diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,35785 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 5.0,
+  "global_step": 5960,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9996644295302016e-05,
+      "loss": 1.4355,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9993288590604027e-05,
+      "loss": 1.5078,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.998993288590604e-05,
+      "loss": 1.4473,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9986577181208055e-05,
+      "loss": 1.4355,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.998322147651007e-05,
+      "loss": 1.4541,
+      "step": 5
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.997986577181208e-05,
+      "loss": 1.376,
+      "step": 6
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9976510067114098e-05,
+      "loss": 1.3252,
+      "step": 7
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.997315436241611e-05,
+      "loss": 1.2822,
+      "step": 8
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9969798657718123e-05,
+      "loss": 1.3066,
+      "step": 9
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9966442953020134e-05,
+      "loss": 1.2588,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.996308724832215e-05,
+      "loss": 1.3457,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9959731543624163e-05,
+      "loss": 1.2363,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9956375838926177e-05,
+      "loss": 1.2139,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9953020134228188e-05,
+      "loss": 1.2939,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9949664429530202e-05,
+      "loss": 1.1904,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9946308724832216e-05,
+      "loss": 1.2822,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.994295302013423e-05,
+      "loss": 1.2344,
+      "step": 17
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.993959731543624e-05,
+      "loss": 1.2275,
+      "step": 18
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9936241610738256e-05,
+      "loss": 1.2227,
+      "step": 19
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.993288590604027e-05,
+      "loss": 1.1826,
+      "step": 20
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9929530201342284e-05,
+      "loss": 1.2441,
+      "step": 21
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.99261744966443e-05,
+      "loss": 1.2578,
+      "step": 22
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.992281879194631e-05,
+      "loss": 1.2051,
+      "step": 23
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9919463087248324e-05,
+      "loss": 1.2812,
+      "step": 24
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9916107382550338e-05,
+      "loss": 1.2236,
+      "step": 25
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9912751677852352e-05,
+      "loss": 1.2031,
+      "step": 26
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9909395973154363e-05,
+      "loss": 1.2266,
+      "step": 27
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.9906040268456378e-05,
+      "loss": 1.1895,
+      "step": 28
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.990268456375839e-05,
+      "loss": 1.2168,
+      "step": 29
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9899328859060406e-05,
+      "loss": 1.2148,
+      "step": 30
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9895973154362417e-05,
+      "loss": 1.1543,
+      "step": 31
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.989261744966443e-05,
+      "loss": 1.2178,
+      "step": 32
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9889261744966442e-05,
+      "loss": 1.2246,
+      "step": 33
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.988590604026846e-05,
+      "loss": 1.2275,
+      "step": 34
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.988255033557047e-05,
+      "loss": 1.1914,
+      "step": 35
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9879194630872485e-05,
+      "loss": 1.2051,
+      "step": 36
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.98758389261745e-05,
+      "loss": 1.1914,
+      "step": 37
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.987248322147651e-05,
+      "loss": 1.1982,
+      "step": 38
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9869127516778524e-05,
+      "loss": 1.1719,
+      "step": 39
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.986577181208054e-05,
+      "loss": 1.1465,
+      "step": 40
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9862416107382553e-05,
+      "loss": 1.1504,
+      "step": 41
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9859060402684564e-05,
+      "loss": 1.1641,
+      "step": 42
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9855704697986578e-05,
+      "loss": 1.1396,
+      "step": 43
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9852348993288592e-05,
+      "loss": 1.1445,
+      "step": 44
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9848993288590607e-05,
+      "loss": 1.1846,
+      "step": 45
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9845637583892618e-05,
+      "loss": 1.1631,
+      "step": 46
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9842281879194632e-05,
+      "loss": 1.1221,
+      "step": 47
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9838926174496646e-05,
+      "loss": 1.1553,
+      "step": 48
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.983557046979866e-05,
+      "loss": 1.1475,
+      "step": 49
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.983221476510067e-05,
+      "loss": 1.1436,
+      "step": 50
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9828859060402686e-05,
+      "loss": 1.1797,
+      "step": 51
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.98255033557047e-05,
+      "loss": 1.1992,
+      "step": 52
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9822147651006714e-05,
+      "loss": 1.1904,
+      "step": 53
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9818791946308725e-05,
+      "loss": 1.1602,
+      "step": 54
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.981543624161074e-05,
+      "loss": 1.1182,
+      "step": 55
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9812080536912754e-05,
+      "loss": 1.1689,
+      "step": 56
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9808724832214768e-05,
+      "loss": 1.1982,
+      "step": 57
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.980536912751678e-05,
+      "loss": 1.1533,
+      "step": 58
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9802013422818793e-05,
+      "loss": 1.1924,
+      "step": 59
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9798657718120807e-05,
+      "loss": 1.1787,
+      "step": 60
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.979530201342282e-05,
+      "loss": 1.1484,
+      "step": 61
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9791946308724836e-05,
+      "loss": 1.1465,
+      "step": 62
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9788590604026847e-05,
+      "loss": 1.1738,
+      "step": 63
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.978523489932886e-05,
+      "loss": 1.1836,
+      "step": 64
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9781879194630872e-05,
+      "loss": 1.1396,
+      "step": 65
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.977852348993289e-05,
+      "loss": 1.1162,
+      "step": 66
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.97751677852349e-05,
+      "loss": 1.166,
+      "step": 67
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9771812080536915e-05,
+      "loss": 1.1641,
+      "step": 68
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9768456375838926e-05,
+      "loss": 1.1689,
+      "step": 69
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9765100671140943e-05,
+      "loss": 1.1377,
+      "step": 70
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9761744966442954e-05,
+      "loss": 1.1553,
+      "step": 71
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.975838926174497e-05,
+      "loss": 1.209,
+      "step": 72
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.975503355704698e-05,
+      "loss": 1.1621,
+      "step": 73
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9751677852348994e-05,
+      "loss": 1.1318,
+      "step": 74
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9748322147651008e-05,
+      "loss": 1.1455,
+      "step": 75
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9744966442953022e-05,
+      "loss": 1.1328,
+      "step": 76
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9741610738255036e-05,
+      "loss": 1.1338,
+      "step": 77
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9738255033557047e-05,
+      "loss": 1.1289,
+      "step": 78
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.973489932885906e-05,
+      "loss": 1.165,
+      "step": 79
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9731543624161076e-05,
+      "loss": 1.1416,
+      "step": 80
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.972818791946309e-05,
+      "loss": 1.0947,
+      "step": 81
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.97248322147651e-05,
+      "loss": 1.167,
+      "step": 82
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9721476510067115e-05,
+      "loss": 1.1245,
+      "step": 83
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.971812080536913e-05,
+      "loss": 1.1406,
+      "step": 84
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9714765100671144e-05,
+      "loss": 1.1611,
+      "step": 85
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9711409395973155e-05,
+      "loss": 1.167,
+      "step": 86
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.970805369127517e-05,
+      "loss": 1.1396,
+      "step": 87
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9704697986577183e-05,
+      "loss": 1.124,
+      "step": 88
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9701342281879197e-05,
+      "loss": 1.1338,
+      "step": 89
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.969798657718121e-05,
+      "loss": 1.1377,
+      "step": 90
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9694630872483223e-05,
+      "loss": 1.1484,
+      "step": 91
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9691275167785237e-05,
+      "loss": 1.1318,
+      "step": 92
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.968791946308725e-05,
+      "loss": 1.1377,
+      "step": 93
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9684563758389262e-05,
+      "loss": 1.1201,
+      "step": 94
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9681208053691276e-05,
+      "loss": 1.1094,
+      "step": 95
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.967785234899329e-05,
+      "loss": 1.1504,
+      "step": 96
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9674496644295305e-05,
+      "loss": 1.1025,
+      "step": 97
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9671140939597316e-05,
+      "loss": 1.1025,
+      "step": 98
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.966778523489933e-05,
+      "loss": 1.1338,
+      "step": 99
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9664429530201344e-05,
+      "loss": 1.167,
+      "step": 100
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.966107382550336e-05,
+      "loss": 1.1543,
+      "step": 101
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9657718120805373e-05,
+      "loss": 1.1475,
+      "step": 102
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9654362416107384e-05,
+      "loss": 1.1162,
+      "step": 103
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9651006711409398e-05,
+      "loss": 1.1289,
+      "step": 104
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.964765100671141e-05,
+      "loss": 1.1392,
+      "step": 105
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9644295302013427e-05,
+      "loss": 1.1045,
+      "step": 106
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9640939597315437e-05,
+      "loss": 1.1646,
+      "step": 107
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9637583892617452e-05,
+      "loss": 1.1294,
+      "step": 108
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9634228187919463e-05,
+      "loss": 1.1074,
+      "step": 109
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.963087248322148e-05,
+      "loss": 1.1934,
+      "step": 110
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.962751677852349e-05,
+      "loss": 1.0933,
+      "step": 111
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9624161073825505e-05,
+      "loss": 1.1348,
+      "step": 112
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.962080536912752e-05,
+      "loss": 1.1816,
+      "step": 113
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.961744966442953e-05,
+      "loss": 1.1143,
+      "step": 114
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9614093959731545e-05,
+      "loss": 1.1367,
+      "step": 115
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.961073825503356e-05,
+      "loss": 1.0859,
+      "step": 116
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9607382550335573e-05,
+      "loss": 1.1787,
+      "step": 117
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9604026845637584e-05,
+      "loss": 1.1357,
+      "step": 118
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.96006711409396e-05,
+      "loss": 1.1338,
+      "step": 119
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9597315436241613e-05,
+      "loss": 1.1475,
+      "step": 120
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9593959731543627e-05,
+      "loss": 1.1123,
+      "step": 121
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9590604026845638e-05,
+      "loss": 1.166,
+      "step": 122
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9587248322147652e-05,
+      "loss": 1.0869,
+      "step": 123
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9583892617449667e-05,
+      "loss": 1.1338,
+      "step": 124
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.958053691275168e-05,
+      "loss": 1.1299,
+      "step": 125
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.957718120805369e-05,
+      "loss": 1.1357,
+      "step": 126
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9573825503355706e-05,
+      "loss": 1.0967,
+      "step": 127
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.957046979865772e-05,
+      "loss": 1.1548,
+      "step": 128
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9567114093959735e-05,
+      "loss": 1.1445,
+      "step": 129
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9563758389261745e-05,
+      "loss": 1.0898,
+      "step": 130
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.956040268456376e-05,
+      "loss": 1.0869,
+      "step": 131
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9557046979865774e-05,
+      "loss": 1.1006,
+      "step": 132
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9553691275167788e-05,
+      "loss": 1.1289,
+      "step": 133
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.95503355704698e-05,
+      "loss": 1.0977,
+      "step": 134
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9546979865771813e-05,
+      "loss": 1.1016,
+      "step": 135
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9543624161073828e-05,
+      "loss": 1.0796,
+      "step": 136
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9540268456375842e-05,
+      "loss": 1.1182,
+      "step": 137
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9536912751677856e-05,
+      "loss": 1.1113,
+      "step": 138
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9533557046979867e-05,
+      "loss": 1.0947,
+      "step": 139
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953020134228188e-05,
+      "loss": 1.1113,
+      "step": 140
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9526845637583892e-05,
+      "loss": 1.1533,
+      "step": 141
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.952348993288591e-05,
+      "loss": 1.1064,
+      "step": 142
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.952013422818792e-05,
+      "loss": 1.0903,
+      "step": 143
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9516778523489935e-05,
+      "loss": 1.1123,
+      "step": 144
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9513422818791946e-05,
+      "loss": 1.0718,
+      "step": 145
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9510067114093964e-05,
+      "loss": 1.1074,
+      "step": 146
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9506711409395974e-05,
+      "loss": 1.0957,
+      "step": 147
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.950335570469799e-05,
+      "loss": 1.0981,
+      "step": 148
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.95e-05,
+      "loss": 1.126,
+      "step": 149
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9496644295302014e-05,
+      "loss": 1.1143,
+      "step": 150
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9493288590604028e-05,
+      "loss": 1.124,
+      "step": 151
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9489932885906042e-05,
+      "loss": 1.1104,
+      "step": 152
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9486577181208057e-05,
+      "loss": 1.0493,
+      "step": 153
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9483221476510068e-05,
+      "loss": 1.1255,
+      "step": 154
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9479865771812082e-05,
+      "loss": 1.1533,
+      "step": 155
