diff --git "a/1B/20250802/trainer_state.json" "b/1B/20250802/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/1B/20250802/trainer_state.json"
@@ -0,0 +1,12030 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 500,
+  "global_step": 2000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6666666666666668e-07,
+      "loss": 2.0515,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.3333333333333335e-07,
+      "loss": 2.0566,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.000000000000001e-07,
+      "loss": 2.0623,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.666666666666667e-07,
+      "loss": 2.0472,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.333333333333333e-07,
+      "loss": 2.0446,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0000000000000002e-06,
+      "loss": 2.0509,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1666666666666668e-06,
+      "loss": 2.0109,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3333333333333334e-06,
+      "loss": 2.0202,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5e-06,
+      "loss": 1.9422,
+      "step": 9
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6666666666666667e-06,
+      "loss": 1.9279,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.8333333333333333e-06,
+      "loss": 1.9032,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 1.7856,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.166666666666667e-06,
+      "loss": 1.7715,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3333333333333336e-06,
+      "loss": 1.7619,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5e-06,
+      "loss": 1.6709,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.666666666666667e-06,
+      "loss": 1.6379,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8333333333333335e-06,
+      "loss": 1.6177,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3e-06,
+      "loss": 1.5808,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.1666666666666667e-06,
+      "loss": 1.5683,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3333333333333333e-06,
+      "loss": 1.5503,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.5e-06,
+      "loss": 1.5162,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6666666666666666e-06,
+      "loss": 1.5339,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.833333333333334e-06,
+      "loss": 1.4775,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 1.4577,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.166666666666667e-06,
+      "loss": 1.4375,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.333333333333334e-06,
+      "loss": 1.3869,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.5e-06,
+      "loss": 1.3853,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.666666666666667e-06,
+      "loss": 1.3675,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.833333333333333e-06,
+      "loss": 1.3401,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5e-06,
+      "loss": 1.2995,
+      "step": 30
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.1666666666666675e-06,
+      "loss": 1.2877,
+      "step": 31
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.333333333333334e-06,
+      "loss": 1.2621,
+      "step": 32
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.500000000000001e-06,
+      "loss": 1.2547,
+      "step": 33
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.666666666666667e-06,
+      "loss": 1.223,
+      "step": 34
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.833333333333334e-06,
+      "loss": 1.233,
+      "step": 35
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6e-06,
+      "loss": 1.2166,
+      "step": 36
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.166666666666667e-06,
+      "loss": 1.2248,
+      "step": 37
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.333333333333333e-06,
+      "loss": 1.2029,
+      "step": 38
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.5000000000000004e-06,
+      "loss": 1.1829,
+      "step": 39
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 1.1727,
+      "step": 40
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.833333333333334e-06,
+      "loss": 1.1787,
+      "step": 41
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7e-06,
+      "loss": 1.1532,
+      "step": 42
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.166666666666667e-06,
+      "loss": 1.1585,
+      "step": 43
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.333333333333333e-06,
+      "loss": 1.1418,
+      "step": 44
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.500000000000001e-06,
+      "loss": 1.1418,
+      "step": 45
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.666666666666667e-06,
+      "loss": 1.1256,
+      "step": 46
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.833333333333333e-06,
+      "loss": 1.1125,
+      "step": 47
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 1.1149,
+      "step": 48
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.166666666666668e-06,
+      "loss": 1.1131,
+      "step": 49
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.333333333333334e-06,
+      "loss": 1.1002,
+      "step": 50
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.5e-06,
+      "loss": 1.0852,
+      "step": 51
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.666666666666668e-06,
+      "loss": 1.0897,
+      "step": 52
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.833333333333334e-06,
+      "loss": 1.0695,
+      "step": 53
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9e-06,
+      "loss": 1.0799,
+      "step": 54
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.166666666666666e-06,
+      "loss": 1.0798,
+      "step": 55
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.333333333333334e-06,
+      "loss": 1.0747,
+      "step": 56
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.5e-06,
+      "loss": 1.0633,
+      "step": 57
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.666666666666667e-06,
+      "loss": 1.0686,
+      "step": 58
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.833333333333333e-06,
+      "loss": 1.0641,
+      "step": 59
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1e-05,
+      "loss": 1.0493,
+      "step": 60
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999993444041447e-06,
+      "loss": 1.0418,
+      "step": 61
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99997377618298e-06,
+      "loss": 1.0428,
+      "step": 62
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999940996476176e-06,
+      "loss": 1.0294,
+      "step": 63
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999895105006995e-06,
+      "loss": 1.0282,
+      "step": 64
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999836101895782e-06,
+      "loss": 1.0151,
+      "step": 65
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999763987297266e-06,
+      "loss": 1.0235,
+      "step": 66
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999678761400562e-06,
+      "loss": 1.0185,
+      "step": 67
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99958042442916e-06,
+      "loss": 1.0082,
+      "step": 68
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99946897664094e-06,
+      "loss": 1.0007,
+      "step": 69
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999344418328161e-06,
+      "loss": 0.9926,
+      "step": 70
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999206749817464e-06,
+      "loss": 0.984,
+      "step": 71
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999055971469864e-06,
+      "loss": 0.9843,
+      "step": 72
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998892083680763e-06,
+      "loss": 0.9904,
+      "step": 73
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998715086879938e-06,
+      "loss": 1.0037,
+      "step": 74
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99852498153154e-06,
+      "loss": 0.9654,
+      "step": 75
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998321768134101e-06,
+      "loss": 0.9702,
+      "step": 76
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998105447220523e-06,
+      "loss": 0.9762,
+      "step": 77
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997876019358083e-06,
+      "loss": 0.9678,
+      "step": 78
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997633485148428e-06,
+      "loss": 0.966,
+      "step": 79
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997377845227577e-06,
+      "loss": 0.9567,
+      "step": 80
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997109100265912e-06,
+      "loss": 0.9671,
+      "step": 81
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99682725096819e-06,
+      "loss": 0.9652,
+      "step": 82
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996532298073526e-06,
+      "loss": 0.9552,
+      "step": 83
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.9962242423554e-06,
+      "loss": 0.9444,
+      "step": 84
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995903084621651e-06,
+      "loss": 0.9568,
+      "step": 85
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995568825714479e-06,
+      "loss": 0.9501,
+      "step": 86
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99522146651044e-06,
+      "loss": 0.9551,
+      "step": 87
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99486100792044e-06,
+      "loss": 0.9428,
+      "step": 88
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994487450889741e-06,
+      "loss": 0.9373,
+      "step": 89
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994100796397954e-06,
+      "loss": 0.941,
+      "step": 90
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993701045459035e-06,
+      "loss": 0.9331,
+      "step": 91
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993288199121283e-06,
+      "loss": 0.9457,
+      "step": 92
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.99286225846734e-06,
+      "loss": 0.9401,
+      "step": 93
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992423224614185e-06,
+      "loss": 0.9465,
+      "step": 94
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991971098713136e-06,
+      "loss": 0.9498,
+      "step": 95
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991505881949837e-06,
+      "loss": 0.9355,
+      "step": 96
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991027575544265e-06,
+      "loss": 0.939,
+      "step": 97
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990536180750724e-06,
+      "loss": 0.927,
+      "step": 98
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990031698857842e-06,
+      "loss": 0.9289,
+      "step": 99
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98951413118856e-06,
+      "loss": 0.9306,
+      "step": 100
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.988983479100138e-06,
+      "loss": 0.9342,
+      "step": 101
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.988439743984155e-06,
+      "loss": 0.9321,
+      "step": 102
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987882927266487e-06,
+      "loss": 0.9387,
+      "step": 103
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987313030407325e-06,
+      "loss": 0.9247,
+      "step": 104
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.986730054901153e-06,
+      "loss": 0.9367,
+      "step": 105
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98613400227676e-06,
+      "loss": 0.9343,
+      "step": 106
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985524874097224e-06,
+      "loss": 0.9211,
+      "step": 107
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.984902671959911e-06,
+      "loss": 0.934,
+      "step": 108
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.984267397496474e-06,
+      "loss": 0.9265,
+      "step": 109
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.983619052372847e-06,
+      "loss": 0.9344,
+      "step": 110
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.98295763828924e-06,
+      "loss": 0.9178,
+      "step": 111
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.982283156980133e-06,
+      "loss": 0.9398,
+      "step": 112
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.981595610214276e-06,
+      "loss": 0.9275,
+      "step": 113
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.980894999794678e-06,
+      "loss": 0.9322,
+      "step": 114
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.98018132755861e-06,
+      "loss": 0.9153,
+      "step": 115
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.979454595377594e-06,
+      "loss": 0.9352,
+      "step": 116
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.978714805157398e-06,
+      "loss": 0.9195,
+      "step": 117
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97796195883804e-06,
+      "loss": 0.9197,
+      "step": 118
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97719605839377e-06,
+      "loss": 0.9209,
+      "step": 119
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97641710583307e-06,
+      "loss": 0.9083,
+      "step": 120
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.975625103198655e-06,
+      "loss": 0.9147,
+      "step": 121
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97482005256746e-06,
+      "loss": 0.9106,
+      "step": 122
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.974001956050637e-06,
+      "loss": 0.9069,
+      "step": 123
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.973170815793543e-06,
+      "loss": 0.9144,
+      "step": 124
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.972326633975754e-06,
+      "loss": 0.9123,
+      "step": 125
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.971469412811032e-06,
+      "loss": 0.9249,
+      "step": 126
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.970599154547344e-06,
+      "loss": 0.915,
+      "step": 127
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.969715861466839e-06,
+      "loss": 0.9113,
+      "step": 128
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.968819535885852e-06,
+      "loss": 0.9144,
+      "step": 129
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.96791018015489e-06,
+      "loss": 0.9105,
+      "step": 130
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.966987796658631e-06,
+      "loss": 0.9102,
+      "step": 131
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.966052387815923e-06,
+      "loss": 0.9188,
+      "step": 132
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.965103956079764e-06,
+      "loss": 0.9123,
+      "step": 133
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.964142503937305e-06,
+      "loss": 0.9217,
+      "step": 134
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.963168033909843e-06,
+      "loss": 0.9212,
+      "step": 135
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.962180548552812e-06,
+      "loss": 0.9254,
+      "step": 136
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.961180050455776e-06,
+      "loss": 0.9122,
+      "step": 137
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.96016654224243e-06,
+      "loss": 0.8962,
+      "step": 138
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.959140026570572e-06,
+      "loss": 0.8944,
+      "step": 139
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.958100506132127e-06,
+      "loss": 0.908,
+      "step": 140
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.957047983653112e-06,
+      "loss": 0.9097,
+      "step": 141
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.955982461893648e-06,
+      "loss": 0.9033,
+      "step": 142
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.95490394364794e-06,
+      "loss": 0.8995,
+      "step": 143
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.953812431744274e-06,
+      "loss": 0.9104,
+      "step": 144
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.952707929045018e-06,
+      "loss": 0.8972,
+      "step": 145
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.951590438446597e-06,
+      "loss": 0.8971,
+      "step": 146
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.9504599628795e-06,
+      "loss": 0.8972,
+      "step": 147
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.94931650530827e-06,
+      "loss": 0.9078,
+      "step": 148
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.948160068731492e-06,
+      "loss": 0.9046,
+      "step": 149
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.946990656181782e-06,
+      "loss": 0.8947,
+      "step": 150
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.94580827072579e-06,
+      "loss": 0.9121,
+      "step": 151
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.944612915464183e-06,
+      "loss": 0.8971,
+      "step": 152
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.943404593531642e-06,
+      "loss": 0.9002,
+      "step": 153
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.942183308096853e-06,
+      "loss": 0.8943,
+      "step": 154
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.94094906236249e-06,
+      "loss": 0.8993,
+      "step": 155
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.93970185956522e-06,
+      "loss": 0.9009,
+      "step": 156
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.938441702975689e-06,
+      "loss": 0.9022,
+      "step": 157
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.93716859589851e-06,
+      "loss": 0.9065,
+      "step": 158
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.935882541672255e-06,
+      "loss": 0.8966,
+      "step": 159
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.934583543669454e-06,
+      "loss": 0.905,
+      "step": 160
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.933271605296577e-06,
+      "loss": 0.9035,
+      "step": 161
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.93194672999403e-06,
+      "loss": 0.8935,
+      "step": 162
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.930608921236145e-06,
+      "loss": 0.8892,
+      "step": 163
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.929258182531167e-06,
+      "loss": 0.9106,
+      "step": 164
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.927894517421252e-06,
+      "loss": 0.9023,
+      "step": 165
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.926517929482454e-06,
+      "loss": 0.8816,
+      "step": 166
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.925128422324711e-06,
+      "loss": 0.8934,
+      "step": 167
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.923725999591846e-06,
+      "loss": 0.8883,
+      "step": 168
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.92231066496155e-06,
+      "loss": 0.8905,
+      "step": 169
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.920882422145372e-06,
+      "loss": 0.8933,
+      "step": 170
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.919441274888713e-06,
+      "loss": 0.9009,
+      "step": 171
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.917987226970811e-06,
+      "loss": 0.8864,
+      "step": 172
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.91652028220474e-06,
+      "loss": 0.8899,
+      "step": 173
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.91504044443739e-06,
+      "loss": 0.8922,
+      "step": 174
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.913547717549463e-06,
+      "loss": 0.885,
+      "step": 175
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.912042105455462e-06,
+      "loss": 0.8833,
+      "step": 176
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.910523612103681e-06,
+      "loss": 0.8799,
+      "step": 177
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.908992241476189e-06,
+      "loss": 0.8855,
+      "step": 178
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.907447997588827e-06,
+      "loss": 0.8794,
+      "step": 179
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.905890884491196e-06,
+      "loss": 0.8879,
+      "step": 180
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.904320906266643e-06,
+      "loss": 0.883,
+      "step": 181
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.902738067032254e-06,
+      "loss": 0.8945,
+      "step": 182
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.901142370938837e-06,
+      "loss": 0.9047,
+      "step": 183
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.899533822170922e-06,
+      "loss": 0.8883,
+      "step": 184
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.897912424946741e-06,
+      "loss": 0.8833,
+      "step": 185
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.896278183518216e-06,
+      "loss": 0.8813,
+      "step": 186
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.894631102170959e-06,
+      "loss": 0.8897,
+      "step": 187
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.892971185224244e-06,
+      "loss": 0.876,
+      "step": 188
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.891298437031014e-06,
+      "loss": 0.8853,
+      "step": 189
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.889612861977855e-06,
+      "loss": 0.901,
+      "step": 190
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.887914464484988e-06,
+      "loss": 0.8839,
+      "step": 191
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.886203249006265e-06,
+      "loss": 0.8868,
+      "step": 192
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.88447922002915e-06,
+      "loss": 0.8829,
+      "step": 193
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.882742382074707e-06,
+      "loss": 0.8673,
+      "step": 194
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.88099273969759e-06,
+      "loss": 0.8664,
+      "step": 195
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.879230297486034e-06,
+      "loss": 0.8856,
+      "step": 196
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.877455060061838e-06,
+      "loss": 0.8779,
+      "step": 197
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.875667032080354e-06,
+      "loss": 0.8784,
+      "step": 198
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.873866218230477e-06,
+      "loss": 0.8854,
+      "step": 199
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.872052623234632e-06,
+      "loss": 0.867,
+      "step": 200
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.870226251848759e-06,
+      "loss": 0.8741,
+      "step": 201
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.868387108862307e-06,
+      "loss": 0.8843,
+      "step": 202
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.866535199098212e-06,
+      "loss": 0.8897,
+      "step": 203
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.864670527412891e-06,
+      "loss": 0.8847,
+      "step": 204
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.86279309869623e-06,
+      "loss": 0.8694,
+      "step": 205
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.860902917871566e-06,
+      "loss": 0.879,
+      "step": 206
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.858999989895678e-06,
+      "loss": 0.8607,
+      "step": 207
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.857084319758772e-06,
+      "loss": 0.8782,
+      "step": 208
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.855155912484471e-06,
+      "loss": 0.8766,
+      "step": 209
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.853214773129796e-06,
+      "loss": 0.8779,
+      "step": 210
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.851260906785161e-06,
+      "loss": 0.8795,
+      "step": 211
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.849294318574353e-06,
+      "loss": 0.876,
+      "step": 212
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.847315013654517e-06,
+      "loss": 0.8622,
+      "step": 213
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.845322997216153e-06,
+      "loss": 0.88,
+      "step": 214
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.843318274483089e-06,
+      "loss": 0.8711,
+      "step": 215
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.841300850712479e-06,
+      "loss": 0.8745,
+      "step": 216
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.839270731194782e-06,
+      "loss": 0.8768,
+      "step": 217
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.837227921253747e-06,
+      "loss": 0.8743,
+      "step": 218
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.835172426246407e-06,
+      "loss": 0.8828,
+      "step": 219
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.833104251563058e-06,
+      "loss": 0.8687,
+      "step": 220
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.831023402627244e-06,
+      "loss": 0.8678,
+      "step": 221
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.828929884895753e-06,
+      "loss": 0.8812,
+      "step": 222
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.82682370385859e-06,
+      "loss": 0.8677,
+      "step": 223
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.824704865038967e-06,
+      "loss": 0.8812,
+      "step": 224
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.822573373993296e-06,
+      "loss": 0.8689,
+      "step": 225
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.820429236311158e-06,
+      "loss": 0.8699,
+      "step": 226
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.81827245761531e-06,
+      "loss": 0.8819,
+      "step": 227
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.816103043561648e-06,
+      "loss": 0.8817,
+      "step": 228
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.813920999839211e-06,
+      "loss": 0.8728,
+      "step": 229
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.811726332170153e-06,
+      "loss": 0.8669,
+      "step": 230
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.809519046309733e-06,
+      "loss": 0.8767,
+      "step": 231
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.807299148046301e-06,
+      "loss": 0.8739,
+      "step": 232
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.805066643201283e-06,
+      "loss": 0.8788,
+      "step": 233
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.802821537629162e-06,
+      "loss": 0.8737,
+      "step": 234
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.800563837217465e-06,
+      "loss": 0.8648,
+      "step": 235
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.798293547886748e-06,
+      "loss": 0.8607,
+      "step": 236
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.796010675590582e-06,
+      "loss": 0.8603,
+      "step": 237
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.79371522631553e-06,
+      "loss": 0.8558,
+      "step": 238
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.791407206081144e-06,
+      "loss": 0.8662,
+      "step": 239
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.789086620939936e-06,
+      "loss": 0.8548,
+      "step": 240
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.78675347697737e-06,
+      "loss": 0.8627,
+      "step": 241
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.784407780311845e-06,
+      "loss": 0.8627,
+      "step": 242
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.782049537094676e-06,
+      "loss": 0.8752,
+      "step": 243
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.779678753510082e-06,
+      "loss": 0.8578,
+      "step": 244
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.777295435775165e-06,
+      "loss": 0.851,
+      "step": 245
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.774899590139897e-06,
+      "loss": 0.8561,
+      "step": 246
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.772491222887108e-06,
+      "loss": 0.8551,
+      "step": 247
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.770070340332457e-06,
+      "loss": 0.8582,
+      "step": 248
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.767636948824428e-06,
+      "loss": 0.8644,
+      "step": 249
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.765191054744305e-06,
+      "loss": 0.8748,
+      "step": 250
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.762732664506161e-06,
+      "loss": 0.8676,
+      "step": 251
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.76026178455684e-06,
+      "loss": 0.8745,
+      "step": 252
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.757778421375931e-06,
+      "loss": 0.8712,
+      "step": 253
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.755282581475769e-06,
+      "loss": 0.8756,
+      "step": 254
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.752774271401403e-06,
+      "loss": 0.8694,
+      "step": 255
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.75025349773058e-06,
+      "loss": 0.8409,
+      "step": 256
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.74772026707374e-06,
+      "loss": 0.862,
+      "step": 257
