,
+      "loss": 3.5273,
+      "step": 225
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.9387831563273775e-05,
+      "loss": 3.4336,
+      "step": 226
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.938096733013829e-05,
+      "loss": 3.4141,
+      "step": 227
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9374066055936004e-05,
+      "loss": 3.6797,
+      "step": 228
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.9367127767916828e-05,
+      "loss": 3.6953,
+      "step": 229
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.9360152493476828e-05,
+      "loss": 3.6797,
+      "step": 230
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9353140260158108e-05,
+      "loss": 3.5938,
+      "step": 231
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.9346091095648712e-05,
+      "loss": 3.9492,
+      "step": 232
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.93390050277825e-05,
+      "loss": 3.6172,
+      "step": 233
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.9331882084539056e-05,
+      "loss": 3.5977,
+      "step": 234
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.932472229404356e-05,
+      "loss": 3.5703,
+      "step": 235
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.9317525684566686e-05,
+      "loss": 3.4336,
+      "step": 236
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.931029228452449e-05,
+      "loss": 3.5508,
+      "step": 237
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.9303022122478303e-05,
+      "loss": 3.7188,
+      "step": 238
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9295715227134595e-05,
+      "loss": 3.4766,
+      "step": 239
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.9288371627344894e-05,
+      "loss": 3.6484,
+      "step": 240
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9280991352105656e-05,
+      "loss": 3.5703,
+      "step": 241
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.9273574430558143e-05,
+      "loss": 3.4336,
+      "step": 242
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.9266120891988326e-05,
+      "loss": 3.5469,
+      "step": 243
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.925863076582674e-05,
+      "loss": 3.2812,
+      "step": 244
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.9251104081648423e-05,
+      "loss": 3.4102,
+      "step": 245
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.9243540869172724e-05,
+      "loss": 3.332,
+      "step": 246
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.9235941158263253e-05,
+      "loss": 3.5039,
+      "step": 247
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.922830497892772e-05,
+      "loss": 3.4883,
+      "step": 248
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9220632361317843e-05,
+      "loss": 3.5664,
+      "step": 249
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.9212923335729206e-05,
+      "loss": 3.5195,
+      "step": 250
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.920517793260116e-05,
+      "loss": 3.4531,
+      "step": 251
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.9197396182516694e-05,
+      "loss": 3.7734,
+      "step": 252
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.918957811620231e-05,
+      "loss": 3.6953,
+      "step": 253
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.9181723764527902e-05,
+      "loss": 3.6133,
+      "step": 254
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.917383315850665e-05,
+      "loss": 3.5391,
+      "step": 255
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.9165906329294875e-05,
+      "loss": 3.5898,
+      "step": 256
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.9157943308191934e-05,
+      "loss": 3.7188,
+      "step": 257
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.914994412664008e-05,
+      "loss": 3.8125,
+      "step": 258
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.9141908816224356e-05,
+      "loss": 3.875,
+      "step": 259
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9133837408672456e-05,
+      "loss": 3.4102,
+      "step": 260
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.9125729935854606e-05,
+      "loss": 3.2344,
+      "step": 261
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.9117586429783433e-05,
+      "loss": 3.7656,
+      "step": 262
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.910940692261385e-05,
+      "loss": 3.6992,
+      "step": 263
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.9101191446642917e-05,
+      "loss": 3.4766,
+      "step": 264
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.909294003430972e-05,
+      "loss": 3.1211,
+      "step": 265
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.9084652718195237e-05,
+      "loss": 3.4102,
+      "step": 266
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.907632953102222e-05,
+      "loss": 3.6602,
+      "step": 267
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.906797050565505e-05,
+      "loss": 3.6836,
+      "step": 268
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.9059575675099622e-05,
+      "loss": 3.582,
+      "step": 269
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9051145072503216e-05,
+      "loss": 3.6172,
+      "step": 270
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.9042678731154337e-05,
+      "loss": 3.457,
+      "step": 271
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9034176684482638e-05,
+      "loss": 3.3398,
+      "step": 272
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.9025638966058722e-05,
+      "loss": 3.4883,
+      "step": 273
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.901706560959407e-05,
+      "loss": 3.6602,
+      "step": 274
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.900845664894086e-05,
+      "loss": 3.6797,
+      "step": 275
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.8999812118091877e-05,
+      "loss": 3.4766,
+      "step": 276
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.8991132051180332e-05,
+      "loss": 3.3945,
+      "step": 277
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.898241648247977e-05,
+      "loss": 3.2461,
+      "step": 278
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.8973665446403902e-05,
+      "loss": 3.4023,
+      "step": 279
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.8964878977506496e-05,
+      "loss": 3.4492,
+      "step": 280
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.895605711048122e-05,
+      "loss": 3.5,
+      "step": 281
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.8947199880161515e-05,
+      "loss": 3.4531,
+      "step": 282
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.8938307321520453e-05,
+      "loss": 3.6523,
+      "step": 283
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.89293794696706e-05,
+      "loss": 3.6445,
+      "step": 284
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.8920416359863885e-05,
+      "loss": 3.3711,
+      "step": 285
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.8911418027491453e-05,
+      "loss": 3.4414,
+      "step": 286
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.8902384508083518e-05,
+      "loss": 3.2656,
+      "step": 287
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.8893315837309235e-05,
+      "loss": 3.6289,
+      "step": 288
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.8884212050976568e-05,
+      "loss": 3.4023,
+      "step": 289
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.8875073185032116e-05,
+      "loss": 3.6914,
+      "step": 290
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.8865899275561003e-05,
+      "loss": 3.3281,
+      "step": 291
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.885669035878672e-05,
+      "loss": 3.7227,
+      "step": 292
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.8847446471070985e-05,
+      "loss": 3.2891,
+      "step": 293
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.8838167648913606e-05,
+      "loss": 3.4844,
+      "step": 294
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.882885392895232e-05,
+      "loss": 3.7617,
+      "step": 295
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.881950534796267e-05,
+      "loss": 3.3945,
+      "step": 296
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.8810121942857848e-05,
+      "loss": 3.5547,
+      "step": 297
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.8800703750688536e-05,
+      "loss": 3.6484,
+      "step": 298
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.8791250808642792e-05,
+      "loss": 3.668,
+      "step": 299
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.8781763154045873e-05,
+      "loss": 3.5664,
+      "step": 300
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.877224082436011e-05,
+      "loss": 3.2695,
+      "step": 301
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.8762683857184738e-05,
+      "loss": 3.5781,
+      "step": 302
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.8753092290255765e-05,
+      "loss": 3.8359,
+      "step": 303
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8743466161445823e-05,
+      "loss": 3.3242,
+      "step": 304
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.8733805508764e-05,
+      "loss": 3.3086,
+      "step": 305
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.872411037035572e-05,
+      "loss": 3.4531,
+      "step": 306
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.8714380784502553e-05,
+      "loss": 3.5586,
+      "step": 307
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.870461678962211e-05,
+      "loss": 3.6797,
+      "step": 308
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.869481842426784e-05,
+      "loss": 3.4609,
+      "step": 309
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.8684985727128936e-05,
+      "loss": 3.6289,
+      "step": 310
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.8675118737030123e-05,
+      "loss": 3.4844,
+      "step": 311
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.866521749293155e-05,
+      "loss": 3.7461,
+      "step": 312
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.8655282033928618e-05,
+      "loss": 3.2852,
+      "step": 313
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.8645312399251818e-05,
+      "loss": 3.6875,
+      "step": 314
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.8635308628266586e-05,
+      "loss": 3.2266,
+      "step": 315
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.8625270760473164e-05,
+      "loss": 3.5977,
+      "step": 316
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.8615198835506393e-05,
+      "loss": 3.6133,
+      "step": 317
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.8605092893135626e-05,
+      "loss": 3.6172,
+      "step": 318
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.8594952973264512e-05,
+      "loss": 3.4766,
+      "step": 319
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.8584779115930866e-05,
+      "loss": 3.4766,
+      "step": 320
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.857457136130651e-05,
+      "loss": 3.6875,
+      "step": 321
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.856432974969711e-05,
+      "loss": 3.3359,
+      "step": 322
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.855405432154201e-05,
+      "loss": 3.5,
+      "step": 323
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.8543745117414094e-05,
+      "loss": 3.5547,
+      "step": 324
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.8533402178019596e-05,
+      "loss": 3.1367,
+      "step": 325
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.8523025544197964e-05,
+      "loss": 3.4141,
+      "step": 326
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.8512615256921692e-05,
+      "loss": 3.0078,
+      "step": 327
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.8502171357296144e-05,
+      "loss": 3.0586,
+      "step": 328
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.8491693886559413e-05,
+      "loss": 3.1953,
+      "step": 329
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.848118288608215e-05,
+      "loss": 3.0625,
+      "step": 330
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.8470638397367397e-05,
+      "loss": 3.25,
+      "step": 331
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.846006046205042e-05,
+      "loss": 3.2422,
+      "step": 332
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.8449449121898552e-05,
+      "loss": 2.9258,
+      "step": 333
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.8438804418811038e-05,
+      "loss": 2.9883,
+      "step": 334
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.842812639481884e-05,
+      "loss": 3.3203,
+      "step": 335
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.84174150920845e-05,
+      "loss": 3.0195,
+      "step": 336
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.8406670552901958e-05,
+      "loss": 2.9375,
+      "step": 337
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.839589281969639e-05,
+      "loss": 3.2578,
+      "step": 338
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.8385081935024044e-05,
+      "loss": 3.0469,
+      "step": 339
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.837423794157206e-05,
+      "loss": 3.1367,
+      "step": 340
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.836336088215831e-05,
+      "loss": 3.0234,
+      "step": 341
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.835245079973124e-05,
+      "loss": 2.8242,
+      "step": 342
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.834150773736967e-05,
+      "loss": 2.9414,
+      "step": 343
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.8330531738282656e-05,
+      "loss": 3.0742,
+      "step": 344
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.8319522845809306e-05,
+      "loss": 3.0625,
+      "step": 345
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.8308481103418597e-05,
+      "loss": 2.8828,
+      "step": 346
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.8297406554709228e-05,
+      "loss": 3.1836,
+      "step": 347
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.8286299243409424e-05,
+      "loss": 2.8086,
+      "step": 348
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.8275159213376783e-05,
+      "loss": 2.9258,
+      "step": 349
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.826398650859809e-05,
+      "loss": 3.0977,
+      "step": 350
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.8252781173189148e-05,
+      "loss": 3.3086,
+      "step": 351
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.82415432513946e-05,
+      "loss": 3.0117,
+      "step": 352
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.823027278758776e-05,
+      "loss": 2.957,
+      "step": 353
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.821896982627044e-05,
+      "loss": 3.2617,
+      "step": 354
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.8207634412072765e-05,
+      "loss": 3.1172,
+      "step": 355
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.8196266589753e-05,
+      "loss": 2.8867,
+      "step": 356
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.818486640419737e-05,
+      "loss": 3.2539,
+      "step": 357
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.81734339004199e-05,
+      "loss": 2.8633,
+      "step": 358
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.816196912356222e-05,
+      "loss": 3.1016,
+      "step": 359
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.8150472118893382e-05,
+      "loss": 3.0898,
+      "step": 360
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.8138942931809702e-05,
+      "loss": 2.9453,
+      "step": 361
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.8127381607834563e-05,
+      "loss": 3.2383,
+      "step": 362
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.8115788192618247e-05,
+      "loss": 3.0703,
+      "step": 363
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.8104162731937746e-05,
+      "loss": 3.0977,
+      "step": 364
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.8092505271696582e-05,
+      "loss": 3.2344,
+      "step": 365
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.808081585792463e-05,
+      "loss": 2.7617,
+      "step": 366
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.8069094536777938e-05,
+      "loss": 3.0898,
+      "step": 367
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.805734135453854e-05,
+      "loss": 3.0781,
+      "step": 368
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.8045556357614273e-05,
+      "loss": 3.4922,
+      "step": 369
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.8033739592538598e-05,
+      "loss": 3.1211,
+      "step": 370
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.8021891105970405e-05,
+      "loss": 2.9453,
+      "step": 371
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.8010010944693846e-05,
+      "loss": 3.1016,
+      "step": 372
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.7998099155618147e-05,
+      "loss": 3.0117,
+      "step": 373
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.7986155785777402e-05,
+      "loss": 3.1523,
+      "step": 374
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.7974180882330413e-05,
+      "loss": 3.0352,
+      "step": 375
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.7962174492560492e-05,
+      "loss": 2.8711,
+      "step": 376
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.7950136663875274e-05,
+      "loss": 3.1953,
+      "step": 377
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.7938067443806538e-05,
+      "loss": 3.2188,
+      "step": 378
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.7925966880009998e-05,
+      "loss": 2.8203,
+      "step": 379
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.791383502026515e-05,
+      "loss": 3.1172,
+      "step": 380
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.790167191247504e-05,
+      "loss": 2.9414,
+      "step": 381
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.7889477604666124e-05,
+      "loss": 2.8398,
+      "step": 382
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.787725214498803e-05,
+      "loss": 3.1836,
+      "step": 383
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.78649955817134e-05,
+      "loss": 3.0625,
+      "step": 384
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.785270796323769e-05,
+      "loss": 2.8945,
+      "step": 385
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.784038933807898e-05,
+      "loss": 2.9688,
+      "step": 386
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.7828039754877778e-05,
+      "loss": 3.0352,
+      "step": 387
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.7815659262396825e-05,
+      "loss": 3.0977,
+      "step": 388
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.780324790952092e-05,
+      "loss": 3.1445,
+      "step": 389
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.7790805745256703e-05,
+      "loss": 2.9766,
+      "step": 390
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.7778332818732492e-05,
+      "loss": 3.0547,
+      "step": 391
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.7765829179198048e-05,
+      "loss": 3.1758,
+      "step": 392
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.7753294876024417e-05,
+      "loss": 3.0625,
+      "step": 393
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.7740729958703725e-05,
+      "loss": 2.9297,
+      "step": 394
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.7728134476848965e-05,
+      "loss": 3.0586,
+      "step": 395
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.7715508480193832e-05,
+      "loss": 3.0039,
+      "step": 396
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.7702852018592493e-05,
+      "loss": 2.8086,
+      "step": 397
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.769016514201942e-05,
+      "loss": 2.9336,
+      "step": 398
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7677447900569166e-05,
+      "loss": 3.4219,
+      "step": 399
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.7664700344456198e-05,
+      "loss": 3.0625,
+      "step": 400
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.765192252401467e-05,
+      "loss": 3.2617,
+      "step": 401
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.7639114489698238e-05,
+      "loss": 3.0977,
+      "step": 402
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.762627629207986e-05,
+      "loss": 3.0703,
+      "step": 403
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.7613407981851586e-05,
+      "loss": 3.0938,
+      "step": 404
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.760050960982439e-05,
+      "loss": 3.3047,
+      "step": 405
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.758758122692791e-05,
+      "loss": 2.8867,
+      "step": 406
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.757462288421032e-05,
+      "loss": 3.2148,