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9476510067114096e-05,
+      "loss": 1.1279,
+      "step": 156
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.947315436241611e-05,
+      "loss": 1.1104,
+      "step": 157
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946979865771812e-05,
+      "loss": 1.1396,
+      "step": 158
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9466442953020136e-05,
+      "loss": 1.0879,
+      "step": 159
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946308724832215e-05,
+      "loss": 1.166,
+      "step": 160
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9459731543624164e-05,
+      "loss": 1.0713,
+      "step": 161
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9456375838926175e-05,
+      "loss": 1.1284,
+      "step": 162
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.945302013422819e-05,
+      "loss": 1.0938,
+      "step": 163
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9449664429530204e-05,
+      "loss": 1.1016,
+      "step": 164
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9446308724832218e-05,
+      "loss": 1.0718,
+      "step": 165
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.944295302013423e-05,
+      "loss": 1.1279,
+      "step": 166
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9439597315436243e-05,
+      "loss": 1.1523,
+      "step": 167
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9436241610738257e-05,
+      "loss": 1.1348,
+      "step": 168
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.943288590604027e-05,
+      "loss": 1.1709,
+      "step": 169
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9429530201342282e-05,
+      "loss": 1.1006,
+      "step": 170
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9426174496644297e-05,
+      "loss": 1.0947,
+      "step": 171
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.942281879194631e-05,
+      "loss": 1.1172,
+      "step": 172
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9419463087248325e-05,
+      "loss": 1.1162,
+      "step": 173
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9416107382550336e-05,
+      "loss": 1.0898,
+      "step": 174
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.941275167785235e-05,
+      "loss": 1.124,
+      "step": 175
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9409395973154365e-05,
+      "loss": 1.1226,
+      "step": 176
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9406040268456376e-05,
+      "loss": 1.1558,
+      "step": 177
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9402684563758393e-05,
+      "loss": 1.1533,
+      "step": 178
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9399328859060404e-05,
+      "loss": 1.1191,
+      "step": 179
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.939597315436242e-05,
+      "loss": 1.0898,
+      "step": 180
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.939261744966443e-05,
+      "loss": 1.0908,
+      "step": 181
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9389261744966447e-05,
+      "loss": 1.0996,
+      "step": 182
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9385906040268458e-05,
+      "loss": 1.1367,
+      "step": 183
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9382550335570472e-05,
+      "loss": 1.1436,
+      "step": 184
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9379194630872483e-05,
+      "loss": 1.0781,
+      "step": 185
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9375838926174497e-05,
+      "loss": 1.0918,
+      "step": 186
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.937248322147651e-05,
+      "loss": 1.1191,
+      "step": 187
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9369127516778526e-05,
+      "loss": 1.1279,
+      "step": 188
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9365771812080537e-05,
+      "loss": 1.1074,
+      "step": 189
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.936241610738255e-05,
+      "loss": 1.1416,
+      "step": 190
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9359060402684565e-05,
+      "loss": 1.0737,
+      "step": 191
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.935570469798658e-05,
+      "loss": 1.0977,
+      "step": 192
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9352348993288594e-05,
+      "loss": 1.084,
+      "step": 193
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9348993288590605e-05,
+      "loss": 1.1543,
+      "step": 194
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.934563758389262e-05,
+      "loss": 1.1045,
+      "step": 195
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9342281879194633e-05,
+      "loss": 1.126,
+      "step": 196
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9338926174496648e-05,
+      "loss": 1.0952,
+      "step": 197
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.933557046979866e-05,
+      "loss": 1.0581,
+      "step": 198
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9332214765100673e-05,
+      "loss": 1.0801,
+      "step": 199
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9328859060402687e-05,
+      "loss": 1.1133,
+      "step": 200
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.93255033557047e-05,
+      "loss": 1.0854,
+      "step": 201
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9322147651006712e-05,
+      "loss": 1.1279,
+      "step": 202
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9318791946308726e-05,
+      "loss": 1.1348,
+      "step": 203
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9315436241610737e-05,
+      "loss": 1.0859,
+      "step": 204
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9312080536912755e-05,
+      "loss": 1.1338,
+      "step": 205
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9308724832214766e-05,
+      "loss": 1.0898,
+      "step": 206
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.930536912751678e-05,
+      "loss": 1.1104,
+      "step": 207
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9302013422818794e-05,
+      "loss": 1.1113,
+      "step": 208
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.929865771812081e-05,
+      "loss": 1.0913,
+      "step": 209
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.929530201342282e-05,
+      "loss": 1.1299,
+      "step": 210
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9291946308724834e-05,
+      "loss": 1.1318,
+      "step": 211
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9288590604026848e-05,
+      "loss": 1.0542,
+      "step": 212
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.928523489932886e-05,
+      "loss": 1.1416,
+      "step": 213
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9281879194630873e-05,
+      "loss": 1.1201,
+      "step": 214
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9278523489932887e-05,
+      "loss": 1.1357,
+      "step": 215
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9275167785234902e-05,
+      "loss": 1.0986,
+      "step": 216
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9271812080536913e-05,
+      "loss": 1.083,
+      "step": 217
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.926845637583893e-05,
+      "loss": 1.126,
+      "step": 218
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.926510067114094e-05,
+      "loss": 1.125,
+      "step": 219
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9261744966442955e-05,
+      "loss": 1.0459,
+      "step": 220
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9258389261744966e-05,
+      "loss": 1.0986,
+      "step": 221
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.925503355704698e-05,
+      "loss": 1.1055,
+      "step": 222
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9251677852348995e-05,
+      "loss": 1.1099,
+      "step": 223
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.924832214765101e-05,
+      "loss": 1.1221,
+      "step": 224
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.924496644295302e-05,
+      "loss": 1.1211,
+      "step": 225
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9241610738255034e-05,
+      "loss": 1.1299,
+      "step": 226
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.923825503355705e-05,
+      "loss": 1.0869,
+      "step": 227
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9234899328859063e-05,
+      "loss": 1.0845,
+      "step": 228
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9231543624161074e-05,
+      "loss": 1.0967,
+      "step": 229
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9228187919463088e-05,
+      "loss": 1.0771,
+      "step": 230
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9224832214765102e-05,
+      "loss": 1.124,
+      "step": 231
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9221476510067117e-05,
+      "loss": 1.0732,
+      "step": 232
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.921812080536913e-05,
+      "loss": 1.1455,
+      "step": 233
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9214765100671142e-05,
+      "loss": 1.0986,
+      "step": 234
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9211409395973156e-05,
+      "loss": 1.0908,
+      "step": 235
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.920805369127517e-05,
+      "loss": 1.0757,
+      "step": 236
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9204697986577185e-05,
+      "loss": 1.1406,
+      "step": 237
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9201342281879195e-05,
+      "loss": 1.1055,
+      "step": 238
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.919798657718121e-05,
+      "loss": 1.0542,
+      "step": 239
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.919463087248322e-05,
+      "loss": 1.0591,
+      "step": 240
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9191275167785238e-05,
+      "loss": 1.1104,
+      "step": 241
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.918791946308725e-05,
+      "loss": 1.103,
+      "step": 242
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9184563758389263e-05,
+      "loss": 1.0752,
+      "step": 243
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9181208053691274e-05,
+      "loss": 1.0508,
+      "step": 244
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9177852348993292e-05,
+      "loss": 1.0928,
+      "step": 245
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9174496644295303e-05,
+      "loss": 1.082,
+      "step": 246
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9171140939597317e-05,
+      "loss": 1.1143,
+      "step": 247
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.916778523489933e-05,
+      "loss": 1.1177,
+      "step": 248
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9164429530201342e-05,
+      "loss": 1.1201,
+      "step": 249
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9161073825503357e-05,
+      "loss": 1.0361,
+      "step": 250
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.915771812080537e-05,
+      "loss": 1.0225,
+      "step": 251
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9154362416107385e-05,
+      "loss": 1.0825,
+      "step": 252
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9151006711409396e-05,
+      "loss": 1.0996,
+      "step": 253
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.914765100671141e-05,
+      "loss": 1.0967,
+      "step": 254
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9144295302013425e-05,
+      "loss": 1.0693,
+      "step": 255
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.914093959731544e-05,
+      "loss": 1.1289,
+      "step": 256
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.913758389261745e-05,
+      "loss": 1.103,
+      "step": 257
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9134228187919464e-05,
+      "loss": 1.1074,
+      "step": 258
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9130872483221478e-05,
+      "loss": 1.0718,
+      "step": 259
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9127516778523493e-05,
+      "loss": 1.0986,
+      "step": 260
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9124161073825503e-05,
+      "loss": 1.0713,
+      "step": 261
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9120805369127518e-05,
+      "loss": 1.104,
+      "step": 262
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9117449664429532e-05,
+      "loss": 1.124,
+      "step": 263
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9114093959731546e-05,
+      "loss": 1.1055,
+      "step": 264
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9110738255033557e-05,
+      "loss": 1.1006,
+      "step": 265
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.910738255033557e-05,
+      "loss": 1.0991,
+      "step": 266
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9104026845637586e-05,
+      "loss": 1.0811,
+      "step": 267
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.91006711409396e-05,
+      "loss": 1.0879,
+      "step": 268
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.909731543624161e-05,
+      "loss": 1.1187,
+      "step": 269
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9093959731543625e-05,
+      "loss": 1.1221,
+      "step": 270
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.909060402684564e-05,
+      "loss": 1.0835,
+      "step": 271
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9087248322147654e-05,
+      "loss": 1.0659,
+      "step": 272
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9083892617449668e-05,
+      "loss": 1.085,
+      "step": 273
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.908053691275168e-05,
+      "loss": 1.0898,
+      "step": 274
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9077181208053693e-05,
+      "loss": 1.1299,
+      "step": 275
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9073825503355704e-05,
+      "loss": 1.084,
+      "step": 276
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.907046979865772e-05,
+      "loss": 1.0952,
+      "step": 277
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9067114093959733e-05,
+      "loss": 1.0532,
+      "step": 278
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9063758389261747e-05,
+      "loss": 1.1006,
+      "step": 279
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9060402684563758e-05,
+      "loss": 1.1025,
+      "step": 280
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9057046979865775e-05,
+      "loss": 1.084,
+      "step": 281
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9053691275167786e-05,
+      "loss": 1.106,
+      "step": 282
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.90503355704698e-05,
+      "loss": 1.1201,
+      "step": 283
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.904697986577181e-05,
+      "loss": 1.0664,
+      "step": 284
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.904362416107383e-05,
+      "loss": 1.0986,
+      "step": 285
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.904026845637584e-05,
+      "loss": 1.126,
+      "step": 286
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9036912751677854e-05,
+      "loss": 1.1299,
+      "step": 287
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.903355704697987e-05,
+      "loss": 1.0669,
+      "step": 288
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.903020134228188e-05,
+      "loss": 1.1104,
+      "step": 289
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9026845637583894e-05,
+      "loss": 1.0605,
+      "step": 290
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9023489932885908e-05,
+      "loss": 1.0962,
+      "step": 291
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9020134228187922e-05,
+      "loss": 1.0566,
+      "step": 292
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9016778523489933e-05,
+      "loss": 1.0762,
+      "step": 293
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9013422818791947e-05,
+      "loss": 1.1064,
+      "step": 294
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.901006711409396e-05,
+      "loss": 1.1064,
+      "step": 295
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9006711409395976e-05,
+      "loss": 1.1143,
+      "step": 296
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9003355704697987e-05,
+      "loss": 1.0898,
+      "step": 297
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9e-05,
+      "loss": 1.0898,
+      "step": 298
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.8996644295302015e-05,
+      "loss": 1.04,
+      "step": 299
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.899328859060403e-05,
+      "loss": 1.062,
+      "step": 300
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.898993288590604e-05,
+      "loss": 1.1191,
+      "step": 301
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.8986577181208055e-05,
+      "loss": 1.0957,
+      "step": 302
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.898322147651007e-05,
+      "loss": 1.0752,
+      "step": 303
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.8979865771812083e-05,
+      "loss": 1.0298,
+      "step": 304
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.8976510067114094e-05,
+      "loss": 1.0786,
+      "step": 305
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.897315436241611e-05,
+      "loss": 1.0732,
+      "step": 306
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.8969798657718123e-05,
+      "loss": 1.0806,
+      "step": 307
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.8966442953020137e-05,
+      "loss": 1.103,
+      "step": 308
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.8963087248322148e-05,