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.745174586073982e-06,
+      "loss": 0.8625,
+      "step": 258
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.742616461407059e-06,
+      "loss": 0.86,
+      "step": 259
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.740045899781353e-06,
+      "loss": 0.8584,
+      "step": 260
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.737462907937864e-06,
+      "loss": 0.8663,
+      "step": 261
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.734867492650187e-06,
+      "loss": 0.8564,
+      "step": 262
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.732259660724494e-06,
+      "loss": 0.8579,
+      "step": 263
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.729639418999524e-06,
+      "loss": 0.867,
+      "step": 264
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.727006774346551e-06,
+      "loss": 0.8584,
+      "step": 265
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.724361733669383e-06,
+      "loss": 0.8601,
+      "step": 266
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.721704303904325e-06,
+      "loss": 0.8609,
+      "step": 267
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.719034492020183e-06,
+      "loss": 0.8597,
+      "step": 268
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.716352305018224e-06,
+      "loss": 0.8654,
+      "step": 269
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.713657749932172e-06,
+      "loss": 0.8596,
+      "step": 270
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.710950833828182e-06,
+      "loss": 0.8716,
+      "step": 271
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.708231563804828e-06,
+      "loss": 0.8497,
+      "step": 272
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.705499946993078e-06,
+      "loss": 0.8633,
+      "step": 273
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.702755990556277e-06,
+      "loss": 0.876,
+      "step": 274
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.699999701690133e-06,
+      "loss": 0.8614,
+      "step": 275
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.697231087622691e-06,
+      "loss": 0.8611,
+      "step": 276
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.694450155614318e-06,
+      "loss": 0.8597,
+      "step": 277
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.691656912957686e-06,
+      "loss": 0.8523,
+      "step": 278
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.688851366977747e-06,
+      "loss": 0.8591,
+      "step": 279
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.68603352503172e-06,
+      "loss": 0.8516,
+      "step": 280
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.683203394509064e-06,
+      "loss": 0.868,
+      "step": 281
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.680360982831467e-06,
+      "loss": 0.8545,
+      "step": 282
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.677506297452823e-06,
+      "loss": 0.8581,
+      "step": 283
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.674639345859213e-06,
+      "loss": 0.8421,
+      "step": 284
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.671760135568882e-06,
+      "loss": 0.8557,
+      "step": 285
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.668868674132224e-06,
+      "loss": 0.8584,
+      "step": 286
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.665964969131757e-06,
+      "loss": 0.8686,
+      "step": 287
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.663049028182112e-06,
+      "loss": 0.8564,
+      "step": 288
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.660120858930003e-06,
+      "loss": 0.8649,
+      "step": 289
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.657180469054213e-06,
+      "loss": 0.8621,
+      "step": 290
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.65422786626557e-06,
+      "loss": 0.8415,
+      "step": 291
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.651263058306932e-06,
+      "loss": 0.8641,
+      "step": 292
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.648286052953161e-06,
+      "loss": 0.8508,
+      "step": 293
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.645296858011109e-06,
+      "loss": 0.843,
+      "step": 294
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.64229548131959e-06,
+      "loss": 0.87,
+      "step": 295
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.639281930749363e-06,
+      "loss": 0.8579,
+      "step": 296
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.636256214203115e-06,
+      "loss": 0.8479,
+      "step": 297
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.633218339615433e-06,
+      "loss": 0.851,
+      "step": 298
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.63016831495279e-06,
+      "loss": 0.8631,
+      "step": 299
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.627106148213521e-06,
+      "loss": 0.8479,
+      "step": 300
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.624031847427801e-06,
+      "loss": 0.858,
+      "step": 301
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.620945420657625e-06,
+      "loss": 0.8503,
+      "step": 302
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.617846875996785e-06,
+      "loss": 0.8606,
+      "step": 303
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.61473622157086e-06,
+      "loss": 0.8515,
+      "step": 304
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.61161346553717e-06,
+      "loss": 0.8512,
+      "step": 305
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.608478616084784e-06,
+      "loss": 0.8455,
+      "step": 306
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.605331681434478e-06,
+      "loss": 0.8509,
+      "step": 307
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.602172669838721e-06,
+      "loss": 0.8527,
+      "step": 308
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.599001589581654e-06,
+      "loss": 0.851,
+      "step": 309
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.595818448979061e-06,
+      "loss": 0.846,
+      "step": 310
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.592623256378364e-06,
+      "loss": 0.8567,
+      "step": 311
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.589416020158577e-06,
+      "loss": 0.8448,
+      "step": 312
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.586196748730305e-06,
+      "loss": 0.8355,
+      "step": 313
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.582965450535716e-06,
+      "loss": 0.8503,
+      "step": 314
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.579722134048506e-06,
+      "loss": 0.8477,
+      "step": 315
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.5764668077739e-06,
+      "loss": 0.8355,
+      "step": 316
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.573199480248607e-06,
+      "loss": 0.8441,
+      "step": 317
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.569920160040815e-06,
+      "loss": 0.8411,
+      "step": 318
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.566628855750159e-06,
+      "loss": 0.8501,
+      "step": 319
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.563325576007702e-06,
+      "loss": 0.8509,
+      "step": 320
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.560010329475907e-06,
+      "loss": 0.8432,
+      "step": 321
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.556683124848624e-06,
+      "loss": 0.8429,
+      "step": 322
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.55334397085106e-06,
+      "loss": 0.8501,
+      "step": 323
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.549992876239753e-06,
+      "loss": 0.8449,
+      "step": 324
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.546629849802562e-06,
+      "loss": 0.8502,
+      "step": 325
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.54325490035863e-06,
+      "loss": 0.8548,
+      "step": 326
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.539868036758368e-06,
+      "loss": 0.8563,
+      "step": 327
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.536469267883432e-06,
+      "loss": 0.8571,
+      "step": 328
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.533058602646697e-06,
+      "loss": 0.8358,
+      "step": 329
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.529636049992235e-06,
+      "loss": 0.8501,
+      "step": 330
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.52620161889529e-06,
+      "loss": 0.8445,
+      "step": 331
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.52275531836226e-06,
+      "loss": 0.8362,
+      "step": 332
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.519297157430664e-06,
+      "loss": 0.8482,
+      "step": 333
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.515827145169128e-06,
+      "loss": 0.8529,
+      "step": 334
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.51234529067735e-06,
+      "loss": 0.8363,
+      "step": 335
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.508851603086094e-06,
+      "loss": 0.8338,
+      "step": 336
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.505346091557143e-06,
+      "loss": 0.8391,
+      "step": 337
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.501828765283295e-06,
+      "loss": 0.8203,
+      "step": 338
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.498299633488329e-06,
+      "loss": 0.8498,
+      "step": 339
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.494758705426978e-06,
+      "loss": 0.8525,
+      "step": 340
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.491205990384915e-06,
+      "loss": 0.8402,
+      "step": 341
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.487641497678724e-06,
+      "loss": 0.8412,
+      "step": 342
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.484065236655866e-06,
+      "loss": 0.8445,
+      "step": 343
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.480477216694674e-06,
+      "loss": 0.8439,
+      "step": 344
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.476877447204308e-06,
+      "loss": 0.8345,
+      "step": 345
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.473265937624748e-06,
+      "loss": 0.8438,
+      "step": 346
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.46964269742675e-06,
+      "loss": 0.8547,
+      "step": 347
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.466007736111846e-06,
+      "loss": 0.8452,
+      "step": 348
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.462361063212295e-06,
+      "loss": 0.8472,
+      "step": 349
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.458702688291072e-06,
+      "loss": 0.8433,
+      "step": 350
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.45503262094184e-06,
+      "loss": 0.8525,
+      "step": 351
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.451350870788922e-06,
+      "loss": 0.8442,
+      "step": 352
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.447657447487276e-06,
+      "loss": 0.8444,
+      "step": 353
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.443952360722477e-06,
+      "loss": 0.8365,
+      "step": 354
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.440235620210682e-06,
+      "loss": 0.8424,
+      "step": 355
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.436507235698613e-06,
+      "loss": 0.8489,
+      "step": 356
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.432767216963517e-06,
+      "loss": 0.8455,
+      "step": 357
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.429015573813163e-06,
+      "loss": 0.8558,
+      "step": 358
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.425252316085796e-06,
+      "loss": 0.8453,
+      "step": 359
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.421477453650118e-06,
+      "loss": 0.8383,
+      "step": 360
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.41769099640527e-06,
+      "loss": 0.8481,
+      "step": 361
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.413892954280793e-06,
+      "loss": 0.8452,
+      "step": 362
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.410083337236608e-06,
+      "loss": 0.8284,
+      "step": 363
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.406262155262995e-06,
+      "loss": 0.8386,
+      "step": 364
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.402429418380555e-06,
+      "loss": 0.8372,
+      "step": 365
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.398585136640195e-06,
+      "loss": 0.8399,
+      "step": 366
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.394729320123096e-06,
+      "loss": 0.8366,
+      "step": 367
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.390861978940687e-06,
+      "loss": 0.8347,
+      "step": 368
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.386983123234619e-06,
+      "loss": 0.8462,
+      "step": 369
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.38309276317674e-06,
+      "loss": 0.8301,
+      "step": 370
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.379190908969063e-06,
+      "loss": 0.8342,
+      "step": 371
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.37527757084375e-06,
+      "loss": 0.8317,
+      "step": 372
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.37135275906307e-06,
+      "loss": 0.8389,
+      "step": 373
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.367416483919387e-06,
+      "loss": 0.8377,
+      "step": 374
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.363468755735123e-06,
+      "loss": 0.8394,
+      "step": 375
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.359509584862735e-06,
+      "loss": 0.839,
+      "step": 376
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.355538981684688e-06,
+      "loss": 0.8372,
+      "step": 377
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.351556956613423e-06,
+      "loss": 0.8437,
+      "step": 378
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.347563520091339e-06,
+      "loss": 0.8348,
+      "step": 379
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.343558682590757e-06,
+      "loss": 0.8472,
+      "step": 380
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.339542454613895e-06,
+      "loss": 0.8233,
+      "step": 381
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.335514846692846e-06,
+      "loss": 0.8381,
+      "step": 382
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.331475869389539e-06,
+      "loss": 0.8411,
+      "step": 383
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.327425533295725e-06,
+      "loss": 0.8354,
+      "step": 384
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.323363849032933e-06,
+      "loss": 0.8366,
+      "step": 385
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.31929082725246e-06,
+      "loss": 0.849,
+      "step": 386
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.31520647863533e-06,
+      "loss": 0.8483,
+      "step": 387
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.31111081389227e-06,
+      "loss": 0.8395,
+      "step": 388
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.307003843763687e-06,
+      "loss": 0.8337,
+      "step": 389
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.302885579019626e-06,
+      "loss": 0.8352,
+      "step": 390
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.298756030459762e-06,
+      "loss": 0.8542,
+      "step": 391
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.29461520891335e-06,
+      "loss": 0.8448,
+      "step": 392
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.290463125239212e-06,
+      "loss": 0.8261,
+      "step": 393
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.286299790325708e-06,
+      "loss": 0.8372,
+      "step": 394
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.282125215090694e-06,
+      "loss": 0.8238,
+      "step": 395
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.277939410481507e-06,
+      "loss": 0.8328,
+      "step": 396
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.273742387474933e-06,
+      "loss": 0.8243,
+      "step": 397
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.269534157077177e-06,
+      "loss": 0.832,
+      "step": 398
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.265314730323829e-06,
+      "loss": 0.8434,
+      "step": 399
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.261084118279846e-06,
+      "loss": 0.8209,
+      "step": 400
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.256842332039516e-06,
+      "loss": 0.8427,
+      "step": 401
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.252589382726426e-06,
+      "loss": 0.8452,
+      "step": 402
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.248325281493444e-06,
+      "loss": 0.8184,
+      "step": 403
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.244050039522673e-06,
+      "loss": 0.8384,
+      "step": 404
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.23976366802544e-06,
+      "loss": 0.8282,
+      "step": 405
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.235466178242255e-06,
+      "loss": 0.8382,
+      "step": 406
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.231157581442781e-06,
+      "loss": 0.8307,
+      "step": 407
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.226837888925813e-06,
+      "loss": 0.8301,
+      "step": 408
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.222507112019243e-06,
+      "loss": 0.8342,
+      "step": 409
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.218165262080024e-06,
+      "loss": 0.8264,
+      "step": 410
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.213812350494154e-06,
+      "loss": 0.8432,
+      "step": 411
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.209448388676636e-06,
+      "loss": 0.8252,
+      "step": 412
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.20507338807145e-06,
+      "loss": 0.8435,
+      "step": 413
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.200687360151527e-06,
+      "loss": 0.8294,
+      "step": 414
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.196290316418712e-06,
+      "loss": 0.8278,
+      "step": 415
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.191882268403743e-06,
+      "loss": 0.8134,
+      "step": 416
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.187463227666205e-06,
+      "loss": 0.8245,
+      "step": 417
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.183033205794525e-06,
+      "loss": 0.8378,
+      "step": 418
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.178592214405915e-06,
+      "loss": 0.8381,
+      "step": 419
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.174140265146356e-06,
+      "loss": 0.8459,
+      "step": 420
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.169677369690569e-06,
+      "loss": 0.8304,
+      "step": 421
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.165203539741976e-06,
+      "loss": 0.8292,
+      "step": 422
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.160718787032673e-06,
+      "loss": 0.8273,
+      "step": 423
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.156223123323405e-06,
+      "loss": 0.8453,
+      "step": 424
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.151716560403521e-06,
+      "loss": 0.8373,
+      "step": 425
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.14719911009096e-06,
+      "loss": 0.8323,
+      "step": 426
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.142670784232208e-06,
+      "loss": 0.8303,
+      "step": 427
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.13813159470227e-06,
+      "loss": 0.8238,
+      "step": 428
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.133581553404644e-06,
+      "loss": 0.8234,
+      "step": 429
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.129020672271283e-06,
+      "loss": 0.8263,
+      "step": 430
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.124448963262564e-06,
+      "loss": 0.818,
+      "step": 431
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.119866438367263e-06,
+      "loss": 0.8333,
+      "step": 432
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.115273109602516e-06,
+      "loss": 0.8437,
+      "step": 433
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.11066898901379e-06,
+      "loss": 0.8318,
+      "step": 434
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.106054088674861e-06,
+      "loss": 0.837,
+      "step": 435
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.101428420687759e-06,
+      "loss": 0.8285,
+      "step": 436
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.096791997182766e-06,
+      "loss": 0.8318,
+      "step": 437
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.092144830318357e-06,
+      "loss": 0.8341,
+      "step": 438
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.08748693228119e-06,
+      "loss": 0.8329,
+      "step": 439
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.082818315286054e-06,
+      "loss": 0.8332,
+      "step": 440
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.07813899157586e-06,
+      "loss": 0.8409,
+      "step": 441
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.073448973421581e-06,
+      "loss": 0.8408,
+      "step": 442
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.06874827312225e-06,
+      "loss": 0.8351,
+      "step": 443
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.0640369030049e-06,
+      "loss": 0.817,
+      "step": 444
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.059314875424553e-06,
+      "loss": 0.8363,
+      "step": 445
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.054582202764175e-06,
+      "loss": 0.8313,
+      "step": 446
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.049838897434648e-06,
+      "loss": 0.8353,
+      "step": 447
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.045084971874738e-06,
+      "loss": 0.8252,
+      "step": 448
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.04032043855106e-06,
+      "loss": 0.8236,
+      "step": 449
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.035545309958048e-06,
+      "loss": 0.828,
+      "step": 450
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.030759598617918e-06,
+      "loss": 0.8328,
+      "step": 451
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.025963317080641e-06,
+      "loss": 0.8357,
+      "step": 452
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.02115647792391e-06,
+      "loss": 0.8334,
+      "step": 453
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.016339093753093e-06,
+      "loss": 0.8406,
+      "step": 454
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.011511177201226e-06,
+      "loss": 0.8306,
+      "step": 455
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.006672740928952e-06,
+      "loss": 0.8288,
+      "step": 456
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.001823797624508e-06,
+      "loss": 0.8467,
+      "step": 457
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.99696436000368e-06,
+      "loss": 0.8177,
+      "step": 458
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.992094440809781e-06,
+      "loss": 0.824,
+      "step": 459
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.987214052813605e-06,
+      "loss": 0.8147,
+      "step": 460
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.9823232088134e-06,
+      "loss": 0.8235,
+      "step": 461
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.977421921634833e-06,
+      "loss": 0.8244,
+      "step": 462
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.97251020413096e-06,
+      "loss": 0.829,
+      "step": 463
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.967588069182184e-06,
+      "loss": 0.834,
+      "step": 464
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.962655529696236e-06,
+      "loss": 0.8226,
+      "step": 465
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.957712598608123e-06,
+      "loss": 0.8284,
+      "step": 466
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.952759288880103e-06,
+      "loss": 0.819,
+      "step": 467
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.947795613501658e-06,
+      "loss": 0.8338,
+      "step": 468
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.942821585489445e-06,
+      "loss": 0.8314,
+      "step": 469
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.937837217887273e-06,
+      "loss": 0.8247,
+      "step": 470
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.932842523766066e-06,
+      "loss": 0.8284,
+      "step": 471
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.927837516223824e-06,
+      "loss": 0.8235,
+      "step": 472
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.9228222083856e-06,
+      "loss": 0.8333,
+      "step": 473
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.917796613403451e-06,
+      "loss": 0.8307,
+      "step": 474
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.912760744456416e-06,
+      "loss": 0.8249,
+      "step": 475
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.907714614750473e-06,
+      "loss": 0.8304,
+      "step": 476
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.902658237518508e-06,
+      "loss": 0.8158,
+      "step": 477
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.897591626020284e-06,
+      "loss": 0.8324,
+      "step": 478
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.892514793542396e-06,
+      "loss": 0.8169,
+      "step": 479
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.887427753398249e-06,
+      "loss": 0.8193,
+      "step": 480
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.882330518928006e-06,
+      "loss": 0.8223,
+      "step": 481
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.877223103498576e-06,
+      "loss": 0.8302,
+      "step": 482
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.87210552050356e-06,
+      "loss": 0.8278,
+      "step": 483
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.866977783363219e-06,
+      "loss": 0.8275,
+      "step": 484
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.861839905524452e-06,
+      "loss": 0.8334,
+      "step": 485
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.85669190046074e-06,
+      "loss": 0.8354,
+      "step": 486
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.851533781672126e-06,
+      "loss": 0.8188,
+      "step": 487
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.846365562685178e-06,
+      "loss": 0.8344,
+      "step": 488
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.841187257052945e-06,
+      "loss": 0.8335,
+      "step": 489
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.83599887835493e-06,
+      "loss": 0.8304,
+      "step": 490
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.830800440197056e-06,
+      "loss": 0.8276,
+      "step": 491
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.825591956211614e-06,
+      "loss": 0.8249,
+      "step": 492
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.820373440057252e-06,
+      "loss": 0.8272,
+      "step": 493
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.815144905418918e-06,
+      "loss": 0.8334,
+      "step": 494
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.809906366007832e-06,
+      "loss": 0.8337,
+      "step": 495
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.804657835561456e-06,
+      "loss": 0.8246,
+      "step": 496
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.79939932784345e-06,
+      "loss": 0.8273,
+      "step": 497
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.794130856643635e-06,
+      "loss": 0.8283,
+      "step": 498
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.788852435777962e-06,
+      "loss": 0.8273,
+      "step": 499
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.783564079088478e-06,
+      "loss": 0.8262,
+      "step": 500
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.778265800443277e-06,
+      "loss": 0.8288,
+      "step": 501
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.772957613736483e-06,
+      "loss": 0.8148,
+      "step": 502
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.76763953288819e-06,
+      "loss": 0.8302,
+      "step": 503
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.762311571844453e-06,
+      "loss": 0.8407,
+      "step": 504
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.756973744577221e-06,
+      "loss": 0.8348,
+      "step": 505
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.751626065084328e-06,
+      "loss": 0.8247,
+      "step": 506
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.74626854738944e-06,
+      "loss": 0.8035,
+      "step": 507
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.74090120554202e-06,
+      "loss": 0.8266,
+      "step": 508