+      "step": 407
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.7561634632838062e-05,
+      "loss": 3.1172,
+      "step": 408
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.7548616524095697e-05,
+      "loss": 2.9141,
+      "step": 409
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.753556860938566e-05,
+      "loss": 3.0938,
+      "step": 410
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.7522490940228086e-05,
+      "loss": 2.8672,
+      "step": 411
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.7509383568260597e-05,
+      "loss": 3.1641,
+      "step": 412
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.749624654523809e-05,
+      "loss": 2.9883,
+      "step": 413
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.7483079923032543e-05,
+      "loss": 3.0898,
+      "step": 414
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.7469883753632817e-05,
+      "loss": 3.0391,
+      "step": 415
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.745665808914443e-05,
+      "loss": 3.1055,
+      "step": 416
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.744340298178936e-05,
+      "loss": 3.0664,
+      "step": 417
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.743011848390585e-05,
+      "loss": 2.8672,
+      "step": 418
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.7416804647948194e-05,
+      "loss": 3.2891,
+      "step": 419
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.740346152648652e-05,
+      "loss": 2.9805,
+      "step": 420
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.7390089172206594e-05,
+      "loss": 2.7305,
+      "step": 421
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.7376687637909607e-05,
+      "loss": 3.0547,
+      "step": 422
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.7363256976511972e-05,
+      "loss": 2.7773,
+      "step": 423
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.7349797241045115e-05,
+      "loss": 3.2188,
+      "step": 424
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.733630848465525e-05,
+      "loss": 3.0156,
+      "step": 425
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.732279076060319e-05,
+      "loss": 3.1328,
+      "step": 426
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.730924412226413e-05,
+      "loss": 3.0664,
+      "step": 427
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.729566862312742e-05,
+      "loss": 2.9102,
+      "step": 428
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.7282064316796387e-05,
+      "loss": 3.0508,
+      "step": 429
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.726843125698809e-05,
+      "loss": 2.8711,
+      "step": 430
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.7254769497533128e-05,
+      "loss": 2.75,
+      "step": 431
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.724107909237542e-05,
+      "loss": 2.8438,
+      "step": 432
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.7227360095571992e-05,
+      "loss": 2.9883,
+      "step": 433
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.721361256129277e-05,
+      "loss": 3.2461,
+      "step": 434
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.719983654382036e-05,
+      "loss": 3.0781,
+      "step": 435
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.7186032097549822e-05,
+      "loss": 3.1523,
+      "step": 436
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.717219927698849e-05,
+      "loss": 2.832,
+      "step": 437
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.7158338136755724e-05,
+      "loss": 3.2617,
+      "step": 438
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.7144448731582698e-05,
+      "loss": 3.0781,
+      "step": 439
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.7130531116312202e-05,
+      "loss": 3.1641,
+      "step": 440
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.7116585345898413e-05,
+      "loss": 3.1484,
+      "step": 441
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 1.7102611475406676e-05,
+      "loss": 3.2656,
+      "step": 442
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.7088609560013284e-05,
+      "loss": 3.0938,
+      "step": 443
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 1.7074579655005282e-05,
+      "loss": 2.9648,
+      "step": 444
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 1.7060521815780225e-05,
+      "loss": 3.1328,
+      "step": 445
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.704643609784596e-05,
+      "loss": 3.1211,
+      "step": 446
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 1.7032322556820428e-05,
+      "loss": 3.1719,
+      "step": 447
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.7018181248431416e-05,
+      "loss": 2.9883,
+      "step": 448
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 1.700401222851636e-05,
+      "loss": 3.1172,
+      "step": 449
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 1.698981555302212e-05,
+      "loss": 2.9531,
+      "step": 450
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.6975591278004747e-05,
+      "loss": 2.9375,
+      "step": 451
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 1.696133945962927e-05,
+      "loss": 3.1875,
+      "step": 452
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 1.6947060154169473e-05,
+      "loss": 3.0742,
+      "step": 453
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.6932753418007683e-05,
+      "loss": 3.0977,
+      "step": 454
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 1.691841930763453e-05,
+      "loss": 2.9531,
+      "step": 455
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.690405787964873e-05,
+      "loss": 2.9609,
+      "step": 456
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 1.688966919075687e-05,
+      "loss": 2.7578,
+      "step": 457
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 1.687525329777317e-05,
+      "loss": 2.9961,
+      "step": 458
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.686081025761928e-05,
+      "loss": 3.3203,
+      "step": 459
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 1.684634012732403e-05,
+      "loss": 2.9258,
+      "step": 460
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6831842964023212e-05,
+      "loss": 3.1445,
+      "step": 461
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 1.6817318824959375e-05,
+      "loss": 3.2617,
+      "step": 462
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 1.680276776748157e-05,
+      "loss": 2.9883,
+      "step": 463
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.6788189849045135e-05,
+      "loss": 2.9219,
+      "step": 464
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 1.6773585127211478e-05,
+      "loss": 2.8281,
+      "step": 465
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 1.6758953659647838e-05,
+      "loss": 3.0312,
+      "step": 466
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.6744295504127055e-05,
+      "loss": 3.2461,
+      "step": 467
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 1.6729610718527357e-05,
+      "loss": 3.1562,
+      "step": 468
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.6714899360832118e-05,
+      "loss": 2.9023,
+      "step": 469
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 1.6700161489129624e-05,
+      "loss": 3.0898,
+      "step": 470
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 1.668539716161287e-05,
+      "loss": 2.9414,
+      "step": 471
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.667060643657929e-05,
+      "loss": 2.9844,
+      "step": 472
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 1.6655789372430572e-05,
+      "loss": 3.0859,
+      "step": 473
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.6640946027672395e-05,
+      "loss": 3.1758,
+      "step": 474
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 1.66260764609142e-05,
+      "loss": 3.1719,
+      "step": 475
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 1.6611180730868975e-05,
+      "loss": 3.0508,
+      "step": 476
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.6596258896353027e-05,
+      "loss": 3.1406,
+      "step": 477
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 1.658131101628571e-05,
+      "loss": 3.1836,
+      "step": 478
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.656633714968924e-05,
+      "loss": 3.0352,
+      "step": 479
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 1.6551337355688437e-05,
+      "loss": 2.8789,
+      "step": 480
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 1.653631169351049e-05,
+      "loss": 3.1094,
+      "step": 481
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.6521260222484738e-05,
+      "loss": 3.4102,
+      "step": 482
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 1.650618300204242e-05,
+      "loss": 3.293,
+      "step": 483
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 1.6491080091716457e-05,
+      "loss": 2.9922,
+      "step": 484
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.64759515511412e-05,
+      "loss": 3.082,
+      "step": 485
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 1.6460797440052195e-05,
+      "loss": 2.9297,
+      "step": 486
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.6445617818285974e-05,
+      "loss": 2.8906,
+      "step": 487
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 1.643041274577978e-05,
+      "loss": 3.0625,
+      "step": 488
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 1.6415182282571356e-05,
+      "loss": 3.1562,
+      "step": 489
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.6399926488798702e-05,
+      "loss": 2.6367,
+      "step": 490
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 1.6384645424699835e-05,
+      "loss": 2.207,
+      "step": 491
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.6369339150612557e-05,
+      "loss": 2.4844,
+      "step": 492
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 1.6354007726974205e-05,
+      "loss": 2.4219,
+      "step": 493
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 1.6338651214321426e-05,
+      "loss": 2.4531,
+      "step": 494
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.632326967328993e-05,
+      "loss": 2.4961,
+      "step": 495
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 1.630786316461425e-05,
+      "loss": 2.4219,
+      "step": 496
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 1.6292431749127507e-05,
+      "loss": 2.5273,
+      "step": 497
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.627697548776117e-05,
+      "loss": 2.4492,
+      "step": 498
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 1.6261494441544805e-05,
+      "loss": 2.4922,
+      "step": 499
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.624598867160585e-05,
+      "loss": 2.375,
+      "step": 500
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 1.623045823916936e-05,
+      "loss": 2.6914,
+      "step": 501
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 1.6214903205557774e-05,
+      "loss": 2.4141,
+      "step": 502
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.619932363219067e-05,
+      "loss": 2.5742,
+      "step": 503
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 1.6183719580584515e-05,
+      "loss": 2.332,
+      "step": 504
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.6168091112352443e-05,
+      "loss": 2.4727,
+      "step": 505
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 1.6152438289203982e-05,
+      "loss": 2.5352,
+      "step": 506
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 1.6136761172944837e-05,
+      "loss": 2.4375,
+      "step": 507
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.612105982547663e-05,
+      "loss": 2.543,
+      "step": 508
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 1.6105334308796665e-05,
+      "loss": 2.3945,
+      "step": 509
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.6089584684997674e-05,
+      "loss": 2.4531,
+      "step": 510
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 1.607381101626758e-05,
+      "loss": 2.5781,
+      "step": 511
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 1.6058013364889247e-05,
+      "loss": 2.2852,
+      "step": 512
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.6042191793240242e-05,
+      "loss": 2.293,
+      "step": 513
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 1.6026346363792565e-05,
+      "loss": 2.5156,
+      "step": 514
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 1.6010477139112438e-05,
+      "loss": 2.3711,
+      "step": 515
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.5994584181860028e-05,
+      "loss": 2.2891,
+      "step": 516
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 1.5978667554789216e-05,
+      "loss": 2.3867,
+      "step": 517
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.596272732074734e-05,
+      "loss": 2.457,
+      "step": 518
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 1.5946763542674958e-05,
+      "loss": 2.293,
+      "step": 519
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 1.5930776283605585e-05,
+      "loss": 2.4492,
+      "step": 520
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.5914765606665454e-05,
+      "loss": 2.2383,
+      "step": 521
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 1.5898731575073262e-05,
+      "loss": 2.3281,
+      "step": 522
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.5882674252139928e-05,
+      "loss": 2.4688,
+      "step": 523
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 1.5866593701268334e-05,
+      "loss": 2.3125,
+      "step": 524
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 1.5850489985953076e-05,
+      "loss": 2.3672,
+      "step": 525
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.5834363169780227e-05,
+      "loss": 2.4688,
+      "step": 526
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 1.5818213316427056e-05,
+      "loss": 2.375,
+      "step": 527
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 1.5802040489661817e-05,
+      "loss": 2.418,
+      "step": 528
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.578584475334345e-05,
+      "loss": 2.3867,
+      "step": 529
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 1.5769626171421376e-05,
+      "loss": 2.2852,
+      "step": 530
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.5753384807935214e-05,
+      "loss": 2.5234,
+      "step": 531
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 1.5737120727014535e-05,
+      "loss": 2.3828,
+      "step": 532
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 1.572083399287861e-05,
+      "loss": 2.4023,
+      "step": 533
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.570452466983617e-05,
+      "loss": 2.4961,
+      "step": 534
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 1.5688192822285116e-05,
+      "loss": 2.5234,
+      "step": 535
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.567183851471231e-05,
+      "loss": 2.418,
+      "step": 536
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 1.565546181169328e-05,
+      "loss": 2.3555,
+      "step": 537
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 1.5639062777892e-05,
+      "loss": 2.4883,
+      "step": 538
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.5622641478060602e-05,
+      "loss": 2.5586,
+      "step": 539
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 1.5606197977039154e-05,
+      "loss": 2.3359,
+      "step": 540
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 1.5589732339755362e-05,
+      "loss": 2.3398,
+      "step": 541
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.5573244631224364e-05,
+      "loss": 2.2969,
+      "step": 542
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 1.5556734916548432e-05,
+      "loss": 2.375,
+      "step": 543
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.5540203260916728e-05,
+      "loss": 2.3398,
+      "step": 544
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 1.552364972960506e-05,
+      "loss": 2.3516,
+      "step": 545
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 1.5507074387975603e-05,
+      "loss": 2.4805,
+      "step": 546
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.5490477301476648e-05,
+      "loss": 2.4766,
+      "step": 547
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 1.5473858535642365e-05,
+      "loss": 2.4062,
+      "step": 548
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.5457218156092503e-05,
+      "loss": 2.4727,
+      "step": 549
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 1.5440556228532168e-05,
+      "loss": 2.3672,
+      "step": 550
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 1.5423872818751544e-05,
+      "loss": 2.5195,
+      "step": 551
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.5407167992625636e-05,
+      "loss": 2.418,
+      "step": 552
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 1.5390441816114022e-05,
+      "loss": 2.3828,
+      "step": 553
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.5373694355260565e-05,
+      "loss": 2.4336,
+      "step": 554
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 1.5356925676193192e-05,
+      "loss": 2.3086,
+      "step": 555
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 1.534013584512359e-05,
+      "loss": 2.25,
+      "step": 556
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.5323324928346984e-05,
+      "loss": 2.3242,
+      "step": 557
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 1.5306492992241836e-05,
+      "loss": 2.4023,
+      "step": 558
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 1.5289640103269626e-05,
+      "loss": 2.4531,
+      "step": 559
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.527276632797455e-05,
+      "loss": 2.3945,
+      "step": 560
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 1.5255871732983284e-05,
+      "loss": 2.4258,
+      "step": 561
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.5238956385004703e-05,
+      "loss": 2.4766,
+      "step": 562
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 1.5222020350829636e-05,
+      "loss": 2.4141,
+      "step": 563
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 1.5205063697330582e-05,
+      "loss": 2.3359,
+      "step": 564
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.5188086491461467e-05,
+      "loss": 2.3047,
+      "step": 565
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 1.5171088800257354e-05,
+      "loss": 2.5508,
+      "step": 566
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.5154070690834211e-05,
+      "loss": 2.0957,
+      "step": 567
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 1.5137032230388613e-05,
+      "loss": 2.4102,
+      "step": 568
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 1.5119973486197497e-05,
+      "loss": 2.5352,
+      "step": 569
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.5102894525617892e-05,
+      "loss": 2.25,
+      "step": 570
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 1.5085795416086655e-05,
+      "loss": 2.3047,
+      "step": 571
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 1.5068676225120196e-05,
+      "loss": 2.3359,
+      "step": 572
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.5051537020314218e-05,
+      "loss": 2.5508,
+      "step": 573
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 1.5034377869343453e-05,
+      "loss": 2.6211,
+      "step": 574
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.5017198839961388e-05,
+      "loss": 2.5625,
+      "step": 575
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 2.293,
+      "step": 576
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 1.4982781417369496e-05,
+      "loss": 2.5078,
+      "step": 577
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.4965543160058028e-05,
+      "loss": 2.3594,
+      "step": 578
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 1.4948285296131435e-05,
+      "loss": 2.4531,
+      "step": 579
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.4931007893732981e-05,
+      "loss": 2.4961,
+      "step": 580
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 1.4913711021083071e-05,
+      "loss": 2.3672,
+      "step": 581
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 1.4896394746478995e-05,
+      "loss": 2.5469,
+      "step": 582
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.4879059138294647e-05,
+      "loss": 2.5703,
+      "step": 583
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 1.4861704264980264e-05,
+      "loss": 2.5859,
+      "step": 584
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 1.4844330195062145e-05,
+      "loss": 2.4648,
+      "step": 585
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.4826936997142399e-05,
+      "loss": 2.4883,
+      "step": 586
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 1.4809524739898651e-05,
+      "loss": 2.2656,
+      "step": 587
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.4792093492083792e-05,
+      "loss": 2.2734,