+      "loss": 1.1191,
+      "step": 309
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.8959731543624162e-05,
+      "loss": 1.1172,
+      "step": 310
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.8956375838926176e-05,
+      "loss": 1.0557,
+      "step": 311
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.895302013422819e-05,
+      "loss": 1.106,
+      "step": 312
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.8949664429530205e-05,
+      "loss": 1.0781,
+      "step": 313
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.8946308724832216e-05,
+      "loss": 1.0762,
+      "step": 314
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.894295302013423e-05,
+      "loss": 1.085,
+      "step": 315
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.893959731543624e-05,
+      "loss": 1.0781,
+      "step": 316
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.893624161073826e-05,
+      "loss": 1.0859,
+      "step": 317
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.893288590604027e-05,
+      "loss": 1.0391,
+      "step": 318
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.8929530201342284e-05,
+      "loss": 1.1279,
+      "step": 319
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.8926174496644295e-05,
+      "loss": 1.085,
+      "step": 320
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.8922818791946312e-05,
+      "loss": 1.0811,
+      "step": 321
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.8919463087248323e-05,
+      "loss": 1.0601,
+      "step": 322
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.8916107382550338e-05,
+      "loss": 1.1035,
+      "step": 323
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.891275167785235e-05,
+      "loss": 1.0771,
+      "step": 324
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.8909395973154363e-05,
+      "loss": 1.0938,
+      "step": 325
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.8906040268456377e-05,
+      "loss": 1.1006,
+      "step": 326
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.890268456375839e-05,
+      "loss": 1.0688,
+      "step": 327
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8899328859060406e-05,
+      "loss": 1.061,
+      "step": 328
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8895973154362416e-05,
+      "loss": 1.1328,
+      "step": 329
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.889261744966443e-05,
+      "loss": 1.0879,
+      "step": 330
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8889261744966445e-05,
+      "loss": 1.082,
+      "step": 331
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.888590604026846e-05,
+      "loss": 1.0742,
+      "step": 332
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.888255033557047e-05,
+      "loss": 1.0908,
+      "step": 333
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8879194630872484e-05,
+      "loss": 1.1045,
+      "step": 334
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.88758389261745e-05,
+      "loss": 1.1328,
+      "step": 335
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8872483221476513e-05,
+      "loss": 1.0737,
+      "step": 336
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8869127516778524e-05,
+      "loss": 1.1396,
+      "step": 337
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8865771812080538e-05,
+      "loss": 1.0879,
+      "step": 338
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.8862416107382552e-05,
+      "loss": 1.0781,
+      "step": 339
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8859060402684567e-05,
+      "loss": 1.1162,
+      "step": 340
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8855704697986578e-05,
+      "loss": 1.0508,
+      "step": 341
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8852348993288592e-05,
+      "loss": 1.1504,
+      "step": 342
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8848993288590606e-05,
+      "loss": 1.0869,
+      "step": 343
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.884563758389262e-05,
+      "loss": 1.0728,
+      "step": 344
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.884228187919463e-05,
+      "loss": 1.0366,
+      "step": 345
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8838926174496646e-05,
+      "loss": 1.0898,
+      "step": 346
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.883557046979866e-05,
+      "loss": 1.1094,
+      "step": 347
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8832214765100674e-05,
+      "loss": 1.1172,
+      "step": 348
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8828859060402685e-05,
+      "loss": 1.0928,
+      "step": 349
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.88255033557047e-05,
+      "loss": 1.0645,
+      "step": 350
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.8822147651006713e-05,
+      "loss": 1.0503,
+      "step": 351
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8818791946308724e-05,
+      "loss": 1.1289,
+      "step": 352
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8815436241610742e-05,
+      "loss": 1.084,
+      "step": 353
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8812080536912753e-05,
+      "loss": 1.0767,
+      "step": 354
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8808724832214767e-05,
+      "loss": 1.061,
+      "step": 355
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8805369127516778e-05,
+      "loss": 1.0508,
+      "step": 356
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8802013422818796e-05,
+      "loss": 1.1279,
+      "step": 357
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8798657718120807e-05,
+      "loss": 1.123,
+      "step": 358
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.879530201342282e-05,
+      "loss": 1.1172,
+      "step": 359
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8791946308724832e-05,
+      "loss": 1.0635,
+      "step": 360
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8788590604026846e-05,
+      "loss": 1.0459,
+      "step": 361
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.878523489932886e-05,
+      "loss": 1.0786,
+      "step": 362
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.8781879194630875e-05,
+      "loss": 1.0698,
+      "step": 363
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.877852348993289e-05,
+      "loss": 1.0679,
+      "step": 364
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.87751677852349e-05,
+      "loss": 1.0742,
+      "step": 365
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8771812080536914e-05,
+      "loss": 1.0259,
+      "step": 366
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.876845637583893e-05,
+      "loss": 1.063,
+      "step": 367
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8765100671140943e-05,
+      "loss": 1.0928,
+      "step": 368
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8761744966442953e-05,
+      "loss": 1.0586,
+      "step": 369
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8758389261744968e-05,
+      "loss": 1.0825,
+      "step": 370
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8755033557046982e-05,
+      "loss": 1.1045,
+      "step": 371
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8751677852348996e-05,
+      "loss": 1.0801,
+      "step": 372
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8748322147651007e-05,
+      "loss": 1.1064,
+      "step": 373
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.874496644295302e-05,
+      "loss": 1.0322,
+      "step": 374
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.8741610738255036e-05,
+      "loss": 1.0986,
+      "step": 375
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.873825503355705e-05,
+      "loss": 1.0713,
+      "step": 376
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.873489932885906e-05,
+      "loss": 1.0908,
+      "step": 377
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8731543624161075e-05,
+      "loss": 1.0586,
+      "step": 378
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.872818791946309e-05,
+      "loss": 1.1045,
+      "step": 379
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8724832214765104e-05,
+      "loss": 1.0786,
+      "step": 380
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8721476510067115e-05,
+      "loss": 1.063,
+      "step": 381
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.871812080536913e-05,
+      "loss": 1.0508,
+      "step": 382
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8714765100671143e-05,
+      "loss": 1.0923,
+      "step": 383
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8711409395973157e-05,
+      "loss": 1.0586,
+      "step": 384
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.870805369127517e-05,
+      "loss": 1.0615,
+      "step": 385
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8704697986577183e-05,
+      "loss": 1.085,
+      "step": 386
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.8701342281879197e-05,
+      "loss": 1.0786,
+      "step": 387
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.8697986577181208e-05,
+      "loss": 1.0205,
+      "step": 388
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.8694630872483225e-05,
+      "loss": 1.0713,
+      "step": 389
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.8691275167785236e-05,
+      "loss": 1.083,
+      "step": 390
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.868791946308725e-05,
+      "loss": 1.0391,
+      "step": 391
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.868456375838926e-05,
+      "loss": 1.0986,
+      "step": 392
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.868120805369128e-05,
+      "loss": 1.0771,
+      "step": 393
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.867785234899329e-05,
+      "loss": 1.0771,
+      "step": 394
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.8674496644295304e-05,
+      "loss": 1.1016,
+      "step": 395
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.8671140939597315e-05,
+      "loss": 1.0938,
+      "step": 396
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.866778523489933e-05,
+      "loss": 1.0879,
+      "step": 397
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.8664429530201344e-05,
+      "loss": 1.0562,
+      "step": 398
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.8661073825503358e-05,
+      "loss": 1.0854,
+      "step": 399
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.865771812080537e-05,
+      "loss": 1.0659,
+      "step": 400
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8654362416107383e-05,
+      "loss": 1.0322,
+      "step": 401
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8651006711409397e-05,
+      "loss": 1.0137,
+      "step": 402
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.864765100671141e-05,
+      "loss": 1.1226,
+      "step": 403
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8644295302013426e-05,
+      "loss": 1.0479,
+      "step": 404
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8640939597315437e-05,
+      "loss": 1.0474,
+      "step": 405
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.863758389261745e-05,
+      "loss": 1.0786,
+      "step": 406
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8634228187919465e-05,
+      "loss": 1.1035,
+      "step": 407
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.863087248322148e-05,
+      "loss": 1.0732,
+      "step": 408
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.862751677852349e-05,
+      "loss": 1.0474,
+      "step": 409
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.8624161073825505e-05,
+      "loss": 1.082,
+      "step": 410
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.862080536912752e-05,
+      "loss": 1.0845,
+      "step": 411
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8617449664429533e-05,
+      "loss": 1.0918,
+      "step": 412
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8614093959731544e-05,
+      "loss": 1.1016,
+      "step": 413
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.861073825503356e-05,
+      "loss": 1.083,
+      "step": 414
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.860738255033557e-05,
+      "loss": 1.0654,
+      "step": 415
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8604026845637587e-05,
+      "loss": 1.0688,
+      "step": 416
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8600671140939598e-05,
+      "loss": 1.0229,
+      "step": 417
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8597315436241612e-05,
+      "loss": 1.0732,
+      "step": 418
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8593959731543627e-05,
+      "loss": 1.1025,
+      "step": 419
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.859060402684564e-05,
+      "loss": 1.0688,
+      "step": 420
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.858724832214765e-05,
+      "loss": 1.1035,
+      "step": 421
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.8583892617449666e-05,
+      "loss": 1.0645,
+      "step": 422
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.858053691275168e-05,
+      "loss": 1.1045,
+      "step": 423
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.857718120805369e-05,
+      "loss": 1.0771,
+      "step": 424
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8573825503355705e-05,
+      "loss": 1.0957,
+      "step": 425
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.857046979865772e-05,
+      "loss": 1.0864,
+      "step": 426
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8567114093959734e-05,
+      "loss": 1.0635,
+      "step": 427
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8563758389261745e-05,
+      "loss": 1.0654,
+      "step": 428
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8560402684563762e-05,
+      "loss": 1.0547,
+      "step": 429
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8557046979865773e-05,
+      "loss": 1.082,
+      "step": 430
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8553691275167788e-05,
+      "loss": 1.1094,
+      "step": 431
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.85503355704698e-05,
+      "loss": 1.0449,
+      "step": 432
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8546979865771813e-05,
+      "loss": 1.0527,
+      "step": 433
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.8543624161073827e-05,
+      "loss": 1.083,
+      "step": 434
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.854026845637584e-05,
+      "loss": 1.0586,
+      "step": 435
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8536912751677852e-05,
+      "loss": 1.1172,
+      "step": 436
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8533557046979866e-05,
+      "loss": 1.0298,
+      "step": 437
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.853020134228188e-05,
+      "loss": 1.0938,
+      "step": 438
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8526845637583895e-05,
+      "loss": 1.0889,
+      "step": 439
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8523489932885906e-05,
+      "loss": 1.0972,
+      "step": 440
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.852013422818792e-05,
+      "loss": 1.0708,
+      "step": 441
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8516778523489934e-05,
+      "loss": 1.0454,
+      "step": 442
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.851342281879195e-05,
+      "loss": 1.0977,
+      "step": 443
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8510067114093963e-05,
+      "loss": 1.0723,
+      "step": 444
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8506711409395974e-05,
+      "loss": 1.0737,
+      "step": 445
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.8503355704697988e-05,
+      "loss": 1.0635,
+      "step": 446
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8500000000000002e-05,
+      "loss": 1.0977,
+      "step": 447
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8496644295302017e-05,
+      "loss": 1.085,
+      "step": 448
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8493288590604028e-05,
+      "loss": 1.103,
+      "step": 449
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8489932885906042e-05,
+      "loss": 1.0532,
+      "step": 450
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8486577181208053e-05,
+      "loss": 1.0566,
+      "step": 451
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.848322147651007e-05,
+      "loss": 1.0693,
+      "step": 452
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.847986577181208e-05,
+      "loss": 1.0283,
+      "step": 453
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8476510067114096e-05,
+      "loss": 1.0601,
+      "step": 454
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8473154362416106e-05,
+      "loss": 1.1021,
+      "step": 455
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8469798657718124e-05,
+      "loss": 1.0845,
+      "step": 456
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.8466442953020135e-05,
+      "loss": 1.082,
+      "step": 457
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.846308724832215e-05,
+      "loss": 1.0576,
+      "step": 458
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8459731543624164e-05,
+      "loss": 1.0615,
+      "step": 459
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8456375838926174e-05,
+      "loss": 1.0103,
+      "step": 460
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.845302013422819e-05,
+      "loss": 1.083,
+      "step": 461
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8449664429530203e-05,
+      "loss": 1.0889,
+      "step": 462