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.7355240536173e-06,
+      "loss": 0.8215,
+      "step": 509
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.730137105716231e-06,
+      "loss": 0.8099,
+      "step": 510
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.724740375965458e-06,
+      "loss": 0.8095,
+      "step": 511
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.719333878517274e-06,
+      "loss": 0.8232,
+      "step": 512
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.713917627549587e-06,
+      "loss": 0.8075,
+      "step": 513
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.708491637265888e-06,
+      "loss": 0.813,
+      "step": 514
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.7030559218952e-06,
+      "loss": 0.823,
+      "step": 515
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.697610495692055e-06,
+      "loss": 0.8292,
+      "step": 516
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.692155372936448e-06,
+      "loss": 0.8294,
+      "step": 517
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.686690567933803e-06,
+      "loss": 0.8236,
+      "step": 518
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.681216095014931e-06,
+      "loss": 0.8191,
+      "step": 519
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.675731968536004e-06,
+      "loss": 0.8139,
+      "step": 520
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.6702382028785e-06,
+      "loss": 0.8147,
+      "step": 521
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.66473481244918e-06,
+      "loss": 0.8085,
+      "step": 522
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.659221811680046e-06,
+      "loss": 0.824,
+      "step": 523
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.653699215028298e-06,
+      "loss": 0.8235,
+      "step": 524
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.648167036976303e-06,
+      "loss": 0.8245,
+      "step": 525
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.64262529203155e-06,
+      "loss": 0.823,
+      "step": 526
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.637073994726623e-06,
+      "loss": 0.8183,
+      "step": 527
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.63151315961915e-06,
+      "loss": 0.8111,
+      "step": 528
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.625942801291774e-06,
+      "loss": 0.8175,
+      "step": 529
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.620362934352109e-06,
+      "loss": 0.8213,
+      "step": 530
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.614773573432705e-06,
+      "loss": 0.8089,
+      "step": 531
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.609174733191012e-06,
+      "loss": 0.8231,
+      "step": 532
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.603566428309334e-06,
+      "loss": 0.8217,
+      "step": 533
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.597948673494794e-06,
+      "loss": 0.8147,
+      "step": 534
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.592321483479305e-06,
+      "loss": 0.8223,
+      "step": 535
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.586684873019513e-06,
+      "loss": 0.8131,
+      "step": 536
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.581038856896773e-06,
+      "loss": 0.8274,
+      "step": 537
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.575383449917103e-06,
+      "loss": 0.8223,
+      "step": 538
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.569718666911152e-06,
+      "loss": 0.8151,
+      "step": 539
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.564044522734147e-06,
+      "loss": 0.8212,
+      "step": 540
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.558361032265875e-06,
+      "loss": 0.8206,
+      "step": 541
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.552668210410624e-06,
+      "loss": 0.8214,
+      "step": 542
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.546966072097156e-06,
+      "loss": 0.8245,
+      "step": 543
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.541254632278667e-06,
+      "loss": 0.8184,
+      "step": 544
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.535533905932739e-06,
+      "loss": 0.8188,
+      "step": 545
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.52980390806131e-06,
+      "loss": 0.8187,
+      "step": 546
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.524064653690633e-06,
+      "loss": 0.8224,
+      "step": 547
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.518316157871232e-06,
+      "loss": 0.8223,
+      "step": 548
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.512558435677868e-06,
+      "loss": 0.8125,
+      "step": 549
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.506791502209497e-06,
+      "loss": 0.8275,
+      "step": 550
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.501015372589227e-06,
+      "loss": 0.8206,
+      "step": 551
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.495230061964289e-06,
+      "loss": 0.8176,
+      "step": 552
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.489435585505981e-06,
+      "loss": 0.8193,
+      "step": 553
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.483631958409644e-06,
+      "loss": 0.819,
+      "step": 554
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.477819195894615e-06,
+      "loss": 0.8179,
+      "step": 555
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.471997313204183e-06,
+      "loss": 0.8053,
+      "step": 556
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.466166325605557e-06,
+      "loss": 0.8265,
+      "step": 557
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.460326248389825e-06,
+      "loss": 0.8287,
+      "step": 558
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.454477096871908e-06,
+      "loss": 0.8248,
+      "step": 559
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.448618886390523e-06,
+      "loss": 0.8071,
+      "step": 560
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.442751632308141e-06,
+      "loss": 0.8246,
+      "step": 561
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.436875350010958e-06,
+      "loss": 0.8199,
+      "step": 562
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.430990054908835e-06,
+      "loss": 0.8242,
+      "step": 563
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.425095762435274e-06,
+      "loss": 0.8314,
+      "step": 564
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.419192488047369e-06,
+      "loss": 0.8142,
+      "step": 565
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.41328024722577e-06,
+      "loss": 0.8101,
+      "step": 566
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.407359055474637e-06,
+      "loss": 0.8157,
+      "step": 567
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.401428928321607e-06,
+      "loss": 0.8164,
+      "step": 568
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.395489881317749e-06,
+      "loss": 0.82,
+      "step": 569
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.389541930037516e-06,
+      "loss": 0.8295,
+      "step": 570
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.383585090078721e-06,
+      "loss": 0.8181,
+      "step": 571
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.377619377062483e-06,
+      "loss": 0.8126,
+      "step": 572
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.371644806633186e-06,
+      "loss": 0.8125,
+      "step": 573
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.365661394458446e-06,
+      "loss": 0.8077,
+      "step": 574
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.359669156229062e-06,
+      "loss": 0.8166,
+      "step": 575
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.353668107658984e-06,
+      "loss": 0.8216,
+      "step": 576
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.347658264485258e-06,
+      "loss": 0.8176,
+      "step": 577
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.341639642468002e-06,
+      "loss": 0.8092,
+      "step": 578
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.335612257390346e-06,
+      "loss": 0.8227,
+      "step": 579
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.329576125058406e-06,
+      "loss": 0.8252,
+      "step": 580
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.323531261301237e-06,
+      "loss": 0.8102,
+      "step": 581
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.317477681970786e-06,
+      "loss": 0.8124,
+      "step": 582
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.311415402941865e-06,
+      "loss": 0.8098,
+      "step": 583
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.305344440112089e-06,
+      "loss": 0.8146,
+      "step": 584
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.29926480940185e-06,
+      "loss": 0.8225,
+      "step": 585
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.293176526754274e-06,
+      "loss": 0.811,
+      "step": 586
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.28707960813517e-06,
+      "loss": 0.8035,
+      "step": 587
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.280974069532999e-06,
+      "loss": 0.8207,
+      "step": 588
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.27485992695882e-06,
+      "loss": 0.8244,
+      "step": 589
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.268737196446264e-06,
+      "loss": 0.8218,
+      "step": 590
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.262605894051473e-06,
+      "loss": 0.819,
+      "step": 591
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.256466035853077e-06,
+      "loss": 0.8157,
+      "step": 592
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.250317637952137e-06,
+      "loss": 0.8141,
+      "step": 593
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.244160716472109e-06,
+      "loss": 0.8158,
+      "step": 594
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.237995287558801e-06,
+      "loss": 0.826,
+      "step": 595
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.231821367380335e-06,
+      "loss": 0.8084,
+      "step": 596
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.225638972127093e-06,
+      "loss": 0.8002,
+      "step": 597
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.219448118011687e-06,
+      "loss": 0.8144,
+      "step": 598
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.213248821268913e-06,
+      "loss": 0.8034,
+      "step": 599
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.207041098155701e-06,
+      "loss": 0.8174,
+      "step": 600
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.20082496495108e-06,
+      "loss": 0.809,
+      "step": 601
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.19460043795614e-06,
+      "loss": 0.8169,
+      "step": 602
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.18836753349397e-06,
+      "loss": 0.814,
+      "step": 603
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.182126267909642e-06,
+      "loss": 0.8296,
+      "step": 604
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.175876657570142e-06,
+      "loss": 0.8133,
+      "step": 605
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.16961871886435e-06,
+      "loss": 0.8099,
+      "step": 606
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.163352468202977e-06,
+      "loss": 0.8129,
+      "step": 607
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.157077922018537e-06,
+      "loss": 0.8018,
+      "step": 608
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.150795096765293e-06,
+      "loss": 0.8137,
+      "step": 609
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.144504008919224e-06,
+      "loss": 0.8274,
+      "step": 610
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.138204674977972e-06,
+      "loss": 0.8027,
+      "step": 611
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.13189711146081e-06,
+      "loss": 0.8232,
+      "step": 612
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.125581334908586e-06,
+      "loss": 0.8148,
+      "step": 613
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.119257361883686e-06,
+      "loss": 0.8091,
+      "step": 614
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.112925208969996e-06,
+      "loss": 0.8233,
+      "step": 615
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.106584892772844e-06,
+      "loss": 0.8203,
+      "step": 616
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.100236429918973e-06,
+      "loss": 0.8273,
+      "step": 617
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.093879837056486e-06,
+      "loss": 0.8167,
+      "step": 618
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.087515130854808e-06,
+      "loss": 0.8027,
+      "step": 619
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.081142328004638e-06,
+      "loss": 0.806,
+      "step": 620
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.074761445217908e-06,
+      "loss": 0.8025,
+      "step": 621
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.068372499227738e-06,
+      "loss": 0.8184,
+      "step": 622
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.061975506788398e-06,
+      "loss": 0.806,
+      "step": 623
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.055570484675252e-06,
+      "loss": 0.8146,
+      "step": 624
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.049157449684723e-06,
+      "loss": 0.8259,
+      "step": 625
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.042736418634252e-06,
+      "loss": 0.8209,
+      "step": 626
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.03630740836224e-06,
+      "loss": 0.8178,
+      "step": 627
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.029870435728018e-06,
+      "loss": 0.8086,
+      "step": 628
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.023425517611797e-06,
+      "loss": 0.8079,
+      "step": 629
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.016972670914624e-06,
+      "loss": 0.8251,
+      "step": 630
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.010511912558336e-06,
+      "loss": 0.8147,
+      "step": 631
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.004043259485519e-06,
+      "loss": 0.818,
+      "step": 632
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.99756672865946e-06,
+      "loss": 0.813,
+      "step": 633
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.99108233706411e-06,
+      "loss": 0.8104,
+      "step": 634
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.984590101704026e-06,
+      "loss": 0.8268,
+      "step": 635
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.978090039604342e-06,
+      "loss": 0.8178,
+      "step": 636
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.971582167810709e-06,
+      "loss": 0.826,
+      "step": 637
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.965066503389264e-06,
+      "loss": 0.8034,
+      "step": 638
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.95854306342658e-06,
+      "loss": 0.8226,
+      "step": 639
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.952011865029614e-06,
+      "loss": 0.8078,
+      "step": 640
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.945472925325674e-06,
+      "loss": 0.8239,
+      "step": 641
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.938926261462366e-06,
+      "loss": 0.8236,
+      "step": 642
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.932371890607556e-06,
+      "loss": 0.8086,
+      "step": 643
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.925809829949312e-06,
+      "loss": 0.8172,
+      "step": 644
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.919240096695877e-06,
+      "loss": 0.8234,
+      "step": 645
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.91266270807561e-06,
+      "loss": 0.816,
+      "step": 646
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.906077681336947e-06,
+      "loss": 0.8269,
+      "step": 647
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.89948503374835e-06,
+      "loss": 0.8171,
+      "step": 648
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.892884782598272e-06,
+      "loss": 0.8127,
+      "step": 649
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.886276945195098e-06,
+      "loss": 0.814,
+      "step": 650
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.879661538867116e-06,
+      "loss": 0.8145,
+      "step": 651
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.873038580962453e-06,
+      "loss": 0.8078,
+      "step": 652
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.866408088849048e-06,
+      "loss": 0.8139,
+      "step": 653
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.859770079914592e-06,
+      "loss": 0.8091,
+      "step": 654
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.853124571566491e-06,
+      "loss": 0.7965,
+      "step": 655
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.846471581231814e-06,
+      "loss": 0.8176,
+      "step": 656
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.839811126357254e-06,
+      "loss": 0.819,
+      "step": 657
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.833143224409076e-06,
+      "loss": 0.8161,
+      "step": 658
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.826467892873074e-06,
+      "loss": 0.8125,
+      "step": 659
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.819785149254534e-06,
+      "loss": 0.8094,
+      "step": 660
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.813095011078164e-06,
+      "loss": 0.7998,
+      "step": 661
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.806397495888074e-06,
+      "loss": 0.812,
+      "step": 662
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.799692621247718e-06,
+      "loss": 0.8151,
+      "step": 663
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.792980404739849e-06,
+      "loss": 0.8164,
+      "step": 664
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.786260863966468e-06,
+      "loss": 0.8084,
+      "step": 665
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.779534016548791e-06,
+      "loss": 0.8124,
+      "step": 666
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.77279988012719e-06,
+      "loss": 0.8054,
+      "step": 667
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.766058472361154e-06,
+      "loss": 0.807,
+      "step": 668
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.759309810929237e-06,
+      "loss": 0.8046,
+      "step": 669
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.752553913529019e-06,
+      "loss": 0.796,
+      "step": 670
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.745790797877051e-06,
+      "loss": 0.813,
+      "step": 671
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.739020481708816e-06,
+      "loss": 0.8103,
+      "step": 672
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.73224298277868e-06,
+      "loss": 0.806,
+      "step": 673
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.725458318859842e-06,
+      "loss": 0.8146,
+      "step": 674
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.718666507744293e-06,
+      "loss": 0.8078,
+      "step": 675
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.711867567242769e-06,
+      "loss": 0.8125,
+      "step": 676
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.705061515184693e-06,
+      "loss": 0.8055,
+      "step": 677
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.698248369418146e-06,
+      "loss": 0.8228,
+      "step": 678
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.691428147809811e-06,
+      "loss": 0.8121,
+      "step": 679
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.68460086824492e-06,
+      "loss": 0.8135,
+      "step": 680
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.67776654862722e-06,
+      "loss": 0.8106,
+      "step": 681
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.670925206878917e-06,
+      "loss": 0.8084,
+      "step": 682
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.66407686094063e-06,
+      "loss": 0.81,
+      "step": 683
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.657221528771352e-06,
+      "loss": 0.8119,
+      "step": 684
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.65035922834839e-06,
+      "loss": 0.8088,
+      "step": 685
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.643489977667327e-06,
+      "loss": 0.8125,
+      "step": 686
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.636613794741973e-06,
+      "loss": 0.8167,
+      "step": 687
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.629730697604314e-06,
+      "loss": 0.8071,
+      "step": 688
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.622840704304473e-06,
+      "loss": 0.7975,
+      "step": 689
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.61594383291065e-06,
+      "loss": 0.8026,
+      "step": 690
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.6090401015090905e-06,
+      "loss": 0.8034,
+      "step": 691
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.602129528204023e-06,
+      "loss": 0.8177,
+      "step": 692
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.5952121311176205e-06,
+      "loss": 0.7963,
+      "step": 693
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.588287928389952e-06,
+      "loss": 0.8128,
+      "step": 694
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.58135693817893e-06,
+      "loss": 0.8067,
+      "step": 695
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.574419178660269e-06,
+      "loss": 0.8062,
+      "step": 696
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.567474668027433e-06,
+      "loss": 0.8006,
+      "step": 697
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.560523424491595e-06,
+      "loss": 0.8063,
+      "step": 698
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.553565466281576e-06,
+      "loss": 0.8057,
+      "step": 699
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.546600811643816e-06,
+      "loss": 0.8076,
+      "step": 700
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.539629478842305e-06,
+      "loss": 0.808,
+      "step": 701
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.532651486158554e-06,
+      "loss": 0.8062,
+      "step": 702
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.525666851891534e-06,
+      "loss": 0.8034,
+      "step": 703
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.5186755943576324e-06,
+      "loss": 0.8034,
+      "step": 704
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.5116777318906105e-06,
+      "loss": 0.7973,
+      "step": 705
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.504673282841544e-06,
+      "loss": 0.8034,
+      "step": 706
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.4976622655787844e-06,
+      "loss": 0.7987,
+      "step": 707
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.490644698487909e-06,
+      "loss": 0.8042,
+      "step": 708
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.483620599971666e-06,
+      "loss": 0.8118,
+      "step": 709
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.476589988449939e-06,
+      "loss": 0.8089,
+      "step": 710
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.4695528823596845e-06,
+      "loss": 0.8026,
+      "step": 711
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.462509300154892e-06,
+      "loss": 0.8051,
+      "step": 712
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.455459260306537e-06,
+      "loss": 0.798,
+      "step": 713
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.448402781302526e-06,
+      "loss": 0.8097,
+      "step": 714
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.441339881647654e-06,
+      "loss": 0.7969,
+      "step": 715
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.434270579863549e-06,
+      "loss": 0.8211,
+      "step": 716
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.427194894488633e-06,
+      "loss": 0.8177,
+      "step": 717
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.420112844078066e-06,
+      "loss": 0.803,
+      "step": 718
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.413024447203698e-06,
+      "loss": 0.8087,
+      "step": 719
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.405929722454026e-06,
+      "loss": 0.8217,
+      "step": 720
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.398828688434137e-06,
+      "loss": 0.8081,
+      "step": 721
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.391721363765664e-06,
+      "loss": 0.7995,
+      "step": 722
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.384607767086738e-06,
+      "loss": 0.8084,
+      "step": 723
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.3774879170519386e-06,
+      "loss": 0.8088,
+      "step": 724
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.370361832332241e-06,
+      "loss": 0.8086,
+      "step": 725
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.363229531614973e-06,
+      "loss": 0.7998,
+      "step": 726
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.356091033603758e-06,
+      "loss": 0.8128,
+      "step": 727
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.348946357018479e-06,
+      "loss": 0.804,
+      "step": 728
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.341795520595216e-06,
+      "loss": 0.8099,
+      "step": 729
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.334638543086203e-06,
+      "loss": 0.8055,
+      "step": 730
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.327475443259781e-06,
+      "loss": 0.8046,
+      "step": 731
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.320306239900343e-06,
+      "loss": 0.809,
+      "step": 732
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.313130951808289e-06,
+      "loss": 0.8133,
+      "step": 733
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.305949597799976e-06,
+      "loss": 0.8005,
+      "step": 734
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.298762196707668e-06,
+      "loss": 0.8111,
+      "step": 735
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.291568767379484e-06,
+      "loss": 0.8099,
+      "step": 736
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.2843693286793575e-06,
+      "loss": 0.7913,
+      "step": 737
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.277163899486975e-06,
+      "loss": 0.806,
+      "step": 738
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.269952498697734e-06,
+      "loss": 0.8056,
+      "step": 739
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.262735145222696e-06,
+      "loss": 0.8121,
+      "step": 740
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.255511857988524e-06,
+      "loss": 0.7992,
+      "step": 741
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.248282655937451e-06,
+      "loss": 0.8027,
+      "step": 742
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.241047558027211e-06,
+      "loss": 0.8003,
+      "step": 743
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.233806583231012e-06,
+      "loss": 0.7989,
+      "step": 744
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.226559750537462e-06,
+      "loss": 0.7992,
+      "step": 745
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.219307078950536e-06,
+      "loss": 0.8089,
+      "step": 746
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.212048587489519e-06,
+      "loss": 0.8,
+      "step": 747
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.204784295188959e-06,
+      "loss": 0.8074,
+      "step": 748
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.197514221098616e-06,
+      "loss": 0.8118,
+      "step": 749
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.190238384283413e-06,
+      "loss": 0.8163,
+      "step": 750
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.182956803823381e-06,
+      "loss": 0.7944,
+      "step": 751
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.1756694988136165e-06,
+      "loss": 0.8025,
+      "step": 752
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.16837648836423e-06,
+      "loss": 0.8023,
+      "step": 753
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.161077791600288e-06,
+      "loss": 0.8051,
+      "step": 754
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.1537734276617744e-06,
+      "loss": 0.8027,
+      "step": 755
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.14646341570353e-06,
+      "loss": 0.795,
+      "step": 756
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.1391477748952095e-06,
+      "loss": 0.8173,
+      "step": 757
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.1318265244212305e-06,
+      "loss": 0.8145,
+      "step": 758
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.124499683480715e-06,
+      "loss": 0.809,
+      "step": 759
