diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,12630 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 4.99776885319054,
+  "eval_steps": 500,
+  "global_step": 2100,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.174603174603175e-07,
+      "loss": 0.6523,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.34920634920635e-07,
+      "loss": 0.8132,
+      "step": 2
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.523809523809525e-07,
+      "loss": 0.7398,
+      "step": 3
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.26984126984127e-06,
+      "loss": 0.6798,
+      "step": 4
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.5873015873015873e-06,
+      "loss": 0.7227,
+      "step": 5
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.904761904761905e-06,
+      "loss": 0.6691,
+      "step": 6
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.222222222222222e-06,
+      "loss": 0.6958,
+      "step": 7
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.53968253968254e-06,
+      "loss": 0.6918,
+      "step": 8
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.8571428571428573e-06,
+      "loss": 0.6525,
+      "step": 9
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.1746031746031746e-06,
+      "loss": 0.6644,
+      "step": 10
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.492063492063492e-06,
+      "loss": 0.6499,
+      "step": 11
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.80952380952381e-06,
+      "loss": 0.6698,
+      "step": 12
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.126984126984127e-06,
+      "loss": 0.6498,
+      "step": 13
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 4.444444444444444e-06,
+      "loss": 0.6223,
+      "step": 14
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.761904761904762e-06,
+      "loss": 0.6785,
+      "step": 15
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.07936507936508e-06,
+      "loss": 0.7827,
+      "step": 16
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.396825396825397e-06,
+      "loss": 0.6577,
+      "step": 17
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 5.7142857142857145e-06,
+      "loss": 0.7202,
+      "step": 18
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.031746031746032e-06,
+      "loss": 0.6581,
+      "step": 19
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.349206349206349e-06,
+      "loss": 0.6078,
+      "step": 20
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 0.6313,
+      "step": 21
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 6.984126984126984e-06,
+      "loss": 0.7503,
+      "step": 22
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 7.301587301587301e-06,
+      "loss": 0.6677,
+      "step": 23
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.61904761904762e-06,
+      "loss": 0.691,
+      "step": 24
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 7.936507936507936e-06,
+      "loss": 0.5952,
+      "step": 25
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.253968253968254e-06,
+      "loss": 0.6705,
+      "step": 26
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 8.571428571428571e-06,
+      "loss": 0.5933,
+      "step": 27
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 8.888888888888888e-06,
+      "loss": 0.6703,
+      "step": 28
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.206349206349207e-06,
+      "loss": 0.5948,
+      "step": 29
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.523809523809525e-06,
+      "loss": 0.6217,
+      "step": 30
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.841269841269842e-06,
+      "loss": 0.7235,
+      "step": 31
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.015873015873016e-05,
+      "loss": 0.729,
+      "step": 32
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0476190476190477e-05,
+      "loss": 0.6614,
+      "step": 33
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.0793650793650794e-05,
+      "loss": 0.6931,
+      "step": 34
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.1111111111111113e-05,
+      "loss": 0.6634,
+      "step": 35
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1428571428571429e-05,
+      "loss": 0.6526,
+      "step": 36
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.1746031746031748e-05,
+      "loss": 0.6564,
+      "step": 37
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2063492063492064e-05,
+      "loss": 0.6396,
+      "step": 38
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.2380952380952383e-05,
+      "loss": 0.6254,
+      "step": 39
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.2698412698412699e-05,
+      "loss": 0.582,
+      "step": 40
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3015873015873018e-05,
+      "loss": 0.6153,
+      "step": 41
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 0.6275,
+      "step": 42
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3650793650793652e-05,
+      "loss": 0.5774,
+      "step": 43
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.3968253968253968e-05,
+      "loss": 0.6301,
+      "step": 44
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4285714285714287e-05,
+      "loss": 0.6618,
+      "step": 45
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4603174603174603e-05,
+      "loss": 0.5367,
+      "step": 46
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.4920634920634922e-05,
+      "loss": 0.6112,
+      "step": 47
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.523809523809524e-05,
+      "loss": 0.5654,
+      "step": 48
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.555555555555556e-05,
+      "loss": 0.5908,
+      "step": 49
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.5873015873015872e-05,
+      "loss": 0.5893,
+      "step": 50
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6190476190476193e-05,
+      "loss": 0.5991,
+      "step": 51
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.6507936507936507e-05,
+      "loss": 0.5822,
+      "step": 52
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.6825396825396828e-05,
+      "loss": 0.5451,
+      "step": 53
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7142857142857142e-05,
+      "loss": 0.5197,
+      "step": 54
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7460317460317463e-05,
+      "loss": 0.526,
+      "step": 55
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.7777777777777777e-05,
+      "loss": 0.6361,
+      "step": 56
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8095238095238097e-05,
+      "loss": 0.5386,
+      "step": 57
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8412698412698415e-05,
+      "loss": 0.5726,
+      "step": 58
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.8730158730158732e-05,
+      "loss": 0.6624,
+      "step": 59
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.904761904761905e-05,
+      "loss": 0.5621,
+      "step": 60
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9365079365079367e-05,
+      "loss": 0.5574,
+      "step": 61
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9682539682539684e-05,
+      "loss": 0.7377,
+      "step": 62
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 2e-05,
+      "loss": 0.6257,
+      "step": 63
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9999988107104428e-05,
+      "loss": 0.676,
+      "step": 64
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9999952428445985e-05,
+      "loss": 0.6078,
+      "step": 65
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999892964109543e-05,
+      "loss": 0.6194,
+      "step": 66
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999809714236547e-05,
+      "loss": 0.5498,
+      "step": 67
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9999702679025005e-05,
+      "loss": 0.5781,
+      "step": 68
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.999957185872951e-05,
+      "loss": 0.5837,
+      "step": 69
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999417253661235e-05,
+      "loss": 0.496,
+      "step": 70
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999238864187916e-05,
+      "loss": 0.4876,
+      "step": 71
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9999036690733865e-05,
+      "loss": 0.6136,
+      "step": 72
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9998810733779966e-05,
+      "loss": 0.5559,
+      "step": 73
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998560993863682e-05,
+      "loss": 0.596,
+      "step": 74
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9998287471579036e-05,
+      "loss": 0.5408,
+      "step": 75
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.999799016757662e-05,
+      "loss": 0.569,
+      "step": 76
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9997669082563597e-05,
+      "loss": 0.5138,
+      "step": 77
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.99973242173037e-05,
+      "loss": 0.5304,
+      "step": 78
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9996955572617202e-05,
+      "loss": 0.5772,
+      "step": 79
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9996563149380965e-05,
+      "loss": 0.5341,
+      "step": 80
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.99961469485284e-05,
+      "loss": 0.5657,
+      "step": 81
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.999570697104947e-05,
+      "loss": 0.505,
+      "step": 82
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9995243217990694e-05,
+      "loss": 0.5829,
+      "step": 83
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9994755690455154e-05,
+      "loss": 0.5099,
+      "step": 84
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.999424438960246e-05,
+      "loss": 0.6575,
+      "step": 85
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.999370931664879e-05,
+      "loss": 0.513,
+      "step": 86
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9993150472866856e-05,
+      "loss": 0.5306,
+      "step": 87
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9992567859585913e-05,
+      "loss": 0.6356,
+      "step": 88
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9991961478191753e-05,
+      "loss": 0.5703,
+      "step": 89
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9991331330126698e-05,
+      "loss": 0.5833,
+      "step": 90
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.999067741688961e-05,
+      "loss": 0.4593,
+      "step": 91
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9989999740035868e-05,
+      "loss": 0.5753,
+      "step": 92
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9989298301177387e-05,
+      "loss": 0.6287,
+      "step": 93
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.998857310198259e-05,
+      "loss": 0.668,
+      "step": 94
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998782414417642e-05,
+      "loss": 0.5812,
+      "step": 95
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9987051429540335e-05,
+      "loss": 0.5141,
+      "step": 96
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9986254959912298e-05,
+      "loss": 0.4938,
+      "step": 97
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.998543473718677e-05,
+      "loss": 0.6415,
+      "step": 98
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9984590763314722e-05,
+      "loss": 0.5731,
+      "step": 99
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9983723040303613e-05,
+      "loss": 0.5559,
+      "step": 100
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9982831570217385e-05,
+      "loss": 0.6427,
+      "step": 101
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9981916355176473e-05,
+      "loss": 0.5678,
+      "step": 102
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.998097739735779e-05,
+      "loss": 0.4719,
+      "step": 103
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9980014698994722e-05,
+      "loss": 0.4595,
+      "step": 104
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.997902826237712e-05,
+      "loss": 0.4887,
+      "step": 105
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.99780180898513e-05,
+      "loss": 0.5473,
+      "step": 106
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.997698418382004e-05,
+      "loss": 0.5778,
+      "step": 107
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9975926546742573e-05,
+      "loss": 0.5957,
+      "step": 108
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.997484518113456e-05,
+      "loss": 0.5472,
+      "step": 109
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9973740089568124e-05,
+      "loss": 0.4816,
+      "step": 110
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9972611274671815e-05,
+      "loss": 0.442,
+      "step": 111
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9971458739130598e-05,
+      "loss": 0.5623,
+      "step": 112
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9970282485685878e-05,
+      "loss": 0.5356,
+      "step": 113
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9969082517135463e-05,
+      "loss": 0.4991,
+      "step": 114
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9967858836333577e-05,
+      "loss": 0.5333,
+      "step": 115
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9966611446190837e-05,
+      "loss": 0.516,
+      "step": 116
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9965340349674262e-05,
+      "loss": 0.4703,
+      "step": 117
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9964045549807257e-05,
+      "loss": 0.5818,
+      "step": 118
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.99627270496696e-05,
+      "loss": 0.5025,
+      "step": 119
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9961384852397456e-05,
+      "loss": 0.605,
+      "step": 120
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.996001896118334e-05,
+      "loss": 0.5918,
+      "step": 121
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9958629379276138e-05,
+      "loss": 0.4741,
+      "step": 122
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.995721610998108e-05,
+      "loss": 0.6345,
+      "step": 123
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9955779156659735e-05,
+      "loss": 0.4638,
+      "step": 124
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.995431852273001e-05,
+      "loss": 0.5137,
+      "step": 125
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.995283421166614e-05,
+      "loss": 0.4976,
+      "step": 126
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.995132622699868e-05,
+      "loss": 0.5273,
+      "step": 127
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9949794572314487e-05,
+      "loss": 0.5501,
+      "step": 128
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.994823925125672e-05,
+      "loss": 0.5044,
+      "step": 129
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9946660267524844e-05,
+      "loss": 0.5513,
+      "step": 130
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9945057624874585e-05,
+      "loss": 0.5665,
+      "step": 131
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9943431327117955e-05,
+      "loss": 0.4962,
+      "step": 132
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9941781378123244e-05,
+      "loss": 0.5165,
+      "step": 133
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9940107781814976e-05,
+      "loss": 0.5116,
+      "step": 134
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9938410542173937e-05,
+      "loss": 0.6118,
+      "step": 135
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.993668966323714e-05,
+      "loss": 0.4785,
+      "step": 136
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9934945149097837e-05,
+      "loss": 0.5269,
+      "step": 137
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9933177003905494e-05,
+      "loss": 0.6104,
+      "step": 138
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.993138523186578e-05,
+      "loss": 0.5308,
+      "step": 139
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9929569837240567e-05,
+      "loss": 0.5316,
+      "step": 140
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9927730824347915e-05,
+      "loss": 0.6087,
+      "step": 141
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9925868197562064e-05,
+      "loss": 0.5385,
+      "step": 142
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9923981961313415e-05,
+      "loss": 0.4575,
+      "step": 143
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9922072120088537e-05,
+      "loss": 0.4845,
+      "step": 144
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9920138678430134e-05,
+      "loss": 0.593,
+      "step": 145
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9918181640937048e-05,
+      "loss": 0.5137,
+      "step": 146
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9916201012264255e-05,
+      "loss": 0.519,
+      "step": 147
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9914196797122827e-05,
+      "loss": 0.4443,
+      "step": 148
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9912169000279952e-05,
+      "loss": 0.5585,
+      "step": 149
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9910117626558908e-05,
+      "loss": 0.6079,
+      "step": 150
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9908042680839047e-05,
+      "loss": 0.5567,
+      "step": 151
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9905944168055798e-05,
+      "loss": 0.5661,
+      "step": 152
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9903822093200625e-05,
+      "loss": 0.5068,
+      "step": 153
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.990167646132107e-05,
+      "loss": 0.5584,
+      "step": 154
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9899507277520673e-05,
+      "loss": 0.5269,
+      "step": 155
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9897314546959015e-05,
+      "loss": 0.5304,
+      "step": 156
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9895098274851678e-05,
+      "loss": 0.5415,
+      "step": 157
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9892858466470242e-05,
+      "loss": 0.4245,
+      "step": 158
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.989059512714227e-05,
+      "loss": 0.529,
+      "step": 159
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9888308262251286e-05,
+      "loss": 0.4809,
+      "step": 160
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9885997877236788e-05,
+      "loss": 0.557,
+      "step": 161
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.988366397759421e-05,
+      "loss": 0.5368,
+      "step": 162
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9881306568874905e-05,
+      "loss": 0.485,
+      "step": 163
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9878925656686167e-05,
+      "loss": 0.5156,
+      "step": 164
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9876521246691183e-05,
+      "loss": 0.5298,
+      "step": 165
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.987409334460903e-05,
+      "loss": 0.4891,
+      "step": 166
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.987164195621466e-05,
+      "loss": 0.5445,
+      "step": 167
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9869167087338908e-05,
+      "loss": 0.4878,
+      "step": 168
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9866668743868437e-05,
+      "loss": 0.5525,
+      "step": 169
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.986414693174575e-05,
+      "loss": 0.5467,
+      "step": 170
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9861601656969185e-05,
+      "loss": 0.5356,
+      "step": 171
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9859032925592873e-05,
+      "loss": 0.5304,
+      "step": 172
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.985644074372675e-05,
+      "loss": 0.5579,
+      "step": 173
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9853825117536522e-05,
+      "loss": 0.529,
+      "step": 174
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9851186053243667e-05,
+      "loss": 0.4364,
+      "step": 175
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9848523557125402e-05,
+      "loss": 0.5245,
+      "step": 176
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9845837635514692e-05,
+      "loss": 0.5085,
+      "step": 177
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9843128294800208e-05,
+      "loss": 0.4717,
+      "step": 178
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9840395541426333e-05,
+      "loss": 0.4568,
+      "step": 179
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9837639381893137e-05,
+      "loss": 0.4859,
+      "step": 180
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9834859822756366e-05,
+      "loss": 0.5218,
+      "step": 181
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.983205687062742e-05,
+      "loss": 0.5183,
+      "step": 182
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9829230532173337e-05,
+      "loss": 0.4808,
+      "step": 183
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9826380814116795e-05,
+      "loss": 0.551,
+      "step": 184
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9823507723236073e-05,
+      "loss": 0.5743,
+      "step": 185
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9820611266365036e-05,
+      "loss": 0.4715,
+      "step": 186
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9817691450393145e-05,
+      "loss": 0.4933,
+      "step": 187
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.981474828226541e-05,
+      "loss": 0.4776,
+      "step": 188
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9811781768982392e-05,
+      "loss": 0.4805,
+      "step": 189
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.980879191760017e-05,
+      "loss": 0.4869,
+      "step": 190
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9805778735230354e-05,
+      "loss": 0.45,
+      "step": 191
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9802742229040027e-05,
+      "loss": 0.4374,
+      "step": 192
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.979968240625176e-05,
+      "loss": 0.5349,
+      "step": 193
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9796599274143586e-05,
+      "loss": 0.5704,
+      "step": 194
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.979349284004898e-05,
+      "loss": 0.4299,
+      "step": 195
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9790363111356838e-05,
+      "loss": 0.6351,
+      "step": 196
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.978721009551147e-05,
+      "loss": 0.4833,
+      "step": 197
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.978403380001257e-05,
+      "loss": 0.4073,
+      "step": 198
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9780834232415214e-05,
+      "loss": 0.449,
+      "step": 199
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.977761140032982e-05,
+      "loss": 0.5037,
+      "step": 200
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9774365311422155e-05,
+      "loss": 0.446,
+      "step": 201
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9771095973413296e-05,
+      "loss": 0.6041,
+      "step": 202
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9767803394079618e-05,
+      "loss": 0.5457,
+      "step": 203
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9764487581252787e-05,
+      "loss": 0.4263,
+      "step": 204
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9761148542819723e-05,
+      "loss": 0.4336,
+      "step": 205
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9757786286722598e-05,
+      "loss": 0.4866,
+      "step": 206
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9754400820958798e-05,
+      "loss": 0.4706,
+      "step": 207
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9750992153580927e-05,
+      "loss": 0.526,
+      "step": 208
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9747560292696763e-05,
+      "loss": 0.4457,
+      "step": 209
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9744105246469264e-05,
+      "loss": 0.5094,
+      "step": 210
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9740627023116524e-05,
+      "loss": 0.4598,
+      "step": 211
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9737125630911786e-05,
+      "loss": 0.5196,
+      "step": 212
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9733601078183377e-05,
+      "loss": 0.399,
+      "step": 213
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9730053373314722e-05,
+      "loss": 0.6013,
+      "step": 214
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.972648252474433e-05,
+      "loss": 0.3791,
+      "step": 215
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9722888540965738e-05,
+      "loss": 0.452,
+      "step": 216
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.971927143052752e-05,
+      "loss": 0.4789,
+      "step": 217
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9715631202033265e-05,
+      "loss": 0.4591,
+      "step": 218
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9711967864141542e-05,
+      "loss": 0.458,
+      "step": 219
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.9708281425565887e-05,
+      "loss": 0.3924,
+      "step": 220
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9704571895074787e-05,
+      "loss": 0.4112,
+      "step": 221
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.970083928149166e-05,
+      "loss": 0.4585,
+      "step": 222
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9697083593694817e-05,
+      "loss": 0.442,
+      "step": 223
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.9693304840617456e-05,
+      "loss": 0.5523,
+      "step": 224
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.968950303124765e-05,
+      "loss": 0.5045,
+      "step": 225
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.968567817462829e-05,
+      "loss": 0.3874,
+      "step": 226
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.968183027985711e-05,
+      "loss": 0.5769,
+      "step": 227
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.967795935608663e-05,
+      "loss": 0.4122,
+      "step": 228
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.9674065412524147e-05,
+      "loss": 0.3931,
+      "step": 229
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.967014845843171e-05,
+      "loss": 0.3551,
+      "step": 230
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9666208503126115e-05,
+      "loss": 0.4508,
+      "step": 231
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9662245555978846e-05,
+      "loss": 0.4093,
+      "step": 232
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.9658259626416093e-05,
+      "loss": 0.5276,
+      "step": 233
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9654250723918706e-05,
+      "loss": 0.4112,
+      "step": 234
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9650218858022173e-05,
+      "loss": 0.4802,
+      "step": 235
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9646164038316603e-05,
+      "loss": 0.5285,
+      "step": 236
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.9642086274446713e-05,
+      "loss": 0.5639,
+      "step": 237
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.963798557611178e-05,
+      "loss": 0.4867,
+      "step": 238
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9633861953065648e-05,
+      "loss": 0.3667,
+      "step": 239
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.962971541511667e-05,
+      "loss": 0.5379,
+      "step": 240
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.9625545972127723e-05,
+      "loss": 0.4941,
+      "step": 241
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9621353634016155e-05,
+      "loss": 0.461,
+      "step": 242
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.961713841075377e-05,
+      "loss": 0.5803,
+      "step": 243
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9612900312366815e-05,
+      "loss": 0.5037,
+      "step": 244
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.9608639348935938e-05,
+      "loss": 0.5103,
+      "step": 245
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9604355530596182e-05,
+      "loss": 0.4924,
+      "step": 246
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.960004886753694e-05,
+      "loss": 0.3351,
+      "step": 247
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9595719370001967e-05,
+      "loss": 0.4222,
+      "step": 248
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9591367048289297e-05,
+      "loss": 0.4314,
+      "step": 249
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.9586991912751282e-05,
+      "loss": 0.4207,
+      "step": 250
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9582593973794527e-05,
+      "loss": 0.4829,
+      "step": 251
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.957817324187987e-05,
+      "loss": 0.4396,
+      "step": 252
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9573729727522383e-05,
+      "loss": 0.4542,
+      "step": 253
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.9569263441291312e-05,
+      "loss": 0.4642,
+      "step": 254
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9564774393810068e-05,
+      "loss": 0.4277,
+      "step": 255
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.956026259575621e-05,
+      "loss": 0.4651,
+      "step": 256
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.955572805786141e-05,
+      "loss": 0.4077,
+      "step": 257
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.9551170790911415e-05,
+      "loss": 0.4347,
+      "step": 258
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9546590805746054e-05,
+      "loss": 0.4321,
+      "step": 259
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.9541988113259182e-05,
+      "loss": 0.4918,
+      "step": 260
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.953736272439866e-05,
+      "loss": 0.4836,
+      "step": 261
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.953271465016635e-05,
+      "loss": 0.3936,
+      "step": 262
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9528043901618066e-05,
+      "loss": 0.3525,
+      "step": 263
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9523350489863545e-05,
+      "loss": 0.581,
+      "step": 264
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9518634426066442e-05,
+      "loss": 0.4813,
+      "step": 265