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8446308724832217e-05,
+      "loss": 1.0791,
+      "step": 463
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8442953020134228e-05,
+      "loss": 1.0293,
+      "step": 464
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8439597315436242e-05,
+      "loss": 1.041,
+      "step": 465
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8436241610738257e-05,
+      "loss": 1.0361,
+      "step": 466
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.843288590604027e-05,
+      "loss": 1.1143,
+      "step": 467
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.8429530201342282e-05,
+      "loss": 1.0664,
+      "step": 468
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.84261744966443e-05,
+      "loss": 1.0757,
+      "step": 469
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.842281879194631e-05,
+      "loss": 1.0444,
+      "step": 470
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8419463087248325e-05,
+      "loss": 1.0532,
+      "step": 471
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8416107382550336e-05,
+      "loss": 1.0283,
+      "step": 472
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.841275167785235e-05,
+      "loss": 1.0757,
+      "step": 473
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8409395973154364e-05,
+      "loss": 1.0845,
+      "step": 474
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.840604026845638e-05,
+      "loss": 1.0664,
+      "step": 475
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.840268456375839e-05,
+      "loss": 1.1245,
+      "step": 476
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8399328859060404e-05,
+      "loss": 1.1079,
+      "step": 477
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8395973154362418e-05,
+      "loss": 1.0996,
+      "step": 478
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8392617449664432e-05,
+      "loss": 1.0347,
+      "step": 479
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8389261744966443e-05,
+      "loss": 1.0684,
+      "step": 480
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.8385906040268457e-05,
+      "loss": 1.0361,
+      "step": 481
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.838255033557047e-05,
+      "loss": 1.0596,
+      "step": 482
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8379194630872486e-05,
+      "loss": 1.0552,
+      "step": 483
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.83758389261745e-05,
+      "loss": 1.0557,
+      "step": 484
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.837248322147651e-05,
+      "loss": 1.1152,
+      "step": 485
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8369127516778525e-05,
+      "loss": 1.0483,
+      "step": 486
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.836577181208054e-05,
+      "loss": 1.0293,
+      "step": 487
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8362416107382554e-05,
+      "loss": 1.0723,
+      "step": 488
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8359060402684565e-05,
+      "loss": 1.0537,
+      "step": 489
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.835570469798658e-05,
+      "loss": 1.0527,
+      "step": 490
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.835234899328859e-05,
+      "loss": 1.0977,
+      "step": 491
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8348993288590607e-05,
+      "loss": 1.0703,
+      "step": 492
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.834563758389262e-05,
+      "loss": 1.0425,
+      "step": 493
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.8342281879194633e-05,
+      "loss": 1.0601,
+      "step": 494
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8338926174496644e-05,
+      "loss": 1.0693,
+      "step": 495
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.833557046979866e-05,
+      "loss": 1.063,
+      "step": 496
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8332214765100672e-05,
+      "loss": 1.0776,
+      "step": 497
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8328859060402686e-05,
+      "loss": 1.0615,
+      "step": 498
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.83255033557047e-05,
+      "loss": 1.0889,
+      "step": 499
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.832214765100671e-05,
+      "loss": 0.9907,
+      "step": 500
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8318791946308726e-05,
+      "loss": 1.0537,
+      "step": 501
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.831543624161074e-05,
+      "loss": 1.0513,
+      "step": 502
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8312080536912754e-05,
+      "loss": 1.0718,
+      "step": 503
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8308724832214765e-05,
+      "loss": 1.1025,
+      "step": 504
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.830536912751678e-05,
+      "loss": 1.0278,
+      "step": 505
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.8302013422818794e-05,
+      "loss": 1.0898,
+      "step": 506
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8298657718120808e-05,
+      "loss": 1.0396,
+      "step": 507
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.829530201342282e-05,
+      "loss": 1.0986,
+      "step": 508
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8291946308724833e-05,
+      "loss": 1.0742,
+      "step": 509
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8288590604026847e-05,
+      "loss": 1.0288,
+      "step": 510
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8285234899328862e-05,
+      "loss": 1.0566,
+      "step": 511
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8281879194630873e-05,
+      "loss": 1.0576,
+      "step": 512
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8278523489932887e-05,
+      "loss": 1.084,
+      "step": 513
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.82751677852349e-05,
+      "loss": 1.0415,
+      "step": 514
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8271812080536915e-05,
+      "loss": 1.0557,
+      "step": 515
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8268456375838926e-05,
+      "loss": 1.0713,
+      "step": 516
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.826510067114094e-05,
+      "loss": 1.0415,
+      "step": 517
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.8261744966442955e-05,
+      "loss": 1.0513,
+      "step": 518
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.825838926174497e-05,
+      "loss": 1.0581,
+      "step": 519
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.825503355704698e-05,
+      "loss": 1.0806,
+      "step": 520
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8251677852348994e-05,
+      "loss": 1.0845,
+      "step": 521
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.824832214765101e-05,
+      "loss": 1.1162,
+      "step": 522
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8244966442953023e-05,
+      "loss": 1.0479,
+      "step": 523
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8241610738255037e-05,
+      "loss": 1.0684,
+      "step": 524
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8238255033557048e-05,
+      "loss": 1.0713,
+      "step": 525
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8234899328859062e-05,
+      "loss": 1.084,
+      "step": 526
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8231543624161073e-05,
+      "loss": 1.0181,
+      "step": 527
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.822818791946309e-05,
+      "loss": 1.042,
+      "step": 528
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8224832214765102e-05,
+      "loss": 1.0889,
+      "step": 529
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.8221476510067116e-05,
+      "loss": 1.0371,
+      "step": 530
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8218120805369127e-05,
+      "loss": 1.0503,
+      "step": 531
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8214765100671145e-05,
+      "loss": 1.0493,
+      "step": 532
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8211409395973155e-05,
+      "loss": 1.0591,
+      "step": 533
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.820805369127517e-05,
+      "loss": 1.0459,
+      "step": 534
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.820469798657718e-05,
+      "loss": 1.083,
+      "step": 535
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8201342281879195e-05,
+      "loss": 1.0747,
+      "step": 536
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.819798657718121e-05,
+      "loss": 1.0547,
+      "step": 537
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8194630872483223e-05,
+      "loss": 1.0288,
+      "step": 538
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8191275167785238e-05,
+      "loss": 1.0771,
+      "step": 539
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.818791946308725e-05,
+      "loss": 1.0435,
+      "step": 540
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8184563758389263e-05,
+      "loss": 1.0605,
+      "step": 541
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.8181208053691277e-05,
+      "loss": 1.0601,
+      "step": 542
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.817785234899329e-05,
+      "loss": 1.0415,
+      "step": 543
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8174496644295302e-05,
+      "loss": 1.0908,
+      "step": 544
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8171140939597317e-05,
+      "loss": 1.0557,
+      "step": 545
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.816778523489933e-05,
+      "loss": 1.0337,
+      "step": 546
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8164429530201345e-05,
+      "loss": 1.0522,
+      "step": 547
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8161073825503356e-05,
+      "loss": 1.0371,
+      "step": 548
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.815771812080537e-05,
+      "loss": 1.0557,
+      "step": 549
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8154362416107385e-05,
+      "loss": 1.0459,
+      "step": 550
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.81510067114094e-05,
+      "loss": 1.0977,
+      "step": 551
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.814765100671141e-05,
+      "loss": 1.0215,
+      "step": 552
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8144295302013424e-05,
+      "loss": 1.0933,
+      "step": 553
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.8140939597315438e-05,
+      "loss": 1.0562,
+      "step": 554
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8137583892617453e-05,
+      "loss": 1.0874,
+      "step": 555
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8134228187919463e-05,
+      "loss": 1.0591,
+      "step": 556
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8130872483221478e-05,
+      "loss": 1.0693,
+      "step": 557
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8127516778523492e-05,
+      "loss": 1.0771,
+      "step": 558
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8124161073825506e-05,
+      "loss": 1.0723,
+      "step": 559
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8120805369127517e-05,
+      "loss": 1.0337,
+      "step": 560
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.811744966442953e-05,
+      "loss": 1.0811,
+      "step": 561
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8114093959731546e-05,
+      "loss": 1.0625,
+      "step": 562
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8110738255033557e-05,
+      "loss": 1.0332,
+      "step": 563
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8107382550335574e-05,
+      "loss": 1.0669,
+      "step": 564
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.8104026845637585e-05,
+      "loss": 1.0464,
+      "step": 565
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.81006711409396e-05,
+      "loss": 1.0532,
+      "step": 566
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.809731543624161e-05,
+      "loss": 1.0376,
+      "step": 567
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.8093959731543628e-05,
+      "loss": 1.1084,
+      "step": 568
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.809060402684564e-05,
+      "loss": 1.0703,
+      "step": 569
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.8087248322147653e-05,
+      "loss": 1.0312,
+      "step": 570
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.8083892617449664e-05,
+      "loss": 1.0244,
+      "step": 571
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.8080536912751678e-05,
+      "loss": 1.0923,
+      "step": 572
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.8077181208053692e-05,
+      "loss": 1.0542,
+      "step": 573
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.8073825503355707e-05,
+      "loss": 1.0625,
+      "step": 574
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.8070469798657718e-05,
+      "loss": 1.001,
+      "step": 575
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.8067114093959732e-05,
+      "loss": 1.0669,
+      "step": 576
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.8063758389261746e-05,
+      "loss": 1.0132,
+      "step": 577
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.806040268456376e-05,
+      "loss": 1.0454,
+      "step": 578
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8057046979865775e-05,
+      "loss": 1.0869,
+      "step": 579
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8053691275167786e-05,
+      "loss": 1.0815,
+      "step": 580
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.80503355704698e-05,
+      "loss": 1.0469,
+      "step": 581
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8046979865771814e-05,
+      "loss": 1.0625,
+      "step": 582
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.804362416107383e-05,
+      "loss": 1.0664,
+      "step": 583
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.804026845637584e-05,
+      "loss": 1.0566,
+      "step": 584
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8036912751677854e-05,
+      "loss": 1.0806,
+      "step": 585
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8033557046979868e-05,
+      "loss": 1.0557,
+      "step": 586
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8030201342281882e-05,
+      "loss": 1.1113,
+      "step": 587
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8026845637583893e-05,
+      "loss": 1.0986,
+      "step": 588
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.8023489932885907e-05,
+      "loss": 1.0312,
+      "step": 589
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.802013422818792e-05,
+      "loss": 1.0146,
+      "step": 590
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8016778523489936e-05,
+      "loss": 1.0674,
+      "step": 591
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8013422818791947e-05,
+      "loss": 1.0791,
+      "step": 592
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.801006711409396e-05,
+      "loss": 1.0518,
+      "step": 593
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8006711409395975e-05,
+      "loss": 1.0513,
+      "step": 594
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.800335570469799e-05,
+      "loss": 1.0679,
+      "step": 595
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.8e-05,
+      "loss": 1.0234,
+      "step": 596
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7996644295302015e-05,
+      "loss": 1.0537,
+      "step": 597
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.799328859060403e-05,
+      "loss": 1.0576,
+      "step": 598
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.798993288590604e-05,
+      "loss": 1.0591,
+      "step": 599
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.7986577181208054e-05,
+      "loss": 1.0205,
+      "step": 600
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.798322147651007e-05,
+      "loss": 1.0522,
+      "step": 601
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7979865771812083e-05,
+      "loss": 1.0669,
+      "step": 602
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7976510067114094e-05,
+      "loss": 1.0835,
+      "step": 603
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.797315436241611e-05,
+      "loss": 1.0352,
+      "step": 604
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7969798657718122e-05,
+      "loss": 1.0225,
+      "step": 605
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7966442953020136e-05,
+      "loss": 1.0791,
+      "step": 606
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7963087248322147e-05,
+      "loss": 1.0835,
+      "step": 607
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.795973154362416e-05,
+      "loss": 1.064,
+      "step": 608
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7956375838926176e-05,
+      "loss": 1.0269,
+      "step": 609
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.795302013422819e-05,
+      "loss": 1.0146,
+      "step": 610
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.79496644295302e-05,
+      "loss": 1.0532,
+      "step": 611
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.7946308724832215e-05,
+      "loss": 1.085,
+      "step": 612
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.794295302013423e-05,
+      "loss": 1.0376,
+      "step": 613
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7939597315436244e-05,
+      "loss": 1.0518,
+      "step": 614
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7936241610738258e-05,
+      "loss": 1.0498,
+      "step": 615
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.793288590604027e-05,