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.117167271287453e-06,
+      "loss": 0.8142,
+      "step": 760
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.109829307069837e-06,
+      "loss": 0.8041,
+      "step": 761
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.102485810070824e-06,
+      "loss": 0.8091,
+      "step": 762
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.095136799547881e-06,
+      "loss": 0.8022,
+      "step": 763
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.0877822947729265e-06,
+      "loss": 0.8046,
+      "step": 764
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.080422315032298e-06,
+      "loss": 0.8086,
+      "step": 765
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.073056879626681e-06,
+      "loss": 0.8147,
+      "step": 766
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.065686007871071e-06,
+      "loss": 0.7925,
+      "step": 767
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.05830971909472e-06,
+      "loss": 0.8004,
+      "step": 768
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.050928032641087e-06,
+      "loss": 0.8064,
+      "step": 769
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.043540967867782e-06,
+      "loss": 0.8014,
+      "step": 770
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.036148544146522e-06,
+      "loss": 0.794,
+      "step": 771
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.028750780863078e-06,
+      "loss": 0.8123,
+      "step": 772
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.021347697417218e-06,
+      "loss": 0.7975,
+      "step": 773
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.013939313222669e-06,
+      "loss": 0.8077,
+      "step": 774
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.006525647707054e-06,
+      "loss": 0.8018,
+      "step": 775
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.999106720311846e-06,
+      "loss": 0.8007,
+      "step": 776
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.991682550492317e-06,
+      "loss": 0.7955,
+      "step": 777
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.9842531577174865e-06,
+      "loss": 0.7969,
+      "step": 778
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.976818561470073e-06,
+      "loss": 0.8038,
+      "step": 779
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.969378781246436e-06,
+      "loss": 0.8078,
+      "step": 780
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.961933836556533e-06,
+      "loss": 0.8035,
+      "step": 781
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.954483746923865e-06,
+      "loss": 0.8083,
+      "step": 782
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.9470285318854205e-06,
+      "loss": 0.7981,
+      "step": 783
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.939568210991633e-06,
+      "loss": 0.7997,
+      "step": 784
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.932102803806324e-06,
+      "loss": 0.8138,
+      "step": 785
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.924632329906657e-06,
+      "loss": 0.8021,
+      "step": 786
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.917156808883073e-06,
+      "loss": 0.7938,
+      "step": 787
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.9096762603392595e-06,
+      "loss": 0.7931,
+      "step": 788
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.90219070389208e-06,
+      "loss": 0.8024,
+      "step": 789
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.894700159171535e-06,
+      "loss": 0.801,
+      "step": 790
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.8872046458207035e-06,
+      "loss": 0.799,
+      "step": 791
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.8797041834956955e-06,
+      "loss": 0.7893,
+      "step": 792
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.872198791865602e-06,
+      "loss": 0.8057,
+      "step": 793
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.8646884906124345e-06,
+      "loss": 0.801,
+      "step": 794
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.857173299431084e-06,
+      "loss": 0.8151,
+      "step": 795
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.849653238029261e-06,
+      "loss": 0.8069,
+      "step": 796
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.842128326127453e-06,
+      "loss": 0.7955,
+      "step": 797
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.834598583458862e-06,
+      "loss": 0.8062,
+      "step": 798
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.82706402976936e-06,
+      "loss": 0.809,
+      "step": 799
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.819524684817439e-06,
+      "loss": 0.8039,
+      "step": 800
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.811980568374148e-06,
+      "loss": 0.7905,
+      "step": 801
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.804431700223057e-06,
+      "loss": 0.8031,
+      "step": 802
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.7968781001601865e-06,
+      "loss": 0.8106,
+      "step": 803
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.78931978799398e-06,
+      "loss": 0.8006,
+      "step": 804
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.781756783545225e-06,
+      "loss": 0.7955,
+      "step": 805
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.774189106647021e-06,
+      "loss": 0.8091,
+      "step": 806
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.766616777144716e-06,
+      "loss": 0.8155,
+      "step": 807
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.7590398148958625e-06,
+      "loss": 0.8036,
+      "step": 808
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.751458239770163e-06,
+      "loss": 0.7991,
+      "step": 809
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.743872071649411e-06,
+      "loss": 0.8115,
+      "step": 810
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.736281330427451e-06,
+      "loss": 0.809,
+      "step": 811
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.728686036010115e-06,
+      "loss": 0.8038,
+      "step": 812
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.721086208315177e-06,
+      "loss": 0.7955,
+      "step": 813
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.7134818672723005e-06,
+      "loss": 0.8003,
+      "step": 814
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.705873032822981e-06,
+      "loss": 0.8018,
+      "step": 815
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.698259724920503e-06,
+      "loss": 0.8017,
+      "step": 816
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.690641963529876e-06,
+      "loss": 0.8104,
+      "step": 817
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.6830197686277945e-06,
+      "loss": 0.8045,
+      "step": 818
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.675393160202572e-06,
+      "loss": 0.7986,
+      "step": 819
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.667762158254104e-06,
+      "loss": 0.7983,
+      "step": 820
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.660126782793801e-06,
+      "loss": 0.8022,
+      "step": 821
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.652487053844544e-06,
+      "loss": 0.8083,
+      "step": 822
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.644842991440635e-06,
+      "loss": 0.7947,
+      "step": 823
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.637194615627733e-06,
+      "loss": 0.7961,
+      "step": 824
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.629541946462816e-06,
+      "loss": 0.7984,
+      "step": 825
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.621885004014113e-06,
+      "loss": 0.7936,
+      "step": 826
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.614223808361064e-06,
+      "loss": 0.8031,
+      "step": 827
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.6065583795942625e-06,
+      "loss": 0.8074,
+      "step": 828
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.5988887378154e-06,
+      "loss": 0.8096,
+      "step": 829
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.591214903137221e-06,
+      "loss": 0.8074,
+      "step": 830
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.583536895683458e-06,
+      "loss": 0.8067,
+      "step": 831
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.5758547355887944e-06,
+      "loss": 0.8052,
+      "step": 832
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.5681684429987956e-06,
+      "loss": 0.8064,
+      "step": 833
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.560478038069873e-06,
+      "loss": 0.7982,
+      "step": 834
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.552783540969212e-06,
+      "loss": 0.7903,
+      "step": 835
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.545084971874738e-06,
+      "loss": 0.8151,
+      "step": 836
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.537382350975048e-06,
+      "loss": 0.7999,
+      "step": 837
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.52967569846937e-06,
+      "loss": 0.8051,
+      "step": 838
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.521965034567499e-06,
+      "loss": 0.8023,
+      "step": 839
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.514250379489754e-06,
+      "loss": 0.8165,
+      "step": 840
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.506531753466919e-06,
+      "loss": 0.812,
+      "step": 841
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.49880917674019e-06,
+      "loss": 0.8103,
+      "step": 842
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.491082669561124e-06,
+      "loss": 0.8152,
+      "step": 843
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.483352252191585e-06,
+      "loss": 0.8092,
+      "step": 844
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.475617944903693e-06,
+      "loss": 0.7955,
+      "step": 845
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.467879767979764e-06,
+      "loss": 0.798,
+      "step": 846
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.460137741712269e-06,
+      "loss": 0.8028,
+      "step": 847
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.452391886403767e-06,
+      "loss": 0.804,
+      "step": 848
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.444642222366861e-06,
+      "loss": 0.7862,
+      "step": 849
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.436888769924142e-06,
+      "loss": 0.8147,
+      "step": 850
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.429131549408133e-06,
+      "loss": 0.7977,
+      "step": 851
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.421370581161244e-06,
+      "loss": 0.799,
+      "step": 852
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.413605885535705e-06,
+      "loss": 0.7918,
+      "step": 853
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.405837482893529e-06,
+      "loss": 0.8058,
+      "step": 854
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.398065393606445e-06,
+      "loss": 0.8037,
+      "step": 855
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.390289638055851e-06,
+      "loss": 0.7918,
+      "step": 856
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.38251023663276e-06,
+      "loss": 0.7958,
+      "step": 857
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.374727209737743e-06,
+      "loss": 0.8038,
+      "step": 858
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.366940577780885e-06,
+      "loss": 0.8059,
+      "step": 859
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.3591503611817155e-06,
+      "loss": 0.8031,
+      "step": 860
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.351356580369172e-06,
+      "loss": 0.7965,
+      "step": 861
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.343559255781538e-06,
+      "loss": 0.8096,
+      "step": 862
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.335758407866383e-06,
+      "loss": 0.8012,
+      "step": 863
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.3279540570805265e-06,
+      "loss": 0.8032,
+      "step": 864
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.320146223889966e-06,
+      "loss": 0.8011,
+      "step": 865
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.3123349287698345e-06,
+      "loss": 0.8027,
+      "step": 866
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.304520192204342e-06,
+      "loss": 0.8102,
+      "step": 867
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.296702034686726e-06,
+      "loss": 0.8027,
+      "step": 868
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.288880476719191e-06,
+      "loss": 0.7901,
+      "step": 869
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.281055538812861e-06,
+      "loss": 0.7991,
+      "step": 870
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.273227241487727e-06,
+      "loss": 0.8016,
+      "step": 871
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.265395605272581e-06,
+      "loss": 0.7965,
+      "step": 872
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.25756065070498e-06,
+      "loss": 0.8045,
+      "step": 873
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.249722398331177e-06,
+      "loss": 0.8015,
+      "step": 874
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.241880868706075e-06,
+      "loss": 0.7877,
+      "step": 875
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.234036082393171e-06,
+      "loss": 0.8067,
+      "step": 876
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.226188059964504e-06,
+      "loss": 0.8101,
+      "step": 877
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.218336822000598e-06,
+      "loss": 0.8088,
+      "step": 878
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.210482389090408e-06,
+      "loss": 0.7931,
+      "step": 879
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.202624781831269e-06,
+      "loss": 0.8028,
+      "step": 880
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.194764020828839e-06,
+      "loss": 0.7985,
+      "step": 881
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.18690012669705e-06,
+      "loss": 0.7843,
+      "step": 882
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.179033120058046e-06,
+      "loss": 0.7974,
+      "step": 883
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.171163021542134e-06,
+      "loss": 0.8004,
+      "step": 884
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.163289851787732e-06,
+      "loss": 0.8095,
+      "step": 885
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.155413631441307e-06,
+      "loss": 0.8092,
+      "step": 886
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.1475343811573305e-06,
+      "loss": 0.7904,
+      "step": 887
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.139652121598219e-06,
+      "loss": 0.7946,
+      "step": 888
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.131766873434276e-06,
+      "loss": 0.7843,
+      "step": 889
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.123878657343648e-06,
+      "loss": 0.8045,
+      "step": 890
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.115987494012261e-06,
+      "loss": 0.8088,
+      "step": 891
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.108093404133772e-06,
+      "loss": 0.8046,
+      "step": 892
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.100196408409511e-06,
+      "loss": 0.7992,
+      "step": 893
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.092296527548427e-06,
+      "loss": 0.8063,
+      "step": 894
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.084393782267039e-06,
+      "loss": 0.8051,
+      "step": 895
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.076488193289375e-06,
+      "loss": 0.801,
+      "step": 896
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.068579781346919e-06,
+      "loss": 0.806,
+      "step": 897
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.060668567178561e-06,
+      "loss": 0.8107,
+      "step": 898
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.052754571530536e-06,
+      "loss": 0.7979,
+      "step": 899
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.044837815156377e-06,
+      "loss": 0.7933,
+      "step": 900
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.036918318816853e-06,
+      "loss": 0.8064,
+      "step": 901
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.028996103279918e-06,
+      "loss": 0.8012,
+      "step": 902
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.021071189320665e-06,
+      "loss": 0.795,
+      "step": 903
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.013143597721252e-06,
+      "loss": 0.7915,
+      "step": 904
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.005213349270864e-06,
+      "loss": 0.8013,
+      "step": 905
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.997280464765655e-06,
+      "loss": 0.817,
+      "step": 906
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.98934496500869e-06,
+      "loss": 0.7893,
+      "step": 907
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.981406870809889e-06,
+      "loss": 0.7918,
+      "step": 908
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.97346620298598e-06,
+      "loss": 0.7996,
+      "step": 909
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.965522982360441e-06,
+      "loss": 0.7955,
+      "step": 910
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.957577229763438e-06,
+      "loss": 0.7983,
+      "step": 911
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.949628966031785e-06,
+      "loss": 0.7995,
+      "step": 912
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.941678212008875e-06,
+      "loss": 0.8007,
+      "step": 913
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.933724988544632e-06,
+      "loss": 0.7905,
+      "step": 914
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.925769316495461e-06,
+      "loss": 0.7931,
+      "step": 915
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9178112167241805e-06,
+      "loss": 0.7902,
+      "step": 916
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.909850710099985e-06,
+      "loss": 0.7933,
+      "step": 917
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9018878174983674e-06,
+      "loss": 0.7962,
+      "step": 918
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.893922559801092e-06,
+      "loss": 0.7993,
+      "step": 919
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.885954957896115e-06,
+      "loss": 0.7864,
+      "step": 920
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8779850326775436e-06,
+      "loss": 0.7913,
+      "step": 921
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.87001280504558e-06,
+      "loss": 0.8036,
+      "step": 922
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8620382959064596e-06,
+      "loss": 0.787,
+      "step": 923
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.854061526172402e-06,
+      "loss": 0.8044,
+      "step": 924
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.846082516761558e-06,
+      "loss": 0.7853,
+      "step": 925
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.838101288597951e-06,
+      "loss": 0.8018,
+      "step": 926
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.830117862611417e-06,
+      "loss": 0.8068,
+      "step": 927
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.822132259737565e-06,
+      "loss": 0.7991,
+      "step": 928
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.814144500917703e-06,
+      "loss": 0.7922,
+      "step": 929
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.806154607098799e-06,
+      "loss": 0.7981,
+      "step": 930
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7981625992334205e-06,
+      "loss": 0.7954,
+      "step": 931
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7901684982796716e-06,
+      "loss": 0.8054,
+      "step": 932
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.782172325201155e-06,
+      "loss": 0.7831,
+      "step": 933
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.774174100966899e-06,
+      "loss": 0.805,
+      "step": 934
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.766173846551316e-06,
+      "loss": 0.7958,
+      "step": 935
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.75817158293414e-06,
+      "loss": 0.788,
+      "step": 936
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.750167331100375e-06,
+      "loss": 0.7905,
+      "step": 937
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.742161112040237e-06,
+      "loss": 0.7906,
+      "step": 938
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.734152946749102e-06,
+      "loss": 0.8046,
+      "step": 939
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.726142856227453e-06,
+      "loss": 0.7968,
+      "step": 940
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7181308614808136e-06,
+      "loss": 0.793,
+      "step": 941
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7101169835197115e-06,
+      "loss": 0.8038,
+      "step": 942
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.702101243359602e-06,
+      "loss": 0.7881,
+      "step": 943
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.694083662020835e-06,
+      "loss": 0.7976,
+      "step": 944
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.686064260528577e-06,
+      "loss": 0.7892,
+      "step": 945
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.678043059912776e-06,
+      "loss": 0.7861,
+      "step": 946
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.6700200812080984e-06,
+      "loss": 0.7892,
+      "step": 947
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.661995345453867e-06,
+      "loss": 0.8029,
+      "step": 948
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.653968873694017e-06,
+      "loss": 0.782,
+      "step": 949
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.645940686977033e-06,
+      "loss": 0.7925,
+      "step": 950
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.6379108063559015e-06,
+      "loss": 0.7915,
+      "step": 951
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.629879252888046e-06,
+      "loss": 0.8032,
+      "step": 952
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.621846047635282e-06,
+      "loss": 0.7939,
+      "step": 953
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.613811211663751e-06,
+      "loss": 0.7964,
+      "step": 954
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.6057747660438745e-06,
+      "loss": 0.7951,
+      "step": 955
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.597736731850295e-06,
+      "loss": 0.793,
+      "step": 956
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.589697130161819e-06,
+      "loss": 0.7921,
+      "step": 957
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.581655982061367e-06,
+      "loss": 0.7786,
+      "step": 958
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.57361330863591e-06,
+      "loss": 0.7925,
+      "step": 959
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5655691309764225e-06,
+      "loss": 0.797,
+      "step": 960
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.557523470177821e-06,
+      "loss": 0.7994,
+      "step": 961
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.549476347338915e-06,
+      "loss": 0.7956,
+      "step": 962
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5414277835623465e-06,
+      "loss": 0.801,
+      "step": 963
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.533377799954532e-06,
+      "loss": 0.8043,
+      "step": 964
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.52532641762562e-06,
+      "loss": 0.7994,
+      "step": 965
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.517273657689419e-06,
+      "loss": 0.7992,
+      "step": 966
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.509219541263355e-06,
+      "loss": 0.8041,
+      "step": 967
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.501164089468406e-06,
+      "loss": 0.8005,
+      "step": 968
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.493107323429057e-06,
+      "loss": 0.8075,
+      "step": 969
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.485049264273241e-06,
+      "loss": 0.8074,
+      "step": 970
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.476989933132274e-06,
+      "loss": 0.799,
+      "step": 971
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4689293511408155e-06,
+      "loss": 0.8019,
+      "step": 972
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.460867539436799e-06,
+      "loss": 0.789,
+      "step": 973
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.45280451916139e-06,
+      "loss": 0.7902,
+      "step": 974
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.444740311458915e-06,
+      "loss": 0.7939,
+      "step": 975
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.43667493747682e-06,
+      "loss": 0.7935,
+      "step": 976
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.428608418365611e-06,
+      "loss": 0.8011,
+      "step": 977
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4205407752787884e-06,
+      "loss": 0.8028,
+      "step": 978
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.412472029372812e-06,
+      "loss": 0.8042,
+      "step": 979
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.404402201807022e-06,
+      "loss": 0.789,
+      "step": 980
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.396331313743603e-06,
+      "loss": 0.7917,
+      "step": 981
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.388259386347518e-06,
+      "loss": 0.7991,
+      "step": 982
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.380186440786452e-06,
+      "loss": 0.8002,
+      "step": 983
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.372112498230771e-06,
+      "loss": 0.7824,
+      "step": 984
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.364037579853441e-06,
+      "loss": 0.7994,
+      "step": 985
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.355961706829997e-06,
+      "loss": 0.8024,
+      "step": 986
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.347884900338475e-06,
+      "loss": 0.7999,
+      "step": 987
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.339807181559359e-06,
+      "loss": 0.8051,
+      "step": 988
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.331728571675522e-06,
+      "loss": 0.7973,
+      "step": 989
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.323649091872179e-06,
+      "loss": 0.7908,
+      "step": 990
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.315568763336823e-06,
+      "loss": 0.79,
+      "step": 991
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.307487607259175e-06,
+      "loss": 0.7912,
+      "step": 992
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.299405644831122e-06,
+      "loss": 0.7973,
+      "step": 993
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.291322897246669e-06,
+      "loss": 0.7882,
+      "step": 994
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.283239385701882e-06,
+      "loss": 0.8051,
+      "step": 995
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.275155131394825e-06,
+      "loss": 0.8001,
+      "step": 996
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.267070155525513e-06,
+      "loss": 0.7981,
+      "step": 997
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.258984479295853e-06,
+      "loss": 0.776,
+      "step": 998
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.250898123909588e-06,
+      "loss": 0.7869,
+      "step": 999
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.242811110572243e-06,
+      "loss": 0.7871,
+      "step": 1000
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.234723460491065e-06,
+      "loss": 0.7965,
+      "step": 1001
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.226635194874978e-06,
+      "loss": 0.7996,
+      "step": 1002
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.218546334934513e-06,
+      "loss": 0.805,
+      "step": 1003
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.210456901881761e-06,
+      "loss": 0.7967,
+      "step": 1004
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.202366916930319e-06,
+      "loss": 0.7816,
+      "step": 1005
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.194276401295231e-06,
+      "loss": 0.7863,
+      "step": 1006
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.186185376192927e-06,
+      "loss": 0.7987,
+      "step": 1007
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.1780938628411795e-06,
+      "loss": 0.7951,
+      "step": 1008
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.170001882459039e-06,
+      "loss": 0.7848,
+      "step": 1009
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.161909456266781e-06,