+      "step": 588
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 1.4774643322525691e-05,
+      "loss": 2.5156,
+      "step": 589
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 1.4757174300126935e-05,
+      "loss": 2.6797,
+      "step": 590
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.473968649386455e-05,
+      "loss": 2.3398,
+      "step": 591
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 1.4722179972789725e-05,
+      "loss": 2.2539,
+      "step": 592
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.4704654806027558e-05,
+      "loss": 2.5781,
+      "step": 593
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 1.4687111062776758e-05,
+      "loss": 2.5352,
+      "step": 594
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 1.466954881230939e-05,
+      "loss": 2.5195,
+      "step": 595
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.4651968123970592e-05,
+      "loss": 2.3945,
+      "step": 596
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 1.4634369067178312e-05,
+      "loss": 2.4922,
+      "step": 597
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.4616751711423016e-05,
+      "loss": 2.4922,
+      "step": 598
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 1.4599116126267431e-05,
+      "loss": 2.4961,
+      "step": 599
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 1.4581462381346261e-05,
+      "loss": 2.4922,
+      "step": 600
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.4563790546365914e-05,
+      "loss": 2.5,
+      "step": 601
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 1.454610069110423e-05,
+      "loss": 2.4219,
+      "step": 602
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 1.45283928854102e-05,
+      "loss": 2.418,
+      "step": 603
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.4510667199203697e-05,
+      "loss": 2.5488,
+      "step": 604
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 1.4492923702475183e-05,
+      "loss": 2.5312,
+      "step": 605
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.4475162465285463e-05,
+      "loss": 2.5273,
+      "step": 606
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 1.4457383557765385e-05,
+      "loss": 2.4141,
+      "step": 607
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 1.443958705011556e-05,
+      "loss": 2.4453,
+      "step": 608
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.4421773012606104e-05,
+      "loss": 2.293,
+      "step": 609
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 1.4403941515576344e-05,
+      "loss": 2.4258,
+      "step": 610
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.4386092629434551e-05,
+      "loss": 2.4648,
+      "step": 611
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 1.4368226424657661e-05,
+      "loss": 2.3438,
+      "step": 612
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 1.4350342971790979e-05,
+      "loss": 2.2168,
+      "step": 613
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.4332442341447926e-05,
+      "loss": 2.3828,
+      "step": 614
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 1.4314524604309748e-05,
+      "loss": 2.5117,
+      "step": 615
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 1.4296589831125234e-05,
+      "loss": 2.4961,
+      "step": 616
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.4278638092710446e-05,
+      "loss": 2.5391,
+      "step": 617
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 1.4260669459948429e-05,
+      "loss": 2.3828,
+      "step": 618
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.4242684003788934e-05,
+      "loss": 2.4102,
+      "step": 619
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 1.4224681795248149e-05,
+      "loss": 2.457,
+      "step": 620
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 1.42066629054084e-05,
+      "loss": 2.5,
+      "step": 621
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.418862740541788e-05,
+      "loss": 2.4102,
+      "step": 622
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 1.4170575366490376e-05,
+      "loss": 2.1758,
+      "step": 623
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.415250685990497e-05,
+      "loss": 2.6445,
+      "step": 624
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 1.4134421957005775e-05,
+      "loss": 2.043,
+      "step": 625
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 1.4116320729201642e-05,
+      "loss": 2.457,
+      "step": 626
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.4098203247965876e-05,
+      "loss": 2.1992,
+      "step": 627
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 1.4080069584835971e-05,
+      "loss": 2.2891,
+      "step": 628
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 1.4061919811413305e-05,
+      "loss": 2.2227,
+      "step": 629
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.4043753999362872e-05,
+      "loss": 2.2305,
+      "step": 630
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 1.4025572220412998e-05,
+      "loss": 2.625,
+      "step": 631
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.400737454635505e-05,
+      "loss": 2.4219,
+      "step": 632
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 1.398916104904316e-05,
+      "loss": 2.6133,
+      "step": 633
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 1.3970931800393943e-05,
+      "loss": 2.5625,
+      "step": 634
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.3952686872386195e-05,
+      "loss": 2.4531,
+      "step": 635
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 1.3934426337060638e-05,
+      "loss": 2.6016,
+      "step": 636
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.391615026651961e-05,
+      "loss": 2.3789,
+      "step": 637
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 1.3897858732926794e-05,
+      "loss": 2.3281,
+      "step": 638
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 1.3879551808506932e-05,
+      "loss": 2.2031,
+      "step": 639
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.3861229565545532e-05,
+      "loss": 2.5352,
+      "step": 640
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 1.384289207638859e-05,
+      "loss": 2.3008,
+      "step": 641
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.3824539413442304e-05,
+      "loss": 2.5352,
+      "step": 642
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 1.3806171649172782e-05,
+      "loss": 2.4922,
+      "step": 643
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 1.3787788856105762e-05,
+      "loss": 2.3945,
+      "step": 644
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.3769391106826326e-05,
+      "loss": 2.6016,
+      "step": 645
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 1.3750978473978611e-05,
+      "loss": 2.4375,
+      "step": 646
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 1.3732551030265514e-05,
+      "loss": 2.5195,
+      "step": 647
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.371410884844843e-05,
+      "loss": 2.5391,
+      "step": 648
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 1.3695652001346928e-05,
+      "loss": 2.4102,
+      "step": 649
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.3677180561838501e-05,
+      "loss": 2.4727,
+      "step": 650
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 1.3658694602858247e-05,
+      "loss": 2.6055,
+      "step": 651
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 1.36401941973986e-05,
+      "loss": 2.2852,
+      "step": 652
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.362167941850904e-05,
+      "loss": 1.9121,
+      "step": 653
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.3603150339295797e-05,
+      "loss": 2.0977,
+      "step": 654
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.3584607032921566e-05,
+      "loss": 1.9668,
+      "step": 655
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.3566049572605222e-05,
+      "loss": 1.8398,
+      "step": 656
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.3547478031621517e-05,
+      "loss": 1.7559,
+      "step": 657
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.3528892483300821e-05,
+      "loss": 2.0586,
+      "step": 658
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.3510293001028792e-05,
+      "loss": 1.8984,
+      "step": 659
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.3491679658246114e-05,
+      "loss": 1.6895,
+      "step": 660
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.3473052528448203e-05,
+      "loss": 1.7812,
+      "step": 661
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.3454411685184913e-05,
+      "loss": 1.7539,
+      "step": 662
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.3435757202060242e-05,
+      "loss": 1.9492,
+      "step": 663
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.3417089152732049e-05,
+      "loss": 1.7031,
+      "step": 664
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.3398407610911752e-05,
+      "loss": 1.791,
+      "step": 665
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.3379712650364061e-05,
+      "loss": 1.8066,
+      "step": 666
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.3361004344906652e-05,
+      "loss": 1.6992,
+      "step": 667
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.3342282768409904e-05,
+      "loss": 1.8965,
+      "step": 668
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.3323547994796597e-05,
+      "loss": 1.7832,
+      "step": 669
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.330480009804162e-05,
+      "loss": 1.8633,
+      "step": 670
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.3286039152171667e-05,
+      "loss": 1.6055,
+      "step": 671
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.3267265231264982e-05,
+      "loss": 1.8164,
+      "step": 672
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.3248478409451017e-05,
+      "loss": 1.9805,
+      "step": 673
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.3229678760910174e-05,
+      "loss": 1.666,
+      "step": 674
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.3210866359873506e-05,
+      "loss": 1.8867,
+      "step": 675
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.3192041280622409e-05,
+      "loss": 1.9473,
+      "step": 676
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.3173203597488348e-05,
+      "loss": 1.9375,
+      "step": 677
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.3154353384852559e-05,
+      "loss": 1.8145,
+      "step": 678
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.3135490717145726e-05,
+      "loss": 1.7539,
+      "step": 679
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.3116615668847749e-05,
+      "loss": 1.7734,
+      "step": 680
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3097728314487385e-05,
+      "loss": 1.7656,
+      "step": 681
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3078828728641994e-05,
+      "loss": 1.8672,
+      "step": 682
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.305991698593723e-05,
+      "loss": 1.7656,
+      "step": 683
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3040993161046749e-05,
+      "loss": 1.8789,
+      "step": 684
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3022057328691915e-05,
+      "loss": 1.627,
+      "step": 685
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.3003109563641499e-05,
+      "loss": 1.7695,
+      "step": 686
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.298414994071139e-05,
+      "loss": 1.709,
+      "step": 687
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2965178534764311e-05,
+      "loss": 1.7383,
+      "step": 688
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.294619542070949e-05,
+      "loss": 1.6523,
+      "step": 689
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2927200673502399e-05,
+      "loss": 1.8145,
+      "step": 690
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.2908194368144437e-05,
+      "loss": 1.7949,
+      "step": 691
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.288917657968265e-05,
+      "loss": 1.7422,
+      "step": 692
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.287014738320941e-05,
+      "loss": 1.9102,
+      "step": 693
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.285110685386215e-05,
+      "loss": 1.6523,
+      "step": 694
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.283205506682304e-05,
+      "loss": 1.5938,
+      "step": 695
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.2812992097318711e-05,
+      "loss": 1.6797,
+      "step": 696
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.2793918020619937e-05,
+      "loss": 1.8164,
+      "step": 697
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.2774832912041356e-05,
+      "loss": 1.6328,
+      "step": 698
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.2755736846941167e-05,
+      "loss": 1.9219,
+      "step": 699
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.2736629900720832e-05,
+      "loss": 1.8496,
+      "step": 700
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.2717512148824764e-05,
+      "loss": 1.7031,
+      "step": 701
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.2698383666740064e-05,
+      "loss": 1.7266,
+      "step": 702
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 1.2679244529996182e-05,
+      "loss": 1.9102,
+      "step": 703
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 1.2660094814164653e-05,
+      "loss": 1.6855,
+      "step": 704
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.2640934594858773e-05,
+      "loss": 1.6641,
+      "step": 705
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 1.262176394773332e-05,
+      "loss": 1.8672,
+      "step": 706
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.2602582948484243e-05,
+      "loss": 1.7383,
+      "step": 707
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 1.2583391672848361e-05,
+      "loss": 2.0586,
+      "step": 708
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 1.256419019660308e-05,
+      "loss": 1.8281,
+      "step": 709
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.2544978595566078e-05,
+      "loss": 1.7207,
+      "step": 710
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 1.2525756945595006e-05,
+      "loss": 1.6328,
+      "step": 711
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.2506525322587207e-05,
+      "loss": 1.8379,
+      "step": 712
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 1.2487283802479389e-05,
+      "loss": 1.8828,
+      "step": 713
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 1.246803246124735e-05,
+      "loss": 1.916,
+      "step": 714
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.2448771374905655e-05,
+      "loss": 1.7852,
+      "step": 715
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 1.2429500619507362e-05,
+      "loss": 2.0391,
+      "step": 716
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.2410220271143693e-05,
+      "loss": 1.7422,
+      "step": 717
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 1.2390930405943766e-05,
+      "loss": 1.8672,
+      "step": 718
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 1.237163110007426e-05,
+      "loss": 1.8457,
+      "step": 719
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.2352322429739134e-05,
+      "loss": 1.7402,
+      "step": 720
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 1.233300447117933e-05,
+      "loss": 1.6465,
+      "step": 721
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 1.2313677300672463e-05,
+      "loss": 1.6777,
+      "step": 722
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.2294340994532511e-05,
+      "loss": 1.7656,
+      "step": 723
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 1.2274995629109545e-05,
+      "loss": 1.8066,
+      "step": 724
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.2255641280789385e-05,
+      "loss": 1.8809,
+      "step": 725
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 1.2236278025993334e-05,
+      "loss": 1.8223,
+      "step": 726
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 1.2216905941177854e-05,
+      "loss": 1.7656,
+      "step": 727
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.2197525102834284e-05,
+      "loss": 1.8066,
+      "step": 728
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 1.2178135587488515e-05,
+      "loss": 1.7207,
+      "step": 729
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.215873747170071e-05,
+      "loss": 1.8535,
+      "step": 730
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 1.2139330832064975e-05,
+      "loss": 1.7949,
+      "step": 731
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 1.2119915745209092e-05,
+      "loss": 1.8926,
+      "step": 732
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.2100492287794186e-05,
+      "loss": 1.6777,
+      "step": 733
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.2081060536514432e-05,
+      "loss": 1.7773,
+      "step": 734
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.206162056809676e-05,
+      "loss": 1.6699,
+      "step": 735
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.2042172459300546e-05,
+      "loss": 1.709,
+      "step": 736
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.2022716286917298e-05,
+      "loss": 1.8887,
+      "step": 737
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.2003252127770378e-05,
+      "loss": 1.9219,
+      "step": 738
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.198378005871467e-05,
+      "loss": 1.8535,
+      "step": 739
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.1964300156636304e-05,
+      "loss": 1.7051,
+      "step": 740
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.1944812498452329e-05,
+      "loss": 1.7578,
+      "step": 741
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.192531716111042e-05,
+      "loss": 1.8203,
+      "step": 742
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.1905814221588581e-05,
+      "loss": 1.6016,
+      "step": 743
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.1886303756894828e-05,
+      "loss": 1.543,
+      "step": 744
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.1866785844066884e-05,
+      "loss": 1.8145,
+      "step": 745
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.1847260560171895e-05,
+      "loss": 1.6719,
+      "step": 746
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.18277279823061e-05,
+      "loss": 1.6953,
+      "step": 747
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.1808188187594549e-05,
+      "loss": 1.6406,
+      "step": 748
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.1788641253190779e-05,
+      "loss": 1.7246,
+      "step": 749
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.176908725627652e-05,
+      "loss": 1.6992,
+      "step": 750
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.1749526274061394e-05,
+      "loss": 1.916,
+      "step": 751
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.1729958383782598e-05,
+      "loss": 1.6543,
+      "step": 752
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.1710383662704608e-05,
+      "loss": 1.707,
+      "step": 753
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.1690802188118878e-05,
+      "loss": 1.6953,
+      "step": 754
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.1671214037343515e-05,
+      "loss": 1.6875,
+      "step": 755
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.1651619287723e-05,
+      "loss": 1.7969,
+      "step": 756
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.1632018016627859e-05,
+      "loss": 1.7461,
+      "step": 757
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.1612410301454384e-05,
+      "loss": 1.8887,
+      "step": 758
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.1592796219624292e-05,
+      "loss": 1.9414,
+      "step": 759
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.1573175848584455e-05,
+      "loss": 1.8711,
+      "step": 760
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.1553549265806567e-05,
+      "loss": 1.7246,
+      "step": 761
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.1533916548786856e-05,
+      "loss": 1.8496,
+      "step": 762
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.1514277775045768e-05,
+      "loss": 1.918,
+      "step": 763
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.1494633022127669e-05,
+      "loss": 1.8574,
+      "step": 764
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.1474982367600524e-05,
+      "loss": 1.668,
+      "step": 765
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.1455325889055616e-05,
+      "loss": 1.7031,
+      "step": 766
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.1435663664107204e-05,
+      "loss": 1.7754,
+      "step": 767
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.141599577039226e-05,
+      "loss": 1.7129,
+      "step": 768
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.1396322285570119e-05,
+      "loss": 1.6582,
+      "step": 769