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.9513895721444286e-05,
+      "loss": 0.3635,
+      "step": 266
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9509134387268463e-05,
+      "loss": 0.4591,
+      "step": 267
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9504350434864187e-05,
+      "loss": 0.4305,
+      "step": 268
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.949954387561046e-05,
+      "loss": 0.4302,
+      "step": 269
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.9494714720940065e-05,
+      "loss": 0.4583,
+      "step": 270
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.948986298233953e-05,
+      "loss": 0.5054,
+      "step": 271
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.94849886713491e-05,
+      "loss": 0.4474,
+      "step": 272
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.9480091799562706e-05,
+      "loss": 0.441,
+      "step": 273
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.947517237862795e-05,
+      "loss": 0.5154,
+      "step": 274
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.947023042024606e-05,
+      "loss": 0.3683,
+      "step": 275
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9465265936171875e-05,
+      "loss": 0.5619,
+      "step": 276
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9460278938213815e-05,
+      "loss": 0.4841,
+      "step": 277
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9455269438233846e-05,
+      "loss": 0.5081,
+      "step": 278
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.9450237448147463e-05,
+      "loss": 0.4346,
+      "step": 279
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9445182979923657e-05,
+      "loss": 0.4058,
+      "step": 280
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.944010604558487e-05,
+      "loss": 0.459,
+      "step": 281
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9435006657206998e-05,
+      "loss": 0.4237,
+      "step": 282
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.9429884826919338e-05,
+      "loss": 0.5279,
+      "step": 283
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9424740566904572e-05,
+      "loss": 0.4191,
+      "step": 284
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.941957388939873e-05,
+      "loss": 0.4296,
+      "step": 285
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.9414384806691154e-05,
+      "loss": 0.3797,
+      "step": 286
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.94091733311245e-05,
+      "loss": 0.4491,
+      "step": 287
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9403939475094664e-05,
+      "loss": 0.425,
+      "step": 288
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9398683251050796e-05,
+      "loss": 0.4891,
+      "step": 289
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9393404671495232e-05,
+      "loss": 0.3898,
+      "step": 290
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9388103748983496e-05,
+      "loss": 0.3425,
+      "step": 291
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.9382780496124255e-05,
+      "loss": 0.3748,
+      "step": 292
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9377434925579283e-05,
+      "loss": 0.5272,
+      "step": 293
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.937206705006344e-05,
+      "loss": 0.376,
+      "step": 294
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.9366676882344643e-05,
+      "loss": 0.4875,
+      "step": 295
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.936126443524384e-05,
+      "loss": 0.4761,
+      "step": 296
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9355829721634952e-05,
+      "loss": 0.4184,
+      "step": 297
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9350372754444887e-05,
+      "loss": 0.4083,
+      "step": 298
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.934489354665347e-05,
+      "loss": 0.3693,
+      "step": 299
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9339392111293425e-05,
+      "loss": 0.4208,
+      "step": 300
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.933386846145036e-05,
+      "loss": 0.3714,
+      "step": 301
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9328322610262704e-05,
+      "loss": 0.3947,
+      "step": 302
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9322754570921713e-05,
+      "loss": 0.3724,
+      "step": 303
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9317164356671395e-05,
+      "loss": 0.3416,
+      "step": 304
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9311551980808533e-05,
+      "loss": 0.3642,
+      "step": 305
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.930591745668259e-05,
+      "loss": 0.4707,
+      "step": 306
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9300260797695744e-05,
+      "loss": 0.4067,
+      "step": 307
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.9294582017302797e-05,
+      "loss": 0.3781,
+      "step": 308
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9288881129011177e-05,
+      "loss": 0.4427,
+      "step": 309
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.92831581463809e-05,
+      "loss": 0.454,
+      "step": 310
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.927741308302453e-05,
+      "loss": 0.3751,
+      "step": 311
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9271645952607164e-05,
+      "loss": 0.4375,
+      "step": 312
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.9265856768846366e-05,
+      "loss": 0.477,
+      "step": 313
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9260045545512174e-05,
+      "loss": 0.3936,
+      "step": 314
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9254212296427043e-05,
+      "loss": 0.3455,
+      "step": 315
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9248357035465814e-05,
+      "loss": 0.3754,
+      "step": 316
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.9242479776555693e-05,
+      "loss": 0.2681,
+      "step": 317
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.92365805336762e-05,
+      "loss": 0.5,
+      "step": 318
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.9230659320859157e-05,
+      "loss": 0.407,
+      "step": 319
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.922471615218863e-05,
+      "loss": 0.3495,
+      "step": 320
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.921875104180092e-05,
+      "loss": 0.3546,
+      "step": 321
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.921276400388451e-05,
+      "loss": 0.3648,
+      "step": 322
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.920675505268005e-05,
+      "loss": 0.458,
+      "step": 323
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9200724202480305e-05,
+      "loss": 0.3965,
+      "step": 324
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.9194671467630127e-05,
+      "loss": 0.4098,
+      "step": 325
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.918859686252642e-05,
+      "loss": 0.393,
+      "step": 326
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9182500401618122e-05,
+      "loss": 0.316,
+      "step": 327
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9176382099406142e-05,
+      "loss": 0.3247,
+      "step": 328
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.9170241970443344e-05,
+      "loss": 0.4197,
+      "step": 329
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9164080029334517e-05,
+      "loss": 0.4314,
+      "step": 330
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.915789629073632e-05,
+      "loss": 0.374,
+      "step": 331
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9151690769357266e-05,
+      "loss": 0.3801,
+      "step": 332
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.914546347995768e-05,
+      "loss": 0.3246,
+      "step": 333
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.9139214437349663e-05,
+      "loss": 0.3872,
+      "step": 334
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9132943656397055e-05,
+      "loss": 0.3532,
+      "step": 335
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9126651152015404e-05,
+      "loss": 0.3991,
+      "step": 336
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.912033693917193e-05,
+      "loss": 0.3002,
+      "step": 337
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9114001032885488e-05,
+      "loss": 0.337,
+      "step": 338
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9107643448226536e-05,
+      "loss": 0.386,
+      "step": 339
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9101264200317093e-05,
+      "loss": 0.6428,
+      "step": 340
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9094863304330692e-05,
+      "loss": 0.4307,
+      "step": 341
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9088440775492386e-05,
+      "loss": 0.4289,
+      "step": 342
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9081996629078655e-05,
+      "loss": 0.3972,
+      "step": 343
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9075530880417422e-05,
+      "loss": 0.3303,
+      "step": 344
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.9069043544887978e-05,
+      "loss": 0.3198,
+      "step": 345
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.906253463792096e-05,
+      "loss": 0.409,
+      "step": 346
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9056004174998324e-05,
+      "loss": 0.384,
+      "step": 347
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.904945217165329e-05,
+      "loss": 0.4378,
+      "step": 348
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9042878643470313e-05,
+      "loss": 0.2538,
+      "step": 349
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.9036283606085057e-05,
+      "loss": 0.5109,
+      "step": 350
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9029667075184333e-05,
+      "loss": 0.4353,
+      "step": 351
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9023029066506086e-05,
+      "loss": 0.2687,
+      "step": 352
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.901636959583935e-05,
+      "loss": 0.3496,
+      "step": 353
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.900968867902419e-05,
+      "loss": 0.2978,
+      "step": 354
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.9002986331951713e-05,
+      "loss": 0.4561,
+      "step": 355
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8996262570563967e-05,
+      "loss": 0.3518,
+      "step": 356
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8989517410853956e-05,
+      "loss": 0.3126,
+      "step": 357
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.8982750868865574e-05,
+      "loss": 0.5896,
+      "step": 358
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.897596296069358e-05,
+      "loss": 0.2792,
+      "step": 359
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.896915370248355e-05,
+      "loss": 0.4081,
+      "step": 360
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.896232311043184e-05,
+      "loss": 0.3621,
+      "step": 361
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8955471200785554e-05,
+      "loss": 0.3421,
+      "step": 362
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.8948597989842504e-05,
+      "loss": 0.3931,
+      "step": 363
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8941703493951163e-05,
+      "loss": 0.3424,
+      "step": 364
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.893478772951064e-05,
+      "loss": 0.3649,
+      "step": 365
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.8927850712970626e-05,
+      "loss": 0.2754,
+      "step": 366
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.892089246083136e-05,
+      "loss": 0.466,
+      "step": 367
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8913912989643597e-05,
+      "loss": 0.4167,
+      "step": 368
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.890691231600856e-05,
+      "loss": 0.3557,
+      "step": 369
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.889989045657791e-05,
+      "loss": 0.3821,
+      "step": 370
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.8892847428053692e-05,
+      "loss": 0.3587,
+      "step": 371
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8885783247188305e-05,
+      "loss": 0.371,
+      "step": 372
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8878697930784463e-05,
+      "loss": 0.4077,
+      "step": 373
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8871591495695156e-05,
+      "loss": 0.5043,
+      "step": 374
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8864463958823598e-05,
+      "loss": 0.3466,
+      "step": 375
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.8857315337123197e-05,
+      "loss": 0.3545,
+      "step": 376
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.885014564759752e-05,
+      "loss": 0.4091,
+      "step": 377
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8842954907300236e-05,
+      "loss": 0.3306,
+      "step": 378
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.8835743133335096e-05,
+      "loss": 0.2366,
+      "step": 379
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.882851034285587e-05,
+      "loss": 0.3058,
+      "step": 380
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8821256553066326e-05,
+      "loss": 0.4342,
+      "step": 381
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8813981781220172e-05,
+      "loss": 0.3734,
+      "step": 382
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.880668604462103e-05,
+      "loss": 0.271,
+      "step": 383
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8799369360622394e-05,
+      "loss": 0.3004,
+      "step": 384
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8792031746627563e-05,
+      "loss": 0.3272,
+      "step": 385
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8784673220089642e-05,
+      "loss": 0.2885,
+      "step": 386
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8777293798511466e-05,
+      "loss": 0.4055,
+      "step": 387
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.876989349944557e-05,
+      "loss": 0.3248,
+      "step": 388
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.876247234049416e-05,
+      "loss": 0.2952,
+      "step": 389
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8755030339309037e-05,
+      "loss": 0.3231,
+      "step": 390
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.8747567513591597e-05,
+      "loss": 0.3077,
+      "step": 391
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.874008388109276e-05,
+      "loss": 0.4491,
+      "step": 392
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8732579459612935e-05,
+      "loss": 0.3087,
+      "step": 393
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8725054267001992e-05,
+      "loss": 0.3941,
+      "step": 394
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8717508321159186e-05,
+      "loss": 0.2441,
+      "step": 395
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8709941640033153e-05,
+      "loss": 0.3098,
+      "step": 396
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8702354241621837e-05,
+      "loss": 0.2794,
+      "step": 397
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.869474614397247e-05,
+      "loss": 0.4059,
+      "step": 398
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.8687117365181514e-05,
+      "loss": 0.2528,
+      "step": 399
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.867946792339462e-05,
+      "loss": 0.3647,
+      "step": 400
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.867179783680659e-05,
+      "loss": 0.4391,
+      "step": 401
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8664107123661337e-05,
+      "loss": 0.2288,
+      "step": 402
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8656395802251826e-05,
+      "loss": 0.3154,
+      "step": 403
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.864866389092005e-05,
+      "loss": 0.2354,
+      "step": 404
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.8640911408056963e-05,
+      "loss": 0.2924,
+      "step": 405
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.863313837210247e-05,
+      "loss": 0.3689,
+      "step": 406
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8625344801545343e-05,
+      "loss": 0.446,
+      "step": 407
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.861753071492321e-05,
+      "loss": 0.3061,
+      "step": 408
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.860969613082249e-05,
+      "loss": 0.2869,
+      "step": 409
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.860184106787837e-05,
+      "loss": 0.3621,
+      "step": 410
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.859396554477473e-05,
+      "loss": 0.3412,
+      "step": 411
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.858606958024413e-05,
+      "loss": 0.2804,
+      "step": 412
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8578153193067746e-05,
+      "loss": 0.3961,
+      "step": 413
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8570216402075326e-05,
+      "loss": 0.209,
+      "step": 414
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.856225922614516e-05,
+      "loss": 0.315,
+      "step": 415
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.855428168420402e-05,
+      "loss": 0.3927,
+      "step": 416
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.854628379522712e-05,
+      "loss": 0.307,
+      "step": 417
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8538265578238076e-05,
+      "loss": 0.3467,
+      "step": 418
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8530227052308843e-05,
+      "loss": 0.3571,
+      "step": 419
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8522168236559693e-05,
+      "loss": 0.2836,
+      "step": 420
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.851408915015916e-05,
+      "loss": 0.3063,
+      "step": 421
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8505989812323995e-05,
+      "loss": 0.2435,
+      "step": 422
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8497870242319103e-05,
+      "loss": 0.316,
+      "step": 423
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.848973045945753e-05,
+      "loss": 0.2644,
+      "step": 424
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.8481570483100394e-05,
+      "loss": 0.261,
+      "step": 425
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.847339033265684e-05,
+      "loss": 0.2738,
+      "step": 426
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8465190027584007e-05,
+      "loss": 0.1964,
+      "step": 427
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8456969587386962e-05,
+      "loss": 0.2179,
+      "step": 428
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8448729031618687e-05,
+      "loss": 0.2587,
+      "step": 429
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.8440468379879982e-05,
+      "loss": 0.2478,
+      "step": 430
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.8432187651819467e-05,
+      "loss": 0.1628,
+      "step": 431
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.842388686713351e-05,
+      "loss": 0.188,
+      "step": 432
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.841556604556618e-05,
+      "loss": 0.1627,
+      "step": 433
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.840722520690921e-05,
+      "loss": 0.2472,
+      "step": 434
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8398864371001945e-05,
+      "loss": 0.2497,
+      "step": 435
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8390483557731297e-05,
+      "loss": 0.2002,
+      "step": 436
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8382082787031694e-05,
+      "loss": 0.2376,
+      "step": 437
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8373662078885028e-05,
+      "loss": 0.1666,
+      "step": 438
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.8365221453320625e-05,
+      "loss": 0.3223,
+      "step": 439
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.835676093041518e-05,
+      "loss": 0.2438,
+      "step": 440
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.8348280530292712e-05,
+      "loss": 0.2836,
+      "step": 441
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.833978027312453e-05,
+      "loss": 0.3688,
+      "step": 442
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.833126017912916e-05,
+      "loss": 0.1153,
+      "step": 443
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8322720268572333e-05,
+      "loss": 0.283,
+      "step": 444
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8314160561766886e-05,
+      "loss": 0.3348,
+      "step": 445
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.8305581079072766e-05,
+      "loss": 0.2184,
+      "step": 446
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.829698184089695e-05,
+      "loss": 0.3867,
+      "step": 447
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8288362867693414e-05,
+      "loss": 0.3718,
+      "step": 448
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.827972417996306e-05,
+      "loss": 0.2912,
+      "step": 449
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.8271065798253688e-05,
+      "loss": 0.2192,
+      "step": 450
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.826238774315995e-05,
+      "loss": 0.1406,
+      "step": 451
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8253690035323287e-05,
+      "loss": 0.2058,
+      "step": 452
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8244972695431874e-05,
+      "loss": 0.1916,
+      "step": 453
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.823623574422061e-05,
+      "loss": 0.3711,
+      "step": 454
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.8227479202471016e-05,
+      "loss": 0.1448,
+      "step": 455
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8218703091011214e-05,
+      "loss": 0.2001,
+      "step": 456
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.820990743071589e-05,
+      "loss": 0.2368,
+      "step": 457
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8201092242506213e-05,
+      "loss": 0.226,
+      "step": 458
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.8192257547349805e-05,
+      "loss": 0.3707,
+      "step": 459
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.818340336626069e-05,
+      "loss": 0.1656,
+      "step": 460
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8174529720299237e-05,
+      "loss": 0.146,
+      "step": 461
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.816563663057211e-05,
+      "loss": 0.2701,
+      "step": 462
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.8156724118232234e-05,
+      "loss": 0.198,
+      "step": 463
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.814779220447872e-05,
+      "loss": 0.1934,
+      "step": 464
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8138840910556836e-05,
+      "loss": 0.2659,
+      "step": 465
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.812987025775794e-05,
+      "loss": 0.1695,
+      "step": 466
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8120880267419435e-05,
+      "loss": 0.2478,
+      "step": 467
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.8111870960924733e-05,
+      "loss": 0.3167,
+      "step": 468
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8102842359703177e-05,
+      "loss": 0.1002,
+      "step": 469
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.809379448523001e-05,
+      "loss": 0.2506,
+      "step": 470
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8084727359026316e-05,
+      "loss": 0.2196,
+      "step": 471
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.8075641002658978e-05,
+      "loss": 0.242,
+      "step": 472
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8066535437740603e-05,
+      "loss": 0.1986,
+      "step": 473
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8057410685929505e-05,
+      "loss": 0.1902,
+      "step": 474
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.804826676892963e-05,
+      "loss": 0.199,
+      "step": 475
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.8039103708490503e-05,
+      "loss": 0.284,
+      "step": 476
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8029921526407184e-05,
+      "loss": 0.249,
+      "step": 477
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8020720244520234e-05,
+      "loss": 0.2072,
+      "step": 478
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8011499884715616e-05,
+      "loss": 0.1783,
+      "step": 479
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.8002260468924696e-05,
+      "loss": 0.195,
+      "step": 480
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.7993002019124145e-05,
+      "loss": 0.3398,
+      "step": 481
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7983724557335923e-05,
+      "loss": 0.2159,
+      "step": 482
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.797442810562721e-05,
+      "loss": 0.306,
+      "step": 483
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.7965112686110346e-05,
+      "loss": 0.2499,
+      "step": 484
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.79557783209428e-05,
+      "loss": 0.3041,
+      "step": 485
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.794642503232709e-05,
+      "loss": 0.1226,
+      "step": 486
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7937052842510764e-05,
+      "loss": 0.231,
+      "step": 487
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7927661773786306e-05,
+      "loss": 0.2679,
+      "step": 488
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.7918251848491118e-05,
+      "loss": 0.2362,
+      "step": 489
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.790882308900746e-05,
+      "loss": 0.1802,
+      "step": 490
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.789937551776237e-05,
+      "loss": 0.1848,
+      "step": 491
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7889909157227652e-05,
+      "loss": 0.1625,
+      "step": 492
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.7880424029919794e-05,
+      "loss": 0.3255,
+      "step": 493
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7870920158399918e-05,
+      "loss": 0.2169,
+      "step": 494
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7861397565273736e-05,
+      "loss": 0.2115,
+      "step": 495
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7851856273191487e-05,
+      "loss": 0.1283,
+      "step": 496
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.7842296304847892e-05,
+      "loss": 0.279,
+      "step": 497
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7832717682982092e-05,
+      "loss": 0.237,
+      "step": 498
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7823120430377593e-05,
+      "loss": 0.2531,
+      "step": 499
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.7813504569862224e-05,
+      "loss": 0.2454,
+      "step": 500
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.780387012430807e-05,
+      "loss": 0.1734,
+      "step": 501
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.779421711663142e-05,
+      "loss": 0.1549,
+      "step": 502
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7784545569792713e-05,
+      "loss": 0.2078,
+      "step": 503
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7774855506796497e-05,
+      "loss": 0.1691,
+      "step": 504
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.7765146950691345e-05,
+      "loss": 0.2838,
+      "step": 505
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.775541992456983e-05,
+      "loss": 0.2426,
+      "step": 506
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7745674451568454e-05,
+      "loss": 0.1459,
+      "step": 507
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7735910554867592e-05,
+      "loss": 0.213,
+      "step": 508
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7726128257691447e-05,
+      "loss": 0.2372,
+      "step": 509
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.7716327583307988e-05,
+      "loss": 0.2032,
+      "step": 510
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7706508555028895e-05,
+      "loss": 0.2153,
+      "step": 511
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7696671196209497e-05,
+      "loss": 0.1324,
+      "step": 512
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.768681553024874e-05,
+      "loss": 0.3756,
+      "step": 513
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.7676941580589097e-05,
+      "loss": 0.161,
+      "step": 514
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.766704937071654e-05,
+      "loss": 0.2129,
+      "step": 515
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.765713892416048e-05,
+      "loss": 0.2059,
+      "step": 516
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7647210264493685e-05,
+      "loss": 0.1708,
+      "step": 517
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.7637263415332272e-05,
+      "loss": 0.1642,
+      "step": 518
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.76272984003356e-05,
+      "loss": 0.2026,
+      "step": 519
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7617315243206245e-05,
+      "loss": 0.2589,
+      "step": 520
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.760731396768994e-05,
+      "loss": 0.3203,
+      "step": 521
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.759729459757551e-05,
+      "loss": 0.1849,
+      "step": 522
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.7587257156694813e-05,
+      "loss": 0.1574,
+      "step": 523
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7577201668922702e-05,
+      "loss": 0.2613,
+      "step": 524
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7567128158176955e-05,
+      "loss": 0.1033,
+      "step": 525
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.7557036648418205e-05,
+      "loss": 0.155,
+      "step": 526
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.754692716364991e-05,
+      "loss": 0.2272,
+      "step": 527
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.753679972791828e-05,
+      "loss": 0.2682,
+      "step": 528
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.7526654365312222e-05,
+      "loss": 0.3655,
+      "step": 529
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.751649109996328e-05,