+      "loss": 0.8035,
+      "step": 1010
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.1538166054858485e-06,
+      "loss": 0.782,
+      "step": 1011
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.145723351338799e-06,
+      "loss": 0.8002,
+      "step": 1012
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.137629715049251e-06,
+      "loss": 0.797,
+      "step": 1013
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.129535717841818e-06,
+      "loss": 0.7976,
+      "step": 1014
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.121441380942066e-06,
+      "loss": 0.8054,
+      "step": 1015
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.11334672557645e-06,
+      "loss": 0.801,
+      "step": 1016
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.105251772972261e-06,
+      "loss": 0.7874,
+      "step": 1017
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.097156544357567e-06,
+      "loss": 0.7917,
+      "step": 1018
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.089061060961162e-06,
+      "loss": 0.8093,
+      "step": 1019
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.080965344012509e-06,
+      "loss": 0.7819,
+      "step": 1020
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.072869414741677e-06,
+      "loss": 0.7834,
+      "step": 1021
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.064773294379302e-06,
+      "loss": 0.7793,
+      "step": 1022
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0566770041565124e-06,
+      "loss": 0.798,
+      "step": 1023
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.048580565304887e-06,
+      "loss": 0.7848,
+      "step": 1024
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.040483999056393e-06,
+      "loss": 0.7959,
+      "step": 1025
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.032387326643331e-06,
+      "loss": 0.7923,
+      "step": 1026
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.024290569298281e-06,
+      "loss": 0.7826,
+      "step": 1027
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.016193748254045e-06,
+      "loss": 0.8006,
+      "step": 1028
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.00809688474359e-06,
+      "loss": 0.7941,
+      "step": 1029
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 5e-06,
+      "loss": 0.7951,
+      "step": 1030
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.99190311525641e-06,
+      "loss": 0.7794,
+      "step": 1031
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.983806251745958e-06,
+      "loss": 0.7902,
+      "step": 1032
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.97570943070172e-06,
+      "loss": 0.7942,
+      "step": 1033
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9676126733566705e-06,
+      "loss": 0.8032,
+      "step": 1034
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9595160009436075e-06,
+      "loss": 0.802,
+      "step": 1035
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.951419434695115e-06,
+      "loss": 0.792,
+      "step": 1036
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.943322995843488e-06,
+      "loss": 0.7967,
+      "step": 1037
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.935226705620699e-06,
+      "loss": 0.7949,
+      "step": 1038
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.927130585258324e-06,
+      "loss": 0.8017,
+      "step": 1039
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.919034655987493e-06,
+      "loss": 0.793,
+      "step": 1040
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.910938939038839e-06,
+      "loss": 0.7864,
+      "step": 1041
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9028434556424335e-06,
+      "loss": 0.7924,
+      "step": 1042
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.894748227027741e-06,
+      "loss": 0.8013,
+      "step": 1043
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.886653274423551e-06,
+      "loss": 0.787,
+      "step": 1044
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8785586190579355e-06,
+      "loss": 0.7918,
+      "step": 1045
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.870464282158184e-06,
+      "loss": 0.7873,
+      "step": 1046
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.862370284950752e-06,
+      "loss": 0.7992,
+      "step": 1047
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8542766486612035e-06,
+      "loss": 0.7756,
+      "step": 1048
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.846183394514154e-06,
+      "loss": 0.7918,
+      "step": 1049
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.838090543733222e-06,
+      "loss": 0.7986,
+      "step": 1050
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.829998117540963e-06,
+      "loss": 0.7937,
+      "step": 1051
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.821906137158822e-06,
+      "loss": 0.7963,
+      "step": 1052
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.813814623807076e-06,
+      "loss": 0.7978,
+      "step": 1053
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.805723598704772e-06,
+      "loss": 0.782,
+      "step": 1054
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.797633083069683e-06,
+      "loss": 0.7878,
+      "step": 1055
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7895430981182415e-06,
+      "loss": 0.7949,
+      "step": 1056
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.781453665065491e-06,
+      "loss": 0.7882,
+      "step": 1057
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.773364805125025e-06,
+      "loss": 0.7912,
+      "step": 1058
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.765276539508936e-06,
+      "loss": 0.7878,
+      "step": 1059
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.757188889427761e-06,
+      "loss": 0.8024,
+      "step": 1060
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.749101876090414e-06,
+      "loss": 0.7922,
+      "step": 1061
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.741015520704148e-06,
+      "loss": 0.7993,
+      "step": 1062
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.732929844474489e-06,
+      "loss": 0.7865,
+      "step": 1063
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.724844868605176e-06,
+      "loss": 0.7978,
+      "step": 1064
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.716760614298118e-06,
+      "loss": 0.788,
+      "step": 1065
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.708677102753331e-06,
+      "loss": 0.7953,
+      "step": 1066
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.700594355168878e-06,
+      "loss": 0.8057,
+      "step": 1067
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6925123927408265e-06,
+      "loss": 0.7909,
+      "step": 1068
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6844312366631775e-06,
+      "loss": 0.7896,
+      "step": 1069
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6763509081278215e-06,
+      "loss": 0.802,
+      "step": 1070
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.668271428324479e-06,
+      "loss": 0.7821,
+      "step": 1071
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.660192818440642e-06,
+      "loss": 0.7827,
+      "step": 1072
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.652115099661525e-06,
+      "loss": 0.7894,
+      "step": 1073
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.644038293170003e-06,
+      "loss": 0.796,
+      "step": 1074
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.63596242014656e-06,
+      "loss": 0.7898,
+      "step": 1075
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.627887501769231e-06,
+      "loss": 0.7989,
+      "step": 1076
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.619813559213549e-06,
+      "loss": 0.8079,
+      "step": 1077
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.611740613652485e-06,
+      "loss": 0.7829,
+      "step": 1078
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.603668686256398e-06,
+      "loss": 0.7925,
+      "step": 1079
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.59559779819298e-06,
+      "loss": 0.7851,
+      "step": 1080
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.58752797062719e-06,
+      "loss": 0.8047,
+      "step": 1081
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.579459224721212e-06,
+      "loss": 0.7769,
+      "step": 1082
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.571391581634391e-06,
+      "loss": 0.7994,
+      "step": 1083
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5633250625231806e-06,
+      "loss": 0.7893,
+      "step": 1084
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.555259688541087e-06,
+      "loss": 0.7982,
+      "step": 1085
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.547195480838612e-06,
+      "loss": 0.7999,
+      "step": 1086
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.539132460563202e-06,
+      "loss": 0.7734,
+      "step": 1087
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.531070648859186e-06,
+      "loss": 0.7963,
+      "step": 1088
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.523010066867727e-06,
+      "loss": 0.8027,
+      "step": 1089
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.51495073572676e-06,
+      "loss": 0.7986,
+      "step": 1090
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.5068926765709435e-06,
+      "loss": 0.7755,
+      "step": 1091
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.498835910531595e-06,
+      "loss": 0.7901,
+      "step": 1092
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.490780458736646e-06,
+      "loss": 0.7991,
+      "step": 1093
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.482726342310582e-06,
+      "loss": 0.7979,
+      "step": 1094
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.474673582374381e-06,
+      "loss": 0.7879,
+      "step": 1095
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4666222000454685e-06,
+      "loss": 0.7827,
+      "step": 1096
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.458572216437655e-06,
+      "loss": 0.7989,
+      "step": 1097
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.450523652661086e-06,
+      "loss": 0.791,
+      "step": 1098
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.44247652982218e-06,
+      "loss": 0.7966,
+      "step": 1099
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.434430869023579e-06,
+      "loss": 0.7984,
+      "step": 1100
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.426386691364092e-06,
+      "loss": 0.7983,
+      "step": 1101
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.418344017938634e-06,
+      "loss": 0.8067,
+      "step": 1102
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4103028698381815e-06,
+      "loss": 0.7854,
+      "step": 1103
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.402263268149707e-06,
+      "loss": 0.7937,
+      "step": 1104
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.394225233956127e-06,
+      "loss": 0.7874,
+      "step": 1105
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.386188788336251e-06,
+      "loss": 0.8001,
+      "step": 1106
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.37815395236472e-06,
+      "loss": 0.7931,
+      "step": 1107
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.370120747111956e-06,
+      "loss": 0.7955,
+      "step": 1108
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.362089193644101e-06,
+      "loss": 0.794,
+      "step": 1109
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3540593130229695e-06,
+      "loss": 0.7894,
+      "step": 1110
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3460311263059864e-06,
+      "loss": 0.7915,
+      "step": 1111
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.338004654546136e-06,
+      "loss": 0.7898,
+      "step": 1112
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.329979918791904e-06,
+      "loss": 0.7969,
+      "step": 1113
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3219569400872244e-06,
+      "loss": 0.7888,
+      "step": 1114
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.313935739471425e-06,
+      "loss": 0.7912,
+      "step": 1115
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3059163379791676e-06,
+      "loss": 0.7831,
+      "step": 1116
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.297898756640399e-06,
+      "loss": 0.794,
+      "step": 1117
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.289883016480291e-06,
+      "loss": 0.7816,
+      "step": 1118
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.281869138519188e-06,
+      "loss": 0.7904,
+      "step": 1119
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.27385714377255e-06,
+      "loss": 0.7961,
+      "step": 1120
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.265847053250898e-06,
+      "loss": 0.7905,
+      "step": 1121
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.257838887959764e-06,
+      "loss": 0.7941,
+      "step": 1122
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.249832668899627e-06,
+      "loss": 0.79,
+      "step": 1123
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.24182841706586e-06,
+      "loss": 0.7837,
+      "step": 1124
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.233826153448685e-06,
+      "loss": 0.7918,
+      "step": 1125
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2258258990331015e-06,
+      "loss": 0.7826,
+      "step": 1126
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.217827674798845e-06,
+      "loss": 0.7955,
+      "step": 1127
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.209831501720328e-06,
+      "loss": 0.7918,
+      "step": 1128
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.20183740076658e-06,
+      "loss": 0.7952,
+      "step": 1129
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.1938453929012014e-06,
+      "loss": 0.7858,
+      "step": 1130
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.185855499082297e-06,
+      "loss": 0.7921,
+      "step": 1131
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.177867740262437e-06,
+      "loss": 0.7943,
+      "step": 1132
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.169882137388583e-06,
+      "loss": 0.781,
+      "step": 1133
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.16189871140205e-06,
+      "loss": 0.7975,
+      "step": 1134
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1539174832384425e-06,
+      "loss": 0.7832,
+      "step": 1135
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.145938473827598e-06,
+      "loss": 0.7875,
+      "step": 1136
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.137961704093543e-06,
+      "loss": 0.7849,
+      "step": 1137
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.129987194954421e-06,
+      "loss": 0.7929,
+      "step": 1138
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.122014967322458e-06,
+      "loss": 0.7947,
+      "step": 1139
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1140450421038865e-06,
+      "loss": 0.7917,
+      "step": 1140
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.106077440198909e-06,
+      "loss": 0.7961,
+      "step": 1141
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.098112182501633e-06,
+      "loss": 0.7913,
+      "step": 1142
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.090149289900017e-06,
+      "loss": 0.794,
+      "step": 1143
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.08218878327582e-06,
+      "loss": 0.789,
+      "step": 1144
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.074230683504541e-06,
+      "loss": 0.7774,
+      "step": 1145
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.066275011455369e-06,
+      "loss": 0.7805,
+      "step": 1146
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.058321787991127e-06,
+      "loss": 0.7888,
+      "step": 1147
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.050371033968216e-06,
+      "loss": 0.794,
+      "step": 1148
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0424227702365634e-06,
+      "loss": 0.7909,
+      "step": 1149
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.034477017639561e-06,
+      "loss": 0.788,
+      "step": 1150
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.026533797014022e-06,
+      "loss": 0.7847,
+      "step": 1151
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.018593129190113e-06,
+      "loss": 0.7767,
+      "step": 1152
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.010655034991312e-06,
+      "loss": 0.7878,
+      "step": 1153
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.002719535234346e-06,
+      "loss": 0.8035,
+      "step": 1154
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.994786650729137e-06,
+      "loss": 0.7951,
+      "step": 1155
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.98685640227875e-06,
+      "loss": 0.7891,
+      "step": 1156
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.978928810679337e-06,
+      "loss": 0.7826,
+      "step": 1157
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9710038967200825e-06,
+      "loss": 0.809,
+      "step": 1158
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.96308168118315e-06,
+      "loss": 0.7932,
+      "step": 1159
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.955162184843625e-06,
+      "loss": 0.7922,
+      "step": 1160
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.947245428469466e-06,
+      "loss": 0.7951,
+      "step": 1161
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.93933143282144e-06,
+      "loss": 0.7829,
+      "step": 1162
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.931420218653082e-06,
+      "loss": 0.7782,
+      "step": 1163
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9235118067106255e-06,
+      "loss": 0.786,
+      "step": 1164
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9156062177329624e-06,
+      "loss": 0.785,
+      "step": 1165
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.907703472451574e-06,
+      "loss": 0.7973,
+      "step": 1166
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.899803591590491e-06,
+      "loss": 0.8005,
+      "step": 1167
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.89190659586623e-06,
+      "loss": 0.7849,
+      "step": 1168
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.884012505987741e-06,
+      "loss": 0.7842,
+      "step": 1169
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.8761213426563546e-06,
+      "loss": 0.7651,
+      "step": 1170
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8682331265657254e-06,
+      "loss": 0.7865,
+      "step": 1171
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8603478784017845e-06,
+      "loss": 0.7927,
+      "step": 1172
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.852465618842671e-06,
+      "loss": 0.7782,
+      "step": 1173
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8445863685586946e-06,
+      "loss": 0.787,
+      "step": 1174
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.836710148212271e-06,
+      "loss": 0.7933,
+      "step": 1175
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.828836978457868e-06,
+      "loss": 0.7922,
+      "step": 1176
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.820966879941957e-06,
+      "loss": 0.7859,
+      "step": 1177
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8130998733029517e-06,
+      "loss": 0.7917,
+      "step": 1178
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.805235979171163e-06,
+      "loss": 0.7781,
+      "step": 1179
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7973752181687336e-06,
+      "loss": 0.7856,
+      "step": 1180
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.789517610909594e-06,
+      "loss": 0.7967,
+      "step": 1181
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7816631779994018e-06,
+      "loss": 0.7911,
+      "step": 1182
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.773811940035496e-06,
+      "loss": 0.7891,
+      "step": 1183
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7659639176068287e-06,
+      "loss": 0.7916,
+      "step": 1184
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7581191312939256e-06,
+      "loss": 0.7854,
+      "step": 1185
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7502776016688234e-06,
+      "loss": 0.7999,
+      "step": 1186
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.74243934929502e-06,
+      "loss": 0.782,
+      "step": 1187
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.734604394727419e-06,
+      "loss": 0.7952,
+      "step": 1188
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7267727585122738e-06,
+      "loss": 0.786,
+      "step": 1189
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7189444611871383e-06,
+      "loss": 0.7914,
+      "step": 1190
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7111195232808093e-06,
+      "loss": 0.7856,
+      "step": 1191
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.703297965313275e-06,
+      "loss": 0.7854,
+      "step": 1192
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6954798077956576e-06,
+      "loss": 0.7914,
+      "step": 1193
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6876650712301654e-06,
+      "loss": 0.7997,
+      "step": 1194
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.679853776110035e-06,
+      "loss": 0.7941,
+      "step": 1195
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6720459429194743e-06,
+      "loss": 0.7978,
+      "step": 1196
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.664241592133618e-06,
+      "loss": 0.7916,
+      "step": 1197
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.656440744218464e-06,
+      "loss": 0.7953,
+      "step": 1198
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6486434196308296e-06,
+      "loss": 0.786,
+      "step": 1199
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6408496388182857e-06,
+      "loss": 0.7778,
+      "step": 1200
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6330594222191172e-06,
+      "loss": 0.7874,
+      "step": 1201
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6252727902622575e-06,
+      "loss": 0.7982,
+      "step": 1202
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.617489763367241e-06,
+      "loss": 0.7849,
+      "step": 1203
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6097103619441505e-06,
+      "loss": 0.7812,
+      "step": 1204
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.601934606393556e-06,
+      "loss": 0.7961,
+      "step": 1205
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.594162517106472e-06,
+      "loss": 0.7845,
+      "step": 1206
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5863941144642957e-06,
+      "loss": 0.7867,
+      "step": 1207
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.578629418838757e-06,
+      "loss": 0.788,
+      "step": 1208
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5708684505918678e-06,
+      "loss": 0.7847,
+      "step": 1209
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5631112300758595e-06,
+      "loss": 0.7746,
+      "step": 1210
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5553577776331405e-06,
+      "loss": 0.786,
+      "step": 1211
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5476081135962335e-06,
+      "loss": 0.7925,
+      "step": 1212
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5398622582877327e-06,
+      "loss": 0.7787,
+      "step": 1213
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.532120232020236e-06,
+      "loss": 0.778,
+      "step": 1214
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.524382055096309e-06,
+      "loss": 0.7879,
+      "step": 1215
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.516647747808417e-06,
+      "loss": 0.7852,
+      "step": 1216
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.508917330438878e-06,
+      "loss": 0.7797,
+      "step": 1217
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5011908232598124e-06,
+      "loss": 0.7964,
+      "step": 1218
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4934682465330826e-06,
+      "loss": 0.7882,
+      "step": 1219
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4857496205102475e-06,
+      "loss": 0.7815,
+      "step": 1220
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4780349654325023e-06,
+      "loss": 0.775,
+      "step": 1221
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4703243015306314e-06,
+      "loss": 0.8014,
+      "step": 1222
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.462617649024953e-06,
+      "loss": 0.7995,
+      "step": 1223
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4549150281252635e-06,
+      "loss": 0.7865,
+      "step": 1224
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4472164590307895e-06,
+      "loss": 0.7832,
+      "step": 1225
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4395219619301288e-06,
+      "loss": 0.7763,
+      "step": 1226
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4318315570012057e-06,
+      "loss": 0.7818,
+      "step": 1227
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4241452644112085e-06,
+      "loss": 0.7914,
+      "step": 1228
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4164631043165432e-06,
+      "loss": 0.7891,
+      "step": 1229
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4087850968627823e-06,
+      "loss": 0.7858,
+      "step": 1230
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.401111262184601e-06,
+      "loss": 0.7898,
+      "step": 1231
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3934416204057396e-06,
+      "loss": 0.7873,
+      "step": 1232
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3857761916389377e-06,
+      "loss": 0.799,
+      "step": 1233
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3781149959858894e-06,
+      "loss": 0.7907,
+      "step": 1234
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.370458053537186e-06,
+      "loss": 0.7941,
+      "step": 1235
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3628053843722674e-06,
+      "loss": 0.7922,
+      "step": 1236
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.355157008559367e-06,
+      "loss": 0.7855,
+      "step": 1237
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3475129461554567e-06,
+      "loss": 0.7756,
+      "step": 1238
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3398732172062023e-06,
+      "loss": 0.7799,
+      "step": 1239
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3322378417458985e-06,
+      "loss": 0.7929,
+      "step": 1240
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3246068397974297e-06,
+      "loss": 0.8017,
+      "step": 1241
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3169802313722076e-06,
+      "loss": 0.7901,
+      "step": 1242
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.309358036470124e-06,
+      "loss": 0.7856,
+      "step": 1243
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3017402750794976e-06,
+      "loss": 0.7825,
+      "step": 1244
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2941269671770193e-06,
+      "loss": 0.7848,
+      "step": 1245
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2865181327277007e-06,
+      "loss": 0.7859,
+      "step": 1246
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2789137916848242e-06,
+      "loss": 0.79,
+      "step": 1247
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.271313963989886e-06,
+      "loss": 0.793,
+      "step": 1248
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2637186695725486e-06,
+      "loss": 0.7893,
+      "step": 1249
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2561279283505888e-06,
+      "loss": 0.7865,
+      "step": 1250
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.248541760229837e-06,
+      "loss": 0.7942,
+      "step": 1251
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.240960185104137e-06,
+      "loss": 0.7806,
+      "step": 1252
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.233383222855284e-06,
+      "loss": 0.7834,
+      "step": 1253
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2258108933529808e-06,
+      "loss": 0.7723,
+      "step": 1254
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.218243216454775e-06,
+      "loss": 0.7843,
+      "step": 1255
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2106802120060197e-06,
+      "loss": 0.7862,
+      "step": 1256
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2031218998398126e-06,
+      "loss": 0.7802,
+      "step": 1257
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.195568299776945e-06,
+      "loss": 0.7827,
+      "step": 1258
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.188019431625853e-06,