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.1376643287322202e-05,
+      "loss": 1.8672,
+      "step": 770
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.1356958853351705e-05,
+      "loss": 1.8867,
+      "step": 771
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.1337269061383278e-05,
+      "loss": 1.8359,
+      "step": 772
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.1317573989162727e-05,
+      "loss": 1.8535,
+      "step": 773
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.129787371445672e-05,
+      "loss": 1.7793,
+      "step": 774
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.1278168315052445e-05,
+      "loss": 1.834,
+      "step": 775
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.1258457868757352e-05,
+      "loss": 1.8906,
+      "step": 776
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1238742453398794e-05,
+      "loss": 1.9512,
+      "step": 777
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1219022146823762e-05,
+      "loss": 1.8047,
+      "step": 778
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.1199297026898547e-05,
+      "loss": 1.627,
+      "step": 779
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.1179567171508463e-05,
+      "loss": 1.8242,
+      "step": 780
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 1.1159832658557498e-05,
+      "loss": 1.7129,
+      "step": 781
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.1140093565968055e-05,
+      "loss": 1.7012,
+      "step": 782
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 1.1120349971680605e-05,
+      "loss": 1.8145,
+      "step": 783
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 1.1100601953653393e-05,
+      "loss": 1.6426,
+      "step": 784
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.1080849589862142e-05,
+      "loss": 1.8574,
+      "step": 785
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 1.1061092958299727e-05,
+      "loss": 1.752,
+      "step": 786
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.1041332136975874e-05,
+      "loss": 1.9531,
+      "step": 787
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 1.1021567203916861e-05,
+      "loss": 1.7676,
+      "step": 788
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 1.1001798237165185e-05,
+      "loss": 1.7656,
+      "step": 789
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.0982025314779287e-05,
+      "loss": 1.9512,
+      "step": 790
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 1.0962248514833218e-05,
+      "loss": 1.791,
+      "step": 791
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 1.0942467915416342e-05,
+      "loss": 1.8398,
+      "step": 792
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.092268359463302e-05,
+      "loss": 1.6797,
+      "step": 793
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 1.090289563060232e-05,
+      "loss": 1.7871,
+      "step": 794
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.088310410145768e-05,
+      "loss": 1.6738,
+      "step": 795
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 1.086330908534663e-05,
+      "loss": 1.8711,
+      "step": 796
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 1.0843510660430447e-05,
+      "loss": 1.752,
+      "step": 797
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.0823708904883898e-05,
+      "loss": 1.9297,
+      "step": 798
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.0803903896894877e-05,
+      "loss": 1.9141,
+      "step": 799
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.0784095714664124e-05,
+      "loss": 1.7188,
+      "step": 800
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.0764284436404924e-05,
+      "loss": 1.7441,
+      "step": 801
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.0744470140342775e-05,
+      "loss": 1.7266,
+      "step": 802
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.0724652904715091e-05,
+      "loss": 1.832,
+      "step": 803
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.0704832807770909e-05,
+      "loss": 1.6152,
+      "step": 804
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.0685009927770542e-05,
+      "loss": 1.8281,
+      "step": 805
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 1.0665184342985306e-05,
+      "loss": 1.7812,
+      "step": 806
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.064535613169719e-05,
+      "loss": 1.875,
+      "step": 807
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.0625525372198564e-05,
+      "loss": 1.748,
+      "step": 808
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.0605692142791846e-05,
+      "loss": 1.7148,
+      "step": 809
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.0585856521789215e-05,
+      "loss": 1.7715,
+      "step": 810
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.056601858751229e-05,
+      "loss": 1.7676,
+      "step": 811
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 1.0546178418291833e-05,
+      "loss": 1.7852,
+      "step": 812
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.0526336092467414e-05,
+      "loss": 1.9141,
+      "step": 813
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.0506491688387128e-05,
+      "loss": 1.6602,
+      "step": 814
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.0486645284407282e-05,
+      "loss": 1.75,
+      "step": 815
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.0466796958892071e-05,
+      "loss": 1.5469,
+      "step": 816
+    },
+    {
+      "epoch": 5.01,
+      "learning_rate": 1.0446946790213275e-05,
+      "loss": 1.2852,
+      "step": 817
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.0427094856749966e-05,
+      "loss": 1.3926,
+      "step": 818
+    },
+    {
+      "epoch": 5.02,
+      "learning_rate": 1.0407241236888164e-05,
+      "loss": 1.293,
+      "step": 819
+    },
+    {
+      "epoch": 5.03,
+      "learning_rate": 1.0387386009020569e-05,
+      "loss": 1.2559,
+      "step": 820
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 1.0367529251546208e-05,
+      "loss": 1.3379,
+      "step": 821
+    },
+    {
+      "epoch": 5.04,
+      "learning_rate": 1.034767104287017e-05,
+      "loss": 1.3047,
+      "step": 822
+    },
+    {
+      "epoch": 5.05,
+      "learning_rate": 1.032781146140326e-05,
+      "loss": 1.3105,
+      "step": 823
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 1.0307950585561705e-05,
+      "loss": 1.3203,
+      "step": 824
+    },
+    {
+      "epoch": 5.06,
+      "learning_rate": 1.0288088493766846e-05,
+      "loss": 1.2461,
+      "step": 825
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 1.0268225264444829e-05,
+      "loss": 1.3281,
+      "step": 826
+    },
+    {
+      "epoch": 5.07,
+      "learning_rate": 1.0248360976026279e-05,
+      "loss": 1.1758,
+      "step": 827
+    },
+    {
+      "epoch": 5.08,
+      "learning_rate": 1.0228495706946015e-05,
+      "loss": 1.1465,
+      "step": 828
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 1.0208629535642726e-05,
+      "loss": 1.1836,
+      "step": 829
+    },
+    {
+      "epoch": 5.09,
+      "learning_rate": 1.0188762540558657e-05,
+      "loss": 1.1504,
+      "step": 830
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 1.0168894800139311e-05,
+      "loss": 1.1641,
+      "step": 831
+    },
+    {
+      "epoch": 5.1,
+      "learning_rate": 1.0149026392833137e-05,
+      "loss": 1.1504,
+      "step": 832
+    },
+    {
+      "epoch": 5.11,
+      "learning_rate": 1.0129157397091208e-05,
+      "loss": 1.2832,
+      "step": 833
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 1.010928789136693e-05,
+      "loss": 1.25,
+      "step": 834
+    },
+    {
+      "epoch": 5.12,
+      "learning_rate": 1.0089417954115715e-05,
+      "loss": 1.2207,
+      "step": 835
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 1.0069547663794682e-05,
+      "loss": 1.1855,
+      "step": 836
+    },
+    {
+      "epoch": 5.13,
+      "learning_rate": 1.0049677098862347e-05,
+      "loss": 1.1289,
+      "step": 837
+    },
+    {
+      "epoch": 5.14,
+      "learning_rate": 1.002980633777831e-05,
+      "loss": 1.1562,
+      "step": 838
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 1.0009935459002935e-05,
+      "loss": 1.3242,
+      "step": 839
+    },
+    {
+      "epoch": 5.15,
+      "learning_rate": 9.990064540997066e-06,
+      "loss": 1.3105,
+      "step": 840
+    },
+    {
+      "epoch": 5.16,
+      "learning_rate": 9.970193662221694e-06,
+      "loss": 1.3145,
+      "step": 841
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 9.950322901137655e-06,
+      "loss": 1.2441,
+      "step": 842
+    },
+    {
+      "epoch": 5.17,
+      "learning_rate": 9.93045233620532e-06,
+      "loss": 1.3262,
+      "step": 843
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 9.910582045884292e-06,
+      "loss": 1.2656,
+      "step": 844
+    },
+    {
+      "epoch": 5.18,
+      "learning_rate": 9.890712108633076e-06,
+      "loss": 1.3633,
+      "step": 845
+    },
+    {
+      "epoch": 5.19,
+      "learning_rate": 9.870842602908794e-06,
+      "loss": 1.2734,
+      "step": 846
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 9.850973607166865e-06,
+      "loss": 1.2656,
+      "step": 847
+    },
+    {
+      "epoch": 5.2,
+      "learning_rate": 9.83110519986069e-06,
+      "loss": 1.2949,
+      "step": 848
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 9.811237459441346e-06,
+      "loss": 1.2227,
+      "step": 849
+    },
+    {
+      "epoch": 5.21,
+      "learning_rate": 9.791370464357279e-06,
+      "loss": 1.2793,
+      "step": 850
+    },
+    {
+      "epoch": 5.22,
+      "learning_rate": 9.771504293053985e-06,
+      "loss": 1.3633,
+      "step": 851
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 9.751639023973724e-06,
+      "loss": 1.207,
+      "step": 852
+    },
+    {
+      "epoch": 5.23,
+      "learning_rate": 9.731774735555174e-06,
+      "loss": 1.252,
+      "step": 853
+    },
+    {
+      "epoch": 5.24,
+      "learning_rate": 9.711911506233157e-06,
+      "loss": 1.1992,
+      "step": 854
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 9.692049414438298e-06,
+      "loss": 1.3516,
+      "step": 855
+    },
+    {
+      "epoch": 5.25,
+      "learning_rate": 9.672188538596746e-06,
+      "loss": 1.3574,
+      "step": 856
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 9.652328957129831e-06,
+      "loss": 1.4062,
+      "step": 857
+    },
+    {
+      "epoch": 5.26,
+      "learning_rate": 9.632470748453794e-06,
+      "loss": 1.3223,
+      "step": 858
+    },
+    {
+      "epoch": 5.27,
+      "learning_rate": 9.612613990979436e-06,
+      "loss": 1.2207,
+      "step": 859
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 9.59275876311184e-06,
+      "loss": 1.2441,
+      "step": 860
+    },
+    {
+      "epoch": 5.28,
+      "learning_rate": 9.572905143250039e-06,
+      "loss": 1.0586,
+      "step": 861
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 9.553053209786725e-06,
+      "loss": 1.2148,
+      "step": 862
+    },
+    {
+      "epoch": 5.29,
+      "learning_rate": 9.53320304110793e-06,
+      "loss": 1.2402,
+      "step": 863
+    },
+    {
+      "epoch": 5.3,
+      "learning_rate": 9.513354715592721e-06,
+      "loss": 1.1338,
+      "step": 864
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 9.493508311612874e-06,
+      "loss": 1.332,
+      "step": 865
+    },
+    {
+      "epoch": 5.31,
+      "learning_rate": 9.473663907532593e-06,
+      "loss": 1.2715,
+      "step": 866
+    },
+    {
+      "epoch": 5.32,
+      "learning_rate": 9.453821581708174e-06,
+      "loss": 1.2793,
+      "step": 867
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 9.433981412487711e-06,
+      "loss": 1.2969,
+      "step": 868
+    },
+    {
+      "epoch": 5.33,
+      "learning_rate": 9.414143478210786e-06,
+      "loss": 1.1074,
+      "step": 869
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 9.394307857208158e-06,
+      "loss": 1.1924,
+      "step": 870
+    },
+    {
+      "epoch": 5.34,
+      "learning_rate": 9.374474627801439e-06,
+      "loss": 1.2188,
+      "step": 871
+    },
+    {
+      "epoch": 5.35,
+      "learning_rate": 9.354643868302813e-06,
+      "loss": 1.2246,
+      "step": 872
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 9.334815657014696e-06,
+      "loss": 1.2109,
+      "step": 873
+    },
+    {
+      "epoch": 5.36,
+      "learning_rate": 9.314990072229461e-06,
+      "loss": 1.2832,
+      "step": 874
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 9.295167192229093e-06,
+      "loss": 1.2666,
+      "step": 875
+    },
+    {
+      "epoch": 5.37,
+      "learning_rate": 9.27534709528491e-06,
+      "loss": 1.3066,
+      "step": 876
+    },
+    {
+      "epoch": 5.38,
+      "learning_rate": 9.25552985965723e-06,
+      "loss": 1.5352,
+      "step": 877
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 9.235715563595082e-06,
+      "loss": 1.2305,
+      "step": 878
+    },
+    {
+      "epoch": 5.39,
+      "learning_rate": 9.215904285335876e-06,
+      "loss": 1.1113,
+      "step": 879
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 9.196096103105127e-06,
+      "loss": 1.2285,
+      "step": 880
+    },
+    {
+      "epoch": 5.4,
+      "learning_rate": 9.176291095116104e-06,
+      "loss": 1.2871,
+      "step": 881
+    },
+    {
+      "epoch": 5.41,
+      "learning_rate": 9.156489339569555e-06,
+      "loss": 1.2539,
+      "step": 882
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 9.136690914653377e-06,
+      "loss": 1.2666,
+      "step": 883
+    },
+    {
+      "epoch": 5.42,
+      "learning_rate": 9.11689589854232e-06,
+      "loss": 1.2539,
+      "step": 884
+    },
+    {
+      "epoch": 5.43,
+      "learning_rate": 9.097104369397681e-06,
+      "loss": 1.1562,
+      "step": 885
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 9.07731640536698e-06,
+      "loss": 1.2148,
+      "step": 886
+    },
+    {
+      "epoch": 5.44,
+      "learning_rate": 9.057532084583662e-06,
+      "loss": 1.3848,
+      "step": 887
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 9.037751485166785e-06,
+      "loss": 1.2832,
+      "step": 888
+    },
+    {
+      "epoch": 5.45,
+      "learning_rate": 9.017974685220716e-06,
+      "loss": 1.2832,
+      "step": 889
+    },
+    {
+      "epoch": 5.46,
+      "learning_rate": 8.998201762834815e-06,
+      "loss": 1.3906,
+      "step": 890
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 8.97843279608314e-06,
+      "loss": 1.2539,
+      "step": 891
+    },
+    {
+      "epoch": 5.47,
+      "learning_rate": 8.958667863024127e-06,
+      "loss": 1.168,
+      "step": 892
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 8.938907041700275e-06,
+      "loss": 1.3086,
+      "step": 893
+    },
+    {
+      "epoch": 5.48,
+      "learning_rate": 8.919150410137862e-06,
+      "loss": 1.2656,
+      "step": 894
+    },
+    {
+      "epoch": 5.49,
+      "learning_rate": 8.899398046346608e-06,
+      "loss": 1.209,
+      "step": 895
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 8.8796500283194e-06,
+      "loss": 1.2852,
+      "step": 896
+    },
+    {
+      "epoch": 5.5,
+      "learning_rate": 8.859906434031947e-06,
+      "loss": 1.1504,
+      "step": 897
+    },
+    {
+      "epoch": 5.51,
+      "learning_rate": 8.840167341442505e-06,
+      "loss": 1.0957,
+      "step": 898
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 8.820432828491542e-06,
+      "loss": 1.2148,
+      "step": 899
+    },
+    {
+      "epoch": 5.52,
+      "learning_rate": 8.800702973101454e-06,
+      "loss": 1.2832,
+      "step": 900
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 8.78097785317624e-06,
+      "loss": 1.252,
+      "step": 901
+    },
+    {
+      "epoch": 5.53,
+      "learning_rate": 8.761257546601209e-06,
+      "loss": 1.3633,
+      "step": 902
+    },
+    {
+      "epoch": 5.54,
+      "learning_rate": 8.741542131242652e-06,
+      "loss": 1.2246,
+      "step": 903
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 8.721831684947557e-06,
+      "loss": 1.2148,
+      "step": 904
+    },
+    {
+      "epoch": 5.55,
+      "learning_rate": 8.702126285543286e-06,
+      "loss": 1.127,
+      "step": 905
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 8.682426010837274e-06,
+      "loss": 1.25,
+      "step": 906
+    },
+    {
+      "epoch": 5.56,
+      "learning_rate": 8.662730938616724e-06,
+      "loss": 1.2031,
+      "step": 907
+    },
+    {
+      "epoch": 5.57,
+      "learning_rate": 8.643041146648299e-06,
+      "loss": 1.2246,
+      "step": 908
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 8.6233567126778e-06,
+      "loss": 1.3438,
+      "step": 909
+    },
+    {
+      "epoch": 5.58,
+      "learning_rate": 8.603677714429888e-06,
+      "loss": 1.2852,
+      "step": 910
+    },
+    {
+      "epoch": 5.59,
+      "learning_rate": 8.584004229607747e-06,
+      "loss": 1.418,
+      "step": 911
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 8.564336335892798e-06,
+      "loss": 1.3105,
+      "step": 912
+    },
+    {
+      "epoch": 5.6,
+      "learning_rate": 8.54467411094439e-06,
+      "loss": 1.2422,
+      "step": 913
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 8.52501763239948e-06,
+      "loss": 1.2373,
+      "step": 914
+    },
+    {
+      "epoch": 5.61,
+      "learning_rate": 8.505366977872336e-06,
+      "loss": 1.2637,
+      "step": 915
+    },
+    {
+      "epoch": 5.62,
+      "learning_rate": 8.485722224954237e-06,
+      "loss": 1.3906,
+      "step": 916
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 8.466083451213145e-06,
+      "loss": 1.1748,
+      "step": 917
+    },
+    {
+      "epoch": 5.63,
+      "learning_rate": 8.446450734193437e-06,
+      "loss": 1.2949,
+      "step": 918
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 8.426824151415548e-06,
+      "loss": 1.125,
+      "step": 919
+    },
+    {
+      "epoch": 5.64,
+      "learning_rate": 8.407203780375711e-06,
+      "loss": 1.2539,
+      "step": 920
+    },
+    {
+      "epoch": 5.65,
+      "learning_rate": 8.38758969854562e-06,
+      "loss": 1.2305,
+      "step": 921
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 8.367981983372143e-06,
+      "loss": 1.1523,
+      "step": 922
+    },
+    {
+      "epoch": 5.66,
+      "learning_rate": 8.348380712277002e-06,
+      "loss": 1.2285,
+      "step": 923
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 8.32878596265649e-06,
+      "loss": 1.3281,
+      "step": 924
+    },
+    {
+      "epoch": 5.67,
+      "learning_rate": 8.309197811881128e-06,
+      "loss": 1.3379,
+      "step": 925
+    },
+    {
+      "epoch": 5.68,
+      "learning_rate": 8.289616337295396e-06,
+      "loss": 1.2891,
+      "step": 926
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 8.270041616217407e-06,
+      "loss": 1.2441,
+      "step": 927
+    },
+    {
+      "epoch": 5.69,
+      "learning_rate": 8.250473725938608e-06,
+      "loss": 1.3652,
+      "step": 928
+    },
+    {
+      "epoch": 5.7,
+      "learning_rate": 8.23091274372348e-06,
+      "loss": 1.1523,
+      "step": 929
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 8.211358746809225e-06,
+      "loss": 1.2637,
+      "step": 930
+    },
+    {
+      "epoch": 5.71,
+      "learning_rate": 8.191811812405453e-06,
+      "loss": 1.3184,
+      "step": 931
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 8.172272017693903e-06,
+      "loss": 1.2676,
+      "step": 932
+    },
+    {
+      "epoch": 5.72,
+      "learning_rate": 8.15273943982811e-06,
+      "loss": 1.1836,
+      "step": 933
+    },
+    {
+      "epoch": 5.73,
+      "learning_rate": 8.133214155933118e-06,
+      "loss": 1.1533,
+      "step": 934
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 8.113696243105175e-06,
+      "loss": 1.1562,
+      "step": 935
+    },
+    {
+      "epoch": 5.74,
+      "learning_rate": 8.09418577841142e-06,
+      "loss": 1.3008,
+      "step": 936
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 8.074682838889581e-06,
+      "loss": 1.3379,
+      "step": 937
+    },
+    {
+      "epoch": 5.75,
+      "learning_rate": 8.055187501547674e-06,
+      "loss": 1.2012,
+      "step": 938
+    },
+    {
+      "epoch": 5.76,
+      "learning_rate": 8.035699843363696e-06,
+      "loss": 1.1484,
+      "step": 939
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 8.01621994128533e-06,
+      "loss": 1.293,
+      "step": 940
+    },
+    {
+      "epoch": 5.77,
+      "learning_rate": 7.996747872229624e-06,
+      "loss": 1.3223,
+      "step": 941
+    },
+    {
+      "epoch": 5.78,
+      "learning_rate": 7.977283713082706e-06,
+      "loss": 1.3105,
+      "step": 942
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 7.95782754069946e-06,
+      "loss": 1.207,
+      "step": 943
+    },
+    {
+      "epoch": 5.79,
+      "learning_rate": 7.938379431903243e-06,
+      "loss": 1.1992,
+      "step": 944
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 7.91893946348557e-06,
+      "loss": 1.1582,