+      "loss": 1.083,
+      "step": 616
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7929530201342283e-05,
+      "loss": 1.0317,
+      "step": 617
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7926174496644298e-05,
+      "loss": 1.0244,
+      "step": 618
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7922818791946312e-05,
+      "loss": 1.0312,
+      "step": 619
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7919463087248323e-05,
+      "loss": 1.0986,
+      "step": 620
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7916107382550337e-05,
+      "loss": 1.0537,
+      "step": 621
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.791275167785235e-05,
+      "loss": 1.043,
+      "step": 622
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7909395973154366e-05,
+      "loss": 1.0566,
+      "step": 623
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.7906040268456376e-05,
+      "loss": 1.0308,
+      "step": 624
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.790268456375839e-05,
+      "loss": 1.0322,
+      "step": 625
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.78993288590604e-05,
+      "loss": 1.1055,
+      "step": 626
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.789597315436242e-05,
+      "loss": 1.0029,
+      "step": 627
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.789261744966443e-05,
+      "loss": 1.0518,
+      "step": 628
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7889261744966444e-05,
+      "loss": 1.0811,
+      "step": 629
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.788590604026846e-05,
+      "loss": 1.0366,
+      "step": 630
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7882550335570473e-05,
+      "loss": 1.0654,
+      "step": 631
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7879194630872484e-05,
+      "loss": 1.0029,
+      "step": 632
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7875838926174498e-05,
+      "loss": 1.0625,
+      "step": 633
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7872483221476512e-05,
+      "loss": 1.0366,
+      "step": 634
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7869127516778523e-05,
+      "loss": 1.0645,
+      "step": 635
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7865771812080538e-05,
+      "loss": 1.0527,
+      "step": 636
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.7862416107382552e-05,
+      "loss": 1.0181,
+      "step": 637
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7859060402684566e-05,
+      "loss": 1.0679,
+      "step": 638
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7855704697986577e-05,
+      "loss": 1.021,
+      "step": 639
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7852348993288595e-05,
+      "loss": 0.9844,
+      "step": 640
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7848993288590605e-05,
+      "loss": 1.0312,
+      "step": 641
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.784563758389262e-05,
+      "loss": 1.0317,
+      "step": 642
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.784228187919463e-05,
+      "loss": 1.0186,
+      "step": 643
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7838926174496645e-05,
+      "loss": 1.0532,
+      "step": 644
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.783557046979866e-05,
+      "loss": 1.0728,
+      "step": 645
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7832214765100673e-05,
+      "loss": 1.0654,
+      "step": 646
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7828859060402684e-05,
+      "loss": 1.0366,
+      "step": 647
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.78255033557047e-05,
+      "loss": 1.0557,
+      "step": 648
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.7822147651006713e-05,
+      "loss": 1.0366,
+      "step": 649
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7818791946308727e-05,
+      "loss": 1.0527,
+      "step": 650
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7815436241610738e-05,
+      "loss": 1.0322,
+      "step": 651
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7812080536912752e-05,
+      "loss": 1.0444,
+      "step": 652
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7808724832214767e-05,
+      "loss": 1.0576,
+      "step": 653
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.780536912751678e-05,
+      "loss": 1.0591,
+      "step": 654
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7802013422818795e-05,
+      "loss": 1.0229,
+      "step": 655
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7798657718120806e-05,
+      "loss": 1.0967,
+      "step": 656
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.779530201342282e-05,
+      "loss": 1.04,
+      "step": 657
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7791946308724835e-05,
+      "loss": 0.999,
+      "step": 658
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.778859060402685e-05,
+      "loss": 1.0688,
+      "step": 659
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.778523489932886e-05,
+      "loss": 1.0684,
+      "step": 660
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.7781879194630874e-05,
+      "loss": 1.0859,
+      "step": 661
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7778523489932888e-05,
+      "loss": 1.0801,
+      "step": 662
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7775167785234903e-05,
+      "loss": 1.1021,
+      "step": 663
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7771812080536913e-05,
+      "loss": 1.0757,
+      "step": 664
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7768456375838928e-05,
+      "loss": 1.0229,
+      "step": 665
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.776510067114094e-05,
+      "loss": 1.0498,
+      "step": 666
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7761744966442956e-05,
+      "loss": 1.0698,
+      "step": 667
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7758389261744967e-05,
+      "loss": 1.0322,
+      "step": 668
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.775503355704698e-05,
+      "loss": 1.0283,
+      "step": 669
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7751677852348996e-05,
+      "loss": 1.002,
+      "step": 670
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.774832214765101e-05,
+      "loss": 1.0264,
+      "step": 671
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.774496644295302e-05,
+      "loss": 1.1006,
+      "step": 672
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.7741610738255035e-05,
+      "loss": 1.0122,
+      "step": 673
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.773825503355705e-05,
+      "loss": 1.0581,
+      "step": 674
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.773489932885906e-05,
+      "loss": 1.0171,
+      "step": 675
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7731543624161075e-05,
+      "loss": 1.0928,
+      "step": 676
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.772818791946309e-05,
+      "loss": 1.0117,
+      "step": 677
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7724832214765103e-05,
+      "loss": 1.1152,
+      "step": 678
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7721476510067114e-05,
+      "loss": 1.0332,
+      "step": 679
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.771812080536913e-05,
+      "loss": 1.0957,
+      "step": 680
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7714765100671143e-05,
+      "loss": 1.0503,
+      "step": 681
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7711409395973157e-05,
+      "loss": 1.0513,
+      "step": 682
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7708053691275168e-05,
+      "loss": 1.0542,
+      "step": 683
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7704697986577182e-05,
+      "loss": 1.0747,
+      "step": 684
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.7701342281879196e-05,
+      "loss": 1.0308,
+      "step": 685
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.769798657718121e-05,
+      "loss": 1.022,
+      "step": 686
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.769463087248322e-05,
+      "loss": 1.0293,
+      "step": 687
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7691275167785236e-05,
+      "loss": 1.0527,
+      "step": 688
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.768791946308725e-05,
+      "loss": 1.0742,
+      "step": 689
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7684563758389264e-05,
+      "loss": 1.0151,
+      "step": 690
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7681208053691275e-05,
+      "loss": 1.0542,
+      "step": 691
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.767785234899329e-05,
+      "loss": 1.0615,
+      "step": 692
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7674496644295304e-05,
+      "loss": 1.0522,
+      "step": 693
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7671140939597318e-05,
+      "loss": 1.0444,
+      "step": 694
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7667785234899332e-05,
+      "loss": 1.0444,
+      "step": 695
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7664429530201343e-05,
+      "loss": 1.0464,
+      "step": 696
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.7661073825503357e-05,
+      "loss": 1.0693,
+      "step": 697
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.765771812080537e-05,
+      "loss": 1.0635,
+      "step": 698
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7654362416107386e-05,
+      "loss": 1.062,
+      "step": 699
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7651006711409397e-05,
+      "loss": 1.0723,
+      "step": 700
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.764765100671141e-05,
+      "loss": 1.0693,
+      "step": 701
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7644295302013422e-05,
+      "loss": 1.0293,
+      "step": 702
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.764093959731544e-05,
+      "loss": 1.0767,
+      "step": 703
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.763758389261745e-05,
+      "loss": 1.0337,
+      "step": 704
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7634228187919465e-05,
+      "loss": 1.0742,
+      "step": 705
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7630872483221476e-05,
+      "loss": 1.0337,
+      "step": 706
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7627516778523493e-05,
+      "loss": 1.0347,
+      "step": 707
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.7624161073825504e-05,
+      "loss": 1.0747,
+      "step": 708
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.762080536912752e-05,
+      "loss": 1.0254,
+      "step": 709
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7617449664429533e-05,
+      "loss": 1.0537,
+      "step": 710
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7614093959731544e-05,
+      "loss": 1.0249,
+      "step": 711
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7610738255033558e-05,
+      "loss": 1.0205,
+      "step": 712
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7607382550335572e-05,
+      "loss": 1.022,
+      "step": 713
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7604026845637586e-05,
+      "loss": 1.0742,
+      "step": 714
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7600671140939597e-05,
+      "loss": 1.0244,
+      "step": 715
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.759731543624161e-05,
+      "loss": 1.0454,
+      "step": 716
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7593959731543626e-05,
+      "loss": 1.0308,
+      "step": 717
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.759060402684564e-05,
+      "loss": 1.043,
+      "step": 718
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.758724832214765e-05,
+      "loss": 1.0171,
+      "step": 719
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.7583892617449665e-05,
+      "loss": 1.0015,
+      "step": 720
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.758053691275168e-05,
+      "loss": 1.0073,
+      "step": 721
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7577181208053694e-05,
+      "loss": 1.0356,
+      "step": 722
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7573825503355705e-05,
+      "loss": 1.0547,
+      "step": 723
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.757046979865772e-05,
+      "loss": 1.041,
+      "step": 724
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7567114093959733e-05,
+      "loss": 1.042,
+      "step": 725
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7563758389261748e-05,
+      "loss": 1.0488,
+      "step": 726
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.756040268456376e-05,
+      "loss": 1.0498,
+      "step": 727
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7557046979865773e-05,
+      "loss": 1.0547,
+      "step": 728
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7553691275167787e-05,
+      "loss": 1.0469,
+      "step": 729
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.75503355704698e-05,
+      "loss": 1.042,
+      "step": 730
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7546979865771812e-05,
+      "loss": 0.9951,
+      "step": 731
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.7543624161073826e-05,
+      "loss": 1.0488,
+      "step": 732
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.754026845637584e-05,
+      "loss": 1.0107,
+      "step": 733
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.7536912751677855e-05,
+      "loss": 1.0298,
+      "step": 734
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.753355704697987e-05,
+      "loss": 1.0312,
+      "step": 735
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.753020134228188e-05,
+      "loss": 1.02,
+      "step": 736
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.7526845637583894e-05,
+      "loss": 1.042,
+      "step": 737
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.7523489932885905e-05,
+      "loss": 1.0249,
+      "step": 738
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.7520134228187923e-05,
+      "loss": 1.0098,
+      "step": 739
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.7516778523489934e-05,
+      "loss": 1.0161,
+      "step": 740
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.7513422818791948e-05,
+      "loss": 1.0381,
+      "step": 741
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.751006711409396e-05,
+      "loss": 1.0596,
+      "step": 742
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.7506711409395977e-05,
+      "loss": 1.0562,
+      "step": 743
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.7503355704697988e-05,
+      "loss": 1.0469,
+      "step": 744
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.7500000000000002e-05,
+      "loss": 1.0488,
+      "step": 745
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7496644295302013e-05,
+      "loss": 1.0723,
+      "step": 746
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7493288590604027e-05,
+      "loss": 1.0459,
+      "step": 747
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.748993288590604e-05,
+      "loss": 1.0107,
+      "step": 748
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7486577181208056e-05,
+      "loss": 1.0342,
+      "step": 749
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.748322147651007e-05,
+      "loss": 1.0537,
+      "step": 750
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.747986577181208e-05,
+      "loss": 1.0737,
+      "step": 751
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7476510067114095e-05,
+      "loss": 1.0801,
+      "step": 752
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.747315436241611e-05,
+      "loss": 1.0132,
+      "step": 753
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7469798657718124e-05,
+      "loss": 1.0693,
+      "step": 754
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.7466442953020134e-05,
+      "loss": 1.0444,
+      "step": 755
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.746308724832215e-05,
+      "loss": 1.0161,
+      "step": 756
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7459731543624163e-05,
+      "loss": 1.0225,
+      "step": 757
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7456375838926177e-05,
+      "loss": 1.0488,
+      "step": 758
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7453020134228188e-05,
+      "loss": 1.0586,
+      "step": 759
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7449664429530202e-05,
+      "loss": 1.0386,
+      "step": 760
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7446308724832217e-05,
+      "loss": 1.0439,
+      "step": 761
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.744295302013423e-05,
+      "loss": 1.022,
+      "step": 762
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7439597315436242e-05,
+      "loss": 1.0586,
+      "step": 763
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7436241610738256e-05,
+      "loss": 1.0278,
+      "step": 764
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.743288590604027e-05,
+      "loss": 1.0703,
+      "step": 765
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7429530201342285e-05,
+      "loss": 1.0801,
+      "step": 766
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.7426174496644296e-05,
+      "loss": 1.0264,
+      "step": 767
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.742281879194631e-05,
+      "loss": 1.0361,
+      "step": 768
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7419463087248324e-05,
+      "loss": 1.0742,