+      "loss": 0.1427,
+      "step": 530
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.750630995604559e-05,
+      "loss": 0.1654,
+      "step": 531
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.749611095777581e-05,
+      "loss": 0.1772,
+      "step": 532
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7485894129413053e-05,
+      "loss": 0.1462,
+      "step": 533
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.7475659495258864e-05,
+      "loss": 0.1412,
+      "step": 534
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.746540707965713e-05,
+      "loss": 0.2159,
+      "step": 535
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7455136906994027e-05,
+      "loss": 0.0936,
+      "step": 536
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7444849001697976e-05,
+      "loss": 0.3473,
+      "step": 537
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7434543388239572e-05,
+      "loss": 0.1598,
+      "step": 538
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.7424220091131536e-05,
+      "loss": 0.2956,
+      "step": 539
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7413879134928644e-05,
+      "loss": 0.302,
+      "step": 540
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.740352054422768e-05,
+      "loss": 0.1745,
+      "step": 541
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7393144343667367e-05,
+      "loss": 0.1999,
+      "step": 542
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7382750557928323e-05,
+      "loss": 0.2452,
+      "step": 543
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.7372339211732988e-05,
+      "loss": 0.0971,
+      "step": 544
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7361910329845578e-05,
+      "loss": 0.2826,
+      "step": 545
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7351463937072008e-05,
+      "loss": 0.188,
+      "step": 546
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7341000058259846e-05,
+      "loss": 0.2045,
+      "step": 547
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.7330518718298263e-05,
+      "loss": 0.2253,
+      "step": 548
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7320019942117954e-05,
+      "loss": 0.1643,
+      "step": 549
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7309503754691088e-05,
+      "loss": 0.2374,
+      "step": 550
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.729897018103125e-05,
+      "loss": 0.2405,
+      "step": 551
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.7288419246193373e-05,
+      "loss": 0.2194,
+      "step": 552
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7277850975273694e-05,
+      "loss": 0.2181,
+      "step": 553
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7267265393409684e-05,
+      "loss": 0.209,
+      "step": 554
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7256662525779984e-05,
+      "loss": 0.3263,
+      "step": 555
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.7246042397604356e-05,
+      "loss": 0.1577,
+      "step": 556
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7235405034143608e-05,
+      "loss": 0.1709,
+      "step": 557
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7224750460699558e-05,
+      "loss": 0.1724,
+      "step": 558
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7214078702614946e-05,
+      "loss": 0.1573,
+      "step": 559
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.7203389785273402e-05,
+      "loss": 0.2831,
+      "step": 560
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7192683734099355e-05,
+      "loss": 0.1179,
+      "step": 561
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7181960574557993e-05,
+      "loss": 0.1681,
+      "step": 562
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7171220332155202e-05,
+      "loss": 0.1585,
+      "step": 563
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.71604630324375e-05,
+      "loss": 0.2118,
+      "step": 564
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.7149688700991972e-05,
+      "loss": 0.1141,
+      "step": 565
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.713889736344622e-05,
+      "loss": 0.2,
+      "step": 566
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7128089045468294e-05,
+      "loss": 0.185,
+      "step": 567
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.711726377276663e-05,
+      "loss": 0.1666,
+      "step": 568
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.7106421571090003e-05,
+      "loss": 0.2227,
+      "step": 569
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.709556246622744e-05,
+      "loss": 0.1354,
+      "step": 570
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.708468648400819e-05,
+      "loss": 0.2438,
+      "step": 571
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7073793650301624e-05,
+      "loss": 0.1741,
+      "step": 572
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.7062883991017217e-05,
+      "loss": 0.2801,
+      "step": 573
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.705195753210446e-05,
+      "loss": 0.0871,
+      "step": 574
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7041014299552792e-05,
+      "loss": 0.248,
+      "step": 575
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.7030054319391563e-05,
+      "loss": 0.2151,
+      "step": 576
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.701907761768995e-05,
+      "loss": 0.2158,
+      "step": 577
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.7008084220556904e-05,
+      "loss": 0.1386,
+      "step": 578
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6997074154141097e-05,
+      "loss": 0.2256,
+      "step": 579
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.6986047444630838e-05,
+      "loss": 0.0921,
+      "step": 580
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.697500411825403e-05,
+      "loss": 0.2807,
+      "step": 581
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6963944201278097e-05,
+      "loss": 0.1708,
+      "step": 582
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.695286772000993e-05,
+      "loss": 0.3033,
+      "step": 583
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.694177470079581e-05,
+      "loss": 0.3607,
+      "step": 584
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.6930665170021364e-05,
+      "loss": 0.1635,
+      "step": 585
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.691953915411149e-05,
+      "loss": 0.1188,
+      "step": 586
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.69083966795303e-05,
+      "loss": 0.2395,
+      "step": 587
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6897237772781046e-05,
+      "loss": 0.1687,
+      "step": 588
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.688606246040607e-05,
+      "loss": 0.2563,
+      "step": 589
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.6874870768986743e-05,
+      "loss": 0.1256,
+      "step": 590
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6863662725143383e-05,
+      "loss": 0.2804,
+      "step": 591
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6852438355535214e-05,
+      "loss": 0.1268,
+      "step": 592
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.6841197686860283e-05,
+      "loss": 0.1213,
+      "step": 593
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.682994074585541e-05,
+      "loss": 0.3132,
+      "step": 594
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.681866755929612e-05,
+      "loss": 0.2123,
+      "step": 595
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6807378153996584e-05,
+      "loss": 0.2708,
+      "step": 596
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6796072556809536e-05,
+      "loss": 0.1616,
+      "step": 597
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.6784750794626242e-05,
+      "loss": 0.1794,
+      "step": 598
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6773412894376404e-05,
+      "loss": 0.1405,
+      "step": 599
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.676205888302812e-05,
+      "loss": 0.2685,
+      "step": 600
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.6750688787587798e-05,
+      "loss": 0.1762,
+      "step": 601
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.673930263510011e-05,
+      "loss": 0.1584,
+      "step": 602
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6727900452647925e-05,
+      "loss": 0.1439,
+      "step": 603
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6716482267352234e-05,
+      "loss": 0.1494,
+      "step": 604
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6705048106372096e-05,
+      "loss": 0.2946,
+      "step": 605
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.6693597996904566e-05,
+      "loss": 0.15,
+      "step": 606
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.668213196618463e-05,
+      "loss": 0.065,
+      "step": 607
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.667065004148516e-05,
+      "loss": 0.2172,
+      "step": 608
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.665915225011681e-05,
+      "loss": 0.1882,
+      "step": 609
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6647638619427998e-05,
+      "loss": 0.188,
+      "step": 610
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.6636109176804798e-05,
+      "loss": 0.1402,
+      "step": 611
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6624563949670903e-05,
+      "loss": 0.1747,
+      "step": 612
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.661300296548755e-05,
+      "loss": 0.1696,
+      "step": 613
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.660142625175346e-05,
+      "loss": 0.3405,
+      "step": 614
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.6589833836004753e-05,
+      "loss": 0.1921,
+      "step": 615
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6578225745814907e-05,
+      "loss": 0.242,
+      "step": 616
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.656660200879469e-05,
+      "loss": 0.1817,
+      "step": 617
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6554962652592078e-05,
+      "loss": 0.1747,
+      "step": 618
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.6543307704892196e-05,
+      "loss": 0.333,
+      "step": 619
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6531637193417264e-05,
+      "loss": 0.3019,
+      "step": 620
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6519951145926515e-05,
+      "loss": 0.1871,
+      "step": 621
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.650824959021614e-05,
+      "loss": 0.1533,
+      "step": 622
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.6496532554119214e-05,
+      "loss": 0.1917,
+      "step": 623
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6484800065505627e-05,
+      "loss": 0.1844,
+      "step": 624
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6473052152282045e-05,
+      "loss": 0.1749,
+      "step": 625
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6461288842391798e-05,
+      "loss": 0.2345,
+      "step": 626
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6449510163814854e-05,
+      "loss": 0.1345,
+      "step": 627
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.6437716144567732e-05,
+      "loss": 0.2127,
+      "step": 628
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6425906812703435e-05,
+      "loss": 0.3082,
+      "step": 629
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6414082196311402e-05,
+      "loss": 0.2044,
+      "step": 630
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.640224232351741e-05,
+      "loss": 0.1174,
+      "step": 631
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.6390387222483537e-05,
+      "loss": 0.217,
+      "step": 632
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6378516921408077e-05,
+      "loss": 0.1287,
+      "step": 633
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6366631448525486e-05,
+      "loss": 0.1816,
+      "step": 634
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6354730832106292e-05,
+      "loss": 0.1867,
+      "step": 635
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.6342815100457064e-05,
+      "loss": 0.26,
+      "step": 636
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.63308842819203e-05,
+      "loss": 0.2845,
+      "step": 637
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6318938404874405e-05,
+      "loss": 0.1425,
+      "step": 638
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.630697749773359e-05,
+      "loss": 0.1233,
+      "step": 639
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.6295001588947817e-05,
+      "loss": 0.1382,
+      "step": 640
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6283010707002738e-05,
+      "loss": 0.0944,
+      "step": 641
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6271004880419606e-05,
+      "loss": 0.1019,
+      "step": 642
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6258984137755236e-05,
+      "loss": 0.0991,
+      "step": 643
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.6246948507601915e-05,
+      "loss": 0.287,
+      "step": 644
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6234898018587336e-05,
+      "loss": 0.2029,
+      "step": 645
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.6222832699374546e-05,
+      "loss": 0.2005,
+      "step": 646
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.621075257866186e-05,
+      "loss": 0.1554,
+      "step": 647
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.61986576851828e-05,
+      "loss": 0.1416,
+      "step": 648
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.618654804770603e-05,
+      "loss": 0.1498,
+      "step": 649
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.617442369503528e-05,
+      "loss": 0.1838,
+      "step": 650
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6162284656009276e-05,
+      "loss": 0.1611,
+      "step": 651
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.6150130959501686e-05,
+      "loss": 0.2353,
+      "step": 652
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.613796263442104e-05,
+      "loss": 0.1755,
+      "step": 653
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6125779709710668e-05,
+      "loss": 0.1497,
+      "step": 654
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6113582214348608e-05,
+      "loss": 0.1474,
+      "step": 655
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6101370177347573e-05,
+      "loss": 0.2424,
+      "step": 656
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.6089143627754863e-05,
+      "loss": 0.2187,
+      "step": 657
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.607690259465229e-05,
+      "loss": 0.1445,
+      "step": 658
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.606464710715612e-05,
+      "loss": 0.2885,
+      "step": 659
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6052377194417e-05,
+      "loss": 0.1884,
+      "step": 660
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.6040092885619886e-05,
+      "loss": 0.1909,
+      "step": 661
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.602779420998398e-05,
+      "loss": 0.2289,
+      "step": 662
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.6015481196762658e-05,
+      "loss": 0.1435,
+      "step": 663
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.600315387524339e-05,
+      "loss": 0.1335,
+      "step": 664
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.5990812274747695e-05,
+      "loss": 0.3038,
+      "step": 665
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5978456424631033e-05,
+      "loss": 0.1885,
+      "step": 666
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.596608635428278e-05,
+      "loss": 0.1,
+      "step": 667
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5953702093126128e-05,
+      "loss": 0.2145,
+      "step": 668
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5941303670618018e-05,
+      "loss": 0.1545,
+      "step": 669
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.5928891116249076e-05,
+      "loss": 0.1383,
+      "step": 670
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5916464459543554e-05,
+      "loss": 0.1229,
+      "step": 671
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5904023730059227e-05,
+      "loss": 0.1245,
+      "step": 672
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.5891568957387362e-05,
+      "loss": 0.1707,
+      "step": 673
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.587910017115262e-05,
+      "loss": 0.1776,
+      "step": 674
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5866617401012995e-05,
+      "loss": 0.2129,
+      "step": 675
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.585412067665974e-05,
+      "loss": 0.12,
+      "step": 676
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.584161002781731e-05,
+      "loss": 0.1492,
+      "step": 677
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.5829085484243264e-05,
+      "loss": 0.1752,
+      "step": 678
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5816547075728227e-05,
+      "loss": 0.0922,
+      "step": 679
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5803994832095792e-05,
+      "loss": 0.1219,
+      "step": 680
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5791428783202465e-05,
+      "loss": 0.1809,
+      "step": 681
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.5778848958937587e-05,
+      "loss": 0.0931,
+      "step": 682
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5766255389223263e-05,
+      "loss": 0.1359,
+      "step": 683
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.57536481040143e-05,
+      "loss": 0.1478,
+      "step": 684
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5741027133298118e-05,
+      "loss": 0.126,
+      "step": 685
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.57283925070947e-05,
+      "loss": 0.199,
+      "step": 686
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.5715744255456497e-05,
+      "loss": 0.1431,
+      "step": 687
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.570308240846838e-05,
+      "loss": 0.1063,
+      "step": 688
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5690406996247557e-05,
+      "loss": 0.1424,
+      "step": 689
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5677718048943495e-05,
+      "loss": 0.1406,
+      "step": 690
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.5665015596737857e-05,
+      "loss": 0.1532,
+      "step": 691
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.565229966984444e-05,
+      "loss": 0.1779,
+      "step": 692
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5639570298509067e-05,
+      "loss": 0.1187,
+      "step": 693
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5626827513009565e-05,
+      "loss": 0.1457,
+      "step": 694
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.5614071343655657e-05,
+      "loss": 0.1826,
+      "step": 695
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.56013018207889e-05,
+      "loss": 0.1555,
+      "step": 696
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5588518974782613e-05,
+      "loss": 0.1255,
+      "step": 697
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5575722836041808e-05,
+      "loss": 0.0715,
+      "step": 698
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.5562913435003113e-05,
+      "loss": 0.0912,
+      "step": 699
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.55500908021347e-05,
+      "loss": 0.1494,
+      "step": 700
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.553725496793622e-05,
+      "loss": 0.175,
+      "step": 701
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5524405962938717e-05,
+      "loss": 0.2097,
+      "step": 702
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.5511543817704563e-05,
+      "loss": 0.1811,
+      "step": 703
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5498668562827397e-05,
+      "loss": 0.0532,
+      "step": 704
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.548578022893202e-05,
+      "loss": 0.1484,
+      "step": 705
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5472878846674366e-05,
+      "loss": 0.1005,
+      "step": 706
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.545996444674138e-05,
+      "loss": 0.1958,
+      "step": 707
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.5447037059851e-05,
+      "loss": 0.0538,
+      "step": 708
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5434096716752023e-05,
+      "loss": 0.1529,
+      "step": 709
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5421143448224087e-05,
+      "loss": 0.1015,
+      "step": 710
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.5408177285077564e-05,
+      "loss": 0.1087,
+      "step": 711
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.53951982581535e-05,
+      "loss": 0.2389,
+      "step": 712
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5382206398323535e-05,
+      "loss": 0.1022,
+      "step": 713
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.536920173648984e-05,
+      "loss": 0.1802,
+      "step": 714
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.5356184303585028e-05,
+      "loss": 0.1475,
+      "step": 715
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.534315413057209e-05,
+      "loss": 0.1398,
+      "step": 716
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5330111248444332e-05,
+      "loss": 0.1536,
+      "step": 717
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5317055688225277e-05,
+      "loss": 0.0968,
+      "step": 718
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5303987480968607e-05,
+      "loss": 0.2838,
+      "step": 719
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.5290906657758085e-05,
+      "loss": 0.1439,
+      "step": 720
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5277813249707488e-05,
+      "loss": 0.1159,
+      "step": 721
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.526470728796052e-05,
+      "loss": 0.1306,
+      "step": 722
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5251588803690753e-05,
+      "loss": 0.0931,
+      "step": 723
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.5238457828101531e-05,
+      "loss": 0.0967,
+      "step": 724
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5225314392425926e-05,
+      "loss": 0.1626,
+      "step": 725
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5212158527926635e-05,
+      "loss": 0.1343,
+      "step": 726
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5198990265895925e-05,
+      "loss": 0.1107,
+      "step": 727
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5185809637655548e-05,
+      "loss": 0.1219,
+      "step": 728
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.5172616674556673e-05,
+      "loss": 0.1008,
+      "step": 729
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.51594114079798e-05,
+      "loss": 0.2645,
+      "step": 730
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5146193869334711e-05,
+      "loss": 0.2534,
+      "step": 731
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5132964090060358e-05,
+      "loss": 0.1861,
+      "step": 732
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.5119722101624823e-05,
+      "loss": 0.1227,
+      "step": 733
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.510646793552522e-05,
+      "loss": 0.1986,
+      "step": 734
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5093201623287631e-05,
+      "loss": 0.2203,
+      "step": 735
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5079923196467035e-05,
+      "loss": 0.2191,
+      "step": 736
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.5066632686647213e-05,
+      "loss": 0.1116,
+      "step": 737
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5053330125440701e-05,
+      "loss": 0.1799,
+      "step": 738
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5040015544488689e-05,
+      "loss": 0.2609,
+      "step": 739
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5026688975460962e-05,
+      "loss": 0.3747,
+      "step": 740
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.5013350450055821e-05,
+      "loss": 0.2313,
+      "step": 741
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 0.2275,
+      "step": 742
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4986637657048606e-05,
+      "loss": 0.1506,
+      "step": 743
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4973263452985023e-05,
+      "loss": 0.0972,
+      "step": 744
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.4959877419620858e-05,
+      "loss": 0.1736,
+      "step": 745
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4946479588795848e-05,
+      "loss": 0.2365,
+      "step": 746
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4933069992377793e-05,
+      "loss": 0.1381,
+      "step": 747
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4919648662262481e-05,
+      "loss": 0.0563,
+      "step": 748
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4906215630373606e-05,
+      "loss": 0.3345,
+      "step": 749
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.4892770928662699e-05,
+      "loss": 0.092,
+      "step": 750
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4879314589109045e-05,
+      "loss": 0.1623,
+      "step": 751
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.486584664371961e-05,
+      "loss": 0.0802,
+      "step": 752
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4852367124528975e-05,
+      "loss": 0.166,
+      "step": 753
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.4838876063599234e-05,
+      "loss": 0.1218,
+      "step": 754
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.482537349301995e-05,
+      "loss": 0.1762,
+      "step": 755
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4811859444908053e-05,
+      "loss": 0.0821,
+      "step": 756
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.4798333951407771e-05,
+      "loss": 0.0903,
+      "step": 757
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.478479704469057e-05,
+      "loss": 0.2427,
+      "step": 758
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4771248756955042e-05,
+      "loss": 0.1602,
+      "step": 759
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4757689120426868e-05,
+      "loss": 0.0621,
+      "step": 760
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4744118167358719e-05,
+      "loss": 0.1585,
+      "step": 761
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.4730535930030172e-05,
+      "loss": 0.1719,
+      "step": 762
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4716942440747663e-05,
+      "loss": 0.0688,
+      "step": 763
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4703337731844374e-05,
+      "loss": 0.146,
+      "step": 764
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4689721835680183e-05,
+      "loss": 0.1637,
+      "step": 765
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.4676094784641581e-05,
+      "loss": 0.1372,
+      "step": 766
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4662456611141582e-05,
+      "loss": 0.105,
+      "step": 767
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4648807347619662e-05,
+      "loss": 0.0572,
+      "step": 768
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4635147026541674e-05,
+      "loss": 0.2206,
+      "step": 769
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4621475680399771e-05,
+      "loss": 0.2389,
+      "step": 770
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.4607793341712336e-05,
+      "loss": 0.1736,
+      "step": 771
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4594100043023888e-05,
+      "loss": 0.1253,
+      "step": 772
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4580395816905027e-05,
+      "loss": 0.1144,
+      "step": 773
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4566680695952333e-05,
+      "loss": 0.248,
+      "step": 774
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.4552954712788312e-05,
+      "loss": 0.1726,
+      "step": 775
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4539217900061302e-05,
+      "loss": 0.0874,
+      "step": 776
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4525470290445392e-05,
+      "loss": 0.174,
+      "step": 777
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.451171191664036e-05,
+      "loss": 0.1,
+      "step": 778
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.4497942811371592e-05,
+      "loss": 0.0684,
+      "step": 779
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4484163007389991e-05,
+      "loss": 0.1305,
+      "step": 780
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.447037253747191e-05,
+      "loss": 0.1833,
+      "step": 781
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4456571434419076e-05,
+      "loss": 0.1494,
+      "step": 782
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.4442759731058502e-05,
+      "loss": 0.1779,
+      "step": 783
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4428937460242417e-05,
+      "loss": 0.1246,
+      "step": 784
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.441510465484819e-05,
+      "loss": 0.0674,
+      "step": 785
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.4401261347778235e-05,
+      "loss": 0.159,
+      "step": 786
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.438740757195996e-05,
+      "loss": 0.1424,
+      "step": 787
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4373543360345661e-05,
+      "loss": 0.0471,
+      "step": 788
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4359668745912472e-05,
+      "loss": 0.178,
+      "step": 789
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4345783761662249e-05,
+      "loss": 0.2033,
+      "step": 790
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4331888440621533e-05,
+      "loss": 0.1475,
+      "step": 791
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.4317982815841442e-05,
+      "loss": 0.1786,
+      "step": 792
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4304066920397604e-05,
+      "loss": 0.1677,
+      "step": 793
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4290140787390083e-05,
+      "loss": 0.0866,
+      "step": 794