+      "step": 945
+    },
+    {
+      "epoch": 5.8,
+      "learning_rate": 7.899507712205818e-06,
+      "loss": 1.168,
+      "step": 946
+    },
+    {
+      "epoch": 5.81,
+      "learning_rate": 7.880084254790911e-06,
+      "loss": 1.3105,
+      "step": 947
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 7.860669167935028e-06,
+      "loss": 1.2988,
+      "step": 948
+    },
+    {
+      "epoch": 5.82,
+      "learning_rate": 7.841262528299296e-06,
+      "loss": 1.1211,
+      "step": 949
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 7.821864412511485e-06,
+      "loss": 1.2832,
+      "step": 950
+    },
+    {
+      "epoch": 5.83,
+      "learning_rate": 7.802474897165716e-06,
+      "loss": 1.0977,
+      "step": 951
+    },
+    {
+      "epoch": 5.84,
+      "learning_rate": 7.783094058822147e-06,
+      "loss": 1.0918,
+      "step": 952
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 7.76372197400667e-06,
+      "loss": 1.2617,
+      "step": 953
+    },
+    {
+      "epoch": 5.85,
+      "learning_rate": 7.74435871921062e-06,
+      "loss": 1.2793,
+      "step": 954
+    },
+    {
+      "epoch": 5.86,
+      "learning_rate": 7.72500437089046e-06,
+      "loss": 1.2402,
+      "step": 955
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 7.705659005467489e-06,
+      "loss": 1.2344,
+      "step": 956
+    },
+    {
+      "epoch": 5.87,
+      "learning_rate": 7.68632269932754e-06,
+      "loss": 1.2832,
+      "step": 957
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 7.666995528820673e-06,
+      "loss": 1.2402,
+      "step": 958
+    },
+    {
+      "epoch": 5.88,
+      "learning_rate": 7.647677570260868e-06,
+      "loss": 1.3262,
+      "step": 959
+    },
+    {
+      "epoch": 5.89,
+      "learning_rate": 7.628368899925744e-06,
+      "loss": 1.2695,
+      "step": 960
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 7.609069594056234e-06,
+      "loss": 1.2031,
+      "step": 961
+    },
+    {
+      "epoch": 5.9,
+      "learning_rate": 7.589779728856307e-06,
+      "loss": 1.1484,
+      "step": 962
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 7.570499380492641e-06,
+      "loss": 1.3203,
+      "step": 963
+    },
+    {
+      "epoch": 5.91,
+      "learning_rate": 7.551228625094349e-06,
+      "loss": 1.2754,
+      "step": 964
+    },
+    {
+      "epoch": 5.92,
+      "learning_rate": 7.5319675387526555e-06,
+      "loss": 1.2559,
+      "step": 965
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 7.512716197520614e-06,
+      "loss": 1.209,
+      "step": 966
+    },
+    {
+      "epoch": 5.93,
+      "learning_rate": 7.493474677412795e-06,
+      "loss": 1.1875,
+      "step": 967
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 7.4742430544049945e-06,
+      "loss": 1.2168,
+      "step": 968
+    },
+    {
+      "epoch": 5.94,
+      "learning_rate": 7.4550214044339256e-06,
+      "loss": 1.209,
+      "step": 969
+    },
+    {
+      "epoch": 5.95,
+      "learning_rate": 7.435809803396923e-06,
+      "loss": 1.25,
+      "step": 970
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 7.416608327151642e-06,
+      "loss": 1.1211,
+      "step": 971
+    },
+    {
+      "epoch": 5.96,
+      "learning_rate": 7.397417051515758e-06,
+      "loss": 1.1113,
+      "step": 972
+    },
+    {
+      "epoch": 5.97,
+      "learning_rate": 7.37823605226668e-06,
+      "loss": 1.2422,
+      "step": 973
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 7.359065405141228e-06,
+      "loss": 1.2363,
+      "step": 974
+    },
+    {
+      "epoch": 5.98,
+      "learning_rate": 7.33990518583535e-06,
+      "loss": 1.1338,
+      "step": 975
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 7.320755470003822e-06,
+      "loss": 1.0918,
+      "step": 976
+    },
+    {
+      "epoch": 5.99,
+      "learning_rate": 7.301616333259942e-06,
+      "loss": 1.3027,
+      "step": 977
+    },
+    {
+      "epoch": 6.0,
+      "learning_rate": 7.282487851175237e-06,
+      "loss": 1.0625,
+      "step": 978
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 7.263370099279173e-06,
+      "loss": 0.792,
+      "step": 979
+    },
+    {
+      "epoch": 6.01,
+      "learning_rate": 7.244263153058835e-06,
+      "loss": 0.9102,
+      "step": 980
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 7.225167087958647e-06,
+      "loss": 0.832,
+      "step": 981
+    },
+    {
+      "epoch": 6.02,
+      "learning_rate": 7.2060819793800665e-06,
+      "loss": 0.8662,
+      "step": 982
+    },
+    {
+      "epoch": 6.03,
+      "learning_rate": 7.187007902681289e-06,
+      "loss": 0.8164,
+      "step": 983
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 7.16794493317696e-06,
+      "loss": 0.8496,
+      "step": 984
+    },
+    {
+      "epoch": 6.04,
+      "learning_rate": 7.148893146137852e-06,
+      "loss": 0.9854,
+      "step": 985
+    },
+    {
+      "epoch": 6.05,
+      "learning_rate": 7.129852616790594e-06,
+      "loss": 0.8486,
+      "step": 986
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 7.110823420317356e-06,
+      "loss": 0.8359,
+      "step": 987
+    },
+    {
+      "epoch": 6.06,
+      "learning_rate": 7.091805631855566e-06,
+      "loss": 0.7695,
+      "step": 988
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 7.072799326497603e-06,
+      "loss": 0.8828,
+      "step": 989
+    },
+    {
+      "epoch": 6.07,
+      "learning_rate": 7.053804579290513e-06,
+      "loss": 0.9307,
+      "step": 990
+    },
+    {
+      "epoch": 6.08,
+      "learning_rate": 7.034821465235693e-06,
+      "loss": 0.7568,
+      "step": 991
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 7.0158500592886115e-06,
+      "loss": 0.8779,
+      "step": 992
+    },
+    {
+      "epoch": 6.09,
+      "learning_rate": 6.996890436358505e-06,
+      "loss": 0.9648,
+      "step": 993
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 6.977942671308087e-06,
+      "loss": 0.7734,
+      "step": 994
+    },
+    {
+      "epoch": 6.1,
+      "learning_rate": 6.95900683895325e-06,
+      "loss": 0.8066,
+      "step": 995
+    },
+    {
+      "epoch": 6.11,
+      "learning_rate": 6.9400830140627705e-06,
+      "loss": 0.9189,
+      "step": 996
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 6.921171271358007e-06,
+      "loss": 0.8271,
+      "step": 997
+    },
+    {
+      "epoch": 6.12,
+      "learning_rate": 6.902271685512616e-06,
+      "loss": 0.9258,
+      "step": 998
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 6.883384331152254e-06,
+      "loss": 0.9004,
+      "step": 999
+    },
+    {
+      "epoch": 6.13,
+      "learning_rate": 6.864509282854272e-06,
+      "loss": 0.8652,
+      "step": 1000
+    },
+    {
+      "epoch": 6.14,
+      "learning_rate": 6.845646615147445e-06,
+      "loss": 0.8779,
+      "step": 1001
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 6.826796402511653e-06,
+      "loss": 0.8105,
+      "step": 1002
+    },
+    {
+      "epoch": 6.15,
+      "learning_rate": 6.8079587193775935e-06,
+      "loss": 0.9023,
+      "step": 1003
+    },
+    {
+      "epoch": 6.16,
+      "learning_rate": 6.789133640126498e-06,
+      "loss": 0.8877,
+      "step": 1004
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 6.770321239089825e-06,
+      "loss": 0.9209,
+      "step": 1005
+    },
+    {
+      "epoch": 6.17,
+      "learning_rate": 6.751521590548986e-06,
+      "loss": 0.8389,
+      "step": 1006
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 6.732734768735021e-06,
+      "loss": 0.8125,
+      "step": 1007
+    },
+    {
+      "epoch": 6.18,
+      "learning_rate": 6.713960847828335e-06,
+      "loss": 0.8408,
+      "step": 1008
+    },
+    {
+      "epoch": 6.19,
+      "learning_rate": 6.695199901958386e-06,
+      "loss": 0.9258,
+      "step": 1009
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 6.6764520052034054e-06,
+      "loss": 0.8213,
+      "step": 1010
+    },
+    {
+      "epoch": 6.2,
+      "learning_rate": 6.657717231590095e-06,
+      "loss": 0.8838,
+      "step": 1011
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 6.638995655093351e-06,
+      "loss": 0.667,
+      "step": 1012
+    },
+    {
+      "epoch": 6.21,
+      "learning_rate": 6.620287349635942e-06,
+      "loss": 0.9072,
+      "step": 1013
+    },
+    {
+      "epoch": 6.22,
+      "learning_rate": 6.601592389088251e-06,
+      "loss": 0.8184,
+      "step": 1014
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 6.582910847267957e-06,
+      "loss": 0.9688,
+      "step": 1015
+    },
+    {
+      "epoch": 6.23,
+      "learning_rate": 6.564242797939759e-06,
+      "loss": 0.7861,
+      "step": 1016
+    },
+    {
+      "epoch": 6.24,
+      "learning_rate": 6.545588314815088e-06,
+      "loss": 0.9268,
+      "step": 1017
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 6.526947471551799e-06,
+      "loss": 0.7949,
+      "step": 1018
+    },
+    {
+      "epoch": 6.25,
+      "learning_rate": 6.508320341753889e-06,
+      "loss": 0.8994,
+      "step": 1019
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 6.489706998971212e-06,
+      "loss": 0.8193,
+      "step": 1020
+    },
+    {
+      "epoch": 6.26,
+      "learning_rate": 6.471107516699183e-06,
+      "loss": 0.877,
+      "step": 1021
+    },
+    {
+      "epoch": 6.27,
+      "learning_rate": 6.452521968378482e-06,
+      "loss": 0.8525,
+      "step": 1022
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 6.4339504273947805e-06,
+      "loss": 0.8115,
+      "step": 1023
+    },
+    {
+      "epoch": 6.28,
+      "learning_rate": 6.415392967078438e-06,
+      "loss": 0.8262,
+      "step": 1024
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 6.396849660704205e-06,
+      "loss": 0.9258,
+      "step": 1025
+    },
+    {
+      "epoch": 6.29,
+      "learning_rate": 6.378320581490962e-06,
+      "loss": 0.873,
+      "step": 1026
+    },
+    {
+      "epoch": 6.3,
+      "learning_rate": 6.3598058026013995e-06,
+      "loss": 0.9082,
+      "step": 1027
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 6.3413053971417575e-06,
+      "loss": 0.9756,
+      "step": 1028
+    },
+    {
+      "epoch": 6.31,
+      "learning_rate": 6.322819438161502e-06,
+      "loss": 0.7363,
+      "step": 1029
+    },
+    {
+      "epoch": 6.32,
+      "learning_rate": 6.304347998653074e-06,
+      "loss": 0.835,
+      "step": 1030
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 6.285891151551573e-06,
+      "loss": 0.8457,
+      "step": 1031
+    },
+    {
+      "epoch": 6.33,
+      "learning_rate": 6.267448969734486e-06,
+      "loss": 0.833,
+      "step": 1032
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 6.24902152602139e-06,
+      "loss": 0.7949,
+      "step": 1033
+    },
+    {
+      "epoch": 6.34,
+      "learning_rate": 6.2306088931736766e-06,
+      "loss": 0.9092,
+      "step": 1034
+    },
+    {
+      "epoch": 6.35,
+      "learning_rate": 6.21221114389424e-06,
+      "loss": 0.8643,
+      "step": 1035
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 6.193828350827222e-06,
+      "loss": 0.8809,
+      "step": 1036
+    },
+    {
+      "epoch": 6.36,
+      "learning_rate": 6.175460586557701e-06,
+      "loss": 0.8662,
+      "step": 1037
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 6.157107923611412e-06,
+      "loss": 0.8682,
+      "step": 1038
+    },
+    {
+      "epoch": 6.37,
+      "learning_rate": 6.1387704344544684e-06,
+      "loss": 0.8701,
+      "step": 1039
+    },
+    {
+      "epoch": 6.38,
+      "learning_rate": 6.120448191493071e-06,
+      "loss": 0.791,
+      "step": 1040
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 6.102141267073207e-06,
+      "loss": 0.8857,
+      "step": 1041
+    },
+    {
+      "epoch": 6.39,
+      "learning_rate": 6.083849733480394e-06,
+      "loss": 0.8623,
+      "step": 1042
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 6.065573662939367e-06,
+      "loss": 0.8105,
+      "step": 1043
+    },
+    {
+      "epoch": 6.4,
+      "learning_rate": 6.047313127613808e-06,
+      "loss": 0.9443,
+      "step": 1044
+    },
+    {
+      "epoch": 6.41,
+      "learning_rate": 6.0290681996060605e-06,
+      "loss": 0.7783,
+      "step": 1045
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 6.010838950956841e-06,
+      "loss": 0.8701,
+      "step": 1046
+    },
+    {
+      "epoch": 6.42,
+      "learning_rate": 5.992625453644953e-06,
+      "loss": 0.8672,
+      "step": 1047
+    },
+    {
+      "epoch": 6.43,
+      "learning_rate": 5.974427779587004e-06,
+      "loss": 0.8262,
+      "step": 1048
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 5.9562460006371295e-06,
+      "loss": 0.8818,
+      "step": 1049
+    },
+    {
+      "epoch": 6.44,
+      "learning_rate": 5.938080188586699e-06,
+      "loss": 0.7998,
+      "step": 1050
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 5.919930415164033e-06,
+      "loss": 0.7217,
+      "step": 1051
+    },
+    {
+      "epoch": 6.45,
+      "learning_rate": 5.901796752034128e-06,
+      "loss": 0.8486,
+      "step": 1052
+    },
+    {
+      "epoch": 6.46,
+      "learning_rate": 5.883679270798363e-06,
+      "loss": 0.7949,
+      "step": 1053
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 5.865578042994227e-06,
+      "loss": 0.9209,
+      "step": 1054
+    },
+    {
+      "epoch": 6.47,
+      "learning_rate": 5.84749314009503e-06,
+      "loss": 0.8779,
+      "step": 1055
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 5.829424633509627e-06,
+      "loss": 0.9678,
+      "step": 1056
+    },
+    {
+      "epoch": 6.48,
+      "learning_rate": 5.8113725945821245e-06,
+      "loss": 0.7764,
+      "step": 1057
+    },
+    {
+      "epoch": 6.49,
+      "learning_rate": 5.7933370945916036e-06,
+      "loss": 0.8252,
+      "step": 1058
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 5.775318204751854e-06,
+      "loss": 0.8438,
+      "step": 1059
+    },
+    {
+      "epoch": 6.5,
+      "learning_rate": 5.757315996211066e-06,
+      "loss": 0.7744,
+      "step": 1060
+    },
+    {
+      "epoch": 6.51,
+      "learning_rate": 5.7393305400515755e-06,
+      "loss": 0.8027,
+      "step": 1061
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 5.721361907289556e-06,
+      "loss": 0.834,
+      "step": 1062
+    },
+    {
+      "epoch": 6.52,
+      "learning_rate": 5.703410168874768e-06,
+      "loss": 0.8496,
+      "step": 1063
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 5.685475395690259e-06,
+      "loss": 1.0342,
+      "step": 1064
+    },
+    {
+      "epoch": 6.53,
+      "learning_rate": 5.667557658552078e-06,
+      "loss": 0.8789,
+      "step": 1065
+    },
+    {
+      "epoch": 6.54,
+      "learning_rate": 5.649657028209024e-06,
+      "loss": 0.7568,
+      "step": 1066
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 5.631773575342343e-06,
+      "loss": 0.791,
+      "step": 1067
+    },
+    {
+      "epoch": 6.55,
+      "learning_rate": 5.61390737056545e-06,
+      "loss": 0.9238,
+      "step": 1068
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 5.5960584844236565e-06,
+      "loss": 0.7002,
+      "step": 1069
+    },
+    {
+      "epoch": 6.56,
+      "learning_rate": 5.5782269873939e-06,
+      "loss": 0.8096,
+      "step": 1070
+    },
+    {
+      "epoch": 6.57,
+      "learning_rate": 5.560412949884442e-06,
+      "loss": 0.8545,
+      "step": 1071
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 5.542616442234618e-06,
+      "loss": 0.8203,
+      "step": 1072
+    },
+    {
+      "epoch": 6.58,
+      "learning_rate": 5.52483753471454e-06,
+      "loss": 0.8271,
+      "step": 1073
+    },
+    {
+      "epoch": 6.59,
+      "learning_rate": 5.507076297524818e-06,
+      "loss": 0.8428,
+      "step": 1074
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 5.48933280079631e-06,
+      "loss": 0.8076,
+      "step": 1075
+    },
+    {
+      "epoch": 6.6,
+      "learning_rate": 5.471607114589806e-06,
+      "loss": 0.8057,
+      "step": 1076
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 5.453899308895774e-06,
+      "loss": 0.7715,
+      "step": 1077
+    },
+    {
+      "epoch": 6.61,
+      "learning_rate": 5.436209453634087e-06,
+      "loss": 0.7207,
+      "step": 1078
+    },
+    {
+      "epoch": 6.62,
+      "learning_rate": 5.418537618653743e-06,
+      "loss": 0.7812,
+      "step": 1079
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 5.400883873732574e-06,
+      "loss": 0.8213,
+      "step": 1080
+    },
+    {
+      "epoch": 6.63,
+      "learning_rate": 5.3832482885769855e-06,
+      "loss": 0.7451,
+      "step": 1081
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 5.365630932821688e-06,
+      "loss": 0.835,
+      "step": 1082
+    },
+    {
+      "epoch": 6.64,
+      "learning_rate": 5.3480318760294084e-06,
+      "loss": 0.8604,
+      "step": 1083
+    },
+    {
+      "epoch": 6.65,
+      "learning_rate": 5.330451187690614e-06,
+      "loss": 0.9072,
+      "step": 1084
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 5.3128889372232436e-06,
+      "loss": 0.8721,
+      "step": 1085
+    },
+    {
+      "epoch": 6.66,
+      "learning_rate": 5.295345193972445e-06,
+      "loss": 0.8779,
+      "step": 1086
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 5.277820027210279e-06,
+      "loss": 0.8916,
+      "step": 1087
+    },
+    {
+      "epoch": 6.67,
+      "learning_rate": 5.260313506135452e-06,
+      "loss": 0.8721,
+      "step": 1088
+    },
+    {
+      "epoch": 6.68,
+      "learning_rate": 5.242825699873068e-06,
+      "loss": 0.8613,
+      "step": 1089
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 5.225356677474309e-06,
+      "loss": 0.8379,
+      "step": 1090
+    },
+    {
+      "epoch": 6.69,
+      "learning_rate": 5.2079065079162115e-06,
+      "loss": 0.708,
+      "step": 1091
+    },
+    {
+      "epoch": 6.7,
+      "learning_rate": 5.190475260101353e-06,
+      "loss": 0.873,
+      "step": 1092
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 5.1730630028576055e-06,
+      "loss": 0.7119,
+      "step": 1093
+    },
+    {
+      "epoch": 6.71,
+      "learning_rate": 5.155669804937855e-06,
+      "loss": 0.8848,
+      "step": 1094
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 5.138295735019741e-06,
+      "loss": 0.8633,
+      "step": 1095
+    },
+    {
+      "epoch": 6.72,
+      "learning_rate": 5.120940861705357e-06,
+      "loss": 0.8203,
+      "step": 1096
+    },
+    {
+      "epoch": 6.73,
+      "learning_rate": 5.103605253521007e-06,
+      "loss": 0.8398,
+      "step": 1097
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 5.086288978916931e-06,
+      "loss": 0.9297,
+      "step": 1098
+    },
+    {
+      "epoch": 6.74,
+      "learning_rate": 5.068992106267021e-06,
+      "loss": 0.71,
+      "step": 1099
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 5.051714703868569e-06,
+      "loss": 0.7275,
+      "step": 1100
+    },
+    {
+      "epoch": 6.75,
+      "learning_rate": 5.034456839941979e-06,
+      "loss": 0.8164,
+      "step": 1101
+    },
+    {
+      "epoch": 6.76,
+      "learning_rate": 5.017218582630507e-06,
+      "loss": 0.7363,
+      "step": 1102
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 5.000000000000003e-06,
+      "loss": 0.9561,
+      "step": 1103
+    },
+    {
+      "epoch": 6.77,
+      "learning_rate": 4.982801160038614e-06,
+      "loss": 0.834,
+      "step": 1104
+    },
+    {
+      "epoch": 6.78,
+      "learning_rate": 4.965622130656551e-06,
+      "loss": 0.8418,
+      "step": 1105
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 4.948462979685783e-06,
+      "loss": 0.8418,
+      "step": 1106
+    },
+    {
+      "epoch": 6.79,
+      "learning_rate": 4.931323774879807e-06,
+      "loss": 0.8584,
+      "step": 1107
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 4.914204583913349e-06,
+      "loss": 0.8105,
+      "step": 1108
+    },
+    {
+      "epoch": 6.8,
+      "learning_rate": 4.897105474382109e-06,
+      "loss": 0.9131,
+      "step": 1109
+    },
+    {
+      "epoch": 6.81,
+      "learning_rate": 4.880026513802504e-06,
+      "loss": 0.791,
+      "step": 1110
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 4.862967769611389e-06,
+      "loss": 0.8828,
+      "step": 1111
+    },
+    {
+      "epoch": 6.82,
+      "learning_rate": 4.845929309165793e-06,
+      "loss": 0.8291,
+      "step": 1112
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 4.828911199742646e-06,
+      "loss": 0.8252,
+      "step": 1113
+    },
+    {
+      "epoch": 6.83,
+      "learning_rate": 4.8119135085385375e-06,
+      "loss": 0.7529,
+      "step": 1114
+    },
+    {
+      "epoch": 6.84,
+      "learning_rate": 4.794936302669417e-06,
+      "loss": 0.8613,
+      "step": 1115
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 4.777979649170367e-06,
+      "loss": 0.7803,
+      "step": 1116
+    },
+    {
+      "epoch": 6.85,
+      "learning_rate": 4.7610436149953e-06,
+      "loss": 0.9141,
+      "step": 1117
+    },
+    {
+      "epoch": 6.86,
+      "learning_rate": 4.744128267016719e-06,