+      "step": 769
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.741610738255034e-05,
+      "loss": 1.0059,
+      "step": 770
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.741275167785235e-05,
+      "loss": 1.0518,
+      "step": 771
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7409395973154364e-05,
+      "loss": 0.9893,
+      "step": 772
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7406040268456378e-05,
+      "loss": 1.0386,
+      "step": 773
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.740268456375839e-05,
+      "loss": 1.0342,
+      "step": 774
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7399328859060406e-05,
+      "loss": 1.0518,
+      "step": 775
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7395973154362417e-05,
+      "loss": 0.9956,
+      "step": 776
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.739261744966443e-05,
+      "loss": 1.0234,
+      "step": 777
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.7389261744966442e-05,
+      "loss": 1.0649,
+      "step": 778
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.738590604026846e-05,
+      "loss": 1.0859,
+      "step": 779
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.738255033557047e-05,
+      "loss": 1.0342,
+      "step": 780
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7379194630872485e-05,
+      "loss": 1.0547,
+      "step": 781
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7375838926174496e-05,
+      "loss": 1.0249,
+      "step": 782
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.737248322147651e-05,
+      "loss": 1.0659,
+      "step": 783
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7369127516778525e-05,
+      "loss": 1.0967,
+      "step": 784
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.736577181208054e-05,
+      "loss": 1.0669,
+      "step": 785
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.736241610738255e-05,
+      "loss": 1.0669,
+      "step": 786
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7359060402684564e-05,
+      "loss": 1.0874,
+      "step": 787
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.735570469798658e-05,
+      "loss": 1.0288,
+      "step": 788
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7352348993288593e-05,
+      "loss": 1.0649,
+      "step": 789
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7348993288590607e-05,
+      "loss": 1.0142,
+      "step": 790
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7345637583892618e-05,
+      "loss": 1.0342,
+      "step": 791
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.7342281879194632e-05,
+      "loss": 1.0103,
+      "step": 792
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.7338926174496646e-05,
+      "loss": 1.0317,
+      "step": 793
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.733557046979866e-05,
+      "loss": 1.0596,
+      "step": 794
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.733221476510067e-05,
+      "loss": 1.0439,
+      "step": 795
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.7328859060402686e-05,
+      "loss": 1.0513,
+      "step": 796
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.73255033557047e-05,
+      "loss": 1.0366,
+      "step": 797
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.7322147651006714e-05,
+      "loss": 1.0854,
+      "step": 798
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.7318791946308725e-05,
+      "loss": 1.0371,
+      "step": 799
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.731543624161074e-05,
+      "loss": 1.0298,
+      "step": 800
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.731208053691275e-05,
+      "loss": 1.0879,
+      "step": 801
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.7308724832214768e-05,
+      "loss": 1.0078,
+      "step": 802
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.730536912751678e-05,
+      "loss": 1.0293,
+      "step": 803
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.7302013422818793e-05,
+      "loss": 1.0234,
+      "step": 804
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.7298657718120807e-05,
+      "loss": 1.0239,
+      "step": 805
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.7295302013422822e-05,
+      "loss": 1.0488,
+      "step": 806
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.7291946308724833e-05,
+      "loss": 1.0645,
+      "step": 807
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.7288590604026847e-05,
+      "loss": 1.0508,
+      "step": 808
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.728523489932886e-05,
+      "loss": 1.0825,
+      "step": 809
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.7281879194630872e-05,
+      "loss": 1.0029,
+      "step": 810
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.7278523489932886e-05,
+      "loss": 1.0054,
+      "step": 811
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.72751677852349e-05,
+      "loss": 1.0459,
+      "step": 812
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.7271812080536915e-05,
+      "loss": 0.9868,
+      "step": 813
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.7268456375838926e-05,
+      "loss": 1.0376,
+      "step": 814
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.7265100671140943e-05,
+      "loss": 1.0649,
+      "step": 815
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.7261744966442954e-05,
+      "loss": 1.0361,
+      "step": 816
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.725838926174497e-05,
+      "loss": 1.0771,
+      "step": 817
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.725503355704698e-05,
+      "loss": 1.0098,
+      "step": 818
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7251677852348994e-05,
+      "loss": 1.0483,
+      "step": 819
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7248322147651008e-05,
+      "loss": 1.0249,
+      "step": 820
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7244966442953022e-05,
+      "loss": 1.0278,
+      "step": 821
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7241610738255033e-05,
+      "loss": 1.042,
+      "step": 822
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7238255033557047e-05,
+      "loss": 1.0786,
+      "step": 823
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7234899328859062e-05,
+      "loss": 1.0493,
+      "step": 824
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7231543624161076e-05,
+      "loss": 1.0356,
+      "step": 825
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7228187919463087e-05,
+      "loss": 1.0732,
+      "step": 826
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.72248322147651e-05,
+      "loss": 1.0425,
+      "step": 827
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7221476510067115e-05,
+      "loss": 1.0479,
+      "step": 828
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.721812080536913e-05,
+      "loss": 1.0513,
+      "step": 829
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7214765100671144e-05,
+      "loss": 1.0225,
+      "step": 830
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7211409395973155e-05,
+      "loss": 1.0469,
+      "step": 831
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.720805369127517e-05,
+      "loss": 1.0083,
+      "step": 832
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7204697986577183e-05,
+      "loss": 1.0322,
+      "step": 833
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7201342281879198e-05,
+      "loss": 0.9937,
+      "step": 834
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.719798657718121e-05,
+      "loss": 1.0454,
+      "step": 835
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7194630872483223e-05,
+      "loss": 1.0498,
+      "step": 836
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7191275167785234e-05,
+      "loss": 1.0317,
+      "step": 837
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.718791946308725e-05,
+      "loss": 0.9927,
+      "step": 838
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7184563758389262e-05,
+      "loss": 1.0405,
+      "step": 839
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7181208053691277e-05,
+      "loss": 1.0239,
+      "step": 840
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.717785234899329e-05,
+      "loss": 1.0552,
+      "step": 841
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7174496644295305e-05,
+      "loss": 1.0264,
+      "step": 842
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7171140939597316e-05,
+      "loss": 1.0581,
+      "step": 843
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.716778523489933e-05,
+      "loss": 1.043,
+      "step": 844
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7164429530201345e-05,
+      "loss": 1.0684,
+      "step": 845
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.716107382550336e-05,
+      "loss": 1.0449,
+      "step": 846
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.715771812080537e-05,
+      "loss": 1.0312,
+      "step": 847
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7154362416107384e-05,
+      "loss": 0.9932,
+      "step": 848
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7151006711409398e-05,
+      "loss": 1.0234,
+      "step": 849
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.714765100671141e-05,
+      "loss": 1.0386,
+      "step": 850
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7144295302013427e-05,
+      "loss": 1.0459,
+      "step": 851
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7140939597315438e-05,
+      "loss": 1.0464,
+      "step": 852
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7137583892617452e-05,
+      "loss": 1.0645,
+      "step": 853
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7134228187919463e-05,
+      "loss": 1.0391,
+      "step": 854
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.713087248322148e-05,
+      "loss": 1.0332,
+      "step": 855
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.712751677852349e-05,
+      "loss": 1.0986,
+      "step": 856
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7124161073825506e-05,
+      "loss": 1.0566,
+      "step": 857
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7120805369127517e-05,
+      "loss": 1.0229,
+      "step": 858
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.711744966442953e-05,
+      "loss": 0.9771,
+      "step": 859
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7114093959731545e-05,
+      "loss": 1.0605,
+      "step": 860
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.711073825503356e-05,
+      "loss": 1.0659,
+      "step": 861
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.710738255033557e-05,
+      "loss": 1.0239,
+      "step": 862
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7104026845637584e-05,
+      "loss": 1.0195,
+      "step": 863
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.71006711409396e-05,
+      "loss": 1.0586,
+      "step": 864
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7097315436241613e-05,
+      "loss": 1.0215,
+      "step": 865
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7093959731543627e-05,
+      "loss": 1.0107,
+      "step": 866
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7090604026845638e-05,
+      "loss": 1.0435,
+      "step": 867
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7087248322147652e-05,
+      "loss": 1.0454,
+      "step": 868
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7083892617449667e-05,
+      "loss": 1.0498,
+      "step": 869
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.708053691275168e-05,
+      "loss": 1.0386,
+      "step": 870
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7077181208053692e-05,
+      "loss": 0.9893,
+      "step": 871
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7073825503355706e-05,
+      "loss": 1.0444,
+      "step": 872
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.707046979865772e-05,
+      "loss": 1.062,
+      "step": 873
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7067114093959735e-05,
+      "loss": 1.0649,
+      "step": 874
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7063758389261746e-05,
+      "loss": 1.0273,
+      "step": 875
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.706040268456376e-05,
+      "loss": 1.0723,
+      "step": 876
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.705704697986577e-05,
+      "loss": 1.0239,
+      "step": 877
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.705369127516779e-05,
+      "loss": 1.0,
+      "step": 878
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.70503355704698e-05,
+      "loss": 1.0254,
+      "step": 879
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7046979865771814e-05,
+      "loss": 1.0732,
+      "step": 880
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7043624161073828e-05,
+      "loss": 0.9971,
+      "step": 881
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7040268456375842e-05,
+      "loss": 1.0234,
+      "step": 882
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7036912751677853e-05,
+      "loss": 1.0151,
+      "step": 883
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7033557046979867e-05,
+      "loss": 0.9849,
+      "step": 884
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.703020134228188e-05,
+      "loss": 1.0181,
+      "step": 885
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7026845637583892e-05,
+      "loss": 1.0181,
+      "step": 886
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7023489932885907e-05,
+      "loss": 0.998,
+      "step": 887
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.702013422818792e-05,
+      "loss": 1.0288,
+      "step": 888
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7016778523489935e-05,
+      "loss": 1.0234,
+      "step": 889
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7013422818791946e-05,
+      "loss": 1.0508,
+      "step": 890
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7010067114093964e-05,
+      "loss": 1.0469,
+      "step": 891
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7006711409395975e-05,
+      "loss": 1.0439,
+      "step": 892
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.700335570469799e-05,
+      "loss": 1.0776,
+      "step": 893
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7e-05,
+      "loss": 1.0571,
+      "step": 894
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6996644295302014e-05,
+      "loss": 0.999,
+      "step": 895
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.699328859060403e-05,
+      "loss": 1.0356,
+      "step": 896
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6989932885906043e-05,
+      "loss": 1.0229,
+      "step": 897
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6986577181208054e-05,
+      "loss": 1.0352,
+      "step": 898
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6983221476510068e-05,
+      "loss": 1.0674,
+      "step": 899
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6979865771812082e-05,
+      "loss": 1.0615,
+      "step": 900
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6976510067114096e-05,
+      "loss": 1.0742,
+      "step": 901
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6973154362416107e-05,
+      "loss": 1.0547,
+      "step": 902
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.696979865771812e-05,
+      "loss": 1.0537,
+      "step": 903
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6966442953020136e-05,
+      "loss": 1.0376,
+      "step": 904
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.696308724832215e-05,
+      "loss": 0.9922,
+      "step": 905
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6959731543624164e-05,
+      "loss": 1.0347,
+      "step": 906
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6956375838926175e-05,
+      "loss": 1.0439,
+      "step": 907
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.695302013422819e-05,
+      "loss": 1.0205,
+      "step": 908
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6949664429530204e-05,
+      "loss": 1.0317,
+      "step": 909
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.6946308724832218e-05,
+      "loss": 0.9941,
+      "step": 910
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.694295302013423e-05,
+      "loss": 1.0234,
+      "step": 911
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.6939597315436243e-05,
+      "loss": 0.9937,
+      "step": 912
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.6936241610738254e-05,
+      "loss": 1.0503,
+      "step": 913
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.6932885906040272e-05,
+      "loss": 1.0239,
+      "step": 914
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.6929530201342283e-05,
+      "loss": 1.0278,
+      "step": 915
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.6926174496644297e-05,
+      "loss": 1.022,
+      "step": 916
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.6922818791946308e-05,
+      "loss": 1.0542,
+      "step": 917
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.6919463087248325e-05,
+      "loss": 1.043,
+      "step": 918
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.6916107382550336e-05,
+      "loss": 1.0332,
+      "step": 919
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.691275167785235e-05,
+      "loss": 1.0244,
+      "step": 920
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.6909395973154365e-05,
+      "loss": 1.0439,
+      "step": 921
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.6906040268456376e-05,
+      "loss": 1.0205,
+      "step": 922