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.427620444994328e-05,
+      "loss": 0.1,
+      "step": 795
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.4262257941205882e-05,
+      "loss": 0.2158,
+      "step": 796
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.424830129435076e-05,
+      "loss": 0.1259,
+      "step": 797
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4234334542574906e-05,
+      "loss": 0.0814,
+      "step": 798
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4220357719099338e-05,
+      "loss": 0.1178,
+      "step": 799
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.4206370857169043e-05,
+      "loss": 0.0388,
+      "step": 800
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4192373990052877e-05,
+      "loss": 0.2655,
+      "step": 801
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.4178367151043492e-05,
+      "loss": 0.0989,
+      "step": 802
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.416435037345727e-05,
+      "loss": 0.1686,
+      "step": 803
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.415032369063422e-05,
+      "loss": 0.1091,
+      "step": 804
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4136287135937915e-05,
+      "loss": 0.1923,
+      "step": 805
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4122240742755417e-05,
+      "loss": 0.1298,
+      "step": 806
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.410818454449718e-05,
+      "loss": 0.2154,
+      "step": 807
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.4094118574596984e-05,
+      "loss": 0.151,
+      "step": 808
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.408004286651185e-05,
+      "loss": 0.0864,
+      "step": 809
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4065957453721963e-05,
+      "loss": 0.124,
+      "step": 810
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4051862369730596e-05,
+      "loss": 0.1508,
+      "step": 811
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4037757648064019e-05,
+      "loss": 0.08,
+      "step": 812
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.4023643322271425e-05,
+      "loss": 0.138,
+      "step": 813
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.4009519425924858e-05,
+      "loss": 0.2105,
+      "step": 814
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3995385992619124e-05,
+      "loss": 0.1697,
+      "step": 815
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3981243055971712e-05,
+      "loss": 0.1154,
+      "step": 816
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.3967090649622711e-05,
+      "loss": 0.1642,
+      "step": 817
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3952928807234745e-05,
+      "loss": 0.2067,
+      "step": 818
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3938757562492873e-05,
+      "loss": 0.1188,
+      "step": 819
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3924576949104525e-05,
+      "loss": 0.1014,
+      "step": 820
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.3910387000799409e-05,
+      "loss": 0.1267,
+      "step": 821
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3896187751329438e-05,
+      "loss": 0.0963,
+      "step": 822
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3881979234468652e-05,
+      "loss": 0.11,
+      "step": 823
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3867761484013135e-05,
+      "loss": 0.2152,
+      "step": 824
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.3853534533780932e-05,
+      "loss": 0.2496,
+      "step": 825
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3839298417611964e-05,
+      "loss": 0.078,
+      "step": 826
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3825053169367964e-05,
+      "loss": 0.3288,
+      "step": 827
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.381079882293238e-05,
+      "loss": 0.0808,
+      "step": 828
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.3796535412210301e-05,
+      "loss": 0.172,
+      "step": 829
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3782262971128386e-05,
+      "loss": 0.0468,
+      "step": 830
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3767981533634755e-05,
+      "loss": 0.0696,
+      "step": 831
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3753691133698944e-05,
+      "loss": 0.1215,
+      "step": 832
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3739391805311795e-05,
+      "loss": 0.1017,
+      "step": 833
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.3725083582485397e-05,
+      "loss": 0.1305,
+      "step": 834
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3710766499252986e-05,
+      "loss": 0.1407,
+      "step": 835
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3696440589668879e-05,
+      "loss": 0.0846,
+      "step": 836
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.368210588780838e-05,
+      "loss": 0.1086,
+      "step": 837
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.3667762427767721e-05,
+      "loss": 0.0935,
+      "step": 838
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3653410243663953e-05,
+      "loss": 0.0823,
+      "step": 839
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3639049369634878e-05,
+      "loss": 0.087,
+      "step": 840
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3624679839838972e-05,
+      "loss": 0.0918,
+      "step": 841
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.3610301688455302e-05,
+      "loss": 0.0436,
+      "step": 842
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3595914949683433e-05,
+      "loss": 0.0447,
+      "step": 843
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3581519657743365e-05,
+      "loss": 0.0598,
+      "step": 844
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3567115846875438e-05,
+      "loss": 0.1437,
+      "step": 845
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.3552703551340259e-05,
+      "loss": 0.0528,
+      "step": 846
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.353828280541861e-05,
+      "loss": 0.0577,
+      "step": 847
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3523853643411372e-05,
+      "loss": 0.0379,
+      "step": 848
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3509416099639456e-05,
+      "loss": 0.0429,
+      "step": 849
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.3494970208443699e-05,
+      "loss": 0.0693,
+      "step": 850
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3480516004184795e-05,
+      "loss": 0.0575,
+      "step": 851
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3466053521243214e-05,
+      "loss": 0.1163,
+      "step": 852
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3451582794019116e-05,
+      "loss": 0.0314,
+      "step": 853
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3437103856932266e-05,
+      "loss": 0.052,
+      "step": 854
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.3422616744421966e-05,
+      "loss": 0.0261,
+      "step": 855
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3408121490946961e-05,
+      "loss": 0.048,
+      "step": 856
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3393618130985355e-05,
+      "loss": 0.1219,
+      "step": 857
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3379106699034536e-05,
+      "loss": 0.0405,
+      "step": 858
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.3364587229611095e-05,
+      "loss": 0.0778,
+      "step": 859
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3350059757250737e-05,
+      "loss": 0.0377,
+      "step": 860
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3335524316508208e-05,
+      "loss": 0.0914,
+      "step": 861
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.33209809419572e-05,
+      "loss": 0.0946,
+      "step": 862
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.3306429668190282e-05,
+      "loss": 0.0708,
+      "step": 863
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3291870529818809e-05,
+      "loss": 0.0198,
+      "step": 864
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3277303561472842e-05,
+      "loss": 0.0536,
+      "step": 865
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3262728797801074e-05,
+      "loss": 0.057,
+      "step": 866
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.3248146273470725e-05,
+      "loss": 0.0351,
+      "step": 867
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3233556023167487e-05,
+      "loss": 0.0493,
+      "step": 868
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3218958081595426e-05,
+      "loss": 0.0832,
+      "step": 869
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3204352483476898e-05,
+      "loss": 0.0321,
+      "step": 870
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.3189739263552474e-05,
+      "loss": 0.0281,
+      "step": 871
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3175118456580858e-05,
+      "loss": 0.0441,
+      "step": 872
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.316049009733879e-05,
+      "loss": 0.0581,
+      "step": 873
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3145854220620981e-05,
+      "loss": 0.0278,
+      "step": 874
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.3131210861240027e-05,
+      "loss": 0.0944,
+      "step": 875
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.311656005402631e-05,
+      "loss": 0.0494,
+      "step": 876
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.310190183382794e-05,
+      "loss": 0.0622,
+      "step": 877
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3087236235510649e-05,
+      "loss": 0.0398,
+      "step": 878
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3072563293957725e-05,
+      "loss": 0.0547,
+      "step": 879
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.3057883044069919e-05,
+      "loss": 0.0701,
+      "step": 880
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3043195520765369e-05,
+      "loss": 0.03,
+      "step": 881
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3028500758979507e-05,
+      "loss": 0.0676,
+      "step": 882
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.3013798793664993e-05,
+      "loss": 0.0344,
+      "step": 883
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.299908965979161e-05,
+      "loss": 0.0786,
+      "step": 884
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2984373392346195e-05,
+      "loss": 0.0631,
+      "step": 885
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2969650026332559e-05,
+      "loss": 0.0561,
+      "step": 886
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.2954919596771388e-05,
+      "loss": 0.136,
+      "step": 887
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.294018213870018e-05,
+      "loss": 0.0342,
+      "step": 888
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2925437687173144e-05,
+      "loss": 0.0299,
+      "step": 889
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2910686277261117e-05,
+      "loss": 0.0782,
+      "step": 890
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2895927944051503e-05,
+      "loss": 0.0165,
+      "step": 891
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.2881162722648162e-05,
+      "loss": 0.0466,
+      "step": 892
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2866390648171341e-05,
+      "loss": 0.0312,
+      "step": 893
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2851611755757587e-05,
+      "loss": 0.0458,
+      "step": 894
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2836826080559667e-05,
+      "loss": 0.0686,
+      "step": 895
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.2822033657746478e-05,
+      "loss": 0.0634,
+      "step": 896
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.280723452250297e-05,
+      "loss": 0.071,
+      "step": 897
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2792428710030053e-05,
+      "loss": 0.0359,
+      "step": 898
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2777616255544527e-05,
+      "loss": 0.0234,
+      "step": 899
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2762797194278984e-05,
+      "loss": 0.0503,
+      "step": 900
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.2747971561481735e-05,
+      "loss": 0.0243,
+      "step": 901
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2733139392416722e-05,
+      "loss": 0.0608,
+      "step": 902
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2718300722363431e-05,
+      "loss": 0.0538,
+      "step": 903
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2703455586616811e-05,
+      "loss": 0.0246,
+      "step": 904
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.2688604020487194e-05,
+      "loss": 0.0728,
+      "step": 905
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2673746059300209e-05,
+      "loss": 0.0276,
+      "step": 906
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2658881738396682e-05,
+      "loss": 0.0587,
+      "step": 907
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2644011093132585e-05,
+      "loss": 0.0697,
+      "step": 908
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.2629134158878919e-05,
+      "loss": 0.0541,
+      "step": 909
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.2614250971021658e-05,
+      "loss": 0.0443,
+      "step": 910
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.259936156496163e-05,
+      "loss": 0.1334,
+      "step": 911
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.258446597611447e-05,
+      "loss": 0.0158,
+      "step": 912
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.256956423991052e-05,
+      "loss": 0.046,
+      "step": 913
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.255465639179473e-05,
+      "loss": 0.0737,
+      "step": 914
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.25397424672266e-05,
+      "loss": 0.0324,
+      "step": 915
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.252482250168008e-05,
+      "loss": 0.0274,
+      "step": 916
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2509896530643488e-05,
+      "loss": 0.0466,
+      "step": 917
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.2494964589619424e-05,
+      "loss": 0.0711,
+      "step": 918
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.24800267141247e-05,
+      "loss": 0.0843,
+      "step": 919
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2465082939690225e-05,
+      "loss": 0.0331,
+      "step": 920
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2450133301860951e-05,
+      "loss": 0.2207,
+      "step": 921
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.2435177836195777e-05,
+      "loss": 0.0461,
+      "step": 922
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2420216578267464e-05,
+      "loss": 0.0404,
+      "step": 923
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2405249563662539e-05,
+      "loss": 0.0844,
+      "step": 924
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2390276827981233e-05,
+      "loss": 0.0475,
+      "step": 925
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.2375298406837388e-05,
+      "loss": 0.0432,
+      "step": 926
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.236031433585836e-05,
+      "loss": 0.0236,
+      "step": 927
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2345324650684946e-05,
+      "loss": 0.0627,
+      "step": 928
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.23303293869713e-05,
+      "loss": 0.0246,
+      "step": 929
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.2315328580384841e-05,
+      "loss": 0.1025,
+      "step": 930
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2300322266606176e-05,
+      "loss": 0.09,
+      "step": 931
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2285310481329011e-05,
+      "loss": 0.0731,
+      "step": 932
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2270293260260066e-05,
+      "loss": 0.1367,
+      "step": 933
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.2255270639118984e-05,
+      "loss": 0.0519,
+      "step": 934
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.224024265363826e-05,
+      "loss": 0.1174,
+      "step": 935
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2225209339563144e-05,
+      "loss": 0.1127,
+      "step": 936
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2210170732651571e-05,
+      "loss": 0.0671,
+      "step": 937
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2195126868674052e-05,
+      "loss": 0.0699,
+      "step": 938
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.2180077783413601e-05,
+      "loss": 0.151,
+      "step": 939
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2165023512665672e-05,
+      "loss": 0.0235,
+      "step": 940
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2149964092238031e-05,
+      "loss": 0.1363,
+      "step": 941
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.21348995579507e-05,
+      "loss": 0.0386,
+      "step": 942
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.2119829945635868e-05,
+      "loss": 0.1278,
+      "step": 943
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2104755291137797e-05,
+      "loss": 0.05,
+      "step": 944
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2089675630312755e-05,
+      "loss": 0.0404,
+      "step": 945
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2074590999028895e-05,
+      "loss": 0.0573,
+      "step": 946
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.2059501433166211e-05,
+      "loss": 0.1191,
+      "step": 947
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2044406968616433e-05,
+      "loss": 0.0249,
+      "step": 948
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2029307641282935e-05,
+      "loss": 0.0764,
+      "step": 949
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.2014203487080666e-05,
+      "loss": 0.0326,
+      "step": 950
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.1999094541936047e-05,
+      "loss": 0.0605,
+      "step": 951
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1983980841786899e-05,
+      "loss": 0.0516,
+      "step": 952
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.196886242258236e-05,
+      "loss": 0.0622,
+      "step": 953
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1953739320282778e-05,
+      "loss": 0.2093,
+      "step": 954
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.1938611570859655e-05,
+      "loss": 0.0827,
+      "step": 955
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1923479210295536e-05,
+      "loss": 0.1065,
+      "step": 956
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1908342274583937e-05,
+      "loss": 0.059,
+      "step": 957
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.189320079972926e-05,
+      "loss": 0.0625,
+      "step": 958
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1878054821746703e-05,
+      "loss": 0.0414,
+      "step": 959
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.1862904376662167e-05,
+      "loss": 0.0246,
+      "step": 960
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1847749500512186e-05,
+      "loss": 0.0796,
+      "step": 961
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1832590229343834e-05,
+      "loss": 0.0363,
+      "step": 962
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1817426599214637e-05,
+      "loss": 0.0143,
+      "step": 963
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.1802258646192486e-05,
+      "loss": 0.1474,
+      "step": 964
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1787086406355558e-05,
+      "loss": 0.0495,
+      "step": 965
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.177190991579223e-05,
+      "loss": 0.0708,
+      "step": 966
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.175672921060098e-05,
+      "loss": 0.0786,
+      "step": 967
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.1741544326890317e-05,
+      "loss": 0.0332,
+      "step": 968
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1726355300778693e-05,
+      "loss": 0.0713,
+      "step": 969
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1711162168394404e-05,
+      "loss": 0.0295,
+      "step": 970
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1695964965875515e-05,
+      "loss": 0.0231,
+      "step": 971
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.1680763729369783e-05,
+      "loss": 0.0523,
+      "step": 972
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1665558495034546e-05,
+      "loss": 0.0646,
+      "step": 973
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1650349299036656e-05,
+      "loss": 0.0317,
+      "step": 974
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.163513617755239e-05,
+      "loss": 0.0348,
+      "step": 975
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.1619919166767362e-05,
+      "loss": 0.0828,
+      "step": 976
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1604698302876438e-05,
+      "loss": 0.0362,
+      "step": 977
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1589473622083642e-05,
+      "loss": 0.0342,
+      "step": 978
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1574245160602085e-05,
+      "loss": 0.0414,
+      "step": 979
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1559012954653865e-05,
+      "loss": 0.0702,
+      "step": 980
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.1543777040469994e-05,
+      "loss": 0.0254,
+      "step": 981
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1528537454290295e-05,
+      "loss": 0.0437,
+      "step": 982
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1513294232363329e-05,
+      "loss": 0.0375,
+      "step": 983
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1498047410946307e-05,
+      "loss": 0.1009,
+      "step": 984
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.1482797026305e-05,
+      "loss": 0.0938,
+      "step": 985
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1467543114713654e-05,
+      "loss": 0.0447,
+      "step": 986
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1452285712454905e-05,
+      "loss": 0.0755,
+      "step": 987
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1437024855819689e-05,
+      "loss": 0.1118,
+      "step": 988
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.1421760581107164e-05,
+      "loss": 0.0332,
+      "step": 989
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1406492924624614e-05,
+      "loss": 0.065,
+      "step": 990
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1391221922687371e-05,
+      "loss": 0.0306,
+      "step": 991
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1375947611618716e-05,
+      "loss": 0.0338,
+      "step": 992
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.1360670027749806e-05,
+      "loss": 0.1265,
+      "step": 993
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1345389207419588e-05,
+      "loss": 0.057,
+      "step": 994
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.13301051869747e-05,
+      "loss": 0.0309,
+      "step": 995
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.131481800276939e-05,
+      "loss": 0.1103,
+      "step": 996
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.1299527691165439e-05,
+      "loss": 0.0206,
+      "step": 997
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1284234288532064e-05,
+      "loss": 0.0176,
+      "step": 998
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.126893783124583e-05,
+      "loss": 0.0584,
+      "step": 999
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1253638355690572e-05,
+      "loss": 0.0495,
+      "step": 1000
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1238335898257305e-05,
+      "loss": 0.0972,
+      "step": 1001
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.1223030495344127e-05,
+      "loss": 0.0203,
+      "step": 1002
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1207722183356163e-05,
+      "loss": 0.0872,
+      "step": 1003
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1192410998705432e-05,
+      "loss": 0.0461,
+      "step": 1004
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1177096977810803e-05,
+      "loss": 0.0468,
+      "step": 1005
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.1161780157097889e-05,
+      "loss": 0.0455,
+      "step": 1006
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1146460572998952e-05,
+      "loss": 0.08,
+      "step": 1007
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1131138261952845e-05,
+      "loss": 0.0285,
+      "step": 1008
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1115813260404889e-05,
+      "loss": 0.0839,
+      "step": 1009
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.1100485604806814e-05,
+      "loss": 0.0417,
+      "step": 1010
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1085155331616664e-05,
+      "loss": 0.0351,
+      "step": 1011
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1069822477298706e-05,
+      "loss": 0.0652,
+      "step": 1012
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1054487078323346e-05,
+      "loss": 0.0776,
+      "step": 1013
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.1039149171167046e-05,
+      "loss": 0.0802,
+      "step": 1014
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.1023808792312226e-05,
+      "loss": 0.0135,
+      "step": 1015
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.1008465978247198e-05,
+      "loss": 0.0597,
+      "step": 1016
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0993120765466056e-05,
+      "loss": 0.0335,
+      "step": 1017
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0977773190468599e-05,
+      "loss": 0.0833,
+      "step": 1018
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0962423289760254e-05,
+      "loss": 0.0315,
+      "step": 1019
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0947071099851972e-05,
+      "loss": 0.062,
+      "step": 1020
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.093171665726015e-05,
+      "loss": 0.0836,
+      "step": 1021
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0916359998506549e-05,
+      "loss": 0.0189,
+      "step": 1022
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.090100116011819e-05,
+      "loss": 0.055,
+      "step": 1023
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0885640178627291e-05,
+      "loss": 0.0498,
+      "step": 1024
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0870277090571158e-05,
+      "loss": 0.0491,
+      "step": 1025
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0854911932492114e-05,
+      "loss": 0.0636,
+      "step": 1026
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.08395447409374e-05,
+      "loss": 0.1529,
+      "step": 1027
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.08241755524591e-05,
+      "loss": 0.0412,
+      "step": 1028
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0808804403614044e-05,
+      "loss": 0.0276,
+      "step": 1029
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0793431330963723e-05,
+      "loss": 0.0164,
+      "step": 1030
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0778056371074213e-05,
+      "loss": 0.0169,
+      "step": 1031
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0762679560516067e-05,
+      "loss": 0.11,
+      "step": 1032
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0747300935864245e-05,
+      "loss": 0.0335,
+      "step": 1033
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.073192053369802e-05,
+      "loss": 0.0217,
+      "step": 1034
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0716538390600908e-05,
+      "loss": 0.0512,
+      "step": 1035
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.070115454316054e-05,
+      "loss": 0.1066,
+      "step": 1036
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0685769027968622e-05,
+      "loss": 0.0419,
+      "step": 1037
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0670381881620814e-05,
+      "loss": 0.0253,
+      "step": 1038
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0654993140716665e-05,
+      "loss": 0.0121,
+      "step": 1039
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0639602841859512e-05,
+      "loss": 0.1919,
+      "step": 1040
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0624211021656392e-05,
+      "loss": 0.0376,
+      "step": 1041
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0608817716717976e-05,
+      "loss": 0.0431,
+      "step": 1042
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0593422963658453e-05,
+      "loss": 0.0665,
+      "step": 1043
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0578026799095464e-05,
+      "loss": 0.0221,
+      "step": 1044
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0562629259650002e-05,
+      "loss": 0.072,
+      "step": 1045
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0547230381946331e-05,
+      "loss": 0.0467,
+      "step": 1046
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0531830202611904e-05,
+      "loss": 0.0641,
+      "step": 1047
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0516428758277265e-05,
+      "loss": 0.0705,
+      "step": 1048
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0501026085575967e-05,
+      "loss": 0.0343,
+      "step": 1049
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0485622221144485e-05,
+      "loss": 0.0499,
+      "step": 1050
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0470217201622129e-05,
+      "loss": 0.0808,
+      "step": 1051
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0454811063650959e-05,
+      "loss": 0.0207,
+      "step": 1052
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.043940384387569e-05,
+      "loss": 0.0328,
+      "step": 1053
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0423995578943615e-05,
+      "loss": 0.0241,
+      "step": 1054
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0408586305504514e-05,
+      "loss": 0.0637,
+      "step": 1055
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.0393176060210556e-05,
+      "loss": 0.0683,
+      "step": 1056
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0377764879716234e-05,
+      "loss": 0.0733,
+      "step": 1057
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.036235280067826e-05,
+      "loss": 0.0871,
+      "step": 1058