+      "loss": 0.8291,
+      "step": 1118
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 4.727233672025453e-06,
+      "loss": 0.7451,
+      "step": 1119
+    },
+    {
+      "epoch": 6.87,
+      "learning_rate": 4.710359896730379e-06,
+      "loss": 0.8457,
+      "step": 1120
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 4.693507007758165e-06,
+      "loss": 0.7646,
+      "step": 1121
+    },
+    {
+      "epoch": 6.88,
+      "learning_rate": 4.676675071653019e-06,
+      "loss": 0.8506,
+      "step": 1122
+    },
+    {
+      "epoch": 6.89,
+      "learning_rate": 4.659864154876411e-06,
+      "loss": 0.7246,
+      "step": 1123
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 4.643074323806813e-06,
+      "loss": 0.8555,
+      "step": 1124
+    },
+    {
+      "epoch": 6.9,
+      "learning_rate": 4.626305644739435e-06,
+      "loss": 0.8125,
+      "step": 1125
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 4.609558183885979e-06,
+      "loss": 0.8418,
+      "step": 1126
+    },
+    {
+      "epoch": 6.91,
+      "learning_rate": 4.592832007374364e-06,
+      "loss": 0.8271,
+      "step": 1127
+    },
+    {
+      "epoch": 6.92,
+      "learning_rate": 4.576127181248459e-06,
+      "loss": 0.7979,
+      "step": 1128
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 4.559443771467833e-06,
+      "loss": 0.8438,
+      "step": 1129
+    },
+    {
+      "epoch": 6.93,
+      "learning_rate": 4.542781843907499e-06,
+      "loss": 0.7432,
+      "step": 1130
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 4.5261414643576396e-06,
+      "loss": 0.7852,
+      "step": 1131
+    },
+    {
+      "epoch": 6.94,
+      "learning_rate": 4.509522698523352e-06,
+      "loss": 0.8125,
+      "step": 1132
+    },
+    {
+      "epoch": 6.95,
+      "learning_rate": 4.492925612024402e-06,
+      "loss": 0.7588,
+      "step": 1133
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 4.476350270394942e-06,
+      "loss": 0.751,
+      "step": 1134
+    },
+    {
+      "epoch": 6.96,
+      "learning_rate": 4.4597967390832745e-06,
+      "loss": 0.9287,
+      "step": 1135
+    },
+    {
+      "epoch": 6.97,
+      "learning_rate": 4.4432650834515735e-06,
+      "loss": 0.7432,
+      "step": 1136
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 4.426755368775637e-06,
+      "loss": 0.7783,
+      "step": 1137
+    },
+    {
+      "epoch": 6.98,
+      "learning_rate": 4.4102676602446375e-06,
+      "loss": 0.8613,
+      "step": 1138
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 4.3938020229608506e-06,
+      "loss": 0.8584,
+      "step": 1139
+    },
+    {
+      "epoch": 6.99,
+      "learning_rate": 4.377358521939401e-06,
+      "loss": 0.8105,
+      "step": 1140
+    },
+    {
+      "epoch": 7.0,
+      "learning_rate": 4.360937222108002e-06,
+      "loss": 0.7871,
+      "step": 1141
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 4.344538188306723e-06,
+      "loss": 0.5469,
+      "step": 1142
+    },
+    {
+      "epoch": 7.01,
+      "learning_rate": 4.328161485287693e-06,
+      "loss": 0.6025,
+      "step": 1143
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 4.3118071777148865e-06,
+      "loss": 0.5752,
+      "step": 1144
+    },
+    {
+      "epoch": 7.02,
+      "learning_rate": 4.295475330163832e-06,
+      "loss": 0.6367,
+      "step": 1145
+    },
+    {
+      "epoch": 7.03,
+      "learning_rate": 4.279166007121389e-06,
+      "loss": 0.5527,
+      "step": 1146
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 4.262879272985468e-06,
+      "loss": 0.5439,
+      "step": 1147
+    },
+    {
+      "epoch": 7.04,
+      "learning_rate": 4.246615192064787e-06,
+      "loss": 0.5586,
+      "step": 1148
+    },
+    {
+      "epoch": 7.05,
+      "learning_rate": 4.230373828578626e-06,
+      "loss": 0.6318,
+      "step": 1149
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 4.21415524665655e-06,
+      "loss": 0.6299,
+      "step": 1150
+    },
+    {
+      "epoch": 7.06,
+      "learning_rate": 4.197959510338187e-06,
+      "loss": 0.583,
+      "step": 1151
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 4.181786683572946e-06,
+      "loss": 0.626,
+      "step": 1152
+    },
+    {
+      "epoch": 7.07,
+      "learning_rate": 4.165636830219776e-06,
+      "loss": 0.5845,
+      "step": 1153
+    },
+    {
+      "epoch": 7.08,
+      "learning_rate": 4.149510014046922e-06,
+      "loss": 0.5723,
+      "step": 1154
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 4.1334062987316695e-06,
+      "loss": 0.5391,
+      "step": 1155
+    },
+    {
+      "epoch": 7.09,
+      "learning_rate": 4.117325747860077e-06,
+      "loss": 0.5967,
+      "step": 1156
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 4.101268424926741e-06,
+      "loss": 0.6357,
+      "step": 1157
+    },
+    {
+      "epoch": 7.1,
+      "learning_rate": 4.085234393334551e-06,
+      "loss": 0.5654,
+      "step": 1158
+    },
+    {
+      "epoch": 7.11,
+      "learning_rate": 4.069223716394419e-06,
+      "loss": 0.5889,
+      "step": 1159
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 4.053236457325043e-06,
+      "loss": 0.5615,
+      "step": 1160
+    },
+    {
+      "epoch": 7.12,
+      "learning_rate": 4.0372726792526614e-06,
+      "loss": 0.5459,
+      "step": 1161
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 4.021332445210785e-06,
+      "loss": 0.6182,
+      "step": 1162
+    },
+    {
+      "epoch": 7.13,
+      "learning_rate": 4.005415818139975e-06,
+      "loss": 0.6357,
+      "step": 1163
+    },
+    {
+      "epoch": 7.14,
+      "learning_rate": 3.989522860887567e-06,
+      "loss": 0.5,
+      "step": 1164
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 3.973653636207437e-06,
+      "loss": 0.5625,
+      "step": 1165
+    },
+    {
+      "epoch": 7.15,
+      "learning_rate": 3.95780820675976e-06,
+      "loss": 0.6074,
+      "step": 1166
+    },
+    {
+      "epoch": 7.16,
+      "learning_rate": 3.941986635110754e-06,
+      "loss": 0.6416,
+      "step": 1167
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 3.9261889837324245e-06,
+      "loss": 0.5239,
+      "step": 1168
+    },
+    {
+      "epoch": 7.17,
+      "learning_rate": 3.910415315002328e-06,
+      "loss": 0.5127,
+      "step": 1169
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 3.89466569120334e-06,
+      "loss": 0.5771,
+      "step": 1170
+    },
+    {
+      "epoch": 7.18,
+      "learning_rate": 3.878940174523371e-06,
+      "loss": 0.6367,
+      "step": 1171
+    },
+    {
+      "epoch": 7.19,
+      "learning_rate": 3.8632388270551665e-06,
+      "loss": 0.6191,
+      "step": 1172
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 3.847561710796019e-06,
+      "loss": 0.5928,
+      "step": 1173
+    },
+    {
+      "epoch": 7.2,
+      "learning_rate": 3.8319088876475595e-06,
+      "loss": 0.5742,
+      "step": 1174
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 3.816280419415487e-06,
+      "loss": 0.6201,
+      "step": 1175
+    },
+    {
+      "epoch": 7.21,
+      "learning_rate": 3.8006763678093326e-06,
+      "loss": 0.6885,
+      "step": 1176
+    },
+    {
+      "epoch": 7.22,
+      "learning_rate": 3.785096794442229e-06,
+      "loss": 0.5742,
+      "step": 1177
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 3.7695417608306415e-06,
+      "loss": 0.5352,
+      "step": 1178
+    },
+    {
+      "epoch": 7.23,
+      "learning_rate": 3.7540113283941536e-06,
+      "loss": 0.6123,
+      "step": 1179
+    },
+    {
+      "epoch": 7.24,
+      "learning_rate": 3.7385055584552e-06,
+      "loss": 0.5605,
+      "step": 1180
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 3.723024512238833e-06,
+      "loss": 0.541,
+      "step": 1181
+    },
+    {
+      "epoch": 7.25,
+      "learning_rate": 3.707568250872493e-06,
+      "loss": 0.6328,
+      "step": 1182
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 3.6921368353857524e-06,
+      "loss": 0.5498,
+      "step": 1183
+    },
+    {
+      "epoch": 7.26,
+      "learning_rate": 3.676730326710074e-06,
+      "loss": 0.5938,
+      "step": 1184
+    },
+    {
+      "epoch": 7.27,
+      "learning_rate": 3.6613487856785744e-06,
+      "loss": 0.5742,
+      "step": 1185
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 3.645992273025797e-06,
+      "loss": 0.5493,
+      "step": 1186
+    },
+    {
+      "epoch": 7.28,
+      "learning_rate": 3.630660849387444e-06,
+      "loss": 0.5947,
+      "step": 1187
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 3.6153545753001663e-06,
+      "loss": 0.5522,
+      "step": 1188
+    },
+    {
+      "epoch": 7.29,
+      "learning_rate": 3.6000735112012984e-06,
+      "loss": 0.5967,
+      "step": 1189
+    },
+    {
+      "epoch": 7.3,
+      "learning_rate": 3.584817717428647e-06,
+      "loss": 0.6006,
+      "step": 1190
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 3.569587254220225e-06,
+      "loss": 0.5664,
+      "step": 1191
+    },
+    {
+      "epoch": 7.31,
+      "learning_rate": 3.5543821817140313e-06,
+      "loss": 0.5898,
+      "step": 1192
+    },
+    {
+      "epoch": 7.32,
+      "learning_rate": 3.5392025599478053e-06,
+      "loss": 0.4985,
+      "step": 1193
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 3.5240484488588012e-06,
+      "loss": 0.5273,
+      "step": 1194
+    },
+    {
+      "epoch": 7.33,
+      "learning_rate": 3.5089199082835436e-06,
+      "loss": 0.627,
+      "step": 1195
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 3.493816997957582e-06,
+      "loss": 0.5479,
+      "step": 1196
+    },
+    {
+      "epoch": 7.34,
+      "learning_rate": 3.478739777515264e-06,
+      "loss": 0.5625,
+      "step": 1197
+    },
+    {
+      "epoch": 7.35,
+      "learning_rate": 3.463688306489511e-06,
+      "loss": 0.5649,
+      "step": 1198
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 3.448662644311567e-06,
+      "loss": 0.6064,
+      "step": 1199
+    },
+    {
+      "epoch": 7.36,
+      "learning_rate": 3.433662850310763e-06,
+      "loss": 0.6211,
+      "step": 1200
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 3.418688983714291e-06,
+      "loss": 0.5337,
+      "step": 1201
+    },
+    {
+      "epoch": 7.37,
+      "learning_rate": 3.403741103646977e-06,
+      "loss": 0.6035,
+      "step": 1202
+    },
+    {
+      "epoch": 7.38,
+      "learning_rate": 3.3888192691310262e-06,
+      "loss": 0.5508,
+      "step": 1203
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 3.373923539085805e-06,
+      "loss": 0.5215,
+      "step": 1204
+    },
+    {
+      "epoch": 7.39,
+      "learning_rate": 3.3590539723276083e-06,
+      "loss": 0.5239,
+      "step": 1205
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 3.3442106275694295e-06,
+      "loss": 0.5444,
+      "step": 1206
+    },
+    {
+      "epoch": 7.4,
+      "learning_rate": 3.329393563420713e-06,
+      "loss": 0.6401,
+      "step": 1207
+    },
+    {
+      "epoch": 7.41,
+      "learning_rate": 3.3146028383871363e-06,
+      "loss": 0.5825,
+      "step": 1208
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 3.2998385108703766e-06,
+      "loss": 0.5347,
+      "step": 1209
+    },
+    {
+      "epoch": 7.42,
+      "learning_rate": 3.285100639167883e-06,
+      "loss": 0.5645,
+      "step": 1210
+    },
+    {
+      "epoch": 7.43,
+      "learning_rate": 3.2703892814726436e-06,
+      "loss": 0.5459,
+      "step": 1211
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 3.2557044958729466e-06,
+      "loss": 0.582,
+      "step": 1212
+    },
+    {
+      "epoch": 7.44,
+      "learning_rate": 3.2410463403521653e-06,
+      "loss": 0.6035,
+      "step": 1213
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 3.2264148727885257e-06,
+      "loss": 0.6094,
+      "step": 1214
+    },
+    {
+      "epoch": 7.45,
+      "learning_rate": 3.211810150954867e-06,
+      "loss": 0.5801,
+      "step": 1215
+    },
+    {
+      "epoch": 7.46,
+      "learning_rate": 3.1972322325184347e-06,
+      "loss": 0.6016,
+      "step": 1216
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 3.182681175040625e-06,
+      "loss": 0.5352,
+      "step": 1217
+    },
+    {
+      "epoch": 7.47,
+      "learning_rate": 3.1681570359767875e-06,
+      "loss": 0.5757,
+      "step": 1218
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 3.1536598726759747e-06,
+      "loss": 0.5894,
+      "step": 1219
+    },
+    {
+      "epoch": 7.48,
+      "learning_rate": 3.1391897423807204e-06,
+      "loss": 0.4736,
+      "step": 1220
+    },
+    {
+      "epoch": 7.49,
+      "learning_rate": 3.1247467022268284e-06,
+      "loss": 0.4985,
+      "step": 1221
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 3.110330809243134e-06,
+      "loss": 0.5459,
+      "step": 1222
+    },
+    {
+      "epoch": 7.5,
+      "learning_rate": 3.095942120351276e-06,
+      "loss": 0.4756,
+      "step": 1223
+    },
+    {
+      "epoch": 7.51,
+      "learning_rate": 3.081580692365478e-06,
+      "loss": 0.5908,
+      "step": 1224
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 3.0672465819923215e-06,
+      "loss": 0.583,
+      "step": 1225
+    },
+    {
+      "epoch": 7.52,
+      "learning_rate": 3.052939845830528e-06,
+      "loss": 0.5034,
+      "step": 1226
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 3.0386605403707347e-06,
+      "loss": 0.4697,
+      "step": 1227
+    },
+    {
+      "epoch": 7.53,
+      "learning_rate": 3.0244087219952565e-06,
+      "loss": 0.5146,
+      "step": 1228
+    },
+    {
+      "epoch": 7.54,
+      "learning_rate": 3.0101844469778797e-06,
+      "loss": 0.5674,
+      "step": 1229
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 2.9959877714836406e-06,
+      "loss": 0.542,
+      "step": 1230
+    },
+    {
+      "epoch": 7.55,
+      "learning_rate": 2.981818751568586e-06,
+      "loss": 0.5669,
+      "step": 1231
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 2.9676774431795752e-06,
+      "loss": 0.5244,
+      "step": 1232
+    },
+    {
+      "epoch": 7.56,
+      "learning_rate": 2.95356390215404e-06,
+      "loss": 0.5679,
+      "step": 1233
+    },
+    {
+      "epoch": 7.57,
+      "learning_rate": 2.939478184219777e-06,
+      "loss": 0.4868,
+      "step": 1234
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 2.9254203449947196e-06,
+      "loss": 0.5498,
+      "step": 1235
+    },
+    {
+      "epoch": 7.58,
+      "learning_rate": 2.9113904399867188e-06,
+      "loss": 0.6143,
+      "step": 1236
+    },
+    {
+      "epoch": 7.59,
+      "learning_rate": 2.8973885245933287e-06,
+      "loss": 0.6279,
+      "step": 1237
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 2.8834146541015874e-06,
+      "loss": 0.5552,
+      "step": 1238
+    },
+    {
+      "epoch": 7.6,
+      "learning_rate": 2.869468883687798e-06,
+      "loss": 0.5186,
+      "step": 1239
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 2.855551268417305e-06,
+      "loss": 0.5244,
+      "step": 1240
+    },
+    {
+      "epoch": 7.61,
+      "learning_rate": 2.8416618632442785e-06,
+      "loss": 0.5884,
+      "step": 1241
+    },
+    {
+      "epoch": 7.62,
+      "learning_rate": 2.827800723011508e-06,
+      "loss": 0.6289,
+      "step": 1242
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 2.813967902450179e-06,
+      "loss": 0.5732,
+      "step": 1243
+    },
+    {
+      "epoch": 7.63,
+      "learning_rate": 2.8001634561796463e-06,
+      "loss": 0.5527,
+      "step": 1244
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 2.786387438707231e-06,
+      "loss": 0.5835,
+      "step": 1245
+    },
+    {
+      "epoch": 7.64,
+      "learning_rate": 2.7726399044280107e-06,
+      "loss": 0.5557,
+      "step": 1246
+    },
+    {
+      "epoch": 7.65,
+      "learning_rate": 2.758920907624585e-06,
+      "loss": 0.5322,
+      "step": 1247
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 2.7452305024668747e-06,
+      "loss": 0.54,
+      "step": 1248
+    },
+    {
+      "epoch": 7.66,
+      "learning_rate": 2.7315687430119097e-06,
+      "loss": 0.6719,
+      "step": 1249
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 2.7179356832036142e-06,
+      "loss": 0.6846,
+      "step": 1250
+    },
+    {
+      "epoch": 7.67,
+      "learning_rate": 2.704331376872581e-06,
+      "loss": 0.5723,
+      "step": 1251
+    },
+    {
+      "epoch": 7.68,
+      "learning_rate": 2.6907558777358756e-06,
+      "loss": 0.5562,
+      "step": 1252
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 2.677209239396811e-06,
+      "loss": 0.5967,
+      "step": 1253
+    },
+    {
+      "epoch": 7.69,
+      "learning_rate": 2.6636915153447494e-06,
+      "loss": 0.4829,
+      "step": 1254
+    },
+    {
+      "epoch": 7.7,
+      "learning_rate": 2.650202758954886e-06,
+      "loss": 0.6201,
+      "step": 1255
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 2.6367430234880286e-06,
+      "loss": 0.4766,
+      "step": 1256
+    },
+    {
+      "epoch": 7.71,
+      "learning_rate": 2.6233123620903946e-06,
+      "loss": 0.583,
+      "step": 1257
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 2.6099108277934105e-06,
+      "loss": 0.5054,
+      "step": 1258
+    },
+    {
+      "epoch": 7.72,
+      "learning_rate": 2.5965384735134825e-06,
+      "loss": 0.5459,
+      "step": 1259
+    },
+    {
+      "epoch": 7.73,
+      "learning_rate": 2.583195352051808e-06,
+      "loss": 0.5312,
+      "step": 1260
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 2.5698815160941494e-06,
+      "loss": 0.584,
+      "step": 1261
+    },
+    {
+      "epoch": 7.74,
+      "learning_rate": 2.5565970182106425e-06,
+      "loss": 0.5928,
+      "step": 1262
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 2.5433419108555758e-06,
+      "loss": 0.5205,
+      "step": 1263
+    },
+    {
+      "epoch": 7.75,
+      "learning_rate": 2.5301162463671845e-06,
+      "loss": 0.5303,
+      "step": 1264
+    },
+    {
+      "epoch": 7.76,
+      "learning_rate": 2.516920076967455e-06,
+      "loss": 0.5615,
+      "step": 1265
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 2.5037534547619125e-06,
+      "loss": 0.6182,
+      "step": 1266
+    },
+    {
+      "epoch": 7.77,
+      "learning_rate": 2.4906164317394067e-06,
+      "loss": 0.5088,
+      "step": 1267
+    },
+    {
+      "epoch": 7.78,
+      "learning_rate": 2.4775090597719163e-06,
+      "loss": 0.5264,
+      "step": 1268
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 2.4644313906143414e-06,
+      "loss": 0.5195,
+      "step": 1269
+    },
+    {
+      "epoch": 7.79,
+      "learning_rate": 2.451383475904304e-06,
+      "loss": 0.5332,
+      "step": 1270
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 2.438365367161939e-06,
+      "loss": 0.5718,
+      "step": 1271
+    },
+    {
+      "epoch": 7.8,
+      "learning_rate": 2.4253771157896856e-06,
+      "loss": 0.5269,
+      "step": 1272
+    },
+    {
+      "epoch": 7.81,
+      "learning_rate": 2.4124187730720916e-06,
+      "loss": 0.563,
+      "step": 1273
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 2.3994903901756163e-06,
+      "loss": 0.5156,
+      "step": 1274
+    },
+    {
+      "epoch": 7.82,
+      "learning_rate": 2.3865920181484127e-06,
+      "loss": 0.478,
+      "step": 1275
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 2.3737237079201437e-06,
+      "loss": 0.5879,
+      "step": 1276
+    },
+    {
+      "epoch": 7.83,
+      "learning_rate": 2.3608855103017613e-06,
+      "loss": 0.5972,
+      "step": 1277
+    },
+    {
+      "epoch": 7.84,
+      "learning_rate": 2.3480774759853307e-06,
+      "loss": 0.5254,
+      "step": 1278
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 2.3352996555438036e-06,
+      "loss": 0.5645,
+      "step": 1279
+    },
+    {
+      "epoch": 7.85,
+      "learning_rate": 2.3225520994308382e-06,
+      "loss": 0.5957,
+      "step": 1280
+    },
+    {
+      "epoch": 7.86,
+      "learning_rate": 2.309834857980583e-06,
+      "loss": 0.5371,
+      "step": 1281
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 2.297147981407509e-06,
+      "loss": 0.5508,
+      "step": 1282
+    },
+    {
+      "epoch": 7.87,
+      "learning_rate": 2.2844915198061714e-06,
+      "loss": 0.4985,
+      "step": 1283
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 2.2718655231510368e-06,
+      "loss": 0.5928,
+      "step": 1284
+    },
+    {
+      "epoch": 7.88,
+      "learning_rate": 2.2592700412962775e-06,
+      "loss": 0.5928,
+      "step": 1285
+    },
+    {
+      "epoch": 7.89,
+      "learning_rate": 2.246705123975582e-06,
+      "loss": 0.6377,
+      "step": 1286
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 2.234170820801954e-06,
+      "loss": 0.5674,
+      "step": 1287
+    },
+    {
+      "epoch": 7.9,
+      "learning_rate": 2.2216671812675118e-06,
+      "loss": 0.4785,
+      "step": 1288
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 2.209194254743295e-06,
+      "loss": 0.5767,
+      "step": 1289
+    },
+    {
+      "epoch": 7.91,
+      "learning_rate": 2.196752090479083e-06,