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.690268456375839e-05,
+      "loss": 1.0571,
+      "step": 923
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.6899328859060404e-05,
+      "loss": 1.0298,
+      "step": 924
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.689597315436242e-05,
+      "loss": 1.0156,
+      "step": 925
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.689261744966443e-05,
+      "loss": 1.0586,
+      "step": 926
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.6889261744966444e-05,
+      "loss": 1.0508,
+      "step": 927
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.6885906040268458e-05,
+      "loss": 1.0688,
+      "step": 928
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.6882550335570472e-05,
+      "loss": 1.0728,
+      "step": 929
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.6879194630872483e-05,
+      "loss": 1.0312,
+      "step": 930
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.6875838926174497e-05,
+      "loss": 1.0527,
+      "step": 931
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.6872483221476512e-05,
+      "loss": 1.0381,
+      "step": 932
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.6869127516778526e-05,
+      "loss": 1.04,
+      "step": 933
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.6865771812080537e-05,
+      "loss": 1.0801,
+      "step": 934
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.686241610738255e-05,
+      "loss": 1.0073,
+      "step": 935
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.6859060402684565e-05,
+      "loss": 1.0049,
+      "step": 936
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.685570469798658e-05,
+      "loss": 0.9985,
+      "step": 937
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.685234899328859e-05,
+      "loss": 1.019,
+      "step": 938
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.6848993288590605e-05,
+      "loss": 1.0312,
+      "step": 939
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.684563758389262e-05,
+      "loss": 1.0425,
+      "step": 940
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.6842281879194633e-05,
+      "loss": 1.0479,
+      "step": 941
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.6838926174496644e-05,
+      "loss": 1.0244,
+      "step": 942
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.683557046979866e-05,
+      "loss": 1.0576,
+      "step": 943
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.6832214765100673e-05,
+      "loss": 1.0229,
+      "step": 944
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.6828859060402687e-05,
+      "loss": 1.0405,
+      "step": 945
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.68255033557047e-05,
+      "loss": 1.0498,
+      "step": 946
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.6822147651006712e-05,
+      "loss": 1.0366,
+      "step": 947
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.6818791946308727e-05,
+      "loss": 1.0244,
+      "step": 948
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.6815436241610737e-05,
+      "loss": 1.0234,
+      "step": 949
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.6812080536912755e-05,
+      "loss": 1.0063,
+      "step": 950
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.6808724832214766e-05,
+      "loss": 0.9761,
+      "step": 951
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.680536912751678e-05,
+      "loss": 1.0396,
+      "step": 952
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.680201342281879e-05,
+      "loss": 1.0532,
+      "step": 953
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.679865771812081e-05,
+      "loss": 1.0161,
+      "step": 954
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.679530201342282e-05,
+      "loss": 1.0498,
+      "step": 955
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.6791946308724834e-05,
+      "loss": 1.0586,
+      "step": 956
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.6788590604026845e-05,
+      "loss": 1.0464,
+      "step": 957
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.678523489932886e-05,
+      "loss": 1.0229,
+      "step": 958
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.6781879194630873e-05,
+      "loss": 1.041,
+      "step": 959
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6778523489932888e-05,
+      "loss": 1.063,
+      "step": 960
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6775167785234902e-05,
+      "loss": 0.999,
+      "step": 961
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6771812080536913e-05,
+      "loss": 1.0674,
+      "step": 962
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6768456375838927e-05,
+      "loss": 1.0444,
+      "step": 963
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.676510067114094e-05,
+      "loss": 1.0068,
+      "step": 964
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6761744966442956e-05,
+      "loss": 1.0254,
+      "step": 965
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6758389261744967e-05,
+      "loss": 1.0312,
+      "step": 966
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.675503355704698e-05,
+      "loss": 1.0376,
+      "step": 967
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6751677852348995e-05,
+      "loss": 1.0239,
+      "step": 968
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.674832214765101e-05,
+      "loss": 1.0146,
+      "step": 969
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.674496644295302e-05,
+      "loss": 1.0312,
+      "step": 970
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.6741610738255035e-05,
+      "loss": 1.0757,
+      "step": 971
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.673825503355705e-05,
+      "loss": 1.0127,
+      "step": 972
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6734899328859063e-05,
+      "loss": 1.0376,
+      "step": 973
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6731543624161074e-05,
+      "loss": 1.0317,
+      "step": 974
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6728187919463088e-05,
+      "loss": 1.0317,
+      "step": 975
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6724832214765103e-05,
+      "loss": 1.0327,
+      "step": 976
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6721476510067117e-05,
+      "loss": 1.0332,
+      "step": 977
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6718120805369128e-05,
+      "loss": 1.0225,
+      "step": 978
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6714765100671142e-05,
+      "loss": 1.0098,
+      "step": 979
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6711409395973156e-05,
+      "loss": 1.0552,
+      "step": 980
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.670805369127517e-05,
+      "loss": 1.0215,
+      "step": 981
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.670469798657718e-05,
+      "loss": 1.0044,
+      "step": 982
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6701342281879196e-05,
+      "loss": 0.9995,
+      "step": 983
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.669798657718121e-05,
+      "loss": 1.021,
+      "step": 984
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.669463087248322e-05,
+      "loss": 1.0786,
+      "step": 985
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.669127516778524e-05,
+      "loss": 1.0322,
+      "step": 986
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.668791946308725e-05,
+      "loss": 1.0752,
+      "step": 987
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6684563758389264e-05,
+      "loss": 1.0142,
+      "step": 988
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6681208053691275e-05,
+      "loss": 1.0269,
+      "step": 989
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6677852348993292e-05,
+      "loss": 1.0259,
+      "step": 990
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6674496644295303e-05,
+      "loss": 1.0151,
+      "step": 991
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6671140939597317e-05,
+      "loss": 0.9771,
+      "step": 992
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6667785234899328e-05,
+      "loss": 1.0127,
+      "step": 993
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6664429530201343e-05,
+      "loss": 1.0762,
+      "step": 994
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6661073825503357e-05,
+      "loss": 1.0181,
+      "step": 995
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.665771812080537e-05,
+      "loss": 1.04,
+      "step": 996
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6654362416107382e-05,
+      "loss": 1.0146,
+      "step": 997
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6651006711409396e-05,
+      "loss": 1.0117,
+      "step": 998
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.664765100671141e-05,
+      "loss": 1.0093,
+      "step": 999
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6644295302013425e-05,
+      "loss": 1.0596,
+      "step": 1000
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.664093959731544e-05,
+      "loss": 1.0122,
+      "step": 1001
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.663758389261745e-05,
+      "loss": 1.0273,
+      "step": 1002
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6634228187919464e-05,
+      "loss": 1.0698,
+      "step": 1003
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.663087248322148e-05,
+      "loss": 1.021,
+      "step": 1004
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6627516778523493e-05,
+      "loss": 1.0317,
+      "step": 1005
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6624161073825504e-05,
+      "loss": 1.0518,
+      "step": 1006
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6620805369127518e-05,
+      "loss": 1.0063,
+      "step": 1007
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6617449664429532e-05,
+      "loss": 1.0015,
+      "step": 1008
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6614093959731546e-05,
+      "loss": 1.0088,
+      "step": 1009
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6610738255033557e-05,
+      "loss": 1.0186,
+      "step": 1010
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.660738255033557e-05,
+      "loss": 1.0083,
+      "step": 1011
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6604026845637583e-05,
+      "loss": 1.0684,
+      "step": 1012
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.66006711409396e-05,
+      "loss": 1.0537,
+      "step": 1013
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.659731543624161e-05,
+      "loss": 1.0757,
+      "step": 1014
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6593959731543625e-05,
+      "loss": 1.0059,
+      "step": 1015
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.659060402684564e-05,
+      "loss": 1.0254,
+      "step": 1016
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6587248322147654e-05,
+      "loss": 1.0112,
+      "step": 1017
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6583892617449665e-05,
+      "loss": 1.0649,
+      "step": 1018
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.658053691275168e-05,
+      "loss": 0.9883,
+      "step": 1019
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6577181208053693e-05,
+      "loss": 1.0132,
+      "step": 1020
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6573825503355704e-05,
+      "loss": 0.9985,
+      "step": 1021
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.657046979865772e-05,
+      "loss": 1.0034,
+      "step": 1022
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6567114093959733e-05,
+      "loss": 1.0571,
+      "step": 1023
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6563758389261747e-05,
+      "loss": 1.0024,
+      "step": 1024
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6560402684563758e-05,
+      "loss": 1.0376,
+      "step": 1025
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6557046979865776e-05,
+      "loss": 1.022,
+      "step": 1026
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6553691275167786e-05,
+      "loss": 1.0317,
+      "step": 1027
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.65503355704698e-05,
+      "loss": 1.0513,
+      "step": 1028
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.654697986577181e-05,
+      "loss": 1.0654,
+      "step": 1029
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.654362416107383e-05,
+      "loss": 1.0303,
+      "step": 1030
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.654026845637584e-05,
+      "loss": 0.999,
+      "step": 1031
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6536912751677854e-05,
+      "loss": 1.0371,
+      "step": 1032
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6533557046979865e-05,
+      "loss": 1.0342,
+      "step": 1033
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.653020134228188e-05,
+      "loss": 1.0391,
+      "step": 1034
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6526845637583894e-05,
+      "loss": 1.064,
+      "step": 1035
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6523489932885908e-05,
+      "loss": 1.0469,
+      "step": 1036
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.652013422818792e-05,
+      "loss": 1.0503,
+      "step": 1037
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6516778523489933e-05,
+      "loss": 1.0469,
+      "step": 1038
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6513422818791948e-05,
+      "loss": 1.0312,
+      "step": 1039
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6510067114093962e-05,
+      "loss": 1.0449,
+      "step": 1040
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6506711409395976e-05,
+      "loss": 1.0386,
+      "step": 1041
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6503355704697987e-05,
+      "loss": 1.0571,
+      "step": 1042
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.65e-05,
+      "loss": 1.0273,
+      "step": 1043
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6496644295302016e-05,
+      "loss": 1.0332,
+      "step": 1044
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.649328859060403e-05,
+      "loss": 1.0459,
+      "step": 1045
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.648993288590604e-05,
+      "loss": 1.0186,
+      "step": 1046
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6486577181208055e-05,
+      "loss": 1.0278,
+      "step": 1047
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.648322147651007e-05,
+      "loss": 1.0376,
+      "step": 1048
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6479865771812084e-05,
+      "loss": 1.0732,
+      "step": 1049
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6476510067114094e-05,
+      "loss": 1.0625,
+      "step": 1050
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.647315436241611e-05,
+      "loss": 1.0566,
+      "step": 1051
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.646979865771812e-05,
+      "loss": 1.0747,
+      "step": 1052
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6466442953020137e-05,
+      "loss": 1.0298,
+      "step": 1053
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6463087248322148e-05,
+      "loss": 1.0146,
+      "step": 1054
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6459731543624162e-05,
+      "loss": 1.0176,
+      "step": 1055
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6456375838926177e-05,
+      "loss": 1.0396,
+      "step": 1056
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.645302013422819e-05,
+      "loss": 1.0298,
+      "step": 1057
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6449664429530202e-05,
+      "loss": 1.0542,
+      "step": 1058
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6446308724832216e-05,
+      "loss": 1.002,
+      "step": 1059
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.644295302013423e-05,
+      "loss": 1.0024,
+      "step": 1060
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.643959731543624e-05,
+      "loss": 1.0381,
+      "step": 1061
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6436241610738256e-05,
+      "loss": 1.0127,
+      "step": 1062
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.643288590604027e-05,
+      "loss": 1.0381,
+      "step": 1063
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6429530201342284e-05,
+      "loss": 1.0117,
+      "step": 1064
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6426174496644295e-05,
+      "loss": 1.0342,
+      "step": 1065
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6422818791946313e-05,
+      "loss": 1.0615,
+      "step": 1066
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6419463087248324e-05,
+      "loss": 1.0532,
+      "step": 1067
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6416107382550338e-05,
+      "loss": 1.0215,
+      "step": 1068
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.641275167785235e-05,
+      "loss": 1.0244,
+      "step": 1069
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6409395973154363e-05,
+      "loss": 1.0635,
+      "step": 1070
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6406040268456377e-05,
+      "loss": 0.9912,
+      "step": 1071
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.640268456375839e-05,
+      "loss": 1.0498,
+      "step": 1072
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6399328859060402e-05,
+      "loss": 0.9995,
+      "step": 1073
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6395973154362417e-05,
+      "loss": 1.0396,
+      "step": 1074
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.639261744966443e-05,
+      "loss": 1.0469,
+      "step": 1075