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0346939859755481e-05,
+      "loss": 0.0192,
+      "step": 1059
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.0331526093608796e-05,
+      "loss": 0.015,
+      "step": 1060
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.031611153890107e-05,
+      "loss": 0.0595,
+      "step": 1061
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0300696232297041e-05,
+      "loss": 0.0182,
+      "step": 1062
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0285280210463232e-05,
+      "loss": 0.0367,
+      "step": 1063
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0269863510067872e-05,
+      "loss": 0.0713,
+      "step": 1064
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.0254446167780803e-05,
+      "loss": 0.036,
+      "step": 1065
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0239028220273393e-05,
+      "loss": 0.0162,
+      "step": 1066
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0223609704218452e-05,
+      "loss": 0.0205,
+      "step": 1067
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0208190656290138e-05,
+      "loss": 0.0617,
+      "step": 1068
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.0192771113163875e-05,
+      "loss": 0.0591,
+      "step": 1069
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0177351111516271e-05,
+      "loss": 0.0172,
+      "step": 1070
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.0161930688025018e-05,
+      "loss": 0.0309,
+      "step": 1071
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.014650987936881e-05,
+      "loss": 0.0151,
+      "step": 1072
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.013108872222726e-05,
+      "loss": 0.0375,
+      "step": 1073
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0115667253280817e-05,
+      "loss": 0.1066,
+      "step": 1074
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.0100245509210661e-05,
+      "loss": 0.0998,
+      "step": 1075
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.008482352669863e-05,
+      "loss": 0.0282,
+      "step": 1076
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.006940134242713e-05,
+      "loss": 0.0716,
+      "step": 1077
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0053978993079046e-05,
+      "loss": 0.0438,
+      "step": 1078
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0038556515337654e-05,
+      "loss": 0.0544,
+      "step": 1079
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0023133945886545e-05,
+      "loss": 0.0907,
+      "step": 1080
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.0007711321409512e-05,
+      "loss": 0.0614,
+      "step": 1081
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.992288678590491e-06,
+      "loss": 0.0134,
+      "step": 1082
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.976866054113458e-06,
+      "loss": 0.0421,
+      "step": 1083
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.961443484662349e-06,
+      "loss": 0.0535,
+      "step": 1084
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.946021006920959e-06,
+      "loss": 0.1885,
+      "step": 1085
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.930598657572874e-06,
+      "loss": 0.0748,
+      "step": 1086
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.915176473301373e-06,
+      "loss": 0.0365,
+      "step": 1087
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.89975449078934e-06,
+      "loss": 0.0474,
+      "step": 1088
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.884332746719186e-06,
+      "loss": 0.1515,
+      "step": 1089
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.868911277772742e-06,
+      "loss": 0.043,
+      "step": 1090
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.853490120631194e-06,
+      "loss": 0.054,
+      "step": 1091
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.838069311974986e-06,
+      "loss": 0.0453,
+      "step": 1092
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.82264888848373e-06,
+      "loss": 0.0755,
+      "step": 1093
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.807228886836128e-06,
+      "loss": 0.0201,
+      "step": 1094
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.791809343709866e-06,
+      "loss": 0.0345,
+      "step": 1095
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.776390295781552e-06,
+      "loss": 0.0337,
+      "step": 1096
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.760971779726609e-06,
+      "loss": 0.0692,
+      "step": 1097
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.745553832219202e-06,
+      "loss": 0.0808,
+      "step": 1098
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.730136489932133e-06,
+      "loss": 0.0421,
+      "step": 1099
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.714719789536773e-06,
+      "loss": 0.1047,
+      "step": 1100
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.699303767702964e-06,
+      "loss": 0.0279,
+      "step": 1101
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.683888461098932e-06,
+      "loss": 0.076,
+      "step": 1102
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.668473906391209e-06,
+      "loss": 0.0449,
+      "step": 1103
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.653060140244524e-06,
+      "loss": 0.0156,
+      "step": 1104
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.637647199321744e-06,
+      "loss": 0.0745,
+      "step": 1105
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.622235120283769e-06,
+      "loss": 0.095,
+      "step": 1106
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.606823939789449e-06,
+      "loss": 0.1595,
+      "step": 1107
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.591413694495493e-06,
+      "loss": 0.0651,
+      "step": 1108
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.576004421056389e-06,
+      "loss": 0.0417,
+      "step": 1109
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.560596156124314e-06,
+      "loss": 0.0792,
+      "step": 1110
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.545188936349046e-06,
+      "loss": 0.0501,
+      "step": 1111
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.529782798377876e-06,
+      "loss": 0.0848,
+      "step": 1112
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.514377778855521e-06,
+      "loss": 0.0356,
+      "step": 1113
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.498973914424035e-06,
+      "loss": 0.0381,
+      "step": 1114
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.483571241722738e-06,
+      "loss": 0.0884,
+      "step": 1115
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.468169797388098e-06,
+      "loss": 0.0583,
+      "step": 1116
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.45276961805367e-06,
+      "loss": 0.0899,
+      "step": 1117
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.43737074035e-06,
+      "loss": 0.0511,
+      "step": 1118
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.421973200904538e-06,
+      "loss": 0.0466,
+      "step": 1119
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.406577036341548e-06,
+      "loss": 0.0722,
+      "step": 1120
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.391182283282026e-06,
+      "loss": 0.0346,
+      "step": 1121
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.375788978343608e-06,
+      "loss": 0.0294,
+      "step": 1122
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.36039715814049e-06,
+      "loss": 0.047,
+      "step": 1123
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.345006859283338e-06,
+      "loss": 0.032,
+      "step": 1124
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.329618118379187e-06,
+      "loss": 0.0144,
+      "step": 1125
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.31423097203138e-06,
+      "loss": 0.0401,
+      "step": 1126
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.298845456839459e-06,
+      "loss": 0.0706,
+      "step": 1127
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.283461609399094e-06,
+      "loss": 0.0399,
+      "step": 1128
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.268079466301978e-06,
+      "loss": 0.0493,
+      "step": 1129
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.252699064135759e-06,
+      "loss": 0.055,
+      "step": 1130
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.237320439483936e-06,
+      "loss": 0.0956,
+      "step": 1131
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.221943628925788e-06,
+      "loss": 0.045,
+      "step": 1132
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.206568669036278e-06,
+      "loss": 0.0541,
+      "step": 1133
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.19119559638596e-06,
+      "loss": 0.0224,
+      "step": 1134
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.175824447540903e-06,
+      "loss": 0.0384,
+      "step": 1135
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.160455259062603e-06,
+      "loss": 0.0603,
+      "step": 1136
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.14508806750789e-06,
+      "loss": 0.0722,
+      "step": 1137
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.129722909428847e-06,
+      "loss": 0.0104,
+      "step": 1138
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.114359821372714e-06,
+      "loss": 0.0522,
+      "step": 1139
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.098998839881812e-06,
+      "loss": 0.0731,
+      "step": 1140
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.083640001493455e-06,
+      "loss": 0.0399,
+      "step": 1141
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.068283342739854e-06,
+      "loss": 0.1886,
+      "step": 1142
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.052928900148031e-06,
+      "loss": 0.0158,
+      "step": 1143
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.037576710239748e-06,
+      "loss": 0.027,
+      "step": 1144
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.022226809531403e-06,
+      "loss": 0.0665,
+      "step": 1145
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.006879234533948e-06,
+      "loss": 0.0362,
+      "step": 1146
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.991534021752805e-06,
+      "loss": 0.0206,
+      "step": 1147
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.976191207687775e-06,
+      "loss": 0.0444,
+      "step": 1148
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.960850828832958e-06,
+      "loss": 0.0766,
+      "step": 1149
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.945512921676656e-06,
+      "loss": 0.0791,
+      "step": 1150
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.930177522701298e-06,
+      "loss": 0.0128,
+      "step": 1151
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.91484466838334e-06,
+      "loss": 0.0542,
+      "step": 1152
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.899514395193188e-06,
+      "loss": 0.0259,
+      "step": 1153
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.884186739595114e-06,
+      "loss": 0.0468,
+      "step": 1154
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.868861738047158e-06,
+      "loss": 0.0691,
+      "step": 1155
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.85353942700105e-06,
+      "loss": 0.0557,
+      "step": 1156
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.838219842902116e-06,
+      "loss": 0.1175,
+      "step": 1157
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.8229030221892e-06,
+      "loss": 0.0441,
+      "step": 1158
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.807589001294571e-06,
+      "loss": 0.0339,
+      "step": 1159
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.79227781664384e-06,
+      "loss": 0.0532,
+      "step": 1160
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.776969504655875e-06,
+      "loss": 0.0624,
+      "step": 1161
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.7616641017427e-06,
+      "loss": 0.0624,
+      "step": 1162
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.746361644309431e-06,
+      "loss": 0.0494,
+      "step": 1163
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.731062168754174e-06,
+      "loss": 0.0246,
+      "step": 1164
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.715765711467941e-06,
+      "loss": 0.0942,
+      "step": 1165
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.700472308834566e-06,
+      "loss": 0.0336,
+      "step": 1166
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.685181997230616e-06,
+      "loss": 0.0058,
+      "step": 1167
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.669894813025306e-06,
+      "loss": 0.0286,
+      "step": 1168
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.654610792580415e-06,
+      "loss": 0.1517,
+      "step": 1169
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.639329972250199e-06,
+      "loss": 0.063,
+      "step": 1170
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.624052388381289e-06,
+      "loss": 0.0714,
+      "step": 1171
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.608778077312634e-06,
+      "loss": 0.0195,
+      "step": 1172
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.593507075375388e-06,
+      "loss": 0.0598,
+      "step": 1173
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.57823941889284e-06,
+      "loss": 0.1139,
+      "step": 1174
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.562975144180314e-06,
+      "loss": 0.0654,
+      "step": 1175
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.5477142875451e-06,
+      "loss": 0.0218,
+      "step": 1176
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.532456885286349e-06,
+      "loss": 0.0252,
+      "step": 1177
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.517202973695002e-06,
+      "loss": 0.1029,
+      "step": 1178
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.501952589053694e-06,
+      "loss": 0.0061,
+      "step": 1179
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.486705767636673e-06,
+      "loss": 0.0441,
+      "step": 1180
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.471462545709707e-06,
+      "loss": 0.0411,
+      "step": 1181
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.456222959530008e-06,
+      "loss": 0.0313,
+      "step": 1182
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.440987045346135e-06,
+      "loss": 0.0753,
+      "step": 1183
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.425754839397917e-06,
+      "loss": 0.0139,
+      "step": 1184
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.41052637791636e-06,
+      "loss": 0.0509,
+      "step": 1185
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.395301697123566e-06,
+      "loss": 0.0673,
+      "step": 1186
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.380080833232639e-06,
+      "loss": 0.0437,
+      "step": 1187
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.364863822447612e-06,
+      "loss": 0.139,
+      "step": 1188
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.349650700963346e-06,
+      "loss": 0.0338,
+      "step": 1189
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.334441504965456e-06,
+      "loss": 0.0363,
+      "step": 1190
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.319236270630218e-06,
+      "loss": 0.0805,
+      "step": 1191
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.304035034124485e-06,
+      "loss": 0.0102,
+      "step": 1192
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.288837831605598e-06,
+      "loss": 0.0274,
+      "step": 1193
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.273644699221309e-06,
+      "loss": 0.0822,
+      "step": 1194
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.258455673109686e-06,
+      "loss": 0.0305,
+      "step": 1195
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.243270789399024e-06,
+      "loss": 0.022,
+      "step": 1196
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.228090084207773e-06,
+      "loss": 0.0489,
+      "step": 1197
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.212913593644443e-06,
+      "loss": 0.0481,
+      "step": 1198
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.197741353807515e-06,
+      "loss": 0.0944,
+      "step": 1199
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.182573400785367e-06,
+      "loss": 0.0172,
+      "step": 1200
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.167409770656168e-06,
+      "loss": 0.046,
+      "step": 1201
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.152250499487816e-06,
+      "loss": 0.0058,
+      "step": 1202
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.137095623337836e-06,
+      "loss": 0.0148,
+      "step": 1203
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.1219451782533e-06,
+      "loss": 0.0509,
+      "step": 1204
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.106799200270741e-06,
+      "loss": 0.0276,
+      "step": 1205
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.091657725416065e-06,
+      "loss": 0.0393,
+      "step": 1206
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.076520789704468e-06,
+      "loss": 0.0086,
+      "step": 1207
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.061388429140348e-06,
+      "loss": 0.0693,
+      "step": 1208
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.046260679717225e-06,
+      "loss": 0.0727,
+      "step": 1209
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.031137577417645e-06,
+      "loss": 0.02,
+      "step": 1210
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.016019158213103e-06,
+      "loss": 0.0382,
+      "step": 1211
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.000905458063956e-06,
+      "loss": 0.0209,
+      "step": 1212
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.985796512919337e-06,
+      "loss": 0.0771,
+      "step": 1213
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.970692358717067e-06,
+      "loss": 0.0307,
+      "step": 1214
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.95559303138357e-06,
+      "loss": 0.0777,
+      "step": 1215
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 7.94049856683379e-06,
+      "loss": 0.0293,
+      "step": 1216
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.92540900097111e-06,
+      "loss": 0.0494,
+      "step": 1217
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.91032436968725e-06,
+      "loss": 0.1716,
+      "step": 1218
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.895244708862204e-06,
+      "loss": 0.048,
+      "step": 1219
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 7.880170054364136e-06,
+      "loss": 0.0724,
+      "step": 1220
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.865100442049302e-06,
+      "loss": 0.0237,
+      "step": 1221
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.850035907761972e-06,
+      "loss": 0.0206,
+      "step": 1222
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.834976487334331e-06,
+      "loss": 0.0283,
+      "step": 1223
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 7.8199222165864e-06,
+      "loss": 0.0967,
+      "step": 1224
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.804873131325955e-06,
+      "loss": 0.0755,
+      "step": 1225
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.789829267348432e-06,
+      "loss": 0.0467,
+      "step": 1226
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.774790660436857e-06,
+      "loss": 0.0816,
+      "step": 1227
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.759757346361746e-06,
+      "loss": 0.0815,
+      "step": 1228
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 7.744729360881023e-06,
+      "loss": 0.0522,
+      "step": 1229
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.729706739739939e-06,
+      "loss": 0.0494,
+      "step": 1230
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.71468951867099e-06,
+      "loss": 0.061,
+      "step": 1231
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.699677733393827e-06,
+      "loss": 0.0314,
+      "step": 1232
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 7.684671419615164e-06,
+      "loss": 0.077,
+      "step": 1233
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.669670613028705e-06,
+      "loss": 0.0958,
+      "step": 1234
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.654675349315059e-06,
+      "loss": 0.1203,
+      "step": 1235
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.639685664141644e-06,
+      "loss": 0.0266,
+      "step": 1236
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 7.624701593162617e-06,
+      "loss": 0.0159,
+      "step": 1237
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.60972317201877e-06,
+      "loss": 0.0284,
+      "step": 1238
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.594750436337467e-06,
+      "loss": 0.0604,
+      "step": 1239
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.579783421732542e-06,
+      "loss": 0.0377,
+      "step": 1240
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 7.564822163804223e-06,
+      "loss": 0.1295,
+      "step": 1241
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.549866698139048e-06,
+      "loss": 0.0412,
+      "step": 1242
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.534917060309775e-06,
+      "loss": 0.0698,
+      "step": 1243
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.519973285875303e-06,
+      "loss": 0.1288,
+      "step": 1244
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 7.505035410380574e-06,
+      "loss": 0.0436,
+      "step": 1245
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.490103469356513e-06,
+      "loss": 0.0641,
+      "step": 1246
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.475177498319921e-06,
+      "loss": 0.0565,
+      "step": 1247
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.4602575327734004e-06,
+      "loss": 0.0744,
+      "step": 1248
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.445343608205273e-06,
+      "loss": 0.0334,
+      "step": 1249
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 7.430435760089482e-06,
+      "loss": 0.0761,
+      "step": 1250
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.415534023885529e-06,
+      "loss": 0.0209,
+      "step": 1251
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.400638435038371e-06,
+      "loss": 0.0543,
+      "step": 1252
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.385749028978347e-06,
+      "loss": 0.0528,
+      "step": 1253
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 7.37086584112108e-06,
+      "loss": 0.0277,
+      "step": 1254
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.3559889068674165e-06,
+      "loss": 0.0483,
+      "step": 1255
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.3411182616033195e-06,
+      "loss": 0.0244,
+      "step": 1256
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.326253940699795e-06,
+      "loss": 0.017,
+      "step": 1257
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 7.311395979512807e-06,
+      "loss": 0.0228,
+      "step": 1258
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.2965444133831905e-06,
+      "loss": 0.1125,
+      "step": 1259
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.2816992776365714e-06,
+      "loss": 0.0345,
+      "step": 1260
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.26686060758328e-06,
+      "loss": 0.0529,
+      "step": 1261
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.252028438518266e-06,
+      "loss": 0.0154,
+      "step": 1262
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.23720280572102e-06,
+      "loss": 0.0299,
+      "step": 1263
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.222383744455477e-06,
+      "loss": 0.0138,
+      "step": 1264
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.20757128996995e-06,
+      "loss": 0.0399,
+      "step": 1265
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.192765477497032e-06,
+      "loss": 0.02,
+      "step": 1266
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.1779663422535235e-06,
+      "loss": 0.0061,
+      "step": 1267
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.163173919440336e-06,
+      "loss": 0.0235,
+      "step": 1268
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.148388244242414e-06,
+      "loss": 0.0413,
+      "step": 1269
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.133609351828661e-06,
+      "loss": 0.0256,
+      "step": 1270
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.11883727735184e-06,
+      "loss": 0.0381,
+      "step": 1271
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.1040720559485e-06,
+      "loss": 0.0151,
+      "step": 1272
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.089313722738885e-06,
+      "loss": 0.035,
+      "step": 1273
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.0745623128268605e-06,
+      "loss": 0.0336,
+      "step": 1274
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.05981786129982e-06,
+      "loss": 0.0754,
+      "step": 1275
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.045080403228613e-06,
+      "loss": 0.006,
+      "step": 1276
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.030349973667445e-06,
+      "loss": 0.0724,
+      "step": 1277
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.015626607653807e-06,
+      "loss": 0.0292,
+      "step": 1278
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.000910340208393e-06,
+      "loss": 0.0102,
+      "step": 1279
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.98620120633501e-06,
+      "loss": 0.0606,
+      "step": 1280
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.971499241020495e-06,
+      "loss": 0.0512,
+      "step": 1281
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.956804479234635e-06,
+      "loss": 0.0425,
+      "step": 1282
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 6.942116955930083e-06,
+      "loss": 0.0072,
+      "step": 1283
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.927436706042276e-06,
+      "loss": 0.0196,
+      "step": 1284
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.912763764489352e-06,
+      "loss": 0.02,
+      "step": 1285
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.898098166172063e-06,
+      "loss": 0.0809,
+      "step": 1286
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 6.883439945973692e-06,
+      "loss": 0.0084,
+      "step": 1287
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.868789138759977e-06,
+      "loss": 0.0392,
+      "step": 1288
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.8541457793790204e-06,
+      "loss": 0.0132,
+      "step": 1289
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.839509902661215e-06,
+      "loss": 0.0459,
+      "step": 1290
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.8248815434191485e-06,
+      "loss": 0.0124,
+      "step": 1291
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 6.8102607364475296e-06,
+      "loss": 0.0126,
+      "step": 1292
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.795647516523107e-06,
+      "loss": 0.0336,
+      "step": 1293
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.781041918404578e-06,
+      "loss": 0.0085,
+      "step": 1294
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.766443976832518e-06,
+      "loss": 0.0385,
+      "step": 1295
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 6.75185372652928e-06,
+      "loss": 0.0077,
+      "step": 1296
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.737271202198932e-06,
+      "loss": 0.0049,
+      "step": 1297
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.722696438527161e-06,
+      "loss": 0.0222,
+      "step": 1298
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.708129470181197e-06,
+      "loss": 0.0084,
+      "step": 1299
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 6.6935703318097235e-06,
+      "loss": 0.0152,
+      "step": 1300
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.679019058042805e-06,
+      "loss": 0.0784,
+      "step": 1301
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.664475683491797e-06,
+      "loss": 0.0058,
+      "step": 1302
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.6499402427492664e-06,
+      "loss": 0.0582,
+      "step": 1303
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 6.635412770388911e-06,
+      "loss": 0.0197,
+      "step": 1304
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.620893300965465e-06,
+      "loss": 0.0155,
+      "step": 1305
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.606381869014646e-06,
+      "loss": 0.0161,
+      "step": 1306
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.591878509053041e-06,
+      "loss": 0.0145,
+      "step": 1307
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 6.577383255578034e-06,
+      "loss": 0.0309,
+      "step": 1308
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.562896143067734e-06,
+      "loss": 0.0062,
+      "step": 1309
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.548417205980886e-06,
+      "loss": 0.0187,
+      "step": 1310
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.533946478756789e-06,
+      "loss": 0.015,
+      "step": 1311
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.519483995815205e-06,
+      "loss": 0.0165,
+      "step": 1312
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 6.505029791556302e-06,
+      "loss": 0.0127,
+      "step": 1313
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.490583900360543e-06,
+      "loss": 0.0312,
+      "step": 1314
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.476146356588627e-06,
+      "loss": 0.0141,
+      "step": 1315
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.461717194581394e-06,
+      "loss": 0.0346,
+      "step": 1316
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 6.447296448659744e-06,
+      "loss": 0.0188,
+      "step": 1317
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.4328841531245616e-06,
+      "loss": 0.0669,
+      "step": 1318
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.418480342256635e-06,
+      "loss": 0.0191,
+      "step": 1319
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.404085050316568e-06,
+      "loss": 0.0032,
+      "step": 1320
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 6.3896983115447035e-06,