+      "loss": 0.7859,
+      "step": 1259
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.180475315182563e-06,
+      "loss": 0.7814,
+      "step": 1260
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.172935970230641e-06,
+      "loss": 0.7958,
+      "step": 1261
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.16540141654114e-06,
+      "loss": 0.7704,
+      "step": 1262
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.157871673872549e-06,
+      "loss": 0.7914,
+      "step": 1263
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1503467619707407e-06,
+      "loss": 0.7856,
+      "step": 1264
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1428267005689182e-06,
+      "loss": 0.7969,
+      "step": 1265
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1353115093875676e-06,
+      "loss": 0.7891,
+      "step": 1266
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1278012081343996e-06,
+      "loss": 0.7836,
+      "step": 1267
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1202958165043053e-06,
+      "loss": 0.7878,
+      "step": 1268
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.112795354179298e-06,
+      "loss": 0.7989,
+      "step": 1269
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1052998408284664e-06,
+      "loss": 0.7945,
+      "step": 1270
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0978092961079222e-06,
+      "loss": 0.7847,
+      "step": 1271
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.090323739660742e-06,
+      "loss": 0.7943,
+      "step": 1272
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.082843191116928e-06,
+      "loss": 0.7787,
+      "step": 1273
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0753676700933444e-06,
+      "loss": 0.7702,
+      "step": 1274
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0678971961936765e-06,
+      "loss": 0.7849,
+      "step": 1275
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.060431789008368e-06,
+      "loss": 0.7869,
+      "step": 1276
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0529714681145808e-06,
+      "loss": 0.7889,
+      "step": 1277
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.045516253076137e-06,
+      "loss": 0.7923,
+      "step": 1278
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.038066163443467e-06,
+      "loss": 0.8045,
+      "step": 1279
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0306212187535653e-06,
+      "loss": 0.7849,
+      "step": 1280
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0231814385299287e-06,
+      "loss": 0.7857,
+      "step": 1281
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0157468422825148e-06,
+      "loss": 0.7881,
+      "step": 1282
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0083174495076846e-06,
+      "loss": 0.7819,
+      "step": 1283
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.000893279688155e-06,
+      "loss": 0.7956,
+      "step": 1284
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9934743522929476e-06,
+      "loss": 0.776,
+      "step": 1285
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9860606867773323e-06,
+      "loss": 0.7868,
+      "step": 1286
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9786523025827836e-06,
+      "loss": 0.7688,
+      "step": 1287
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9712492191369245e-06,
+      "loss": 0.79,
+      "step": 1288
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9638514558534804e-06,
+      "loss": 0.778,
+      "step": 1289
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9564590321322206e-06,
+      "loss": 0.7781,
+      "step": 1290
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9490719673589153e-06,
+      "loss": 0.7748,
+      "step": 1291
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9416902809052817e-06,
+      "loss": 0.7913,
+      "step": 1292
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.934313992128931e-06,
+      "loss": 0.7875,
+      "step": 1293
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9269431203733213e-06,
+      "loss": 0.7815,
+      "step": 1294
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9195776849677037e-06,
+      "loss": 0.7921,
+      "step": 1295
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.912217705227075e-06,
+      "loss": 0.7809,
+      "step": 1296
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.904863200452121e-06,
+      "loss": 0.7918,
+      "step": 1297
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8975141899291777e-06,
+      "loss": 0.7879,
+      "step": 1298
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8901706929301666e-06,
+      "loss": 0.7837,
+      "step": 1299
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.882832728712551e-06,
+      "loss": 0.7828,
+      "step": 1300
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8755003165192864e-06,
+      "loss": 0.7868,
+      "step": 1301
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.868173475578772e-06,
+      "loss": 0.7771,
+      "step": 1302
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8608522251047905e-06,
+      "loss": 0.7889,
+      "step": 1303
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8535365842964713e-06,
+      "loss": 0.7901,
+      "step": 1304
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8462265723382255e-06,
+      "loss": 0.7892,
+      "step": 1305
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.838922208399712e-06,
+      "loss": 0.7936,
+      "step": 1306
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8316235116357715e-06,
+      "loss": 0.7882,
+      "step": 1307
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8243305011863843e-06,
+      "loss": 0.7948,
+      "step": 1308
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.81704319617662e-06,
+      "loss": 0.778,
+      "step": 1309
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8097616157165886e-06,
+      "loss": 0.7764,
+      "step": 1310
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8024857789013847e-06,
+      "loss": 0.7886,
+      "step": 1311
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7952157048110406e-06,
+      "loss": 0.7783,
+      "step": 1312
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.787951412510481e-06,
+      "loss": 0.787,
+      "step": 1313
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.780692921049465e-06,
+      "loss": 0.7907,
+      "step": 1314
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7734402494625395e-06,
+      "loss": 0.7807,
+      "step": 1315
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7661934167689887e-06,
+      "loss": 0.7886,
+      "step": 1316
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7589524419727885e-06,
+      "loss": 0.7963,
+      "step": 1317
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.751717344062552e-06,
+      "loss": 0.7848,
+      "step": 1318
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7444881420114766e-06,
+      "loss": 0.7859,
+      "step": 1319
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7372648547773063e-06,
+      "loss": 0.7945,
+      "step": 1320
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7300475013022666e-06,
+      "loss": 0.7893,
+      "step": 1321
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.722836100513027e-06,
+      "loss": 0.7871,
+      "step": 1322
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7156306713206433e-06,
+      "loss": 0.7945,
+      "step": 1323
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7084312326205164e-06,
+      "loss": 0.787,
+      "step": 1324
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.701237803292335e-06,
+      "loss": 0.7932,
+      "step": 1325
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6940504022000248e-06,
+      "loss": 0.7822,
+      "step": 1326
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6868690481917115e-06,
+      "loss": 0.7812,
+      "step": 1327
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6796937600996587e-06,
+      "loss": 0.7818,
+      "step": 1328
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6725245567402208e-06,
+      "loss": 0.7767,
+      "step": 1329
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.665361456913797e-06,
+      "loss": 0.7847,
+      "step": 1330
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6582044794047853e-06,
+      "loss": 0.7874,
+      "step": 1331
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6510536429815224e-06,
+      "loss": 0.7874,
+      "step": 1332
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6439089663962424e-06,
+      "loss": 0.7783,
+      "step": 1333
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6367704683850293e-06,
+      "loss": 0.7787,
+      "step": 1334
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.629638167667761e-06,
+      "loss": 0.7893,
+      "step": 1335
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.622512082948063e-06,
+      "loss": 0.7771,
+      "step": 1336
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6153922329132618e-06,
+      "loss": 0.7893,
+      "step": 1337
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6082786362343377e-06,
+      "loss": 0.7775,
+      "step": 1338
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6011713115658654e-06,
+      "loss": 0.7788,
+      "step": 1339
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.594070277545975e-06,
+      "loss": 0.7719,
+      "step": 1340
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5869755527963027e-06,
+      "loss": 0.7848,
+      "step": 1341
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5798871559219362e-06,
+      "loss": 0.7872,
+      "step": 1342
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.572805105511369e-06,
+      "loss": 0.7865,
+      "step": 1343
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5657294201364526e-06,
+      "loss": 0.7697,
+      "step": 1344
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5586601183523485e-06,
+      "loss": 0.7716,
+      "step": 1345
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.551597218697476e-06,
+      "loss": 0.7864,
+      "step": 1346
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.544540739693464e-06,
+      "loss": 0.7834,
+      "step": 1347
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5374906998451094e-06,
+      "loss": 0.7795,
+      "step": 1348
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5304471176403184e-06,
+      "loss": 0.7814,
+      "step": 1349
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5234100115500643e-06,
+      "loss": 0.7839,
+      "step": 1350
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.516379400028335e-06,
+      "loss": 0.7824,
+      "step": 1351
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5093553015120937e-06,
+      "loss": 0.7856,
+      "step": 1352
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.502337734421218e-06,
+      "loss": 0.7881,
+      "step": 1353
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4953267171584573e-06,
+      "loss": 0.7861,
+      "step": 1354
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4883222681093916e-06,
+      "loss": 0.7884,
+      "step": 1355
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4813244056423692e-06,
+      "loss": 0.7989,
+      "step": 1356
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.474333148108469e-06,
+      "loss": 0.8008,
+      "step": 1357
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.467348513841447e-06,
+      "loss": 0.7887,
+      "step": 1358
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4603705211576962e-06,
+      "loss": 0.7894,
+      "step": 1359
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4533991883561868e-06,
+      "loss": 0.7888,
+      "step": 1360
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4464345337184248e-06,
+      "loss": 0.7838,
+      "step": 1361
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.439476575508408e-06,
+      "loss": 0.7651,
+      "step": 1362
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4325253319725693e-06,
+      "loss": 0.7818,
+      "step": 1363
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.425580821339733e-06,
+      "loss": 0.7822,
+      "step": 1364
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4186430618210704e-06,
+      "loss": 0.7906,
+      "step": 1365
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4117120716100484e-06,
+      "loss": 0.7785,
+      "step": 1366
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4047878688823795e-06,
+      "loss": 0.7846,
+      "step": 1367
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3978704717959777e-06,
+      "loss": 0.7658,
+      "step": 1368
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.390959898490909e-06,
+      "loss": 0.7909,
+      "step": 1369
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.38405616708935e-06,
+      "loss": 0.776,
+      "step": 1370
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3771592956955285e-06,
+      "loss": 0.7987,
+      "step": 1371
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3702693023956853e-06,
+      "loss": 0.7826,
+      "step": 1372
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3633862052580273e-06,
+      "loss": 0.7853,
+      "step": 1373
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.356510022332674e-06,
+      "loss": 0.7854,
+      "step": 1374
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.349640771651611e-06,
+      "loss": 0.7785,
+      "step": 1375
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.342778471228648e-06,
+      "loss": 0.7761,
+      "step": 1376
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.33592313905937e-06,
+      "loss": 0.7807,
+      "step": 1377
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.329074793121085e-06,
+      "loss": 0.7769,
+      "step": 1378
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3222334513727808e-06,
+      "loss": 0.7921,
+      "step": 1379
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.315399131755081e-06,
+      "loss": 0.7717,
+      "step": 1380
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.308571852190191e-06,
+      "loss": 0.7904,
+      "step": 1381
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.301751630581855e-06,
+      "loss": 0.775,
+      "step": 1382
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.294938484815308e-06,
+      "loss": 0.7884,
+      "step": 1383
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2881324327572336e-06,
+      "loss": 0.7844,
+      "step": 1384
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.281333492255708e-06,
+      "loss": 0.7786,
+      "step": 1385
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.274541681140159e-06,
+      "loss": 0.7946,
+      "step": 1386
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.267757017221322e-06,
+      "loss": 0.7834,
+      "step": 1387
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.260979518291186e-06,
+      "loss": 0.7796,
+      "step": 1388
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2542092021229516e-06,
+      "loss": 0.7906,
+      "step": 1389
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2474460864709825e-06,
+      "loss": 0.8012,
+      "step": 1390
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.240690189070764e-06,
+      "loss": 0.7786,
+      "step": 1391
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.233941527638848e-06,
+      "loss": 0.7934,
+      "step": 1392
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2272001198728096e-06,
+      "loss": 0.7785,
+      "step": 1393
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2204659834512095e-06,
+      "loss": 0.7929,
+      "step": 1394
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.213739136033533e-06,
+      "loss": 0.7915,
+      "step": 1395
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.207019595260154e-06,
+      "loss": 0.7902,
+      "step": 1396
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2003073787522824e-06,
+      "loss": 0.776,
+      "step": 1397
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1936025041119268e-06,
+      "loss": 0.7839,
+      "step": 1398
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.186904988921838e-06,
+      "loss": 0.7804,
+      "step": 1399
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1802148507454675e-06,
+      "loss": 0.7836,
+      "step": 1400
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.173532107126925e-06,
+      "loss": 0.7805,
+      "step": 1401
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1668567755909257e-06,
+      "loss": 0.7939,
+      "step": 1402
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1601888736427484e-06,
+      "loss": 0.7811,
+      "step": 1403
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1535284187681866e-06,
+      "loss": 0.7713,
+      "step": 1404
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1468754284335097e-06,
+      "loss": 0.7757,
+      "step": 1405
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.140229920085409e-06,
+      "loss": 0.7758,
+      "step": 1406
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1335919111509516e-06,
+      "loss": 0.7796,
+      "step": 1407
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1269614190375477e-06,
+      "loss": 0.7743,
+      "step": 1408
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.120338461132887e-06,
+      "loss": 0.7852,
+      "step": 1409
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1137230548049042e-06,
+      "loss": 0.7918,
+      "step": 1410
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.10711521740173e-06,
+      "loss": 0.7708,
+      "step": 1411
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1005149662516517e-06,
+      "loss": 0.7974,
+      "step": 1412
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.093922318663055e-06,
+      "loss": 0.793,
+      "step": 1413
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.08733729192439e-06,
+      "loss": 0.7728,
+      "step": 1414
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0807599033041238e-06,
+      "loss": 0.7812,
+      "step": 1415
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.07419017005069e-06,
+      "loss": 0.7846,
+      "step": 1416
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.067628109392448e-06,
+      "loss": 0.7732,
+      "step": 1417
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.061073738537635e-06,
+      "loss": 0.7877,
+      "step": 1418
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0545270746743277e-06,
+      "loss": 0.7789,
+      "step": 1419
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0479881349703885e-06,
+      "loss": 0.7687,
+      "step": 1420
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.041456936573421e-06,
+      "loss": 0.7835,
+      "step": 1421
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0349334966107363e-06,
+      "loss": 0.7874,
+      "step": 1422
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0284178321892926e-06,
+      "loss": 0.7674,
+      "step": 1423
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.021909960395661e-06,
+      "loss": 0.7863,
+      "step": 1424
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.015409898295975e-06,
+      "loss": 0.7811,
+      "step": 1425
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0089176629358904e-06,
+      "loss": 0.7831,
+      "step": 1426
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0024332713405396e-06,
+      "loss": 0.7756,
+      "step": 1427
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9959567405144825e-06,
+      "loss": 0.7825,
+      "step": 1428
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9894880874416644e-06,
+      "loss": 0.7786,
+      "step": 1429
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.983027329085377e-06,
+      "loss": 0.771,
+      "step": 1430
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.976574482388204e-06,
+      "loss": 0.7749,
+      "step": 1431
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9701295642719836e-06,
+      "loss": 0.7825,
+      "step": 1432
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.963692591637761e-06,
+      "loss": 0.7767,
+      "step": 1433
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.957263581365749e-06,
+      "loss": 0.7849,
+      "step": 1434
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.950842550315277e-06,
+      "loss": 0.7721,
+      "step": 1435
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.944429515324749e-06,
+      "loss": 0.7769,
+      "step": 1436
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9380244932116034e-06,
+      "loss": 0.7752,
+      "step": 1437
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.931627500772263e-06,
+      "loss": 0.7807,
+      "step": 1438
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9252385547820953e-06,
+      "loss": 0.777,
+      "step": 1439
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9188576719953635e-06,
+      "loss": 0.7777,
+      "step": 1440
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.912484869145193e-06,
+      "loss": 0.7755,
+      "step": 1441
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.906120162943515e-06,
+      "loss": 0.79,
+      "step": 1442
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8997635700810274e-06,
+      "loss": 0.7775,
+      "step": 1443
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8934151072271573e-06,
+      "loss": 0.774,
+      "step": 1444
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8870747910300064e-06,
+      "loss": 0.7744,
+      "step": 1445
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8807426381163151e-06,
+      "loss": 0.7813,
+      "step": 1446
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8744186650914148e-06,
+      "loss": 0.783,
+      "step": 1447
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8681028885391905e-06,
+      "loss": 0.7965,
+      "step": 1448
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8617953250220288e-06,
+      "loss": 0.7863,
+      "step": 1449
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8554959910807773e-06,
+      "loss": 0.7755,
+      "step": 1450
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8492049032347087e-06,
+      "loss": 0.7715,
+      "step": 1451
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8429220779814654e-06,
+      "loss": 0.7813,
+      "step": 1452
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8366475317970244e-06,
+      "loss": 0.7771,
+      "step": 1453
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8303812811356503e-06,
+      "loss": 0.7739,
+      "step": 1454
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8241233424298582e-06,
+      "loss": 0.781,
+      "step": 1455
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.81787373209036e-06,
+      "loss": 0.766,
+      "step": 1456
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8116324665060304e-06,
+      "loss": 0.7804,
+      "step": 1457
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8053995620438625e-06,
+      "loss": 0.7872,
+      "step": 1458
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7991750350489213e-06,
+      "loss": 0.781,
+      "step": 1459
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7929589018443016e-06,
+      "loss": 0.7834,
+      "step": 1460
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7867511787310882e-06,
+      "loss": 0.7857,
+      "step": 1461
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7805518819883134e-06,
+      "loss": 0.7789,
+      "step": 1462
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.774361027872909e-06,
+      "loss": 0.7979,
+      "step": 1463
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7681786326196665e-06,
+      "loss": 0.7811,
+      "step": 1464
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7620047124412e-06,
+      "loss": 0.7893,
+      "step": 1465
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.755839283527893e-06,
+      "loss": 0.7955,
+      "step": 1466
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7496823620478653e-06,
+      "loss": 0.7788,
+      "step": 1467
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.743533964146924e-06,
+      "loss": 0.7803,
+      "step": 1468
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7373941059485282e-06,
+      "loss": 0.7857,
+      "step": 1469
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7312628035537388e-06,
+      "loss": 0.7859,
+      "step": 1470
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7251400730411804e-06,
+      "loss": 0.7741,
+      "step": 1471
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7190259304670038e-06,
+      "loss": 0.7839,
+      "step": 1472
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7129203918648318e-06,
+      "loss": 0.7694,
+      "step": 1473
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.706823473245729e-06,
+      "loss": 0.7774,
+      "step": 1474
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7007351905981512e-06,
+      "loss": 0.7918,
+      "step": 1475
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6946555598879138e-06,
+      "loss": 0.7712,
+      "step": 1476
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6885845970581377e-06,
+      "loss": 0.7822,
+      "step": 1477
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6825223180292138e-06,
+      "loss": 0.781,
+      "step": 1478
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.676468738698765e-06,
+      "loss": 0.7822,
+      "step": 1479
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6704238749415958e-06,
+      "loss": 0.7771,
+      "step": 1480
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6643877426096572e-06,
+      "loss": 0.7737,
+      "step": 1481
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6583603575320002e-06,
+      "loss": 0.7777,
+      "step": 1482
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6523417355147425e-06,
+      "loss": 0.7884,
+      "step": 1483
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6463318923410183e-06,
+      "loss": 0.78,
+      "step": 1484
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.640330843770938e-06,
+      "loss": 0.7829,
+      "step": 1485
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6343386055415545e-06,
+      "loss": 0.7908,
+      "step": 1486
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6283551933668145e-06,
+      "loss": 0.7803,
+      "step": 1487
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6223806229375182e-06,
+      "loss": 0.7884,
+      "step": 1488
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6164149099212778e-06,
+      "loss": 0.7755,
+      "step": 1489
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6104580699624839e-06,
+      "loss": 0.7694,
+      "step": 1490
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6045101186822526e-06,
+      "loss": 0.7883,
+      "step": 1491
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5985710716783936e-06,
+      "loss": 0.7867,
+      "step": 1492
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5926409445253627e-06,
+      "loss": 0.7729,
+      "step": 1493
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5867197527742312e-06,
+      "loss": 0.7765,
+      "step": 1494
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5808075119526323e-06,
+      "loss": 0.7719,
+      "step": 1495
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5749042375647261e-06,
+      "loss": 0.7856,
+      "step": 1496
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5690099450911656e-06,
+      "loss": 0.7737,
+      "step": 1497
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.563124649989043e-06,
+      "loss": 0.7706,
+      "step": 1498
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5572483676918598e-06,
+      "loss": 0.7668,
+      "step": 1499
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5513811136094786e-06,
+      "loss": 0.7737,
+      "step": 1500
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5455229031280932e-06,
+      "loss": 0.7782,
+      "step": 1501
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5396737516101757e-06,
+      "loss": 0.768,
+      "step": 1502
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5338336743944427e-06,
+      "loss": 0.7769,
+      "step": 1503
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5280026867958186e-06,
+      "loss": 0.7781,
+      "step": 1504
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5221808041053877e-06,
+      "loss": 0.7793,
+      "step": 1505
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.516368041590358e-06,
+      "loss": 0.7871,
+      "step": 1506
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5105644144940202e-06,
+      "loss": 0.7795,
+      "step": 1507