+      "loss": 0.5601,
+      "step": 1290
+    },
+    {
+      "epoch": 7.92,
+      "learning_rate": 2.184340737603178e-06,
+      "loss": 0.4595,
+      "step": 1291
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 2.1719602451222245e-06,
+      "loss": 0.5625,
+      "step": 1292
+    },
+    {
+      "epoch": 7.93,
+      "learning_rate": 2.159610661921018e-06,
+      "loss": 0.5679,
+      "step": 1293
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 2.1472920367623094e-06,
+      "loss": 0.6499,
+      "step": 1294
+    },
+    {
+      "epoch": 7.94,
+      "learning_rate": 2.1350044182866025e-06,
+      "loss": 0.4966,
+      "step": 1295
+    },
+    {
+      "epoch": 7.95,
+      "learning_rate": 2.1227478550119763e-06,
+      "loss": 0.5933,
+      "step": 1296
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 2.1105223953338805e-06,
+      "loss": 0.4814,
+      "step": 1297
+    },
+    {
+      "epoch": 7.96,
+      "learning_rate": 2.09832808752496e-06,
+      "loss": 0.5088,
+      "step": 1298
+    },
+    {
+      "epoch": 7.97,
+      "learning_rate": 2.086164979734856e-06,
+      "loss": 0.5586,
+      "step": 1299
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 2.0740331199900053e-06,
+      "loss": 0.5396,
+      "step": 1300
+    },
+    {
+      "epoch": 7.98,
+      "learning_rate": 2.0619325561934658e-06,
+      "loss": 0.6182,
+      "step": 1301
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 2.0498633361247278e-06,
+      "loss": 0.5537,
+      "step": 1302
+    },
+    {
+      "epoch": 7.99,
+      "learning_rate": 2.0378255074395094e-06,
+      "loss": 0.5107,
+      "step": 1303
+    },
+    {
+      "epoch": 8.0,
+      "learning_rate": 2.0258191176695896e-06,
+      "loss": 0.5176,
+      "step": 1304
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 2.0138442142226e-06,
+      "loss": 0.4658,
+      "step": 1305
+    },
+    {
+      "epoch": 8.01,
+      "learning_rate": 2.001900844381857e-06,
+      "loss": 0.3608,
+      "step": 1306
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 1.9899890553061565e-06,
+      "loss": 0.4785,
+      "step": 1307
+    },
+    {
+      "epoch": 8.02,
+      "learning_rate": 1.978108894029598e-06,
+      "loss": 0.4692,
+      "step": 1308
+    },
+    {
+      "epoch": 8.03,
+      "learning_rate": 1.9662604074614044e-06,
+      "loss": 0.4463,
+      "step": 1309
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 1.954443642385727e-06,
+      "loss": 0.4473,
+      "step": 1310
+    },
+    {
+      "epoch": 8.04,
+      "learning_rate": 1.9426586454614617e-06,
+      "loss": 0.3853,
+      "step": 1311
+    },
+    {
+      "epoch": 8.05,
+      "learning_rate": 1.9309054632220645e-06,
+      "loss": 0.4043,
+      "step": 1312
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 1.919184142075372e-06,
+      "loss": 0.3589,
+      "step": 1313
+    },
+    {
+      "epoch": 8.06,
+      "learning_rate": 1.9074947283034206e-06,
+      "loss": 0.3608,
+      "step": 1314
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 1.895837268062256e-06,
+      "loss": 0.499,
+      "step": 1315
+    },
+    {
+      "epoch": 8.07,
+      "learning_rate": 1.884211807381755e-06,
+      "loss": 0.4058,
+      "step": 1316
+    },
+    {
+      "epoch": 8.08,
+      "learning_rate": 1.8726183921654373e-06,
+      "loss": 0.5142,
+      "step": 1317
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 1.8610570681903018e-06,
+      "loss": 0.3506,
+      "step": 1318
+    },
+    {
+      "epoch": 8.09,
+      "learning_rate": 1.8495278811066197e-06,
+      "loss": 0.4849,
+      "step": 1319
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 1.8380308764377841e-06,
+      "loss": 0.3979,
+      "step": 1320
+    },
+    {
+      "epoch": 8.1,
+      "learning_rate": 1.8265660995801004e-06,
+      "loss": 0.375,
+      "step": 1321
+    },
+    {
+      "epoch": 8.11,
+      "learning_rate": 1.8151335958026317e-06,
+      "loss": 0.4575,
+      "step": 1322
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 1.803733410247006e-06,
+      "loss": 0.3691,
+      "step": 1323
+    },
+    {
+      "epoch": 8.12,
+      "learning_rate": 1.7923655879272395e-06,
+      "loss": 0.4448,
+      "step": 1324
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 1.7810301737295588e-06,
+      "loss": 0.4111,
+      "step": 1325
+    },
+    {
+      "epoch": 8.13,
+      "learning_rate": 1.76972721241224e-06,
+      "loss": 0.3872,
+      "step": 1326
+    },
+    {
+      "epoch": 8.14,
+      "learning_rate": 1.7584567486054039e-06,
+      "loss": 0.4336,
+      "step": 1327
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 1.7472188268108569e-06,
+      "loss": 0.3569,
+      "step": 1328
+    },
+    {
+      "epoch": 8.15,
+      "learning_rate": 1.7360134914019122e-06,
+      "loss": 0.4526,
+      "step": 1329
+    },
+    {
+      "epoch": 8.16,
+      "learning_rate": 1.7248407866232175e-06,
+      "loss": 0.4351,
+      "step": 1330
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 1.7137007565905772e-06,
+      "loss": 0.3394,
+      "step": 1331
+    },
+    {
+      "epoch": 8.17,
+      "learning_rate": 1.7025934452907755e-06,
+      "loss": 0.439,
+      "step": 1332
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 1.6915188965814034e-06,
+      "loss": 0.437,
+      "step": 1333
+    },
+    {
+      "epoch": 8.18,
+      "learning_rate": 1.6804771541906972e-06,
+      "loss": 0.3999,
+      "step": 1334
+    },
+    {
+      "epoch": 8.19,
+      "learning_rate": 1.6694682617173452e-06,
+      "loss": 0.3999,
+      "step": 1335
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 1.6584922626303325e-06,
+      "loss": 0.4165,
+      "step": 1336
+    },
+    {
+      "epoch": 8.2,
+      "learning_rate": 1.6475492002687632e-06,
+      "loss": 0.4141,
+      "step": 1337
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 1.6366391178416918e-06,
+      "loss": 0.397,
+      "step": 1338
+    },
+    {
+      "epoch": 8.21,
+      "learning_rate": 1.6257620584279454e-06,
+      "loss": 0.3926,
+      "step": 1339
+    },
+    {
+      "epoch": 8.22,
+      "learning_rate": 1.6149180649759622e-06,
+      "loss": 0.3926,
+      "step": 1340
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 1.60410718030361e-06,
+      "loss": 0.436,
+      "step": 1341
+    },
+    {
+      "epoch": 8.23,
+      "learning_rate": 1.5933294470980443e-06,
+      "loss": 0.4141,
+      "step": 1342
+    },
+    {
+      "epoch": 8.24,
+      "learning_rate": 1.5825849079155032e-06,
+      "loss": 0.4165,
+      "step": 1343
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 1.5718736051811634e-06,
+      "loss": 0.4912,
+      "step": 1344
+    },
+    {
+      "epoch": 8.25,
+      "learning_rate": 1.5611955811889645e-06,
+      "loss": 0.397,
+      "step": 1345
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 1.5505508781014489e-06,
+      "loss": 0.4297,
+      "step": 1346
+    },
+    {
+      "epoch": 8.26,
+      "learning_rate": 1.539939537949583e-06,
+      "loss": 0.4883,
+      "step": 1347
+    },
+    {
+      "epoch": 8.27,
+      "learning_rate": 1.5293616026326053e-06,
+      "loss": 0.3496,
+      "step": 1348
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 1.5188171139178486e-06,
+      "loss": 0.4014,
+      "step": 1349
+    },
+    {
+      "epoch": 8.28,
+      "learning_rate": 1.5083061134405874e-06,
+      "loss": 0.3706,
+      "step": 1350
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 1.4978286427038602e-06,
+      "loss": 0.4463,
+      "step": 1351
+    },
+    {
+      "epoch": 8.29,
+      "learning_rate": 1.4873847430783118e-06,
+      "loss": 0.4316,
+      "step": 1352
+    },
+    {
+      "epoch": 8.3,
+      "learning_rate": 1.476974455802036e-06,
+      "loss": 0.4258,
+      "step": 1353
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 1.4665978219804056e-06,
+      "loss": 0.3833,
+      "step": 1354
+    },
+    {
+      "epoch": 8.31,
+      "learning_rate": 1.4562548825859092e-06,
+      "loss": 0.3687,
+      "step": 1355
+    },
+    {
+      "epoch": 8.32,
+      "learning_rate": 1.4459456784579917e-06,
+      "loss": 0.4141,
+      "step": 1356
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 1.435670250302892e-06,
+      "loss": 0.4692,
+      "step": 1357
+    },
+    {
+      "epoch": 8.33,
+      "learning_rate": 1.425428638693489e-06,
+      "loss": 0.3999,
+      "step": 1358
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 1.415220884069135e-06,
+      "loss": 0.4443,
+      "step": 1359
+    },
+    {
+      "epoch": 8.34,
+      "learning_rate": 1.405047026735491e-06,
+      "loss": 0.3403,
+      "step": 1360
+    },
+    {
+      "epoch": 8.35,
+      "learning_rate": 1.394907106864375e-06,
+      "loss": 0.4438,
+      "step": 1361
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 1.3848011644936077e-06,
+      "loss": 0.3643,
+      "step": 1362
+    },
+    {
+      "epoch": 8.36,
+      "learning_rate": 1.3747292395268407e-06,
+      "loss": 0.4121,
+      "step": 1363
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 1.3646913717334142e-06,
+      "loss": 0.394,
+      "step": 1364
+    },
+    {
+      "epoch": 8.37,
+      "learning_rate": 1.3546876007481847e-06,
+      "loss": 0.4102,
+      "step": 1365
+    },
+    {
+      "epoch": 8.38,
+      "learning_rate": 1.344717966071385e-06,
+      "loss": 0.3857,
+      "step": 1366
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 1.3347825070684518e-06,
+      "loss": 0.3726,
+      "step": 1367
+    },
+    {
+      "epoch": 8.39,
+      "learning_rate": 1.3248812629698815e-06,
+      "loss": 0.4077,
+      "step": 1368
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 1.3150142728710669e-06,
+      "loss": 0.4009,
+      "step": 1369
+    },
+    {
+      "epoch": 8.4,
+      "learning_rate": 1.3051815757321607e-06,
+      "loss": 0.3789,
+      "step": 1370
+    },
+    {
+      "epoch": 8.41,
+      "learning_rate": 1.295383210377895e-06,
+      "loss": 0.3452,
+      "step": 1371
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 1.2856192154974488e-06,
+      "loss": 0.4043,
+      "step": 1372
+    },
+    {
+      "epoch": 8.42,
+      "learning_rate": 1.2758896296442834e-06,
+      "loss": 0.4385,
+      "step": 1373
+    },
+    {
+      "epoch": 8.43,
+      "learning_rate": 1.266194491235998e-06,
+      "loss": 0.4263,
+      "step": 1374
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 1.2565338385541792e-06,
+      "loss": 0.416,
+      "step": 1375
+    },
+    {
+      "epoch": 8.44,
+      "learning_rate": 1.2469077097442372e-06,
+      "loss": 0.4087,
+      "step": 1376
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 1.2373161428152647e-06,
+      "loss": 0.4033,
+      "step": 1377
+    },
+    {
+      "epoch": 8.45,
+      "learning_rate": 1.2277591756398933e-06,
+      "loss": 0.3394,
+      "step": 1378
+    },
+    {
+      "epoch": 8.46,
+      "learning_rate": 1.2182368459541294e-06,
+      "loss": 0.4214,
+      "step": 1379
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 1.2087491913572103e-06,
+      "loss": 0.4229,
+      "step": 1380
+    },
+    {
+      "epoch": 8.47,
+      "learning_rate": 1.1992962493114645e-06,
+      "loss": 0.3779,
+      "step": 1381
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 1.1898780571421554e-06,
+      "loss": 0.4639,
+      "step": 1382
+    },
+    {
+      "epoch": 8.48,
+      "learning_rate": 1.1804946520373307e-06,
+      "loss": 0.4116,
+      "step": 1383
+    },
+    {
+      "epoch": 8.49,
+      "learning_rate": 1.171146071047683e-06,
+      "loss": 0.3823,
+      "step": 1384
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 1.161832351086396e-06,
+      "loss": 0.4209,
+      "step": 1385
+    },
+    {
+      "epoch": 8.5,
+      "learning_rate": 1.1525535289290168e-06,
+      "loss": 0.3936,
+      "step": 1386
+    },
+    {
+      "epoch": 8.51,
+      "learning_rate": 1.1433096412132838e-06,
+      "loss": 0.3999,
+      "step": 1387
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 1.1341007244390023e-06,
+      "loss": 0.437,
+      "step": 1388
+    },
+    {
+      "epoch": 8.52,
+      "learning_rate": 1.124926814967887e-06,
+      "loss": 0.3521,
+      "step": 1389
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 1.1157879490234346e-06,
+      "loss": 0.4141,
+      "step": 1390
+    },
+    {
+      "epoch": 8.53,
+      "learning_rate": 1.1066841626907633e-06,
+      "loss": 0.418,
+      "step": 1391
+    },
+    {
+      "epoch": 8.54,
+      "learning_rate": 1.097615491916485e-06,
+      "loss": 0.4189,
+      "step": 1392
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 1.088581972508549e-06,
+      "loss": 0.4517,
+      "step": 1393
+    },
+    {
+      "epoch": 8.55,
+      "learning_rate": 1.0795836401361148e-06,
+      "loss": 0.4067,
+      "step": 1394
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 1.0706205303294025e-06,
+      "loss": 0.375,
+      "step": 1395
+    },
+    {
+      "epoch": 8.56,
+      "learning_rate": 1.0616926784795511e-06,
+      "loss": 0.3359,
+      "step": 1396
+    },
+    {
+      "epoch": 8.57,
+      "learning_rate": 1.0528001198384862e-06,
+      "loss": 0.4092,
+      "step": 1397
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 1.043942889518782e-06,
+      "loss": 0.3726,
+      "step": 1398
+    },
+    {
+      "epoch": 8.58,
+      "learning_rate": 1.035121022493506e-06,
+      "loss": 0.4136,
+      "step": 1399
+    },
+    {
+      "epoch": 8.59,
+      "learning_rate": 1.026334553596101e-06,
+      "loss": 0.3877,
+      "step": 1400
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 1.0175835175202341e-06,
+      "loss": 0.4268,
+      "step": 1401
+    },
+    {
+      "epoch": 8.6,
+      "learning_rate": 1.0088679488196695e-06,
+      "loss": 0.4053,
+      "step": 1402
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 1.0001878819081268e-06,
+      "loss": 0.3955,
+      "step": 1403
+    },
+    {
+      "epoch": 8.61,
+      "learning_rate": 9.91543351059141e-07,
+      "loss": 0.3677,
+      "step": 1404
+    },
+    {
+      "epoch": 8.62,
+      "learning_rate": 9.829343904059342e-07,
+      "loss": 0.3691,
+      "step": 1405
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 9.743610339412801e-07,
+      "loss": 0.4097,
+      "step": 1406
+    },
+    {
+      "epoch": 8.63,
+      "learning_rate": 9.658233155173657e-07,
+      "loss": 0.4043,
+      "step": 1407
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 9.573212688456635e-07,
+      "loss": 0.4346,
+      "step": 1408
+    },
+    {
+      "epoch": 8.64,
+      "learning_rate": 9.488549274967873e-07,
+      "loss": 0.3755,
+      "step": 1409
+    },
+    {
+      "epoch": 8.65,
+      "learning_rate": 9.404243249003786e-07,
+      "loss": 0.373,
+      "step": 1410
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 9.320294943449537e-07,
+      "loss": 0.4517,
+      "step": 1411
+    },
+    {
+      "epoch": 8.66,
+      "learning_rate": 9.236704689777842e-07,
+      "loss": 0.4087,
+      "step": 1412
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 9.153472818047627e-07,
+      "loss": 0.4146,
+      "step": 1413
+    },
+    {
+      "epoch": 8.67,
+      "learning_rate": 9.070599656902801e-07,
+      "loss": 0.3848,
+      "step": 1414
+    },
+    {
+      "epoch": 8.68,
+      "learning_rate": 8.988085533570833e-07,
+      "loss": 0.3652,
+      "step": 1415
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 8.905930773861527e-07,
+      "loss": 0.3765,
+      "step": 1416
+    },
+    {
+      "epoch": 8.69,
+      "learning_rate": 8.824135702165693e-07,
+      "loss": 0.395,
+      "step": 1417
+    },
+    {
+      "epoch": 8.7,
+      "learning_rate": 8.74270064145396e-07,
+      "loss": 0.3818,
+      "step": 1418
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 8.661625913275463e-07,
+      "loss": 0.375,
+      "step": 1419
+    },
+    {
+      "epoch": 8.71,
+      "learning_rate": 8.580911837756467e-07,
+      "loss": 0.3896,
+      "step": 1420
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 8.500558733599206e-07,
+      "loss": 0.3535,
+      "step": 1421
+    },
+    {
+      "epoch": 8.72,
+      "learning_rate": 8.420566918080686e-07,
+      "loss": 0.4189,
+      "step": 1422
+    },
+    {
+      "epoch": 8.73,
+      "learning_rate": 8.340936707051273e-07,
+      "loss": 0.4199,
+      "step": 1423
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 8.261668414933521e-07,
+      "loss": 0.4771,
+      "step": 1424
+    },
+    {
+      "epoch": 8.74,
+      "learning_rate": 8.182762354720985e-07,
+      "loss": 0.3779,
+      "step": 1425
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 8.10421883797694e-07,
+      "loss": 0.3979,
+      "step": 1426
+    },
+    {
+      "epoch": 8.75,
+      "learning_rate": 8.026038174833085e-07,
+      "loss": 0.4072,
+      "step": 1427
+    },
+    {
+      "epoch": 8.76,
+      "learning_rate": 7.948220673988427e-07,
+      "loss": 0.4141,
+      "step": 1428
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 7.87076664270795e-07,
+      "loss": 0.3457,
+      "step": 1429
+    },
+    {
+      "epoch": 8.77,
+      "learning_rate": 7.793676386821602e-07,
+      "loss": 0.395,
+      "step": 1430
+    },
+    {
+      "epoch": 8.78,
+      "learning_rate": 7.716950210722818e-07,
+      "loss": 0.4409,
+      "step": 1431
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 7.6405884173675e-07,
+      "loss": 0.4697,
+      "step": 1432
+    },
+    {
+      "epoch": 8.79,
+      "learning_rate": 7.564591308272773e-07,
+      "loss": 0.3926,
+      "step": 1433
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 7.488959183515809e-07,
+      "loss": 0.3809,
+      "step": 1434
+    },
+    {
+      "epoch": 8.8,
+      "learning_rate": 7.413692341732582e-07,
+      "loss": 0.3564,
+      "step": 1435
+    },
+    {
+      "epoch": 8.81,
+      "learning_rate": 7.338791080116792e-07,
+      "loss": 0.3618,
+      "step": 1436
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 7.264255694418576e-07,
+      "loss": 0.4092,
+      "step": 1437
+    },
+    {
+      "epoch": 8.82,
+      "learning_rate": 7.190086478943459e-07,
+      "loss": 0.4375,
+      "step": 1438
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 7.116283726551077e-07,
+      "loss": 0.3667,
+      "step": 1439
+    },
+    {
+      "epoch": 8.83,
+      "learning_rate": 7.042847728654078e-07,
+      "loss": 0.3511,
+      "step": 1440
+    },
+    {
+      "epoch": 8.84,
+      "learning_rate": 6.969778775217007e-07,
+      "loss": 0.3926,
+      "step": 1441
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 6.897077154755094e-07,
+      "loss": 0.4565,
+      "step": 1442
+    },
+    {
+      "epoch": 8.85,
+      "learning_rate": 6.824743154333157e-07,
+      "loss": 0.3608,
+      "step": 1443
+    },
+    {
+      "epoch": 8.86,
+      "learning_rate": 6.752777059564431e-07,
+      "loss": 0.4204,
+      "step": 1444
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 6.681179154609463e-07,
+      "loss": 0.4058,
+      "step": 1445
+    },
+    {
+      "epoch": 8.87,
+      "learning_rate": 6.609949722175013e-07,
+      "loss": 0.3936,
+      "step": 1446
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 6.539089043512914e-07,
+      "loss": 0.4004,
+      "step": 1447
+    },
+    {
+      "epoch": 8.88,
+      "learning_rate": 6.468597398418952e-07,
+      "loss": 0.3545,
+      "step": 1448
+    },
+    {
+      "epoch": 8.89,
+      "learning_rate": 6.398475065231746e-07,
+      "loss": 0.3264,
+      "step": 1449
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 6.328722320831737e-07,
+      "loss": 0.3521,
+      "step": 1450
+    },
+    {
+      "epoch": 8.9,
+      "learning_rate": 6.259339440639966e-07,
+      "loss": 0.3779,
+      "step": 1451
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 6.1903266986171e-07,
+      "loss": 0.397,
+      "step": 1452
+    },
+    {
+      "epoch": 8.91,
+      "learning_rate": 6.121684367262271e-07,
+      "loss": 0.4111,
+      "step": 1453
+    },
+    {
+      "epoch": 8.92,
+      "learning_rate": 6.053412717612061e-07,
+      "loss": 0.373,
+      "step": 1454
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 5.985512019239392e-07,
+      "loss": 0.4199,
+      "step": 1455
+    },
+    {
+      "epoch": 8.93,
+      "learning_rate": 5.917982540252442e-07,
+      "loss": 0.3833,
+      "step": 1456
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 5.850824547293655e-07,
+      "loss": 0.3838,
+      "step": 1457
+    },
+    {
+      "epoch": 8.94,
+      "learning_rate": 5.784038305538653e-07,
+      "loss": 0.4448,
+      "step": 1458
+    },
+    {
+      "epoch": 8.95,
+      "learning_rate": 5.71762407869515e-07,
+      "loss": 0.4224,
+      "step": 1459
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 5.651582129001987e-07,
+      "loss": 0.3784,
+      "step": 1460
+    },
+    {
+      "epoch": 8.96,
+      "learning_rate": 5.585912717228015e-07,
+      "loss": 0.3955,
+      "step": 1461
+    },
+    {
+      "epoch": 8.97,
+      "learning_rate": 5.520616102671128e-07,