+      "loss": 0.0226,
+      "step": 1321
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.375320160161031e-06,
+      "loss": 0.0229,
+      "step": 1322
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.360950630365126e-06,
+      "loss": 0.0208,
+      "step": 1323
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.34658975633605e-06,
+      "loss": 0.0119,
+      "step": 1324
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 6.332237572232281e-06,
+      "loss": 0.0141,
+      "step": 1325
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.317894112191621e-06,
+      "loss": 0.0372,
+      "step": 1326
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.3035594103311255e-06,
+      "loss": 0.0105,
+      "step": 1327
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.289233500747016e-06,
+      "loss": 0.0922,
+      "step": 1328
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 6.274916417514605e-06,
+      "loss": 0.0504,
+      "step": 1329
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.260608194688207e-06,
+      "loss": 0.0307,
+      "step": 1330
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.24630886630106e-06,
+      "loss": 0.0305,
+      "step": 1331
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.232018466365247e-06,
+      "loss": 0.0254,
+      "step": 1332
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.217737028871618e-06,
+      "loss": 0.114,
+      "step": 1333
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 6.2034645877897e-06,
+      "loss": 0.0867,
+      "step": 1334
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.189201177067623e-06,
+      "loss": 0.0278,
+      "step": 1335
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.1749468306320405e-06,
+      "loss": 0.0064,
+      "step": 1336
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.160701582388039e-06,
+      "loss": 0.0065,
+      "step": 1337
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 6.146465466219072e-06,
+      "loss": 0.0691,
+      "step": 1338
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.132238515986868e-06,
+      "loss": 0.2176,
+      "step": 1339
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.118020765531349e-06,
+      "loss": 0.0216,
+      "step": 1340
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.103812248670567e-06,
+      "loss": 0.0094,
+      "step": 1341
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 6.0896129992005955e-06,
+      "loss": 0.0143,
+      "step": 1342
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.075423050895478e-06,
+      "loss": 0.0113,
+      "step": 1343
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.061242437507131e-06,
+      "loss": 0.0109,
+      "step": 1344
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.0470711927652585e-06,
+      "loss": 0.0129,
+      "step": 1345
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 6.032909350377292e-06,
+      "loss": 0.0074,
+      "step": 1346
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.018756944028291e-06,
+      "loss": 0.0144,
+      "step": 1347
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 6.004614007380879e-06,
+      "loss": 0.0222,
+      "step": 1348
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.990480574075143e-06,
+      "loss": 0.0159,
+      "step": 1349
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 5.976356677728578e-06,
+      "loss": 0.0233,
+      "step": 1350
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.962242351935985e-06,
+      "loss": 0.0076,
+      "step": 1351
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.948137630269406e-06,
+      "loss": 0.0141,
+      "step": 1352
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.9340425462780405e-06,
+      "loss": 0.0104,
+      "step": 1353
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.919957133488155e-06,
+      "loss": 0.0158,
+      "step": 1354
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 5.905881425403022e-06,
+      "loss": 0.0153,
+      "step": 1355
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.891815455502824e-06,
+      "loss": 0.0323,
+      "step": 1356
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.877759257244588e-06,
+      "loss": 0.0315,
+      "step": 1357
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.863712864062089e-06,
+      "loss": 0.0066,
+      "step": 1358
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 5.849676309365786e-06,
+      "loss": 0.0189,
+      "step": 1359
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.8356496265427346e-06,
+      "loss": 0.0138,
+      "step": 1360
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.82163284895651e-06,
+      "loss": 0.0212,
+      "step": 1361
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.807626009947127e-06,
+      "loss": 0.0253,
+      "step": 1362
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 5.793629142830959e-06,
+      "loss": 0.042,
+      "step": 1363
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.779642280900668e-06,
+      "loss": 0.0054,
+      "step": 1364
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.765665457425102e-06,
+      "loss": 0.0407,
+      "step": 1365
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.751698705649247e-06,
+      "loss": 0.0272,
+      "step": 1366
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 5.737742058794125e-06,
+      "loss": 0.0228,
+      "step": 1367
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.723795550056721e-06,
+      "loss": 0.0148,
+      "step": 1368
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.709859212609919e-06,
+      "loss": 0.0503,
+      "step": 1369
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.695933079602394e-06,
+      "loss": 0.0463,
+      "step": 1370
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 5.682017184158562e-06,
+      "loss": 0.0863,
+      "step": 1371
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.6681115593784705e-06,
+      "loss": 0.0129,
+      "step": 1372
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.654216238337753e-06,
+      "loss": 0.011,
+      "step": 1373
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.6403312540875325e-06,
+      "loss": 0.0216,
+      "step": 1374
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.626456639654338e-06,
+      "loss": 0.0143,
+      "step": 1375
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 5.612592428040042e-06,
+      "loss": 0.0216,
+      "step": 1376
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.5987386522217645e-06,
+      "loss": 0.0108,
+      "step": 1377
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.584895345151812e-06,
+      "loss": 0.0103,
+      "step": 1378
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.571062539757582e-06,
+      "loss": 0.0158,
+      "step": 1379
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 5.557240268941497e-06,
+      "loss": 0.0357,
+      "step": 1380
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.543428565580926e-06,
+      "loss": 0.0157,
+      "step": 1381
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.5296274625280914e-06,
+      "loss": 0.0329,
+      "step": 1382
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.515836992610011e-06,
+      "loss": 0.0177,
+      "step": 1383
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 5.50205718862841e-06,
+      "loss": 0.0156,
+      "step": 1384
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.4882880833596406e-06,
+      "loss": 0.005,
+      "step": 1385
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.4745297095546125e-06,
+      "loss": 0.0226,
+      "step": 1386
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.460782099938702e-06,
+      "loss": 0.0117,
+      "step": 1387
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 5.447045287211687e-06,
+      "loss": 0.0193,
+      "step": 1388
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.433319304047666e-06,
+      "loss": 0.0269,
+      "step": 1389
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.419604183094977e-06,
+      "loss": 0.0104,
+      "step": 1390
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.405899956976115e-06,
+      "loss": 0.0119,
+      "step": 1391
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 5.3922066582876685e-06,
+      "loss": 0.0332,
+      "step": 1392
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.378524319600231e-06,
+      "loss": 0.0148,
+      "step": 1393
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.36485297345833e-06,
+      "loss": 0.057,
+      "step": 1394
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.351192652380341e-06,
+      "loss": 0.0627,
+      "step": 1395
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.337543388858421e-06,
+      "loss": 0.0264,
+      "step": 1396
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 5.32390521535842e-06,
+      "loss": 0.036,
+      "step": 1397
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.310278164319816e-06,
+      "loss": 0.0313,
+      "step": 1398
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.29666226815563e-06,
+      "loss": 0.032,
+      "step": 1399
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.2830575592523415e-06,
+      "loss": 0.0339,
+      "step": 1400
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 5.26946406996983e-06,
+      "loss": 0.0179,
+      "step": 1401
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.255881832641285e-06,
+      "loss": 0.0682,
+      "step": 1402
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.242310879573134e-06,
+      "loss": 0.0133,
+      "step": 1403
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.228751243044961e-06,
+      "loss": 0.0136,
+      "step": 1404
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 5.215202955309435e-06,
+      "loss": 0.0325,
+      "step": 1405
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.2016660485922295e-06,
+      "loss": 0.0282,
+      "step": 1406
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.18814055509195e-06,
+      "loss": 0.0254,
+      "step": 1407
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.174626506980054e-06,
+      "loss": 0.0155,
+      "step": 1408
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 5.1611239364007694e-06,
+      "loss": 0.0419,
+      "step": 1409
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.1476328754710295e-06,
+      "loss": 0.0135,
+      "step": 1410
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.134153356280391e-06,
+      "loss": 0.04,
+      "step": 1411
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.12068541089096e-06,
+      "loss": 0.0076,
+      "step": 1412
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 5.1072290713373056e-06,
+      "loss": 0.0207,
+      "step": 1413
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.093784369626397e-06,
+      "loss": 0.0221,
+      "step": 1414
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.080351337737521e-06,
+      "loss": 0.0233,
+      "step": 1415
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.066930007622209e-06,
+      "loss": 0.0343,
+      "step": 1416
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.053520411204155e-06,
+      "loss": 0.018,
+      "step": 1417
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 5.040122580379148e-06,
+      "loss": 0.0053,
+      "step": 1418
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.026736547014981e-06,
+      "loss": 0.0224,
+      "step": 1419
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.0133623429513985e-06,
+      "loss": 0.0097,
+      "step": 1420
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 5.000000000000003e-06,
+      "loss": 0.0263,
+      "step": 1421
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 4.9866495499441824e-06,
+      "loss": 0.0086,
+      "step": 1422
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.97331102453904e-06,
+      "loss": 0.0116,
+      "step": 1423
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.959984455511313e-06,
+      "loss": 0.0123,
+      "step": 1424
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.946669874559301e-06,
+      "loss": 0.0324,
+      "step": 1425
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.933367313352788e-06,
+      "loss": 0.0392,
+      "step": 1426
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.920076803532971e-06,
+      "loss": 0.0083,
+      "step": 1427
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.9067983767123736e-06,
+      "loss": 0.0178,
+      "step": 1428
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.893532064474787e-06,
+      "loss": 0.0071,
+      "step": 1429
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.880277898375183e-06,
+      "loss": 0.0162,
+      "step": 1430
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.8670359099396466e-06,
+      "loss": 0.0062,
+      "step": 1431
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.85380613066529e-06,
+      "loss": 0.0153,
+      "step": 1432
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.8405885920201975e-06,
+      "loss": 0.0046,
+      "step": 1433
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.827383325443331e-06,
+      "loss": 0.0261,
+      "step": 1434
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.814190362344454e-06,
+      "loss": 0.0106,
+      "step": 1435
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.801009734104078e-06,
+      "loss": 0.0073,
+      "step": 1436
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.787841472073366e-06,
+      "loss": 0.058,
+      "step": 1437
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.774685607574076e-06,
+      "loss": 0.0248,
+      "step": 1438
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.761542171898469e-06,
+      "loss": 0.0442,
+      "step": 1439
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.748411196309248e-06,
+      "loss": 0.0387,
+      "step": 1440
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.735292712039478e-06,
+      "loss": 0.018,
+      "step": 1441
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.722186750292511e-06,
+      "loss": 0.0118,
+      "step": 1442
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.709093342241917e-06,
+      "loss": 0.0033,
+      "step": 1443
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.696012519031397e-06,
+      "loss": 0.0058,
+      "step": 1444
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.682944311774725e-06,
+      "loss": 0.0081,
+      "step": 1445
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.669888751555669e-06,
+      "loss": 0.0234,
+      "step": 1446
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 4.6568458694279095e-06,
+      "loss": 0.0162,
+      "step": 1447
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.643815696414975e-06,
+      "loss": 0.012,
+      "step": 1448
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.630798263510162e-06,
+      "loss": 0.0127,
+      "step": 1449
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.617793601676464e-06,
+      "loss": 0.0229,
+      "step": 1450
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 4.604801741846501e-06,
+      "loss": 0.0295,
+      "step": 1451
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.591822714922441e-06,
+      "loss": 0.0201,
+      "step": 1452
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.578856551775917e-06,
+      "loss": 0.0066,
+      "step": 1453
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.565903283247981e-06,
+      "loss": 0.0213,
+      "step": 1454
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 4.552962940149005e-06,
+      "loss": 0.0236,
+      "step": 1455
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.54003555325862e-06,
+      "loss": 0.0156,
+      "step": 1456
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.527121153325638e-06,
+      "loss": 0.0174,
+      "step": 1457
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.51421977106798e-06,
+      "loss": 0.029,
+      "step": 1458
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.501331437172606e-06,
+      "loss": 0.0264,
+      "step": 1459
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 4.488456182295437e-06,
+      "loss": 0.0109,
+      "step": 1460
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.475594037061285e-06,
+      "loss": 0.0155,
+      "step": 1461
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.4627450320637845e-06,
+      "loss": 0.0276,
+      "step": 1462
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.449909197865303e-06,
+      "loss": 0.0158,
+      "step": 1463
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 4.437086564996891e-06,
+      "loss": 0.0024,
+      "step": 1464
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.424277163958195e-06,
+      "loss": 0.0319,
+      "step": 1465
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.41148102521739e-06,
+      "loss": 0.0217,
+      "step": 1466
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.398698179211103e-06,
+      "loss": 0.0096,
+      "step": 1467
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 4.385928656344345e-06,
+      "loss": 0.0199,
+      "step": 1468
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.373172486990436e-06,
+      "loss": 0.0081,
+      "step": 1469
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.360429701490935e-06,
+      "loss": 0.0258,
+      "step": 1470
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.347700330155568e-06,
+      "loss": 0.005,
+      "step": 1471
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 4.334984403262146e-06,
+      "loss": 0.038,
+      "step": 1472
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.3222819510565085e-06,
+      "loss": 0.0198,
+      "step": 1473
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.309593003752446e-06,
+      "loss": 0.0134,
+      "step": 1474
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.296917591531621e-06,
+      "loss": 0.0164,
+      "step": 1475
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 4.284255744543505e-06,
+      "loss": 0.008,
+      "step": 1476
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.271607492905303e-06,
+      "loss": 0.0026,
+      "step": 1477
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.258972866701883e-06,
+      "loss": 0.016,
+      "step": 1478
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.246351895985702e-06,
+      "loss": 0.0137,
+      "step": 1479
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.233744610776741e-06,
+      "loss": 0.0112,
+      "step": 1480
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 4.221151041062418e-06,
+      "loss": 0.0522,
+      "step": 1481
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.208571216797539e-06,
+      "loss": 0.0402,
+      "step": 1482
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.196005167904211e-06,
+      "loss": 0.0166,
+      "step": 1483
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.183452924271776e-06,
+      "loss": 0.029,
+      "step": 1484
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 4.170914515756738e-06,
+      "loss": 0.0201,
+      "step": 1485
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.1583899721826935e-06,
+      "loss": 0.0129,
+      "step": 1486
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.145879323340261e-06,
+      "loss": 0.0122,
+      "step": 1487
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.133382598987007e-06,
+      "loss": 0.0231,
+      "step": 1488
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 4.120899828847385e-06,
+      "loss": 0.0419,
+      "step": 1489
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.108431042612642e-06,
+      "loss": 0.0164,
+      "step": 1490
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.095976269940777e-06,
+      "loss": 0.0111,
+      "step": 1491
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.083535540456452e-06,
+      "loss": 0.0195,
+      "step": 1492
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 4.071108883750927e-06,
+      "loss": 0.0282,
+      "step": 1493
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.058696329381987e-06,
+      "loss": 0.0112,
+      "step": 1494
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.046297906873875e-06,
+      "loss": 0.0177,
+      "step": 1495
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.033913645717218e-06,
+      "loss": 0.0189,
+      "step": 1496
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 4.021543575368968e-06,
+      "loss": 0.0485,
+      "step": 1497
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 4.009187725252309e-06,
+      "loss": 0.0115,
+      "step": 1498
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.996846124756609e-06,
+      "loss": 0.0175,
+      "step": 1499
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.984518803237343e-06,
+      "loss": 0.0249,
+      "step": 1500
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.972205790016019e-06,
+      "loss": 0.0101,
+      "step": 1501
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.959907114380114e-06,
+      "loss": 0.0421,
+      "step": 1502
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.9476228055830015e-06,
+      "loss": 0.0195,
+      "step": 1503
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.93535289284388e-06,
+      "loss": 0.0214,
+      "step": 1504
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.923097405347709e-06,
+      "loss": 0.0415,
+      "step": 1505
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.910856372245139e-06,
+      "loss": 0.0032,
+      "step": 1506
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.898629822652427e-06,
+      "loss": 0.0092,
+      "step": 1507
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.886417785651395e-06,
+      "loss": 0.0132,
+      "step": 1508
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.874220290289337e-06,
+      "loss": 0.0077,
+      "step": 1509
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.862037365578959e-06,
+      "loss": 0.0296,
+      "step": 1510
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.849869040498315e-06,
+      "loss": 0.0058,
+      "step": 1511
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.837715343990727e-06,
+      "loss": 0.0077,
+      "step": 1512
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.825576304964723e-06,
+      "loss": 0.0074,
+      "step": 1513
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.8134519522939693e-06,
+      "loss": 0.0158,
+      "step": 1514
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.801342314817201e-06,
+      "loss": 0.0239,
+      "step": 1515
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.7892474213381415e-06,
+      "loss": 0.0227,
+      "step": 1516
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.777167300625455e-06,
+      "loss": 0.0095,
+      "step": 1517
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.7651019814126656e-06,
+      "loss": 0.0099,
+      "step": 1518
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.753051492398089e-06,
+      "loss": 0.017,
+      "step": 1519
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.741015862244766e-06,
+      "loss": 0.0092,
+      "step": 1520
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.7289951195803954e-06,
+      "loss": 0.0123,
+      "step": 1521
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.716989292997264e-06,
+      "loss": 0.0098,
+      "step": 1522
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.704998411052182e-06,
+      "loss": 0.0167,
+      "step": 1523
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6930225022664136e-06,
+      "loss": 0.0138,
+      "step": 1524
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.681061595125598e-06,
+      "loss": 0.008,
+      "step": 1525
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.669115718079702e-06,
+      "loss": 0.0089,
+      "step": 1526
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6571848995429404e-06,
+      "loss": 0.0031,
+      "step": 1527
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6452691678937093e-06,
+      "loss": 0.0121,
+      "step": 1528
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6333685514745165e-06,
+      "loss": 0.0167,
+      "step": 1529
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.621483078591923e-06,
+      "loss": 0.0054,
+      "step": 1530
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6096127775164646e-06,
+      "loss": 0.0072,
+      "step": 1531
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.597757676482592e-06,
+      "loss": 0.0088,
+      "step": 1532
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.585917803688603e-06,
+      "loss": 0.0171,
+      "step": 1533
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.574093187296568e-06,
+      "loss": 0.0366,
+      "step": 1534
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5622838554322724e-06,
+      "loss": 0.0241,
+      "step": 1535
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5504898361851492e-06,
+      "loss": 0.007,
+      "step": 1536
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5387111576082046e-06,
+      "loss": 0.0061,
+      "step": 1537
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5269478477179588e-06,
+      "loss": 0.0334,
+      "step": 1538
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.515199934494373e-06,
+      "loss": 0.0136,
+      "step": 1539
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.5034674458807893e-06,
+      "loss": 0.0265,
+      "step": 1540
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.491750409783862e-06,
+      "loss": 0.0106,
+      "step": 1541
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4800488540734857e-06,
+      "loss": 0.0117,
+      "step": 1542
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.46836280658274e-06,
+      "loss": 0.0182,
+      "step": 1543
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.4566922951078086e-06,
+      "loss": 0.0193,
+      "step": 1544
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4450373474079278e-06,
+      "loss": 0.0062,
+      "step": 1545
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.433397991205314e-06,
+      "loss": 0.0133,
+      "step": 1546
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.421774254185096e-06,
+      "loss": 0.015,
+      "step": 1547
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.4101661639952533e-06,
+      "loss": 0.018,
+      "step": 1548
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.398573748246544e-06,
+      "loss": 0.018,
+      "step": 1549
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.3869970345124494e-06,
+      "loss": 0.0237,
+      "step": 1550
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.3754360503290973e-06,
+      "loss": 0.0256,
+      "step": 1551
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.363890823195206e-06,
+      "loss": 0.0032,
+      "step": 1552
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3523613805720055e-06,
+      "loss": 0.0276,
+      "step": 1553
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3408477498831917e-06,
+      "loss": 0.0198,
+      "step": 1554
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3293499585148446e-06,
+      "loss": 0.0198,
+      "step": 1555
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.3178680338153713e-06,
+      "loss": 0.0097,
+      "step": 1556
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.306402003095438e-06,
+      "loss": 0.0065,
+      "step": 1557
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.294951893627907e-06,
+      "loss": 0.0211,
+      "step": 1558
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.2835177326477675e-06,
+      "loss": 0.0072,
+      "step": 1559
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.2720995473520766e-06,
+      "loss": 0.0207,
+      "step": 1560
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2606973648998918e-06,
+      "loss": 0.0389,
+      "step": 1561
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.249311212412206e-06,
+      "loss": 0.0173,
+      "step": 1562
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2379411169718844e-06,
+      "loss": 0.0121,
+      "step": 1563
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2265871056235974e-06,
+      "loss": 0.0065,
+      "step": 1564
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.2152492053737595e-06,
+      "loss": 0.0049,
+      "step": 1565
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.2039274431904643e-06,
+      "loss": 0.0057,
+      "step": 1566
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1926218460034186e-06,
+      "loss": 0.0078,
+      "step": 1567
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.1813324407038826e-06,
+      "loss": 0.0084,
+      "step": 1568
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.170059254144593e-06,
+      "loss": 0.0199,
+      "step": 1569
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1588023131397206e-06,
+      "loss": 0.0265,
+      "step": 1570
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.147561644464788e-06,
+      "loss": 0.0313,
+      "step": 1571
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.136337274856618e-06,
+      "loss": 0.014,
+      "step": 1572
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.1251292310132584e-06,
+      "loss": 0.0223,
+      "step": 1573
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.113937539593931e-06,
+      "loss": 0.0373,
+      "step": 1574
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.1027622272189572e-06,
+      "loss": 0.0174,
+      "step": 1575
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.0916033204697026e-06,
+      "loss": 0.0061,
+      "step": 1576
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.080460845888513e-06,
+      "loss": 0.0212,
+      "step": 1577
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0693348299786407e-06,
+      "loss": 0.0137,
+      "step": 1578
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.058225299204195e-06,
+      "loss": 0.0088,
+      "step": 1579
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0471322799900747e-06,
+      "loss": 0.0169,
+      "step": 1580
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.0360557987219054e-06,
+      "loss": 0.0114,
+      "step": 1581
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.024995881745972e-06,
+      "loss": 0.0193,
+      "step": 1582
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.013952555369164e-06,