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5047699380357134e-06,
+      "loss": 0.791,
+      "step": 1508
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.498984627410774e-06,
+      "loss": 0.778,
+      "step": 1509
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4932084977905043e-06,
+      "loss": 0.7817,
+      "step": 1510
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.487441564322133e-06,
+      "loss": 0.7627,
+      "step": 1511
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4816838421287693e-06,
+      "loss": 0.7888,
+      "step": 1512
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4759353463093695e-06,
+      "loss": 0.788,
+      "step": 1513
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.470196091938691e-06,
+      "loss": 0.7739,
+      "step": 1514
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4644660940672628e-06,
+      "loss": 0.7897,
+      "step": 1515
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4587453677213348e-06,
+      "loss": 0.7854,
+      "step": 1516
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4530339279028438e-06,
+      "loss": 0.7838,
+      "step": 1517
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4473317895893773e-06,
+      "loss": 0.781,
+      "step": 1518
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.441638967734127e-06,
+      "loss": 0.7835,
+      "step": 1519
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4359554772658551e-06,
+      "loss": 0.7717,
+      "step": 1520
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4302813330888498e-06,
+      "loss": 0.7726,
+      "step": 1521
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4246165500828974e-06,
+      "loss": 0.7816,
+      "step": 1522
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4189611431032286e-06,
+      "loss": 0.7787,
+      "step": 1523
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4133151269804873e-06,
+      "loss": 0.7752,
+      "step": 1524
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4076785165206962e-06,
+      "loss": 0.7813,
+      "step": 1525
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4020513265052072e-06,
+      "loss": 0.7759,
+      "step": 1526
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3964335716906696e-06,
+      "loss": 0.7644,
+      "step": 1527
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.39082526680899e-06,
+      "loss": 0.7837,
+      "step": 1528
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3852264265672965e-06,
+      "loss": 0.7711,
+      "step": 1529
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3796370656478936e-06,
+      "loss": 0.7796,
+      "step": 1530
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3740571987082274e-06,
+      "loss": 0.7772,
+      "step": 1531
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.368486840380851e-06,
+      "loss": 0.7844,
+      "step": 1532
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3629260052733783e-06,
+      "loss": 0.7759,
+      "step": 1533
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.357374707968452e-06,
+      "loss": 0.7854,
+      "step": 1534
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3518329630236987e-06,
+      "loss": 0.7837,
+      "step": 1535
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3463007849717035e-06,
+      "loss": 0.7819,
+      "step": 1536
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3407781883199561e-06,
+      "loss": 0.7899,
+      "step": 1537
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3352651875508204e-06,
+      "loss": 0.7772,
+      "step": 1538
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3297617971215022e-06,
+      "loss": 0.7821,
+      "step": 1539
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3242680314639995e-06,
+      "loss": 0.7921,
+      "step": 1540
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3187839049850714e-06,
+      "loss": 0.7823,
+      "step": 1541
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3133094320662e-06,
+      "loss": 0.7816,
+      "step": 1542
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3078446270635541e-06,
+      "loss": 0.768,
+      "step": 1543
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3023895043079476e-06,
+      "loss": 0.7868,
+      "step": 1544
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2969440781048015e-06,
+      "loss": 0.7772,
+      "step": 1545
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.291508362734113e-06,
+      "loss": 0.7858,
+      "step": 1546
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2860823724504134e-06,
+      "loss": 0.7876,
+      "step": 1547
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2806661214827286e-06,
+      "loss": 0.7792,
+      "step": 1548
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2752596240345432e-06,
+      "loss": 0.7756,
+      "step": 1549
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2698628942837698e-06,
+      "loss": 0.785,
+      "step": 1550
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.264475946382701e-06,
+      "loss": 0.7921,
+      "step": 1551
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2590987944579808e-06,
+      "loss": 0.7972,
+      "step": 1552
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.253731452610561e-06,
+      "loss": 0.7826,
+      "step": 1553
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2483739349156726e-06,
+      "loss": 0.7709,
+      "step": 1554
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.24302625542278e-06,
+      "loss": 0.7857,
+      "step": 1555
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2376884281555485e-06,
+      "loss": 0.7863,
+      "step": 1556
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2323604671118095e-06,
+      "loss": 0.7813,
+      "step": 1557
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2270423862635188e-06,
+      "loss": 0.7733,
+      "step": 1558
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2217341995567238e-06,
+      "loss": 0.7625,
+      "step": 1559
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2164359209115235e-06,
+      "loss": 0.7815,
+      "step": 1560
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2111475642220388e-06,
+      "loss": 0.779,
+      "step": 1561
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2058691433563675e-06,
+      "loss": 0.795,
+      "step": 1562
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.200600672156551e-06,
+      "loss": 0.7813,
+      "step": 1563
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1953421644385444e-06,
+      "loss": 0.7796,
+      "step": 1564
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1900936339921692e-06,
+      "loss": 0.7821,
+      "step": 1565
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.184855094581085e-06,
+      "loss": 0.7746,
+      "step": 1566
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1796265599427487e-06,
+      "loss": 0.7854,
+      "step": 1567
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1744080437883859e-06,
+      "loss": 0.7886,
+      "step": 1568
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1691995598029459e-06,
+      "loss": 0.7804,
+      "step": 1569
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.164001121645069e-06,
+      "loss": 0.7623,
+      "step": 1570
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1588127429470564e-06,
+      "loss": 0.7816,
+      "step": 1571
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1536344373148245e-06,
+      "loss": 0.7772,
+      "step": 1572
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.148466218327876e-06,
+      "loss": 0.7763,
+      "step": 1573
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1433080995392614e-06,
+      "loss": 0.7852,
+      "step": 1574
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1381600944755493e-06,
+      "loss": 0.7831,
+      "step": 1575
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.133022216636781e-06,
+      "loss": 0.783,
+      "step": 1576
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1278944794964407e-06,
+      "loss": 0.7818,
+      "step": 1577
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1227768965014246e-06,
+      "loss": 0.7689,
+      "step": 1578
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.117669481071995e-06,
+      "loss": 0.782,
+      "step": 1579
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1125722466017547e-06,
+      "loss": 0.7706,
+      "step": 1580
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1074852064576041e-06,
+      "loss": 0.7787,
+      "step": 1581
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.102408373979717e-06,
+      "loss": 0.7788,
+      "step": 1582
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.097341762481493e-06,
+      "loss": 0.7833,
+      "step": 1583
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.092285385249528e-06,
+      "loss": 0.787,
+      "step": 1584
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0872392555435857e-06,
+      "loss": 0.7761,
+      "step": 1585
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0822033865965503e-06,
+      "loss": 0.7747,
+      "step": 1586
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0771777916144028e-06,
+      "loss": 0.794,
+      "step": 1587
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0721624837761768e-06,
+      "loss": 0.7747,
+      "step": 1588
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0671574762339366e-06,
+      "loss": 0.7883,
+      "step": 1589
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.062162782112729e-06,
+      "loss": 0.7928,
+      "step": 1590
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0571784145105557e-06,
+      "loss": 0.779,
+      "step": 1591
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0522043864983428e-06,
+      "loss": 0.781,
+      "step": 1592
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0472407111198974e-06,
+      "loss": 0.7771,
+      "step": 1593
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0422874013918793e-06,
+      "loss": 0.7869,
+      "step": 1594
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0373444703037644e-06,
+      "loss": 0.779,
+      "step": 1595
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0324119308178166e-06,
+      "loss": 0.7874,
+      "step": 1596
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0274897958690434e-06,
+      "loss": 0.7762,
+      "step": 1597
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0225780783651689e-06,
+      "loss": 0.7783,
+      "step": 1598
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0176767911866026e-06,
+      "loss": 0.777,
+      "step": 1599
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.012785947186397e-06,
+      "loss": 0.7855,
+      "step": 1600
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0079055591902204e-06,
+      "loss": 0.7797,
+      "step": 1601
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0030356399963204e-06,
+      "loss": 0.7844,
+      "step": 1602
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.981762023754949e-07,
+      "loss": 0.7978,
+      "step": 1603
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.933272590710508e-07,
+      "loss": 0.782,
+      "step": 1604
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.88488822798776e-07,
+      "loss": 0.7841,
+      "step": 1605
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.836609062469066e-07,
+      "loss": 0.7782,
+      "step": 1606
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.788435220760917e-07,
+      "loss": 0.7768,
+      "step": 1607
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.740366829193587e-07,
+      "loss": 0.7843,
+      "step": 1608
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.692404013820833e-07,
+      "loss": 0.7821,
+      "step": 1609
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.644546900419533e-07,
+      "loss": 0.7754,
+      "step": 1610
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.596795614489408e-07,
+      "loss": 0.7668,
+      "step": 1611
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.549150281252633e-07,
+      "loss": 0.7791,
+      "step": 1612
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.501611025653523e-07,
+      "loss": 0.7763,
+      "step": 1613
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.454177972358258e-07,
+      "loss": 0.7893,
+      "step": 1614
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.406851245754478e-07,
+      "loss": 0.791,
+      "step": 1615
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.359630969951012e-07,
+      "loss": 0.7808,
+      "step": 1616
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.31251726877751e-07,
+      "loss": 0.7944,
+      "step": 1617
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.265510265784189e-07,
+      "loss": 0.7761,
+      "step": 1618
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.218610084241419e-07,
+      "loss": 0.7793,
+      "step": 1619
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.171816847139447e-07,
+      "loss": 0.7851,
+      "step": 1620
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.12513067718811e-07,
+      "loss": 0.786,
+      "step": 1621
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.078551696816434e-07,
+      "loss": 0.7664,
+      "step": 1622
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.032080028172357e-07,
+      "loss": 0.7798,
+      "step": 1623
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.985715793122407e-07,
+      "loss": 0.7897,
+      "step": 1624
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.939459113251408e-07,
+      "loss": 0.7796,
+      "step": 1625
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.893310109862102e-07,
+      "loss": 0.7762,
+      "step": 1626
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.847268903974854e-07,
+      "loss": 0.7721,
+      "step": 1627
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.801335616327378e-07,
+      "loss": 0.79,
+      "step": 1628
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.755510367374365e-07,
+      "loss": 0.7873,
+      "step": 1629
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.709793277287182e-07,
+      "loss": 0.7849,
+      "step": 1630
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.664184465953557e-07,
+      "loss": 0.7819,
+      "step": 1631
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.618684052977305e-07,
+      "loss": 0.7824,
+      "step": 1632
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.573292157677943e-07,
+      "loss": 0.7814,
+      "step": 1633
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.528008899090412e-07,
+      "loss": 0.7659,
+      "step": 1634
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.4828343959648e-07,
+      "loss": 0.7818,
+      "step": 1635
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.437768766765975e-07,
+      "loss": 0.7692,
+      "step": 1636
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.392812129673278e-07,
+      "loss": 0.7725,
+      "step": 1637
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.347964602580245e-07,
+      "loss": 0.7927,
+      "step": 1638
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.30322630309432e-07,
+      "loss": 0.7813,
+      "step": 1639
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.258597348536452e-07,
+      "loss": 0.7818,
+      "step": 1640
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.214077855940861e-07,
+      "loss": 0.7851,
+      "step": 1641
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.16966794205476e-07,
+      "loss": 0.7685,
+      "step": 1642
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.125367723337957e-07,
+      "loss": 0.7797,
+      "step": 1643
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.081177315962601e-07,
+      "loss": 0.7764,
+      "step": 1644
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.037096835812885e-07,
+      "loss": 0.7781,
+      "step": 1645
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.993126398484741e-07,
+      "loss": 0.7829,
+      "step": 1646
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.949266119285515e-07,
+      "loss": 0.7807,
+      "step": 1647
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.905516113233652e-07,
+      "loss": 0.7883,
+      "step": 1648
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.861876495058474e-07,
+      "loss": 0.7917,
+      "step": 1649
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.818347379199781e-07,
+      "loss": 0.8002,
+      "step": 1650
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.774928879807598e-07,
+      "loss": 0.7701,
+      "step": 1651
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.731621110741871e-07,
+      "loss": 0.7768,
+      "step": 1652
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.688424185572202e-07,
+      "loss": 0.7739,
+      "step": 1653
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.645338217577474e-07,
+      "loss": 0.7795,
+      "step": 1654
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.60236331974561e-07,
+      "loss": 0.7869,
+      "step": 1655
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.55949960477328e-07,
+      "loss": 0.7642,
+      "step": 1656
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.516747185065581e-07,
+      "loss": 0.7796,
+      "step": 1657
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.474106172735746e-07,
+      "loss": 0.7773,
+      "step": 1658
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.431576679604846e-07,
+      "loss": 0.7775,
+      "step": 1659
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.389158817201541e-07,
+      "loss": 0.7871,
+      "step": 1660
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.346852696761725e-07,
+      "loss": 0.7832,
+      "step": 1661
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.304658429228245e-07,
+      "loss": 0.7741,
+      "step": 1662
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.262576125250681e-07,
+      "loss": 0.776,
+      "step": 1663
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.220605895184946e-07,
+      "loss": 0.7727,
+      "step": 1664
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.178747849093093e-07,
+      "loss": 0.7654,
+      "step": 1665
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.13700209674294e-07,
+      "loss": 0.773,
+      "step": 1666
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.095368747607873e-07,
+      "loss": 0.7704,
+      "step": 1667
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.053847910866513e-07,
+      "loss": 0.785,
+      "step": 1668
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.012439695402395e-07,
+      "loss": 0.7778,
+      "step": 1669
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.971144209803738e-07,
+      "loss": 0.7808,
+      "step": 1670
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.929961562363141e-07,
+      "loss": 0.7758,
+      "step": 1671
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.888891861077301e-07,
+      "loss": 0.7839,
+      "step": 1672
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.847935213646705e-07,
+      "loss": 0.7727,
+      "step": 1673
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.807091727475412e-07,
+      "loss": 0.776,
+      "step": 1674
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.766361509670688e-07,
+      "loss": 0.7793,
+      "step": 1675
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.725744667042778e-07,
+      "loss": 0.7744,
+      "step": 1676
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.685241306104612e-07,
+      "loss": 0.7777,
+      "step": 1677
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.644851533071556e-07,
+      "loss": 0.7816,
+      "step": 1678
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.604575453861062e-07,
+      "loss": 0.7815,
+      "step": 1679
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.564413174092443e-07,
+      "loss": 0.7808,
+      "step": 1680
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.524364799086624e-07,
+      "loss": 0.7712,
+      "step": 1681
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.484430433865785e-07,
+      "loss": 0.7955,
+      "step": 1682
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.444610183153138e-07,
+      "loss": 0.7725,
+      "step": 1683
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.404904151372649e-07,
+      "loss": 0.7778,
+      "step": 1684
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.365312442648768e-07,
+      "loss": 0.7609,
+      "step": 1685
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.325835160806132e-07,
+      "loss": 0.7805,
+      "step": 1686
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.286472409369293e-07,
+      "loss": 0.7716,
+      "step": 1687
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.24722429156251e-07,
+      "loss": 0.7771,
+      "step": 1688
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.208090910309372e-07,
+      "loss": 0.7845,
+      "step": 1689
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.16907236823262e-07,
+      "loss": 0.7816,
+      "step": 1690
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.130168767653815e-07,
+      "loss": 0.7876,
+      "step": 1691
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.091380210593145e-07,
+      "loss": 0.7793,
+      "step": 1692
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.052706798769054e-07,
+      "loss": 0.7671,
+      "step": 1693
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.014148633598055e-07,
+      "loss": 0.7771,
+      "step": 1694
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.97570581619446e-07,
+      "loss": 0.7774,
+      "step": 1695
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.937378447370068e-07,
+      "loss": 0.7873,
+      "step": 1696
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.899166627633928e-07,
+      "loss": 0.7873,
+      "step": 1697
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.861070457192081e-07,
+      "loss": 0.7805,
+      "step": 1698
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.823090035947304e-07,
+      "loss": 0.7832,
+      "step": 1699
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.785225463498828e-07,
+      "loss": 0.7872,
+      "step": 1700
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.747476839142052e-07,
+      "loss": 0.7666,
+      "step": 1701
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.709844261868381e-07,
+      "loss": 0.7719,
+      "step": 1702
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.672327830364838e-07,
+      "loss": 0.7911,
+      "step": 1703
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.634927643013899e-07,
+      "loss": 0.7846,
+      "step": 1704
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.597643797893176e-07,
+      "loss": 0.7722,
+      "step": 1705
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.560476392775239e-07,
+      "loss": 0.7935,
+      "step": 1706
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.523425525127252e-07,
+      "loss": 0.7791,
+      "step": 1707
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.486491292110796e-07,
+      "loss": 0.7752,
+      "step": 1708
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.449673790581611e-07,
+      "loss": 0.7725,
+      "step": 1709
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.412973117089288e-07,
+      "loss": 0.7765,
+      "step": 1710
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.376389367877061e-07,
+      "loss": 0.7666,
+      "step": 1711
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.339922638881545e-07,
+      "loss": 0.7712,
+      "step": 1712
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.303573025732506e-07,
+      "loss": 0.7746,
+      "step": 1713
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.267340623752554e-07,
+      "loss": 0.7877,
+      "step": 1714
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.231225527956924e-07,
+      "loss": 0.7847,
+      "step": 1715
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.195227833053273e-07,
+      "loss": 0.7853,
+      "step": 1716
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.159347633441348e-07,
+      "loss": 0.7775,
+      "step": 1717
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.123585023212785e-07,
+      "loss": 0.7795,
+      "step": 1718
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.087940096150857e-07,
+      "loss": 0.7847,
+      "step": 1719
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.05241294573024e-07,
+      "loss": 0.7824,
+      "step": 1720
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.017003665116743e-07,
+      "loss": 0.7818,
+      "step": 1721
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.981712347167061e-07,
+      "loss": 0.7793,
+      "step": 1722
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.946539084428587e-07,
+      "loss": 0.7864,
+      "step": 1723
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.911483969139086e-07,
+      "loss": 0.7734,
+      "step": 1724
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.876547093226514e-07,
+      "loss": 0.7706,
+      "step": 1725
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.841728548308744e-07,
+      "loss": 0.7799,
+      "step": 1726
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.807028425693372e-07,
+      "loss": 0.7784,
+      "step": 1727
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.772446816377408e-07,
+      "loss": 0.7896,
+      "step": 1728
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7379838110471034e-07,
+      "loss": 0.7763,
+      "step": 1729
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7036395000776556e-07,
+      "loss": 0.7878,
+      "step": 1730
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.669413973533038e-07,
+      "loss": 0.7761,
+      "step": 1731
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.6353073211656886e-07,
+      "loss": 0.7722,
+      "step": 1732
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.6013196324163346e-07,
+      "loss": 0.7723,
+      "step": 1733
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5674509964137136e-07,
+      "loss": 0.7814,
+      "step": 1734
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5337015019743915e-07,
+      "loss": 0.7734,
+      "step": 1735
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5000712376024826e-07,
+      "loss": 0.7812,
+      "step": 1736
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4665602914894137e-07,
+      "loss": 0.7628,
+      "step": 1737
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4331687515137614e-07,
+      "loss": 0.7798,
+      "step": 1738
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3998967052409357e-07,
+      "loss": 0.779,
+      "step": 1739
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3667442399229985e-07,
+      "loss": 0.7969,
+      "step": 1740
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3337114424984115e-07,
+      "loss": 0.7699,
+      "step": 1741
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.30079839959186e-07,
+      "loss": 0.7746,
+      "step": 1742
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.2680051975139526e-07,
+      "loss": 0.7815,
+      "step": 1743
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.2353319222610265e-07,
+      "loss": 0.7892,
+      "step": 1744
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.2027786595149555e-07,
+      "loss": 0.7847,
+      "step": 1745
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1703454946428635e-07,
+      "loss": 0.7746,
+      "step": 1746
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1380325126969545e-07,
+      "loss": 0.7818,
+      "step": 1747
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1058397984142405e-07,
+      "loss": 0.7781,
+      "step": 1748
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.0737674362163804e-07,
+      "loss": 0.7751,
+      "step": 1749
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.041815510209396e-07,
+      "loss": 0.7701,
+      "step": 1750
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.009984104183479e-07,
+      "loss": 0.7634,
+      "step": 1751
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9782733016128006e-07,
+      "loss": 0.7837,
+      "step": 1752
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9466831856552315e-07,
+      "loss": 0.7763,
+      "step": 1753
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9152138391521766e-07,
+      "loss": 0.7728,
+      "step": 1754
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.883865344628307e-07,
+      "loss": 0.7746,
+      "step": 1755
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.852637784291424e-07,