+      "loss": 0.4287,
+      "step": 1462
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 5.455692543157243e-07,
+      "loss": 0.4048,
+      "step": 1463
+    },
+    {
+      "epoch": 8.98,
+      "learning_rate": 5.391142295039209e-07,
+      "loss": 0.4062,
+      "step": 1464
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 5.326965613195867e-07,
+      "loss": 0.4785,
+      "step": 1465
+    },
+    {
+      "epoch": 8.99,
+      "learning_rate": 5.263162751031025e-07,
+      "loss": 0.4512,
+      "step": 1466
+    },
+    {
+      "epoch": 9.0,
+      "learning_rate": 5.199733960472431e-07,
+      "loss": 0.416,
+      "step": 1467
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 5.136679491970809e-07,
+      "loss": 0.3584,
+      "step": 1468
+    },
+    {
+      "epoch": 9.01,
+      "learning_rate": 5.073999594498869e-07,
+      "loss": 0.3274,
+      "step": 1469
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 5.011694515550303e-07,
+      "loss": 0.3901,
+      "step": 1470
+    },
+    {
+      "epoch": 9.02,
+      "learning_rate": 4.949764501138832e-07,
+      "loss": 0.3359,
+      "step": 1471
+    },
+    {
+      "epoch": 9.03,
+      "learning_rate": 4.888209795797205e-07,
+      "loss": 0.3325,
+      "step": 1472
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 4.827030642576236e-07,
+      "loss": 0.3188,
+      "step": 1473
+    },
+    {
+      "epoch": 9.04,
+      "learning_rate": 4.766227283043912e-07,
+      "loss": 0.3936,
+      "step": 1474
+    },
+    {
+      "epoch": 9.05,
+      "learning_rate": 4.7057999572843516e-07,
+      "loss": 0.3057,
+      "step": 1475
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 4.645748903896885e-07,
+      "loss": 0.3564,
+      "step": 1476
+    },
+    {
+      "epoch": 9.06,
+      "learning_rate": 4.5860743599951186e-07,
+      "loss": 0.3252,
+      "step": 1477
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 4.5267765612060253e-07,
+      "loss": 0.355,
+      "step": 1478
+    },
+    {
+      "epoch": 9.07,
+      "learning_rate": 4.4678557416689586e-07,
+      "loss": 0.332,
+      "step": 1479
+    },
+    {
+      "epoch": 9.08,
+      "learning_rate": 4.4093121340347824e-07,
+      "loss": 0.3267,
+      "step": 1480
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 4.3511459694648873e-07,
+      "loss": 0.3574,
+      "step": 1481
+    },
+    {
+      "epoch": 9.09,
+      "learning_rate": 4.2933574776303664e-07,
+      "loss": 0.3354,
+      "step": 1482
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 4.235946886711018e-07,
+      "loss": 0.3193,
+      "step": 1483
+    },
+    {
+      "epoch": 9.1,
+      "learning_rate": 4.1789144233945087e-07,
+      "loss": 0.3301,
+      "step": 1484
+    },
+    {
+      "epoch": 9.11,
+      "learning_rate": 4.122260312875437e-07,
+      "loss": 0.3311,
+      "step": 1485
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 4.0659847788544926e-07,
+      "loss": 0.3257,
+      "step": 1486
+    },
+    {
+      "epoch": 9.12,
+      "learning_rate": 4.010088043537519e-07,
+      "loss": 0.3389,
+      "step": 1487
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 3.954570327634677e-07,
+      "loss": 0.3252,
+      "step": 1488
+    },
+    {
+      "epoch": 9.13,
+      "learning_rate": 3.899431850359503e-07,
+      "loss": 0.3359,
+      "step": 1489
+    },
+    {
+      "epoch": 9.14,
+      "learning_rate": 3.8446728294281865e-07,
+      "loss": 0.3408,
+      "step": 1490
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 3.7902934810585603e-07,
+      "loss": 0.3555,
+      "step": 1491
+    },
+    {
+      "epoch": 9.15,
+      "learning_rate": 3.736294019969311e-07,
+      "loss": 0.3066,
+      "step": 1492
+    },
+    {
+      "epoch": 9.16,
+      "learning_rate": 3.682674659379137e-07,
+      "loss": 0.3354,
+      "step": 1493
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 3.629435611005916e-07,
+      "loss": 0.3721,
+      "step": 1494
+    },
+    {
+      "epoch": 9.17,
+      "learning_rate": 3.5765770850658244e-07,
+      "loss": 0.3271,
+      "step": 1495
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 3.5240992902725204e-07,
+      "loss": 0.2993,
+      "step": 1496
+    },
+    {
+      "epoch": 9.18,
+      "learning_rate": 3.4720024338363633e-07,
+      "loss": 0.3398,
+      "step": 1497
+    },
+    {
+      "epoch": 9.19,
+      "learning_rate": 3.420286721463562e-07,
+      "loss": 0.3213,
+      "step": 1498
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 3.3689523573553597e-07,
+      "loss": 0.3203,
+      "step": 1499
+    },
+    {
+      "epoch": 9.2,
+      "learning_rate": 3.3179995442071956e-07,
+      "loss": 0.3105,
+      "step": 1500
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 3.2674284832080127e-07,
+      "loss": 0.3369,
+      "step": 1501
+    },
+    {
+      "epoch": 9.21,
+      "learning_rate": 3.217239374039338e-07,
+      "loss": 0.3384,
+      "step": 1502
+    },
+    {
+      "epoch": 9.22,
+      "learning_rate": 3.1674324148745827e-07,
+      "loss": 0.2983,
+      "step": 1503
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 3.118007802378198e-07,
+      "loss": 0.374,
+      "step": 1504
+    },
+    {
+      "epoch": 9.23,
+      "learning_rate": 3.0689657317049205e-07,
+      "loss": 0.3257,
+      "step": 1505
+    },
+    {
+      "epoch": 9.24,
+      "learning_rate": 3.020306396499062e-07,
+      "loss": 0.3735,
+      "step": 1506
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 2.972029988893621e-07,
+      "loss": 0.3589,
+      "step": 1507
+    },
+    {
+      "epoch": 9.25,
+      "learning_rate": 2.9241366995096387e-07,
+      "loss": 0.2961,
+      "step": 1508
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 2.8766267174553884e-07,
+      "loss": 0.2913,
+      "step": 1509
+    },
+    {
+      "epoch": 9.26,
+      "learning_rate": 2.8295002303256546e-07,
+      "loss": 0.3169,
+      "step": 1510
+    },
+    {
+      "epoch": 9.27,
+      "learning_rate": 2.7827574242009434e-07,
+      "loss": 0.355,
+      "step": 1511
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 2.736398483646807e-07,
+      "loss": 0.3374,
+      "step": 1512
+    },
+    {
+      "epoch": 9.28,
+      "learning_rate": 2.6904235917131094e-07,
+      "loss": 0.334,
+      "step": 1513
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 2.64483292993325e-07,
+      "loss": 0.3369,
+      "step": 1514
+    },
+    {
+      "epoch": 9.29,
+      "learning_rate": 2.599626678323508e-07,
+      "loss": 0.3076,
+      "step": 1515
+    },
+    {
+      "epoch": 9.3,
+      "learning_rate": 2.554805015382289e-07,
+      "loss": 0.3066,
+      "step": 1516
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 2.5103681180894566e-07,
+      "loss": 0.3735,
+      "step": 1517
+    },
+    {
+      "epoch": 9.31,
+      "learning_rate": 2.4663161619055797e-07,
+      "loss": 0.3203,
+      "step": 1518
+    },
+    {
+      "epoch": 9.32,
+      "learning_rate": 2.422649320771331e-07,
+      "loss": 0.2974,
+      "step": 1519
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 2.3793677671066882e-07,
+      "loss": 0.2905,
+      "step": 1520
+    },
+    {
+      "epoch": 9.33,
+      "learning_rate": 2.3364716718103143e-07,
+      "loss": 0.3438,
+      "step": 1521
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 2.293961204258932e-07,
+      "loss": 0.3091,
+      "step": 1522
+    },
+    {
+      "epoch": 9.34,
+      "learning_rate": 2.2518365323065284e-07,
+      "loss": 0.3037,
+      "step": 1523
+    },
+    {
+      "epoch": 9.35,
+      "learning_rate": 2.2100978222838186e-07,
+      "loss": 0.4043,
+      "step": 1524
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 2.1687452389974829e-07,
+      "loss": 0.3203,
+      "step": 1525
+    },
+    {
+      "epoch": 9.36,
+      "learning_rate": 2.1277789457296306e-07,
+      "loss": 0.4023,
+      "step": 1526
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 2.0871991042370255e-07,
+      "loss": 0.3345,
+      "step": 1527
+    },
+    {
+      "epoch": 9.37,
+      "learning_rate": 2.0470058747505516e-07,
+      "loss": 0.3618,
+      "step": 1528
+    },
+    {
+      "epoch": 9.38,
+      "learning_rate": 2.0071994159745367e-07,
+      "loss": 0.333,
+      "step": 1529
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 1.9677798850861517e-07,
+      "loss": 0.3579,
+      "step": 1530
+    },
+    {
+      "epoch": 9.39,
+      "learning_rate": 1.9287474377347238e-07,
+      "loss": 0.3389,
+      "step": 1531
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 1.8901022280411906e-07,
+      "loss": 0.292,
+      "step": 1532
+    },
+    {
+      "epoch": 9.4,
+      "learning_rate": 1.8518444085974697e-07,
+      "loss": 0.3896,
+      "step": 1533
+    },
+    {
+      "epoch": 9.41,
+      "learning_rate": 1.8139741304658566e-07,
+      "loss": 0.3501,
+      "step": 1534
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 1.776491543178438e-07,
+      "loss": 0.3237,
+      "step": 1535
+    },
+    {
+      "epoch": 9.42,
+      "learning_rate": 1.739396794736481e-07,
+      "loss": 0.334,
+      "step": 1536
+    },
+    {
+      "epoch": 9.43,
+      "learning_rate": 1.7026900316098217e-07,
+      "loss": 0.332,
+      "step": 1537
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 1.6663713987363882e-07,
+      "loss": 0.3452,
+      "step": 1538
+    },
+    {
+      "epoch": 9.44,
+      "learning_rate": 1.6304410395215243e-07,
+      "loss": 0.3301,
+      "step": 1539
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 1.5948990958374543e-07,
+      "loss": 0.3374,
+      "step": 1540
+    },
+    {
+      "epoch": 9.45,
+      "learning_rate": 1.559745708022753e-07,
+      "loss": 0.2935,
+      "step": 1541
+    },
+    {
+      "epoch": 9.46,
+      "learning_rate": 1.5249810148817658e-07,
+      "loss": 0.3643,
+      "step": 1542
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 1.490605153684066e-07,
+      "loss": 0.3765,
+      "step": 1543
+    },
+    {
+      "epoch": 9.47,
+      "learning_rate": 1.4566182601638779e-07,
+      "loss": 0.335,
+      "step": 1544
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 1.4230204685196202e-07,
+      "loss": 0.3569,
+      "step": 1545
+    },
+    {
+      "epoch": 9.48,
+      "learning_rate": 1.3898119114133192e-07,
+      "loss": 0.356,
+      "step": 1546
+    },
+    {
+      "epoch": 9.49,
+      "learning_rate": 1.3569927199700628e-07,
+      "loss": 0.3247,
+      "step": 1547
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 1.3245630237775585e-07,
+      "loss": 0.3125,
+      "step": 1548
+    },
+    {
+      "epoch": 9.5,
+      "learning_rate": 1.292522950885533e-07,
+      "loss": 0.3115,
+      "step": 1549
+    },
+    {
+      "epoch": 9.51,
+      "learning_rate": 1.2608726278053208e-07,
+      "loss": 0.3647,
+      "step": 1550
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 1.2296121795092874e-07,
+      "loss": 0.3447,
+      "step": 1551
+    },
+    {
+      "epoch": 9.52,
+      "learning_rate": 1.1987417294303748e-07,
+      "loss": 0.3105,
+      "step": 1552
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 1.1682613994615788e-07,
+      "loss": 0.3765,
+      "step": 1553
+    },
+    {
+      "epoch": 9.53,
+      "learning_rate": 1.1381713099555381e-07,
+      "loss": 0.3472,
+      "step": 1554
+    },
+    {
+      "epoch": 9.54,
+      "learning_rate": 1.1084715797239798e-07,
+      "loss": 0.2969,
+      "step": 1555
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 1.0791623260372863e-07,
+      "loss": 0.3467,
+      "step": 1556
+    },
+    {
+      "epoch": 9.55,
+      "learning_rate": 1.0502436646240399e-07,
+      "loss": 0.3164,
+      "step": 1557
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 1.0217157096705676e-07,
+      "loss": 0.3633,
+      "step": 1558
+    },
+    {
+      "epoch": 9.56,
+      "learning_rate": 9.935785738204417e-08,
+      "loss": 0.3267,
+      "step": 1559
+    },
+    {
+      "epoch": 9.57,
+      "learning_rate": 9.658323681741133e-08,
+      "loss": 0.3037,
+      "step": 1560
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 9.384772022884015e-08,
+      "loss": 0.3833,
+      "step": 1561
+    },
+    {
+      "epoch": 9.58,
+      "learning_rate": 9.11513184176116e-08,
+      "loss": 0.3452,
+      "step": 1562
+    },
+    {
+      "epoch": 9.59,
+      "learning_rate": 8.8494042030558e-08,
+      "loss": 0.3096,
+      "step": 1563
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 8.587590156002635e-08,
+      "loss": 0.3167,
+      "step": 1564
+    },
+    {
+      "epoch": 9.6,
+      "learning_rate": 8.329690734383278e-08,
+      "loss": 0.3413,
+      "step": 1565
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 8.075706956522156e-08,
+      "loss": 0.3936,
+      "step": 1566
+    },
+    {
+      "epoch": 9.61,
+      "learning_rate": 7.825639825282949e-08,
+      "loss": 0.3364,
+      "step": 1567
+    },
+    {
+      "epoch": 9.62,
+      "learning_rate": 7.579490328064265e-08,
+      "loss": 0.3911,
+      "step": 1568
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 7.33725943679553e-08,
+      "loss": 0.2969,
+      "step": 1569
+    },
+    {
+      "epoch": 9.63,
+      "learning_rate": 7.098948107933656e-08,
+      "loss": 0.3291,
+      "step": 1570
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 6.864557282459162e-08,
+      "loss": 0.3184,
+      "step": 1571
+    },
+    {
+      "epoch": 9.64,
+      "learning_rate": 6.634087885871832e-08,
+      "loss": 0.335,
+      "step": 1572
+    },
+    {
+      "epoch": 9.65,
+      "learning_rate": 6.407540828188175e-08,
+      "loss": 0.3523,
+      "step": 1573
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 6.184917003936752e-08,
+      "loss": 0.2961,
+      "step": 1574
+    },
+    {
+      "epoch": 9.66,
+      "learning_rate": 5.966217292155296e-08,
+      "loss": 0.3701,
+      "step": 1575
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 5.7514425563870436e-08,
+      "loss": 0.3662,
+      "step": 1576
+    },
+    {
+      "epoch": 9.67,
+      "learning_rate": 5.540593644677295e-08,
+      "loss": 0.3115,
+      "step": 1577
+    },
+    {
+      "epoch": 9.68,
+      "learning_rate": 5.333671389569972e-08,
+      "loss": 0.3164,
+      "step": 1578
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 5.1306766081048456e-08,
+      "loss": 0.3003,
+      "step": 1579
+    },
+    {
+      "epoch": 9.69,
+      "learning_rate": 4.931610101813533e-08,
+      "loss": 0.3164,
+      "step": 1580
+    },
+    {
+      "epoch": 9.7,
+      "learning_rate": 4.73647265671684e-08,
+      "loss": 0.3521,
+      "step": 1581
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 4.545265043321645e-08,
+      "loss": 0.2876,
+      "step": 1582
+    },
+    {
+      "epoch": 9.71,
+      "learning_rate": 4.357988016617687e-08,
+      "loss": 0.2947,
+      "step": 1583
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 4.174642316074562e-08,
+      "loss": 0.3423,
+      "step": 1584
+    },
+    {
+      "epoch": 9.72,
+      "learning_rate": 3.9952286656389506e-08,
+      "loss": 0.3438,
+      "step": 1585
+    },
+    {
+      "epoch": 9.73,
+      "learning_rate": 3.819747773731841e-08,
+      "loss": 0.3872,
+      "step": 1586
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 3.648200333245422e-08,
+      "loss": 0.3247,
+      "step": 1587
+    },
+    {
+      "epoch": 9.74,
+      "learning_rate": 3.480587021540527e-08,
+      "loss": 0.3091,
+      "step": 1588
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 3.316908500443972e-08,
+      "loss": 0.3633,
+      "step": 1589
+    },
+    {
+      "epoch": 9.75,
+      "learning_rate": 3.1571654162461107e-08,
+      "loss": 0.3281,
+      "step": 1590
+    },
+    {
+      "epoch": 9.76,
+      "learning_rate": 3.001358399697618e-08,
+      "loss": 0.3545,
+      "step": 1591
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 2.8494880660080437e-08,
+      "loss": 0.3472,
+      "step": 1592
+    },
+    {
+      "epoch": 9.77,
+      "learning_rate": 2.7015550148423718e-08,
+      "loss": 0.3682,
+      "step": 1593
+    },
+    {
+      "epoch": 9.78,
+      "learning_rate": 2.557559830319245e-08,
+      "loss": 0.3105,
+      "step": 1594
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 2.417503081008632e-08,
+      "loss": 0.3003,
+      "step": 1595
+    },
+    {
+      "epoch": 9.79,
+      "learning_rate": 2.2813853199292745e-08,
+      "loss": 0.3608,
+      "step": 1596
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 2.1492070845468005e-08,
+      "loss": 0.2871,
+      "step": 1597
+    },
+    {
+      "epoch": 9.8,
+      "learning_rate": 2.0209688967713914e-08,
+      "loss": 0.3169,
+      "step": 1598
+    },
+    {
+      "epoch": 9.81,
+      "learning_rate": 1.896671262955896e-08,
+      "loss": 0.3218,
+      "step": 1599
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 1.7763146738938307e-08,
+      "loss": 0.332,
+      "step": 1600
+    },
+    {
+      "epoch": 9.82,
+      "learning_rate": 1.659899604816939e-08,
+      "loss": 0.3013,
+      "step": 1601
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 1.5474265153944124e-08,
+      "loss": 0.3262,
+      "step": 1602
+    },
+    {
+      "epoch": 9.83,
+      "learning_rate": 1.4388958497300043e-08,
+      "loss": 0.2925,
+      "step": 1603
+    },
+    {
+      "epoch": 9.84,
+      "learning_rate": 1.3343080363604766e-08,
+      "loss": 0.314,
+      "step": 1604
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 1.2336634882544885e-08,
+      "loss": 0.3696,
+      "step": 1605
+    },
+    {
+      "epoch": 9.85,
+      "learning_rate": 1.1369626028104874e-08,
+      "loss": 0.3647,
+      "step": 1606
+    },
+    {
+      "epoch": 9.86,
+      "learning_rate": 1.0442057618551549e-08,
+      "loss": 0.3306,
+      "step": 1607
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 9.553933316420739e-09,
+      "loss": 0.3916,
+      "step": 1608
+    },
+    {
+      "epoch": 9.87,
+      "learning_rate": 8.705256628499525e-09,
+      "loss": 0.3525,
+      "step": 1609
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 7.896030905818474e-09,
+      "loss": 0.3662,
+      "step": 1610
+    },
+    {
+      "epoch": 9.88,
+      "learning_rate": 7.126259343631648e-09,
+      "loss": 0.3042,
+      "step": 1611
+    },
+    {
+      "epoch": 9.89,
+      "learning_rate": 6.39594498140883e-09,
+      "loss": 0.3257,
+      "step": 1612
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 5.705090702819993e-09,
+      "loss": 0.3237,
+      "step": 1613
+    },
+    {
+      "epoch": 9.9,
+      "learning_rate": 5.053699235726406e-09,
+      "loss": 0.29,
+      "step": 1614
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 4.4417731521717576e-09,
+      "loss": 0.3081,
+      "step": 1615
+    },
+    {
+      "epoch": 9.91,
+      "learning_rate": 3.869314868363283e-09,
+      "loss": 0.2944,
+      "step": 1616
+    },
+    {
+      "epoch": 9.92,
+      "learning_rate": 3.3363266446750918e-09,
+      "loss": 0.2676,
+      "step": 1617
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 2.842810585627076e-09,
+      "loss": 0.3086,
+      "step": 1618
+    },
+    {
+      "epoch": 9.93,
+      "learning_rate": 2.388768639886019e-09,
+      "loss": 0.3047,
+      "step": 1619
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 1.9742026002500526e-09,
+      "loss": 0.3242,
+      "step": 1620
+    },
+    {
+      "epoch": 9.94,
+      "learning_rate": 1.5991141036475478e-09,
+      "loss": 0.3086,
+      "step": 1621
+    },
+    {
+      "epoch": 9.95,
+      "learning_rate": 1.263504631129342e-09,
+      "loss": 0.3174,
+      "step": 1622
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 9.673755078598578e-10,
+      "loss": 0.354,
+      "step": 1623
+    },
+    {
+      "epoch": 9.96,
+      "learning_rate": 7.107279031148828e-10,
+      "loss": 0.3208,
+      "step": 1624
+    },
+    {
+      "epoch": 9.97,
+      "learning_rate": 4.935628302760175e-10,
+      "loss": 0.3721,
+      "step": 1625
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 3.158811468273459e-10,
+      "loss": 0.3354,
+      "step": 1626
+    },
+    {
+      "epoch": 9.98,
+      "learning_rate": 1.776835543509936e-10,
+      "loss": 0.3215,
+      "step": 1627
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 7.897059852490785e-11,
+      "loss": 0.2866,
+      "step": 1628
+    },
+    {
+      "epoch": 9.99,
+      "learning_rate": 1.9742669119526824e-11,
+      "loss": 0.3057,
+      "step": 1629
+    },
+    {
+      "epoch": 10.0,
+      "learning_rate": 0.0,
+      "loss": 0.3101,
+      "step": 1630
+    },
+    {
+      "epoch": 10.0,
+      "step": 1630,
+      "total_flos": 41842376695808.0,
+      "train_loss": 1.8768900678201688,
+      "train_runtime": 1493.3467,
+      "train_samples_per_second": 69.736,
+      "train_steps_per_second": 1.092
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 1630,
+  "num_train_epochs": 10,
+  "save_steps": 10000,
+  "total_flos": 41842376695808.0,
+  "trial_name": null,
+  "trial_params": null
+}