+      "loss": 0.0048,
+      "step": 1583
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.002925845858905e-06,
+      "loss": 0.0178,
+      "step": 1584
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.9919157794430963e-06,
+      "loss": 0.0315,
+      "step": 1585
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 2.9809223823100532e-06,
+      "loss": 0.0135,
+      "step": 1586
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9699456806084427e-06,
+      "loss": 0.0638,
+      "step": 1587
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9589857004472124e-06,
+      "loss": 0.0203,
+      "step": 1588
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.948042467895544e-06,
+      "loss": 0.0145,
+      "step": 1589
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 2.9371160089827842e-06,
+      "loss": 0.0129,
+      "step": 1590
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9262063496983783e-06,
+      "loss": 0.0138,
+      "step": 1591
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9153135159918145e-06,
+      "loss": 0.0187,
+      "step": 1592
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.9044375337725595e-06,
+      "loss": 0.0169,
+      "step": 1593
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 2.893578428909998e-06,
+      "loss": 0.006,
+      "step": 1594
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.882736227233368e-06,
+      "loss": 0.0114,
+      "step": 1595
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8719109545317102e-06,
+      "loss": 0.0018,
+      "step": 1596
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.861102636553783e-06,
+      "loss": 0.0107,
+      "step": 1597
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 2.8503112990080316e-06,
+      "loss": 0.0123,
+      "step": 1598
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.839536967562504e-06,
+      "loss": 0.0027,
+      "step": 1599
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.828779667844801e-06,
+      "loss": 0.008,
+      "step": 1600
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.8180394254420107e-06,
+      "loss": 0.0225,
+      "step": 1601
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.807316265900649e-06,
+      "loss": 0.046,
+      "step": 1602
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 2.7966102147265993e-06,
+      "loss": 0.0047,
+      "step": 1603
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7859212973850535e-06,
+      "loss": 0.0479,
+      "step": 1604
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7752495393004476e-06,
+      "loss": 0.0064,
+      "step": 1605
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7645949658563965e-06,
+      "loss": 0.0099,
+      "step": 1606
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 2.7539576023956503e-06,
+      "loss": 0.0115,
+      "step": 1607
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.743337474220019e-06,
+      "loss": 0.0105,
+      "step": 1608
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.732734606590318e-06,
+      "loss": 0.0033,
+      "step": 1609
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.722149024726307e-06,
+      "loss": 0.0457,
+      "step": 1610
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 2.71158075380663e-06,
+      "loss": 0.0102,
+      "step": 1611
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.7010298189687533e-06,
+      "loss": 0.0234,
+      "step": 1612
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.6904962453089135e-06,
+      "loss": 0.0345,
+      "step": 1613
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.679980057882049e-06,
+      "loss": 0.012,
+      "step": 1614
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 2.669481281701739e-06,
+      "loss": 0.0154,
+      "step": 1615
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.658999941740157e-06,
+      "loss": 0.0092,
+      "step": 1616
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.648536062927999e-06,
+      "loss": 0.0085,
+      "step": 1617
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6380896701544266e-06,
+      "loss": 0.0206,
+      "step": 1618
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 2.6276607882670135e-06,
+      "loss": 0.033,
+      "step": 1619
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.61724944207168e-06,
+      "loss": 0.0596,
+      "step": 1620
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.6068556563326362e-06,
+      "loss": 0.0085,
+      "step": 1621
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.596479455772325e-06,
+      "loss": 0.0047,
+      "step": 1622
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.5861208650713586e-06,
+      "loss": 0.0062,
+      "step": 1623
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 2.5757799088684654e-06,
+      "loss": 0.0104,
+      "step": 1624
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.5654566117604286e-06,
+      "loss": 0.009,
+      "step": 1625
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.555150998302026e-06,
+      "loss": 0.016,
+      "step": 1626
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.544863093005975e-06,
+      "loss": 0.0117,
+      "step": 1627
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 2.534592920342872e-06,
+      "loss": 0.0032,
+      "step": 1628
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5243405047411353e-06,
+      "loss": 0.0376,
+      "step": 1629
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.514105870586946e-06,
+      "loss": 0.0064,
+      "step": 1630
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.5038890422241958e-06,
+      "loss": 0.054,
+      "step": 1631
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 2.493690043954411e-06,
+      "loss": 0.0107,
+      "step": 1632
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.483508900036722e-06,
+      "loss": 0.0069,
+      "step": 1633
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4733456346877817e-06,
+      "loss": 0.0414,
+      "step": 1634
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4632002720817227e-06,
+      "loss": 0.027,
+      "step": 1635
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 2.4530728363500923e-06,
+      "loss": 0.0039,
+      "step": 1636
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4429633515817976e-06,
+      "loss": 0.01,
+      "step": 1637
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.432871841823047e-06,
+      "loss": 0.0177,
+      "step": 1638
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.4227983310772963e-06,
+      "loss": 0.0197,
+      "step": 1639
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 2.41274284330519e-06,
+      "loss": 0.0426,
+      "step": 1640
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.4027054024244954e-06,
+      "loss": 0.008,
+      "step": 1641
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.392686032310064e-06,
+      "loss": 0.0102,
+      "step": 1642
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.3826847567937582e-06,
+      "loss": 0.0029,
+      "step": 1643
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.3727015996644043e-06,
+      "loss": 0.012,
+      "step": 1644
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 2.362736584667731e-06,
+      "loss": 0.0222,
+      "step": 1645
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3527897355063147e-06,
+      "loss": 0.0064,
+      "step": 1646
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.342861075839523e-06,
+      "loss": 0.0096,
+      "step": 1647
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.332950629283459e-06,
+      "loss": 0.0864,
+      "step": 1648
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 2.3230584194109074e-06,
+      "loss": 0.0053,
+      "step": 1649
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.313184469751265e-06,
+      "loss": 0.0115,
+      "step": 1650
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.3033288037905045e-06,
+      "loss": 0.0216,
+      "step": 1651
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.293491444971109e-06,
+      "loss": 0.0097,
+      "step": 1652
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 2.2836724166920134e-06,
+      "loss": 0.0197,
+      "step": 1653
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2738717423085543e-06,
+      "loss": 0.0182,
+      "step": 1654
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.26408944513241e-06,
+      "loss": 0.0235,
+      "step": 1655
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2543255484315486e-06,
+      "loss": 0.0059,
+      "step": 1656
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 2.2445800754301715e-06,
+      "loss": 0.0099,
+      "step": 1657
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2348530493086595e-06,
+      "loss": 0.0098,
+      "step": 1658
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2251444932035094e-06,
+      "loss": 0.015,
+      "step": 1659
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.21545443020729e-06,
+      "loss": 0.0128,
+      "step": 1660
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 2.2057828833685855e-06,
+      "loss": 0.0197,
+      "step": 1661
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1961298756919337e-06,
+      "loss": 0.0267,
+      "step": 1662
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.1864954301377774e-06,
+      "loss": 0.0139,
+      "step": 1663
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.176879569622409e-06,
+      "loss": 0.0145,
+      "step": 1664
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.16728231701791e-06,
+      "loss": 0.0127,
+      "step": 1665
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 2.157703695152109e-06,
+      "loss": 0.0318,
+      "step": 1666
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1481437268085135e-06,
+      "loss": 0.0117,
+      "step": 1667
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1386024347262692e-06,
+      "loss": 0.013,
+      "step": 1668
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.1290798416000857e-06,
+      "loss": 0.0101,
+      "step": 1669
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 2.11957597008021e-06,
+      "loss": 0.0097,
+      "step": 1670
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.11009084277235e-06,
+      "loss": 0.0165,
+      "step": 1671
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.100624482237632e-06,
+      "loss": 0.0142,
+      "step": 1672
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.091176910992545e-06,
+      "loss": 0.0079,
+      "step": 1673
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.081748151508883e-06,
+      "loss": 0.0055,
+      "step": 1674
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0723382262136984e-06,
+      "loss": 0.0058,
+      "step": 1675
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.06294715748924e-06,
+      "loss": 0.0259,
+      "step": 1676
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0535749676729123e-06,
+      "loss": 0.0333,
+      "step": 1677
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.0442216790572046e-06,
+      "loss": 0.0664,
+      "step": 1678
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0348873138896563e-06,
+      "loss": 0.0107,
+      "step": 1679
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.025571894372794e-06,
+      "loss": 0.0089,
+      "step": 1680
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0162754426640795e-06,
+      "loss": 0.0074,
+      "step": 1681
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.0069979808758586e-06,
+      "loss": 0.0058,
+      "step": 1682
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9977395310753087e-06,
+      "loss": 0.0049,
+      "step": 1683
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.988500115284385e-06,
+      "loss": 0.0092,
+      "step": 1684
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.979279755479768e-06,
+      "loss": 0.0027,
+      "step": 1685
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.970078473592817e-06,
+      "loss": 0.0068,
+      "step": 1686
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 1.9608962915095e-06,
+      "loss": 0.0025,
+      "step": 1687
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.951733231070373e-06,
+      "loss": 0.0199,
+      "step": 1688
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.942589314070494e-06,
+      "loss": 0.0034,
+      "step": 1689
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9334645622593973e-06,
+      "loss": 0.0072,
+      "step": 1690
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 1.9243589973410248e-06,
+      "loss": 0.0114,
+      "step": 1691
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9152726409736834e-06,
+      "loss": 0.0034,
+      "step": 1692
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.9062055147699931e-06,
+      "loss": 0.0256,
+      "step": 1693
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.897157640296825e-06,
+      "loss": 0.0139,
+      "step": 1694
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 1.8881290390752682e-06,
+      "loss": 0.0051,
+      "step": 1695
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8791197325805644e-06,
+      "loss": 0.0051,
+      "step": 1696
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8701297422420627e-06,
+      "loss": 0.0048,
+      "step": 1697
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8611590894431653e-06,
+      "loss": 0.0047,
+      "step": 1698
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 1.8522077955212791e-06,
+      "loss": 0.0099,
+      "step": 1699
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8432758817677666e-06,
+      "loss": 0.016,
+      "step": 1700
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8343633694278895e-06,
+      "loss": 0.0073,
+      "step": 1701
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8254702797007684e-06,
+      "loss": 0.0064,
+      "step": 1702
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 1.8165966337393127e-06,
+      "loss": 0.0171,
+      "step": 1703
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.8077424526501964e-06,
+      "loss": 0.0056,
+      "step": 1704
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7989077574937886e-06,
+      "loss": 0.007,
+      "step": 1705
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.790092569284111e-06,
+      "loss": 0.008,
+      "step": 1706
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7812969089887867e-06,
+      "loss": 0.0093,
+      "step": 1707
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 1.7725207975289883e-06,
+      "loss": 0.0054,
+      "step": 1708
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.763764255779392e-06,
+      "loss": 0.0101,
+      "step": 1709
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7550273045681243e-06,
+      "loss": 0.0047,
+      "step": 1710
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7463099646767167e-06,
+      "loss": 0.0024,
+      "step": 1711
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 1.7376122568400533e-06,
+      "loss": 0.0043,
+      "step": 1712
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.728934201746314e-06,
+      "loss": 0.0034,
+      "step": 1713
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.720275820036944e-06,
+      "loss": 0.0045,
+      "step": 1714
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7116371323065883e-06,
+      "loss": 0.013,
+      "step": 1715
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 1.7030181591030493e-06,
+      "loss": 0.0035,
+      "step": 1716
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6944189209272354e-06,
+      "loss": 0.0103,
+      "step": 1717
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6858394382331178e-06,
+      "loss": 0.0234,
+      "step": 1718
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6772797314276712e-06,
+      "loss": 0.0171,
+      "step": 1719
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 1.6687398208708382e-06,
+      "loss": 0.0027,
+      "step": 1720
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6602197268754738e-06,
+      "loss": 0.0036,
+      "step": 1721
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6517194697072903e-06,
+      "loss": 0.0138,
+      "step": 1722
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6432390695848233e-06,
+      "loss": 0.0078,
+      "step": 1723
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 1.6347785466793764e-06,
+      "loss": 0.0034,
+      "step": 1724
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6263379211149733e-06,
+      "loss": 0.0148,
+      "step": 1725
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6179172129683086e-06,
+      "loss": 0.0079,
+      "step": 1726
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6095164422687038e-06,
+      "loss": 0.0115,
+      "step": 1727
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.6011356289980562e-06,
+      "loss": 0.0125,
+      "step": 1728
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 1.5927747930907921e-06,
+      "loss": 0.0037,
+      "step": 1729
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.584433954433825e-06,
+      "loss": 0.0075,
+      "step": 1730
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5761131328664946e-06,
+      "loss": 0.0453,
+      "step": 1731
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5678123481805353e-06,
+      "loss": 0.0079,
+      "step": 1732
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 1.5595316201200205e-06,
+      "loss": 0.0054,
+      "step": 1733
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.5512709683813165e-06,
+      "loss": 0.0076,
+      "step": 1734
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.543030412613038e-06,
+      "loss": 0.0195,
+      "step": 1735
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.534809972415998e-06,
+      "loss": 0.0055,
+      "step": 1736
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 1.526609667343163e-06,
+      "loss": 0.0069,
+      "step": 1737
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5184295168996098e-06,
+      "loss": 0.0047,
+      "step": 1738
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5102695405424738e-06,
+      "loss": 0.0033,
+      "step": 1739
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.5021297576809002e-06,
+      "loss": 0.0069,
+      "step": 1740
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 1.4940101876760094e-06,
+      "loss": 0.0186,
+      "step": 1741
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4859108498408404e-06,
+      "loss": 0.0039,
+      "step": 1742
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4778317634403082e-06,
+      "loss": 0.0152,
+      "step": 1743
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.4697729476911614e-06,
+      "loss": 0.0043,
+      "step": 1744
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 1.461734421761929e-06,
+      "loss": 0.0153,
+      "step": 1745
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4537162047728804e-06,
+      "loss": 0.0042,
+      "step": 1746
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4457183157959808e-06,
+      "loss": 0.0296,
+      "step": 1747
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4377407738548411e-06,
+      "loss": 0.0182,
+      "step": 1748
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4297835979246777e-06,
+      "loss": 0.0023,
+      "step": 1749
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 1.4218468069322576e-06,
+      "loss": 0.0046,
+      "step": 1750
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4139304197558713e-06,
+      "loss": 0.022,
+      "step": 1751
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.4060344552252702e-06,
+      "loss": 0.0048,
+      "step": 1752
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.3981589321216315e-06,
+      "loss": 0.0153,
+      "step": 1753
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 1.3903038691775095e-06,
+      "loss": 0.0048,
+      "step": 1754
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3824692850767928e-06,
+      "loss": 0.0043,
+      "step": 1755
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3746551984546609e-06,
+      "loss": 0.0082,
+      "step": 1756
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3668616278975343e-06,
+      "loss": 0.0015,
+      "step": 1757
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 1.3590885919430385e-06,
+      "loss": 0.0076,
+      "step": 1758
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3513361090799537e-06,
+      "loss": 0.0042,
+      "step": 1759
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3436041977481751e-06,
+      "loss": 0.0065,
+      "step": 1760
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3358928763386647e-06,
+      "loss": 0.0127,
+      "step": 1761
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 1.3282021631934106e-06,
+      "loss": 0.014,
+      "step": 1762
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3205320766053831e-06,
+      "loss": 0.0066,
+      "step": 1763
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.3128826348184886e-06,
+      "loss": 0.0045,
+      "step": 1764
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.305253856027533e-06,
+      "loss": 0.0064,
+      "step": 1765
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 1.2976457583781666e-06,
+      "loss": 0.0075,
+      "step": 1766
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2900583599668515e-06,
+      "loss": 0.007,
+      "step": 1767
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.282491678840816e-06,
+      "loss": 0.0028,
+      "step": 1768
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2749457329980108e-06,
+      "loss": 0.0161,
+      "step": 1769
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2674205403870643e-06,
+      "loss": 0.0099,
+      "step": 1770
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 1.2599161189072428e-06,
+      "loss": 0.0052,
+      "step": 1771
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2524324864084058e-06,
+      "loss": 0.006,
+      "step": 1772
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.2449696606909656e-06,
+      "loss": 0.0089,
+      "step": 1773
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.237527659505846e-06,
+      "loss": 0.0024,
+      "step": 1774
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 1.230106500554431e-06,
+      "loss": 0.004,
+      "step": 1775
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.222706201488536e-06,
+      "loss": 0.0079,
+      "step": 1776
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.2153267799103595e-06,
+      "loss": 0.0086,
+      "step": 1777
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.207968253372438e-06,
+      "loss": 0.0057,
+      "step": 1778
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 1.200630639377609e-06,
+      "loss": 0.0167,
+      "step": 1779
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1933139553789697e-06,
+      "loss": 0.0062,
+      "step": 1780
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1860182187798297e-06,
+      "loss": 0.0078,
+      "step": 1781
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1787434469336767e-06,
+      "loss": 0.0217,
+      "step": 1782
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 1.1714896571441326e-06,
+      "loss": 0.0317,
+      "step": 1783
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1642568666649067e-06,
+      "loss": 0.0047,
+      "step": 1784
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.1570450926997657e-06,
+      "loss": 0.0073,
+      "step": 1785
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.149854352402483e-06,
+      "loss": 0.0024,
+      "step": 1786
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 1.142684662876805e-06,
+      "loss": 0.0092,
+      "step": 1787
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.135536041176406e-06,
+      "loss": 0.0079,
+      "step": 1788
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1284085043048465e-06,
+      "loss": 0.0098,
+      "step": 1789
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1213020692155364e-06,
+      "loss": 0.0053,
+      "step": 1790
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1142167528116965e-06,
+      "loss": 0.0039,
+      "step": 1791
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 1.1071525719463094e-06,
+      "loss": 0.0098,
+      "step": 1792
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.100109543422093e-06,
+      "loss": 0.0045,
+      "step": 1793
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0930876839914418e-06,
+      "loss": 0.0125,
+      "step": 1794
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0860870103564069e-06,
+      "loss": 0.014,
+      "step": 1795
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 1.0791075391686435e-06,
+      "loss": 0.0085,
+      "step": 1796
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.072149287029377e-06,
+      "loss": 0.0128,
+      "step": 1797
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0652122704893608e-06,
+      "loss": 0.01,
+      "step": 1798
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.058296506048836e-06,
+      "loss": 0.0036,
+      "step": 1799
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 1.0514020101574974e-06,
+      "loss": 0.0097,
+      "step": 1800
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.044528799214447e-06,
+      "loss": 0.0041,
+      "step": 1801
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0376768895681655e-06,
+      "loss": 0.0029,
+      "step": 1802
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0308462975164547e-06,
+      "loss": 0.0043,
+      "step": 1803
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 1.0240370393064235e-06,
+      "loss": 0.0227,
+      "step": 1804
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.017249131134428e-06,
+      "loss": 0.0048,
+      "step": 1805
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.010482589146048e-06,
+      "loss": 0.0042,
+      "step": 1806
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 1.0037374294360368e-06,
+      "loss": 0.0059,
+      "step": 1807
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 9.970136680482912e-07,
+      "loss": 0.0038,
+      "step": 1808
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.903113209758098e-07,
+      "loss": 0.0109,
+      "step": 1809
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.836304041606537e-07,
+      "loss": 0.0042,
+      "step": 1810
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.76970933493917e-07,
+      "loss": 0.006,
+      "step": 1811
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.703329248156712e-07,
+      "loss": 0.002,
+      "step": 1812
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 9.637163939149485e-07,
+      "loss": 0.0075,
+      "step": 1813
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.571213565296877e-07,
+      "loss": 0.0149,
+      "step": 1814
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.505478283467129e-07,
+      "loss": 0.0082,
+      "step": 1815
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.439958250016768e-07,
+      "loss": 0.0327,
+      "step": 1816
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 9.374653620790397e-07,
+      "loss": 0.0036,
+      "step": 1817
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.309564551120254e-07,
+      "loss": 0.0047,
+      "step": 1818
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.244691195825794e-07,
+      "loss": 0.017,
+      "step": 1819
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.180033709213454e-07,
+      "loss": 0.0158,
+      "step": 1820
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 9.115592245076177e-07,
+      "loss": 0.0124,
+      "step": 1821
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 9.051366956693097e-07,
+      "loss": 0.0023,
+      "step": 1822
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.987357996829115e-07,
+      "loss": 0.0036,
+      "step": 1823
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.923565517734633e-07,
+      "loss": 0.0039,
+      "step": 1824
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 8.859989671145108e-07,
+      "loss": 0.0061,
+      "step": 1825
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.796630608280709e-07,
+      "loss": 0.0123,
+      "step": 1826
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.733488479845997e-07,
+      "loss": 0.0053,
+      "step": 1827
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.670563436029489e-07,
+      "loss": 0.0055,
+      "step": 1828
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 8.607855626503403e-07,
+      "loss": 0.0057,
+      "step": 1829
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.545365200423216e-07,
+      "loss": 0.0055,
+      "step": 1830
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.483092306427354e-07,
+      "loss": 0.0051,
+      "step": 1831
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.42103709263682e-07,
+      "loss": 0.0049,
+      "step": 1832
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.359199706654852e-07,
+      "loss": 0.0091,
+      "step": 1833
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 8.297580295566576e-07,
+      "loss": 0.0049,
+      "step": 1834
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.236179005938616e-07,
+      "loss": 0.0079,
+      "step": 1835
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.174995983818801e-07,
+      "loss": 0.0105,
+      "step": 1836
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.11403137473582e-07,
+      "loss": 0.0153,
+      "step": 1837
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 8.053285323698779e-07,
+      "loss": 0.0075,
+      "step": 1838
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.992757975196974e-07,
+      "loss": 0.0052,
+      "step": 1839
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.932449473199511e-07,
+      "loss": 0.0068,
+      "step": 1840
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.872359961154907e-07,
+      "loss": 0.014,
+      "step": 1841
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 7.812489581990846e-07,
+      "loss": 0.0069,
+      "step": 1842
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.752838478113744e-07,
+      "loss": 0.005,
+      "step": 1843
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.693406791408476e-07,