+      "loss": 0.7781,
+      "step": 1756
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8215312400321504e-07,
+      "loss": 0.7773,
+      "step": 1757
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.790545793423761e-07,
+      "loss": 0.7724,
+      "step": 1758
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.759681525721998e-07,
+      "loss": 0.7784,
+      "step": 1759
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.728938517864794e-07,
+      "loss": 0.7777,
+      "step": 1760
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6983168504721e-07,
+      "loss": 0.7722,
+      "step": 1761
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.667816603845681e-07,
+      "loss": 0.7686,
+      "step": 1762
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.637437857968873e-07,
+      "loss": 0.7832,
+      "step": 1763
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.60718069250639e-07,
+      "loss": 0.7754,
+      "step": 1764
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.577045186804118e-07,
+      "loss": 0.7818,
+      "step": 1765
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.547031419888919e-07,
+      "loss": 0.7773,
+      "step": 1766
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.517139470468395e-07,
+      "loss": 0.7734,
+      "step": 1767
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.4873694169306915e-07,
+      "loss": 0.7843,
+      "step": 1768
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.4577213373443163e-07,
+      "loss": 0.7891,
+      "step": 1769
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.4281953094578877e-07,
+      "loss": 0.7779,
+      "step": 1770
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3987914106999865e-07,
+      "loss": 0.7689,
+      "step": 1771
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.369509718178887e-07,
+      "loss": 0.7704,
+      "step": 1772
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3403503086824383e-07,
+      "loss": 0.7816,
+      "step": 1773
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3113132586777786e-07,
+      "loss": 0.7815,
+      "step": 1774
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2823986443111856e-07,
+      "loss": 0.7778,
+      "step": 1775
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2536065414078724e-07,
+      "loss": 0.7801,
+      "step": 1776
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2249370254717747e-07,
+      "loss": 0.7855,
+      "step": 1777
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.196390171685343e-07,
+      "loss": 0.7744,
+      "step": 1778
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.167966054909377e-07,
+      "loss": 0.7669,
+      "step": 1779
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1396647496828245e-07,
+      "loss": 0.7955,
+      "step": 1780
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.111486330222541e-07,
+      "loss": 0.7777,
+      "step": 1781
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0834308704231485e-07,
+      "loss": 0.7897,
+      "step": 1782
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.05549844385683e-07,
+      "loss": 0.77,
+      "step": 1783
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0276891237731085e-07,
+      "loss": 0.7847,
+      "step": 1784
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0000029830986933e-07,
+      "loss": 0.7912,
+      "step": 1785
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.97244009443724e-07,
+      "loss": 0.7789,
+      "step": 1786
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.945000530069242e-07,
+      "loss": 0.7926,
+      "step": 1787
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.917684361951728e-07,
+      "loss": 0.7734,
+      "step": 1788
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.890491661718187e-07,
+      "loss": 0.7795,
+      "step": 1789
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8634225006782867e-07,
+      "loss": 0.7925,
+      "step": 1790
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.836476949817762e-07,
+      "loss": 0.78,
+      "step": 1791
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.809655079798179e-07,
+      "loss": 0.7773,
+      "step": 1792
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7829569609567566e-07,
+      "loss": 0.7784,
+      "step": 1793
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.75638266330619e-07,
+      "loss": 0.787,
+      "step": 1794
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7299322565344956e-07,
+      "loss": 0.7622,
+      "step": 1795
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7036058100047723e-07,
+      "loss": 0.7755,
+      "step": 1796
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6774033927550567e-07,
+      "loss": 0.7726,
+      "step": 1797
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.65132507349814e-07,
+      "loss": 0.7738,
+      "step": 1798
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6253709206213676e-07,
+      "loss": 0.7722,
+      "step": 1799
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.599541002186479e-07,
+      "loss": 0.779,
+      "step": 1800
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.573835385929424e-07,
+      "loss": 0.7811,
+      "step": 1801
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5482541392601924e-07,
+      "loss": 0.7742,
+      "step": 1802
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.522797329262616e-07,
+      "loss": 0.772,
+      "step": 1803
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.497465022694207e-07,
+      "loss": 0.776,
+      "step": 1804
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4722572859859907e-07,
+      "loss": 0.7933,
+      "step": 1805
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.447174185242324e-07,
+      "loss": 0.7831,
+      "step": 1806
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.422215786240706e-07,
+      "loss": 0.7863,
+      "step": 1807
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.397382154431621e-07,
+      "loss": 0.7807,
+      "step": 1808
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.372673354938393e-07,
+      "loss": 0.787,
+      "step": 1809
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3480894525569564e-07,
+      "loss": 0.7748,
+      "step": 1810
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3236305117557257e-07,
+      "loss": 0.7707,
+      "step": 1811
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2992965966754378e-07,
+      "loss": 0.7805,
+      "step": 1812
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.275087771128931e-07,
+      "loss": 0.7832,
+      "step": 1813
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.251004098601034e-07,
+      "loss": 0.7813,
+      "step": 1814
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2270456422483655e-07,
+      "loss": 0.7792,
+      "step": 1815
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2032124648992015e-07,
+      "loss": 0.7807,
+      "step": 1816
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1795046290532528e-07,
+      "loss": 0.782,
+      "step": 1817
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1559221968815547e-07,
+      "loss": 0.7722,
+      "step": 1818
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.132465230226305e-07,
+      "loss": 0.7747,
+      "step": 1819
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.109133790600648e-07,
+      "loss": 0.7887,
+      "step": 1820
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.08592793918857e-07,
+      "loss": 0.7707,
+      "step": 1821
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.062847736844703e-07,
+      "loss": 0.7614,
+      "step": 1822
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0398932440941989e-07,
+      "loss": 0.785,
+      "step": 1823
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0170645211325335e-07,
+      "loss": 0.7809,
+      "step": 1824
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.994361627825364e-07,
+      "loss": 0.7986,
+      "step": 1825
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9717846237084005e-07,
+      "loss": 0.776,
+      "step": 1826
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9493335679871883e-07,
+      "loss": 0.7705,
+      "step": 1827
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9270085195370048e-07,
+      "loss": 0.7749,
+      "step": 1828
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9048095369026908e-07,
+      "loss": 0.7744,
+      "step": 1829
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8827366782984913e-07,
+      "loss": 0.7894,
+      "step": 1830
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.860790001607904e-07,
+      "loss": 0.7724,
+      "step": 1831
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.838969564383525e-07,
+      "loss": 0.7821,
+      "step": 1832
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8172754238469148e-07,
+      "loss": 0.7627,
+      "step": 1833
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7957076368884274e-07,
+      "loss": 0.7693,
+      "step": 1834
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7742662600670645e-07,
+      "loss": 0.7677,
+      "step": 1835
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7529513496103322e-07,
+      "loss": 0.7786,
+      "step": 1836
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7317629614141128e-07,
+      "loss": 0.7868,
+      "step": 1837
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7107011510424766e-07,
+      "loss": 0.7796,
+      "step": 1838
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.689765973727564e-07,
+      "loss": 0.7787,
+      "step": 1839
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6689574843694433e-07,
+      "loss": 0.7829,
+      "step": 1840
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6482757375359482e-07,
+      "loss": 0.7782,
+      "step": 1841
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6277207874625444e-07,
+      "loss": 0.786,
+      "step": 1842
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6072926880521922e-07,
+      "loss": 0.7836,
+      "step": 1843
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5869914928752117e-07,
+      "loss": 0.7757,
+      "step": 1844
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5668172551691174e-07,
+      "loss": 0.7685,
+      "step": 1845
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.546770027838479e-07,
+      "loss": 0.7786,
+      "step": 1846
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.526849863454838e-07,
+      "loss": 0.7872,
+      "step": 1847
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5070568142564912e-07,
+      "loss": 0.7695,
+      "step": 1848
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4873909321483915e-07,
+      "loss": 0.7744,
+      "step": 1849
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4678522687020414e-07,
+      "loss": 0.7791,
+      "step": 1850
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4484408751552992e-07,
+      "loss": 0.7836,
+      "step": 1851
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4291568024122848e-07,
+      "loss": 0.7707,
+      "step": 1852
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4100001010432295e-07,
+      "loss": 0.7822,
+      "step": 1853
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.390970821284343e-07,
+      "loss": 0.7877,
+      "step": 1854
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3720690130377024e-07,
+      "loss": 0.7752,
+      "step": 1855
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3532947258710905e-07,
+      "loss": 0.7804,
+      "step": 1856
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.334648009017886e-07,
+      "loss": 0.7805,
+      "step": 1857
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3161289113769405e-07,
+      "loss": 0.7706,
+      "step": 1858
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2977374815124178e-07,
+      "loss": 0.7703,
+      "step": 1859
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2794737676536993e-07,
+      "loss": 0.7859,
+      "step": 1860
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2613378176952462e-07,
+      "loss": 0.7795,
+      "step": 1861
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2433296791964754e-07,
+      "loss": 0.7829,
+      "step": 1862
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.225449399381634e-07,
+      "loss": 0.771,
+      "step": 1863
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2076970251396593e-07,
+      "loss": 0.7826,
+      "step": 1864
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1900726030241005e-07,
+      "loss": 0.7703,
+      "step": 1865
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1725761792529378e-07,
+      "loss": 0.7788,
+      "step": 1866
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1552077997085021e-07,
+      "loss": 0.7809,
+      "step": 1867
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1379675099373489e-07,
+      "loss": 0.7742,
+      "step": 1868
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.12085535515013e-07,
+      "loss": 0.7752,
+      "step": 1869
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1038713802214718e-07,
+      "loss": 0.7722,
+      "step": 1870
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0870156296898637e-07,
+      "loss": 0.7781,
+      "step": 1871
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0702881477575589e-07,
+      "loss": 0.7942,
+      "step": 1872
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0536889782904236e-07,
+      "loss": 0.7662,
+      "step": 1873
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0372181648178436e-07,
+      "loss": 0.7849,
+      "step": 1874
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0208757505326016e-07,
+      "loss": 0.776,
+      "step": 1875
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.004661778290783e-07,
+      "loss": 0.7763,
+      "step": 1876
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.88576290611637e-08,
+      "loss": 0.7605,
+      "step": 1877
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.726193296774767e-08,
+      "loss": 0.7744,
+      "step": 1878
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.567909373335794e-08,
+      "loss": 0.7754,
+      "step": 1879
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.410911550880474e-08,
+      "loss": 0.7735,
+      "step": 1880
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.255200241117424e-08,
+      "loss": 0.7669,
+      "step": 1881
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.100775852381227e-08,
+      "loss": 0.774,
+      "step": 1882
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.947638789632007e-08,
+      "loss": 0.7846,
+      "step": 1883
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.795789454453862e-08,
+      "loss": 0.7785,
+      "step": 1884
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.645228245053761e-08,
+      "loss": 0.7809,
+      "step": 1885
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.495955556261204e-08,
+      "loss": 0.7847,
+      "step": 1886
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.347971779526232e-08,
+      "loss": 0.7803,
+      "step": 1887
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.201277302919086e-08,
+      "loss": 0.7729,
+      "step": 1888
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.055872511128936e-08,
+      "loss": 0.7819,
+      "step": 1889
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 7.911757785462882e-08,
+      "loss": 0.7796,
+      "step": 1890
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.768933503845056e-08,
+      "loss": 0.7804,
+      "step": 1891
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.627400040815414e-08,
+      "loss": 0.7822,
+      "step": 1892
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.487157767529008e-08,
+      "loss": 0.7828,
+      "step": 1893
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.34820705175482e-08,
+      "loss": 0.7732,
+      "step": 1894
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.210548257874927e-08,
+      "loss": 0.7856,
+      "step": 1895
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.074181746883402e-08,
+      "loss": 0.7561,
+      "step": 1896
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.939107876385632e-08,
+      "loss": 0.7729,
+      "step": 1897
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.805327000596995e-08,
+      "loss": 0.7828,
+      "step": 1898
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.672839470342308e-08,
+      "loss": 0.7873,
+      "step": 1899
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.54164563305465e-08,
+      "loss": 0.7641,
+      "step": 1900
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.411745832774596e-08,
+      "loss": 0.7709,
+      "step": 1901
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.283140410149213e-08,
+      "loss": 0.7681,
+      "step": 1902
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.15582970243117e-08,
+      "loss": 0.7813,
+      "step": 1903
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.029814043478022e-08,
+      "loss": 0.7802,
+      "step": 1904
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.905093763751091e-08,
+      "loss": 0.7721,
+      "step": 1905
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.781669190314809e-08,
+      "loss": 0.7849,
+      "step": 1906
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.6595406468357705e-08,
+      "loss": 0.7766,
+      "step": 1907
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.538708453581787e-08,
+      "loss": 0.7714,
+      "step": 1908
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.4191729274212235e-08,
+      "loss": 0.7673,
+      "step": 1909
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.3009343818219985e-08,
+      "loss": 0.7735,
+      "step": 1910
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.183993126850972e-08,
+      "loss": 0.7701,
+      "step": 1911
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.068349469173006e-08,
+      "loss": 0.7708,
+      "step": 1912
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.954003712050015e-08,
+      "loss": 0.7782,
+      "step": 1913
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.840956155340415e-08,
+      "loss": 0.7849,
+      "step": 1914
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.729207095498345e-08,
+      "loss": 0.7799,
+      "step": 1915
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.618756825572612e-08,
+      "loss": 0.7709,
+      "step": 1916
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.5096056352061377e-08,
+      "loss": 0.7786,
+      "step": 1917
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.40175381063529e-08,
+      "loss": 0.7809,
+      "step": 1918
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.295201634688828e-08,
+      "loss": 0.7939,
+      "step": 1919
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.189949386787462e-08,
+      "loss": 0.7827,
+      "step": 1920
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.085997342942849e-08,
+      "loss": 0.7682,
+      "step": 1921
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.9833457757572636e-08,
+      "loss": 0.7704,
+      "step": 1922
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.8819949544223744e-08,
+      "loss": 0.7661,
+      "step": 1923
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.781945144718912e-08,
+      "loss": 0.7681,
+      "step": 1924
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.683196609015782e-08,
+      "loss": 0.7721,
+      "step": 1925
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.585749606269562e-08,
+      "loss": 0.7776,
+      "step": 1926
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.489604392023727e-08,
+      "loss": 0.7688,
+      "step": 1927
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.394761218407705e-08,
+      "loss": 0.773,
+      "step": 1928
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.301220334136879e-08,
+      "loss": 0.7911,
+      "step": 1929
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.2089819845111946e-08,
+      "loss": 0.773,
+      "step": 1930
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.1180464114149437e-08,
+      "loss": 0.7976,
+      "step": 1931
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.0284138533160924e-08,
+      "loss": 0.7709,
+      "step": 1932
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.9400845452656756e-08,
+      "loss": 0.7849,
+      "step": 1933
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8530587188968508e-08,
+      "loss": 0.7729,
+      "step": 1934
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.767336602424786e-08,
+      "loss": 0.7792,
+      "step": 1935
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6829184206457194e-08,
+      "loss": 0.7685,
+      "step": 1936
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.599804394936567e-08,
+      "loss": 0.7795,
+      "step": 1937
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5179947432540376e-08,
+      "loss": 0.7831,
+      "step": 1938
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.437489680134464e-08,
+      "loss": 0.7828,
+      "step": 1939
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.358289416693027e-08,
+      "loss": 0.7704,
+      "step": 1940
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.280394160623145e-08,
+      "loss": 0.7786,
+      "step": 1941
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2038041161960288e-08,
+      "loss": 0.7805,
+      "step": 1942
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.128519484260183e-08,
+      "loss": 0.7707,
+      "step": 1943
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0545404622407396e-08,
+      "loss": 0.7866,
+      "step": 1944
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.981867244139124e-08,
+      "loss": 0.7878,
+      "step": 1945
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.91050002053228e-08,
+      "loss": 0.7802,
+      "step": 1946
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.840438978572556e-08,
+      "loss": 0.7756,
+      "step": 1947
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7716843019867646e-08,
+      "loss": 0.7806,
+      "step": 1948
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.704236171076068e-08,
+      "loss": 0.7773,
+      "step": 1949
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6380947627153143e-08,
+      "loss": 0.7765,
+      "step": 1950
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5732602503526483e-08,
+      "loss": 0.7767,
+      "step": 1951
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.509732804009012e-08,
+      "loss": 0.7564,
+      "step": 1952
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4475125902776999e-08,
+      "loss": 0.7724,
+      "step": 1953
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.386599772324082e-08,
+      "loss": 0.7573,
+      "step": 1954
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3269945098847714e-08,
+      "loss": 0.7772,
+      "step": 1955
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.268696959267679e-08,
+      "loss": 0.7786,
+      "step": 1956
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2117072733514035e-08,
+      "loss": 0.7869,
+      "step": 1957
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.156025601584676e-08,
+      "loss": 0.7848,
+      "step": 1958
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1016520899861938e-08,
+      "loss": 0.7808,
+      "step": 1959
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0485868811441757e-08,
+      "loss": 0.786,
+      "step": 1960
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.968301142159188e-09,
+      "loss": 0.7876,
+      "step": 1961
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.463819249275751e-09,
+      "loss": 0.7762,
+      "step": 1962
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.972424455735429e-09,
+      "loss": 0.7752,
+      "step": 1963
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.494118050164646e-09,
+      "loss": 0.7794,
+      "step": 1964
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.02890128686562e-09,
+      "loss": 0.7827,
+      "step": 1965
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.576775385815249e-09,
+      "loss": 0.7828,
+      "step": 1966
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.137741532660669e-09,
+      "loss": 0.7769,
+      "step": 1967
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.711800878718144e-09,
+      "loss": 0.7725,
+      "step": 1968
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.298954540965851e-09,
+      "loss": 0.7793,
+      "step": 1969
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.899203602046655e-09,
+      "loss": 0.7727,
+      "step": 1970
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.512549110259225e-09,
+      "loss": 0.7853,
+      "step": 1971
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.138992079561367e-09,
+      "loss": 0.773,
+      "step": 1972
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.778533489561699e-09,
+      "loss": 0.7947,
+      "step": 1973
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.431174285521866e-09,
+      "loss": 0.7799,
+      "step": 1974
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.096915378349886e-09,
+      "loss": 0.7837,
+      "step": 1975
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.775757644601808e-09,
+      "loss": 0.7731,
+      "step": 1976
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.4677019264750534e-09,
+      "loss": 0.776,
+      "step": 1977
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.1727490318111953e-09,
+      "loss": 0.7807,
+      "step": 1978
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8908997340887367e-09,
+      "loss": 0.7676,
+      "step": 1979
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.6221547724253337e-09,
+      "loss": 0.769,
+      "step": 1980
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.366514851573354e-09,
+      "loss": 0.7667,
+      "step": 1981
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1239806419176556e-09,
+      "loss": 0.7728,
+      "step": 1982
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8945527794772543e-09,
+      "loss": 0.7774,
+      "step": 1983
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6782318658992159e-09,
+      "loss": 0.7879,
+      "step": 1984
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4750184684597657e-09,
+      "loss": 0.7758,
+      "step": 1985
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2849131200631804e-09,
+      "loss": 0.7757,
+      "step": 1986
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1079163192373455e-09,
+      "loss": 0.7791,
+      "step": 1987
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.440285301370865e-10,
+      "loss": 0.784,
+      "step": 1988
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.932501825380634e-10,
+      "loss": 0.773,
+      "step": 1989
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.555816718389896e-10,
+      "loss": 0.779,
+      "step": 1990
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.31023359060523e-10,
+      "loss": 0.766,
+      "step": 1991
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.1957557084082447e-10,
+      "loss": 0.785,
+      "step": 1992
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.2123859943944404e-10,
+      "loss": 0.783,
+      "step": 1993
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.360127027339898e-10,
+      "loss": 0.7789,
+      "step": 1994
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.6389810421846286e-10,
+      "loss": 0.7775,
+      "step": 1995
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0489499300603279e-10,
+      "loss": 0.7692,
+      "step": 1996
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.900352382515184e-11,
+      "loss": 0.7737,
+      "step": 1997
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.622381702066523e-11,
+      "loss": 0.7709,
+      "step": 1998
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.55595855325597e-12,
+      "loss": 0.785,
+      "step": 1999
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 0.777,
+      "step": 2000
+    },
+    {
+      "epoch": 1.0,
+      "step": 2000,
+      "total_flos": 7.1263985286089015e+19,
+      "train_loss": 0.8273731040954589,
+      "train_runtime": 16224.5195,
+      "train_samples_per_second": 15.779,
+      "train_steps_per_second": 0.123
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 2000,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 9223372036854775807,
+  "save_steps": 1000,
+  "total_flos": 7.1263985286089015e+19,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}