+      "loss": 0.0068,
+      "step": 1844
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.634194663238015e-07,
+      "loss": 0.0045,
+      "step": 1845
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 7.575202234443102e-07,
+      "loss": 0.0033,
+      "step": 1846
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.516429645341872e-07,
+      "loss": 0.0057,
+      "step": 1847
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.457877035729588e-07,
+      "loss": 0.0076,
+      "step": 1848
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.399544544878268e-07,
+      "loss": 0.0043,
+      "step": 1849
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 7.341432311536356e-07,
+      "loss": 0.0056,
+      "step": 1850
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.283540473928386e-07,
+      "loss": 0.0045,
+      "step": 1851
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.225869169754696e-07,
+      "loss": 0.0042,
+      "step": 1852
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.168418536191024e-07,
+      "loss": 0.0213,
+      "step": 1853
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.11118870988825e-07,
+      "loss": 0.0059,
+      "step": 1854
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 7.054179826972074e-07,
+      "loss": 0.016,
+      "step": 1855
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.997392023042593e-07,
+      "loss": 0.0087,
+      "step": 1856
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.940825433174103e-07,
+      "loss": 0.0034,
+      "step": 1857
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.884480191914722e-07,
+      "loss": 0.0126,
+      "step": 1858
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 6.828356433286065e-07,
+      "loss": 0.0043,
+      "step": 1859
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.772454290782926e-07,
+      "loss": 0.0193,
+      "step": 1860
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.716773897372963e-07,
+      "loss": 0.0071,
+      "step": 1861
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.661315385496426e-07,
+      "loss": 0.0074,
+      "step": 1862
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 6.606078887065748e-07,
+      "loss": 0.0099,
+      "step": 1863
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.551064533465335e-07,
+      "loss": 0.0084,
+      "step": 1864
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.496272455551145e-07,
+      "loss": 0.0024,
+      "step": 1865
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.441702783650494e-07,
+      "loss": 0.004,
+      "step": 1866
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 6.387355647561655e-07,
+      "loss": 0.0119,
+      "step": 1867
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.33323117655359e-07,
+      "loss": 0.0093,
+      "step": 1868
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.279329499365649e-07,
+      "loss": 0.0049,
+      "step": 1869
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.225650744207212e-07,
+      "loss": 0.0387,
+      "step": 1870
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 6.172195038757478e-07,
+      "loss": 0.0018,
+      "step": 1871
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.118962510165039e-07,
+      "loss": 0.0053,
+      "step": 1872
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.065953285047688e-07,
+      "loss": 0.0053,
+      "step": 1873
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 6.013167489492089e-07,
+      "loss": 0.0073,
+      "step": 1874
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 5.960605249053386e-07,
+      "loss": 0.0044,
+      "step": 1875
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 5.908266688755049e-07,
+      "loss": 0.0138,
+      "step": 1876
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.856151933088472e-07,
+      "loss": 0.0087,
+      "step": 1877
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.804261106012732e-07,
+      "loss": 0.008,
+      "step": 1878
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.752594330954275e-07,
+      "loss": 0.0109,
+      "step": 1879
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 5.701151730806609e-07,
+      "loss": 0.009,
+      "step": 1880
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.649933427930043e-07,
+      "loss": 0.0053,
+      "step": 1881
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.598939544151328e-07,
+      "loss": 0.0053,
+      "step": 1882
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.54817020076347e-07,
+      "loss": 0.0044,
+      "step": 1883
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 5.497625518525374e-07,
+      "loss": 0.0049,
+      "step": 1884
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.447305617661558e-07,
+      "loss": 0.015,
+      "step": 1885
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.397210617861881e-07,
+      "loss": 0.0105,
+      "step": 1886
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.347340638281273e-07,
+      "loss": 0.0223,
+      "step": 1887
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 5.297695797539426e-07,
+      "loss": 0.0125,
+      "step": 1888
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.248276213720526e-07,
+      "loss": 0.0107,
+      "step": 1889
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.199082004372958e-07,
+      "loss": 0.0037,
+      "step": 1890
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.150113286509028e-07,
+      "loss": 0.0117,
+      "step": 1891
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 5.101370176604725e-07,
+      "loss": 0.0041,
+      "step": 1892
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.052852790599383e-07,
+      "loss": 0.0058,
+      "step": 1893
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 5.004561243895433e-07,
+      "loss": 0.0039,
+      "step": 1894
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.956495651358162e-07,
+      "loss": 0.0065,
+      "step": 1895
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.908656127315359e-07,
+      "loss": 0.0169,
+      "step": 1896
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 4.861042785557147e-07,
+      "loss": 0.0253,
+      "step": 1897
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.813655739335599e-07,
+      "loss": 0.0221,
+      "step": 1898
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.7664951013645875e-07,
+      "loss": 0.0043,
+      "step": 1899
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.7195609838193714e-07,
+      "loss": 0.0074,
+      "step": 1900
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 4.672853498336505e-07,
+      "loss": 0.0099,
+      "step": 1901
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.6263727560134086e-07,
+      "loss": 0.005,
+      "step": 1902
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.580118867408223e-07,
+      "loss": 0.0057,
+      "step": 1903
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.534091942539476e-07,
+      "loss": 0.0116,
+      "step": 1904
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 4.4882920908858595e-07,
+      "loss": 0.0105,
+      "step": 1905
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.4427194213859216e-07,
+      "loss": 0.0283,
+      "step": 1906
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.397374042437885e-07,
+      "loss": 0.0094,
+      "step": 1907
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.352256061899329e-07,
+      "loss": 0.0238,
+      "step": 1908
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 4.3073655870869093e-07,
+      "loss": 0.0072,
+      "step": 1909
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.26270272477618e-07,
+      "loss": 0.0078,
+      "step": 1910
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.218267581201296e-07,
+      "loss": 0.0033,
+      "step": 1911
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.1740602620547687e-07,
+      "loss": 0.0053,
+      "step": 1912
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 4.1300808724871964e-07,
+      "loss": 0.0037,
+      "step": 1913
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.086329517107046e-07,
+      "loss": 0.0154,
+      "step": 1914
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 4.042806299980362e-07,
+      "loss": 0.0058,
+      "step": 1915
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.9995113246305783e-07,
+      "loss": 0.0052,
+      "step": 1916
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.9564446940382087e-07,
+      "loss": 0.0059,
+      "step": 1917
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 3.913606510640644e-07,
+      "loss": 0.0076,
+      "step": 1918
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.8709968763318894e-07,
+      "loss": 0.0263,
+      "step": 1919
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.8286158924623287e-07,
+      "loss": 0.0051,
+      "step": 1920
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.7864636598384906e-07,
+      "loss": 0.0074,
+      "step": 1921
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 3.744540278722797e-07,
+      "loss": 0.0035,
+      "step": 1922
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.702845848833314e-07,
+      "loss": 0.0029,
+      "step": 1923
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.661380469343556e-07,
+      "loss": 0.0106,
+      "step": 1924
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.620144238882206e-07,
+      "loss": 0.021,
+      "step": 1925
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 3.579137255532894e-07,
+      "loss": 0.0088,
+      "step": 1926
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.5383596168339974e-07,
+      "loss": 0.0072,
+      "step": 1927
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.4978114197783184e-07,
+      "loss": 0.0147,
+      "step": 1928
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.457492760812975e-07,
+      "loss": 0.0139,
+      "step": 1929
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 3.4174037358390753e-07,
+      "loss": 0.0075,
+      "step": 1930
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.3775444402115554e-07,
+      "loss": 0.0038,
+      "step": 1931
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.3379149687388866e-07,
+      "loss": 0.016,
+      "step": 1932
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.298515415682901e-07,
+      "loss": 0.0125,
+      "step": 1933
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 3.2593458747585683e-07,
+      "loss": 0.0097,
+      "step": 1934
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.2204064391337277e-07,
+      "loss": 0.0031,
+      "step": 1935
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.181697201428924e-07,
+      "loss": 0.0048,
+      "step": 1936
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.1432182537171287e-07,
+      "loss": 0.003,
+      "step": 1937
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.1049696875235515e-07,
+      "loss": 0.0072,
+      "step": 1938
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 3.0669515938254404e-07,
+      "loss": 0.0042,
+      "step": 1939
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 3.029164063051859e-07,
+      "loss": 0.0022,
+      "step": 1940
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.9916071850834004e-07,
+      "loss": 0.005,
+      "step": 1941
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.954281049252106e-07,
+      "loss": 0.0128,
+      "step": 1942
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 2.9171857443411443e-07,
+      "loss": 0.0084,
+      "step": 1943
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.8803213585846036e-07,
+      "loss": 0.0088,
+      "step": 1944
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.8436879796673647e-07,
+      "loss": 0.0067,
+      "step": 1945
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.807285694724804e-07,
+      "loss": 0.0139,
+      "step": 1946
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 2.7711145903426494e-07,
+      "loss": 0.0081,
+      "step": 1947
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.735174752556724e-07,
+      "loss": 0.0104,
+      "step": 1948
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.699466266852779e-07,
+      "loss": 0.0078,
+      "step": 1949
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.6639892181662744e-07,
+      "loss": 0.0068,
+      "step": 1950
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 2.628743690882163e-07,
+      "loss": 0.0067,
+      "step": 1951
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.593729768834752e-07,
+      "loss": 0.0086,
+      "step": 1952
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.5589475353073987e-07,
+      "loss": 0.0038,
+      "step": 1953
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.524397073032403e-07,
+      "loss": 0.0075,
+      "step": 1954
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 2.4900784641907707e-07,
+      "loss": 0.0049,
+      "step": 1955
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.4559917904120266e-07,
+      "loss": 0.0044,
+      "step": 1956
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.4221371327740384e-07,
+      "loss": 0.0044,
+      "step": 1957
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.388514571802769e-07,
+      "loss": 0.0091,
+      "step": 1958
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.3551241874721353e-07,
+      "loss": 0.0025,
+      "step": 1959
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 2.3219660592038285e-07,
+      "loss": 0.0028,
+      "step": 1960
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2890402658670707e-07,
+      "loss": 0.0059,
+      "step": 1961
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.256346885778482e-07,
+      "loss": 0.0096,
+      "step": 1962
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.2238859967018133e-07,
+      "loss": 0.008,
+      "step": 1963
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 2.1916576758478913e-07,
+      "loss": 0.0283,
+      "step": 1964
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1596619998743075e-07,
+      "loss": 0.0034,
+      "step": 1965
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.1278990448853288e-07,
+      "loss": 0.0061,
+      "step": 1966
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.0963688864316324e-07,
+      "loss": 0.0184,
+      "step": 1967
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 2.0650715995102266e-07,
+      "loss": 0.0334,
+      "step": 1968
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0340072585641523e-07,
+      "loss": 0.004,
+      "step": 1969
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 2.0031759374824155e-07,
+      "loss": 0.0041,
+      "step": 1970
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.9725777095997655e-07,
+      "loss": 0.0079,
+      "step": 1971
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.9422126476964842e-07,
+      "loss": 0.0162,
+      "step": 1972
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.9120808239982857e-07,
+      "loss": 0.0086,
+      "step": 1973
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8821823101760949e-07,
+      "loss": 0.0028,
+      "step": 1974
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.8525171773459028e-07,
+      "loss": 0.003,
+      "step": 1975
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.823085496068555e-07,
+      "loss": 0.0066,
+      "step": 1976
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.7938873363496422e-07,
+      "loss": 0.0069,
+      "step": 1977
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.764922767639299e-07,
+      "loss": 0.0043,
+      "step": 1978
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.736191858832048e-07,
+      "loss": 0.0042,
+      "step": 1979
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.707694678266636e-07,
+      "loss": 0.0037,
+      "step": 1980
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.6794312937258417e-07,
+      "loss": 0.0062,
+      "step": 1981
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6514017724363673e-07,
+      "loss": 0.0213,
+      "step": 1982
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.6236061810686488e-07,
+      "loss": 0.0047,
+      "step": 1983
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.5960445857367003e-07,
+      "loss": 0.0036,
+      "step": 1984
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.568717051997948e-07,
+      "loss": 0.006,
+      "step": 1985
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.5416236448531075e-07,
+      "loss": 0.0071,
+      "step": 1986
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.514764428745985e-07,
+      "loss": 0.0023,
+      "step": 1987
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.4881394675633543e-07,
+      "loss": 0.0019,
+      "step": 1988
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.4617488246348012e-07,
+      "loss": 0.0045,
+      "step": 1989
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.435592562732524e-07,
+      "loss": 0.0093,
+      "step": 1990
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.4096707440712898e-07,
+      "loss": 0.0674,
+      "step": 1991
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.383983430308189e-07,
+      "loss": 0.0035,
+      "step": 1992
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.3585306825425248e-07,
+      "loss": 0.0068,
+      "step": 1993
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.3333125613156695e-07,
+      "loss": 0.0087,
+      "step": 1994
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.30832912661093e-07,
+      "loss": 0.0039,
+      "step": 1995
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.2835804378533823e-07,
+      "loss": 0.0025,
+      "step": 1996
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.2590665539097268e-07,
+      "loss": 0.0086,
+      "step": 1997
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2347875330881886e-07,
+      "loss": 0.0027,
+      "step": 1998
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.2107434331383504e-07,
+      "loss": 0.006,
+      "step": 1999
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.1869343112509757e-07,
+      "loss": 0.0043,
+      "step": 2000
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.1633602240579522e-07,
+      "loss": 0.0044,
+      "step": 2001
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 1.1400212276321377e-07,
+      "loss": 0.0092,
+      "step": 2002
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.1169173774871478e-07,
+      "loss": 0.0089,
+      "step": 2003
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.094048728577346e-07,
+      "loss": 0.0084,
+      "step": 2004
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.0714153352975875e-07,
+      "loss": 0.0052,
+      "step": 2005
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 1.049017251483242e-07,
+      "loss": 0.0054,
+      "step": 2006
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.026854530409882e-07,
+      "loss": 0.0041,
+      "step": 2007
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 1.0049272247932951e-07,
+      "loss": 0.0073,
+      "step": 2008
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 9.832353867893385e-08,
+      "loss": 0.0204,
+      "step": 2009
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 9.617790679937288e-08,
+      "loss": 0.0044,
+      "step": 2010
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.405583194420531e-08,
+      "loss": 0.0051,
+      "step": 2011
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 9.19573191609524e-08,
+      "loss": 0.0152,
+      "step": 2012
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.988237344109251e-08,
+      "loss": 0.0158,
+      "step": 2013
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.783099972004882e-08,
+      "loss": 0.0071,
+      "step": 2014
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.580320287717603e-08,
+      "loss": 0.0111,
+      "step": 2015
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.379898773574924e-08,
+      "loss": 0.0019,
+      "step": 2016
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.181835906295177e-08,
+      "loss": 0.0105,
+      "step": 2017
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 7.986132156986737e-08,
+      "loss": 0.0056,
+      "step": 2018
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 7.792787991146356e-08,
+      "loss": 0.0055,
+      "step": 2019
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.601803868658386e-08,
+      "loss": 0.0035,
+      "step": 2020
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.413180243793783e-08,
+      "loss": 0.0056,
+      "step": 2021
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.226917565208658e-08,
+      "loss": 0.0178,
+      "step": 2022
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.043016275943614e-08,
+      "loss": 0.0176,
+      "step": 2023
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.861476813422419e-08,
+      "loss": 0.018,
+      "step": 2024
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.682299609450993e-08,
+      "loss": 0.0132,
+      "step": 2025
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.505485090216424e-08,
+      "loss": 0.0228,
+      "step": 2026
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 6.33103367628618e-08,
+      "loss": 0.0066,
+      "step": 2027
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.158945782606674e-08,
+      "loss": 0.0037,
+      "step": 2028
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.989221818502478e-08,
+      "loss": 0.0037,
+      "step": 2029
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.821862187675775e-08,
+      "loss": 0.0082,
+      "step": 2030
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 5.6568672882043595e-08,
+      "loss": 0.0194,
+      "step": 2031
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.4942375125418554e-08,
+      "loss": 0.0223,
+      "step": 2032
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.3339732475159444e-08,
+      "loss": 0.0124,
+      "step": 2033
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.176074874327919e-08,
+      "loss": 0.0065,
+      "step": 2034
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 5.0205427685514616e-08,
+      "loss": 0.0061,
+      "step": 2035
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.867377300132092e-08,
+      "loss": 0.0055,
+      "step": 2036
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.716578833386054e-08,
+      "loss": 0.0033,
+      "step": 2037
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.568147726999206e-08,
+      "loss": 0.0057,
+      "step": 2038
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.4220843340269105e-08,
+      "loss": 0.014,
+      "step": 2039
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 4.278389001892369e-08,
+      "loss": 0.0107,
+      "step": 2040
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 4.137062072386289e-08,
+      "loss": 0.0307,
+      "step": 2041
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 3.998103881665993e-08,
+      "loss": 0.0088,
+      "step": 2042
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 3.861514760254648e-08,
+      "loss": 0.0095,
+      "step": 2043
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 3.727295033040035e-08,
+      "loss": 0.007,
+      "step": 2044
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.595445019274668e-08,
+      "loss": 0.0136,
+      "step": 2045
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.465965032574015e-08,
+      "loss": 0.0041,
+      "step": 2046
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.3388553809164945e-08,
+      "loss": 0.005,
+      "step": 2047
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 3.214116366642595e-08,
+      "loss": 0.0114,
+      "step": 2048
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 3.091748286453866e-08,
+      "loss": 0.015,
+      "step": 2049
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.971751431412484e-08,
+      "loss": 0.0041,
+      "step": 2050
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.8541260869403565e-08,
+      "loss": 0.0023,
+      "step": 2051
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 2.7388725328189036e-08,
+      "loss": 0.0063,
+      "step": 2052
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.6259910431876146e-08,
+      "loss": 0.0126,
+      "step": 2053
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.5154818865440466e-08,
+      "loss": 0.0037,
+      "step": 2054
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.407345325743049e-08,
+      "loss": 0.0028,
+      "step": 2055
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 2.301581617995985e-08,
+      "loss": 0.002,
+      "step": 2056
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.1981910148701768e-08,
+      "loss": 0.0043,
+      "step": 2057
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 2.0971737622883515e-08,
+      "loss": 0.0054,
+      "step": 2058
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.9985301005280843e-08,
+      "loss": 0.0035,
+      "step": 2059
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.902260264221023e-08,
+      "loss": 0.0095,
+      "step": 2060
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 1.8083644823526645e-08,
+      "loss": 0.003,
+      "step": 2061
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.7168429782615794e-08,
+      "loss": 0.0082,
+      "step": 2062
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.627695969638854e-08,
+      "loss": 0.0192,
+      "step": 2063
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.5409236685277608e-08,
+      "loss": 0.0066,
+      "step": 2064
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 1.4565262813230896e-08,
+      "loss": 0.0085,
+      "step": 2065
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.374504008770483e-08,
+      "loss": 0.0098,
+      "step": 2066
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.2948570459667687e-08,
+      "loss": 0.0288,
+      "step": 2067
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.2175855823581829e-08,
+      "loss": 0.0113,
+      "step": 2068
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 1.1426898017412591e-08,
+      "loss": 0.0075,
+      "step": 2069
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.070169882261496e-08,
+      "loss": 0.0107,
+      "step": 2070
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 1.0000259964132453e-08,
+      "loss": 0.0466,
+      "step": 2071
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 9.322583110392692e-09,
+      "loss": 0.0122,
+      "step": 2072
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 8.668669873304058e-09,
+      "loss": 0.0277,
+      "step": 2073
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 8.038521808249045e-09,
+      "loss": 0.0047,
+      "step": 2074
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 7.432140414087574e-09,
+      "loss": 0.0137,
+      "step": 2075
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 6.849527133144795e-09,
+      "loss": 0.0054,
+      "step": 2076
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 6.29068335121108e-09,
+      "loss": 0.0138,
+      "step": 2077
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.7556103975420215e-09,
+      "loss": 0.0063,
+      "step": 2078
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 5.2443095448506674e-09,
+      "loss": 0.0023,
+      "step": 2079
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.756782009306404e-09,
+      "loss": 0.008,
+      "step": 2080
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 4.293028950532741e-09,
+      "loss": 0.0018,
+      "step": 2081
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 3.853051471601754e-09,
+      "loss": 0.0103,
+      "step": 2082
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.4368506190352035e-09,
+      "loss": 0.0027,
+      "step": 2083
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 3.0444273828000857e-09,
+      "loss": 0.0023,
+      "step": 2084
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.675782696305307e-09,
+      "loss": 0.0046,
+      "step": 2085
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 2.330917436402791e-09,
+      "loss": 0.0036,
+      "step": 2086
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 2.00983242338082e-09,
+      "loss": 0.0069,
+      "step": 2087
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.712528420966253e-09,
+      "loss": 0.0127,
+      "step": 2088
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.4390061363189767e-09,
+      "loss": 0.0065,
+      "step": 2089
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.1892662200341244e-09,
+      "loss": 0.0143,
+      "step": 2090
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 9.633092661376353e-10,
+      "loss": 0.0137,
+      "step": 2091
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 7.611358120862556e-10,
+      "loss": 0.0321,
+      "step": 2092
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 5.827463387653165e-10,
+      "loss": 0.0037,
+      "step": 2093
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 4.2814127048873553e-10,
+      "loss": 0.007,
+      "step": 2094
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.9732097499679536e-10,
+      "loss": 0.0193,
+      "step": 2095
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.902857634550337e-10,
+      "loss": 0.0021,
+      "step": 2096
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 1.070358904564639e-10,
+      "loss": 0.0125,
+      "step": 2097
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 4.757155401602376e-11,
+      "loss": 0.016,
+      "step": 2098
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.1892895576126606e-11,
+      "loss": 0.0053,
+      "step": 2099
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.0,
+      "loss": 0.008,
+      "step": 2100
+    },
+    {
+      "epoch": 5.0,
+      "step": 2100,
+      "total_flos": 2.044872638152573e+18,
+      "train_loss": 0.15093501650063054,
+      "train_runtime": 32559.9107,
+      "train_samples_per_second": 4.129,
+      "train_steps_per_second": 0.064
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 2100,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 5,
+  "save_steps": 300,
+  "total_flos": 2.044872638152573e+18,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}