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6389261744966445e-05,
+      "loss": 0.9795,
+      "step": 1076
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.638590604026846e-05,
+      "loss": 1.0151,
+      "step": 1077
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.638255033557047e-05,
+      "loss": 1.0215,
+      "step": 1078
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6379194630872485e-05,
+      "loss": 0.98,
+      "step": 1079
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.63758389261745e-05,
+      "loss": 1.0288,
+      "step": 1080
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6372483221476513e-05,
+      "loss": 1.0342,
+      "step": 1081
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6369127516778524e-05,
+      "loss": 0.9966,
+      "step": 1082
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.636577181208054e-05,
+      "loss": 1.0474,
+      "step": 1083
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6362416107382553e-05,
+      "loss": 1.04,
+      "step": 1084
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6359060402684567e-05,
+      "loss": 1.0415,
+      "step": 1085
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6355704697986578e-05,
+      "loss": 1.02,
+      "step": 1086
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6352348993288592e-05,
+      "loss": 1.0117,
+      "step": 1087
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6348993288590603e-05,
+      "loss": 1.0078,
+      "step": 1088
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.634563758389262e-05,
+      "loss": 1.0405,
+      "step": 1089
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.634228187919463e-05,
+      "loss": 1.0283,
+      "step": 1090
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6338926174496646e-05,
+      "loss": 0.9937,
+      "step": 1091
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.633557046979866e-05,
+      "loss": 0.9937,
+      "step": 1092
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6332214765100674e-05,
+      "loss": 1.041,
+      "step": 1093
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6328859060402685e-05,
+      "loss": 1.0205,
+      "step": 1094
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.63255033557047e-05,
+      "loss": 1.0093,
+      "step": 1095
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6322147651006714e-05,
+      "loss": 1.0386,
+      "step": 1096
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6318791946308725e-05,
+      "loss": 1.064,
+      "step": 1097
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.631543624161074e-05,
+      "loss": 0.9888,
+      "step": 1098
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6312080536912753e-05,
+      "loss": 1.0474,
+      "step": 1099
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6308724832214767e-05,
+      "loss": 1.0151,
+      "step": 1100
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.630536912751678e-05,
+      "loss": 0.9932,
+      "step": 1101
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6302013422818796e-05,
+      "loss": 1.0303,
+      "step": 1102
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6298657718120807e-05,
+      "loss": 1.0073,
+      "step": 1103
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.629530201342282e-05,
+      "loss": 1.0547,
+      "step": 1104
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6291946308724832e-05,
+      "loss": 0.9785,
+      "step": 1105
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6288590604026846e-05,
+      "loss": 1.0049,
+      "step": 1106
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.628523489932886e-05,
+      "loss": 1.0112,
+      "step": 1107
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6281879194630875e-05,
+      "loss": 1.0073,
+      "step": 1108
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6278523489932886e-05,
+      "loss": 1.0552,
+      "step": 1109
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.62751677852349e-05,
+      "loss": 1.0234,
+      "step": 1110
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6271812080536914e-05,
+      "loss": 0.9937,
+      "step": 1111
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.626845637583893e-05,
+      "loss": 1.0303,
+      "step": 1112
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.626510067114094e-05,
+      "loss": 1.042,
+      "step": 1113
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6261744966442954e-05,
+      "loss": 1.063,
+      "step": 1114
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6258389261744968e-05,
+      "loss": 0.9692,
+      "step": 1115
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6255033557046982e-05,
+      "loss": 1.0278,
+      "step": 1116
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6251677852348997e-05,
+      "loss": 1.0259,
+      "step": 1117
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6248322147651007e-05,
+      "loss": 1.0337,
+      "step": 1118
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.624496644295302e-05,
+      "loss": 1.0181,
+      "step": 1119
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6241610738255036e-05,
+      "loss": 0.9873,
+      "step": 1120
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.623825503355705e-05,
+      "loss": 1.0083,
+      "step": 1121
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.623489932885906e-05,
+      "loss": 0.9795,
+      "step": 1122
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6231543624161075e-05,
+      "loss": 1.0234,
+      "step": 1123
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6228187919463086e-05,
+      "loss": 1.0356,
+      "step": 1124
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6224832214765104e-05,
+      "loss": 1.0283,
+      "step": 1125
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6221476510067115e-05,
+      "loss": 1.0576,
+      "step": 1126
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.621812080536913e-05,
+      "loss": 1.0361,
+      "step": 1127
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.621476510067114e-05,
+      "loss": 1.0269,
+      "step": 1128
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6211409395973158e-05,
+      "loss": 1.0122,
+      "step": 1129
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.620805369127517e-05,
+      "loss": 1.0376,
+      "step": 1130
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6204697986577183e-05,
+      "loss": 1.0337,
+      "step": 1131
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6201342281879197e-05,
+      "loss": 1.0269,
+      "step": 1132
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6197986577181208e-05,
+      "loss": 0.9941,
+      "step": 1133
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6194630872483222e-05,
+      "loss": 1.0762,
+      "step": 1134
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6191275167785237e-05,
+      "loss": 1.0439,
+      "step": 1135
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.618791946308725e-05,
+      "loss": 1.0229,
+      "step": 1136
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.618456375838926e-05,
+      "loss": 0.998,
+      "step": 1137
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6181208053691276e-05,
+      "loss": 1.0273,
+      "step": 1138
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.617785234899329e-05,
+      "loss": 1.0269,
+      "step": 1139
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6174496644295304e-05,
+      "loss": 1.0239,
+      "step": 1140
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6171140939597315e-05,
+      "loss": 1.0059,
+      "step": 1141
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.616778523489933e-05,
+      "loss": 1.0776,
+      "step": 1142
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6164429530201344e-05,
+      "loss": 1.0718,
+      "step": 1143
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6161073825503358e-05,
+      "loss": 0.9912,
+      "step": 1144
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.615771812080537e-05,
+      "loss": 0.9985,
+      "step": 1145
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6154362416107383e-05,
+      "loss": 1.0483,
+      "step": 1146
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6151006711409398e-05,
+      "loss": 1.0366,
+      "step": 1147
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6147651006711412e-05,
+      "loss": 1.0439,
+      "step": 1148
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6144295302013423e-05,
+      "loss": 1.0376,
+      "step": 1149
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.6140939597315437e-05,
+      "loss": 1.0015,
+      "step": 1150
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.613758389261745e-05,
+      "loss": 1.0625,
+      "step": 1151
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6134228187919466e-05,
+      "loss": 1.0234,
+      "step": 1152
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6130872483221476e-05,
+      "loss": 1.0244,
+      "step": 1153
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.612751677852349e-05,
+      "loss": 1.0376,
+      "step": 1154
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6124161073825505e-05,
+      "loss": 1.0444,
+      "step": 1155
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.612080536912752e-05,
+      "loss": 1.0259,
+      "step": 1156
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6117449664429534e-05,
+      "loss": 1.0273,
+      "step": 1157
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6114093959731544e-05,
+      "loss": 1.0137,
+      "step": 1158
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.611073825503356e-05,
+      "loss": 1.0078,
+      "step": 1159
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.610738255033557e-05,
+      "loss": 1.0005,
+      "step": 1160
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6104026845637587e-05,
+      "loss": 0.9517,
+      "step": 1161
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6100671140939598e-05,
+      "loss": 0.9746,
+      "step": 1162
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6097315436241612e-05,
+      "loss": 1.0483,
+      "step": 1163
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6093959731543623e-05,
+      "loss": 1.022,
+      "step": 1164
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.609060402684564e-05,
+      "loss": 0.9961,
+      "step": 1165
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6087248322147652e-05,
+      "loss": 0.9868,
+      "step": 1166
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6083892617449666e-05,
+      "loss": 0.9868,
+      "step": 1167
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6080536912751677e-05,
+      "loss": 1.0381,
+      "step": 1168
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.607718120805369e-05,
+      "loss": 1.0322,
+      "step": 1169
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6073825503355706e-05,
+      "loss": 1.0396,
+      "step": 1170
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.607046979865772e-05,
+      "loss": 1.0459,
+      "step": 1171
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6067114093959734e-05,
+      "loss": 1.0483,
+      "step": 1172
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6063758389261745e-05,
+      "loss": 1.0015,
+      "step": 1173
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.606040268456376e-05,
+      "loss": 1.0156,
+      "step": 1174
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6057046979865774e-05,
+      "loss": 1.0498,
+      "step": 1175
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6053691275167788e-05,
+      "loss": 1.0356,
+      "step": 1176
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.60503355704698e-05,
+      "loss": 1.0615,
+      "step": 1177
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6046979865771813e-05,
+      "loss": 1.0483,
+      "step": 1178
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6043624161073827e-05,
+      "loss": 0.959,
+      "step": 1179
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.604026845637584e-05,
+      "loss": 1.0347,
+      "step": 1180
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6036912751677852e-05,
+      "loss": 1.0073,
+      "step": 1181
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6033557046979867e-05,
+      "loss": 0.9878,
+      "step": 1182
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.603020134228188e-05,
+      "loss": 1.0464,
+      "step": 1183
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6026845637583895e-05,
+      "loss": 0.9922,
+      "step": 1184
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6023489932885906e-05,
+      "loss": 1.0391,
+      "step": 1185
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.602013422818792e-05,
+      "loss": 1.0259,
+      "step": 1186
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.6016778523489935e-05,
+      "loss": 0.979,
+      "step": 1187
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.601342281879195e-05,
+      "loss": 1.0229,
+      "step": 1188
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.601006711409396e-05,
+      "loss": 1.0083,
+      "step": 1189
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.6006711409395974e-05,
+      "loss": 1.043,
+      "step": 1190
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.600335570469799e-05,
+      "loss": 1.0342,
+      "step": 1191
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 0.9766,
+      "step": 1192
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5996644295302014e-05,
+      "loss": 0.9268,
+      "step": 1193
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5993288590604028e-05,
+      "loss": 0.9507,
+      "step": 1194
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5989932885906042e-05,
+      "loss": 0.981,
+      "step": 1195
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5986577181208053e-05,
+      "loss": 1.0063,
+      "step": 1196
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.598322147651007e-05,
+      "loss": 0.9629,
+      "step": 1197
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.597986577181208e-05,
+      "loss": 0.9829,
+      "step": 1198
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5976510067114096e-05,
+      "loss": 0.9639,
+      "step": 1199
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5973154362416107e-05,
+      "loss": 0.9585,
+      "step": 1200
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5969798657718124e-05,
+      "loss": 1.0088,
+      "step": 1201
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5966442953020135e-05,
+      "loss": 0.978,
+      "step": 1202
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.596308724832215e-05,
+      "loss": 0.9766,
+      "step": 1203
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.595973154362416e-05,
+      "loss": 0.9722,
+      "step": 1204
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5956375838926175e-05,
+      "loss": 0.9858,
+      "step": 1205
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.595302013422819e-05,
+      "loss": 1.002,
+      "step": 1206
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5949664429530203e-05,
+      "loss": 0.9487,
+      "step": 1207
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5946308724832214e-05,
+      "loss": 0.9824,
+      "step": 1208
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.594295302013423e-05,
+      "loss": 0.9917,
+      "step": 1209
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5939597315436243e-05,
+      "loss": 0.9771,
+      "step": 1210
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5936241610738257e-05,
+      "loss": 0.9331,
+      "step": 1211
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.593288590604027e-05,
+      "loss": 0.9849,
+      "step": 1212
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5929530201342282e-05,
+      "loss": 0.9756,
+      "step": 1213
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5926174496644296e-05,
+      "loss": 0.978,
+      "step": 1214
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.592281879194631e-05,
+      "loss": 0.9795,
+      "step": 1215
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5919463087248325e-05,
+      "loss": 1.0293,
+      "step": 1216
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5916107382550336e-05,
+      "loss": 0.958,
+      "step": 1217
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.591275167785235e-05,
+      "loss": 0.9839,
+      "step": 1218
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5909395973154364e-05,
+      "loss": 0.9595,
+      "step": 1219
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.590604026845638e-05,
+      "loss": 0.9868,
+      "step": 1220
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.590268456375839e-05,
+      "loss": 0.9741,
+      "step": 1221
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5899328859060404e-05,
+      "loss": 1.0,
+      "step": 1222
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5895973154362418e-05,
+      "loss": 0.9951,
+      "step": 1223
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5892617449664432e-05,
+      "loss": 0.9805,
+      "step": 1224
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5889261744966443e-05,
+      "loss": 0.9922,
+      "step": 1225
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5885906040268457e-05,
+      "loss": 0.9609,
+      "step": 1226
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5882550335570472e-05,
+      "loss": 0.9648,
+      "step": 1227
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5879194630872486e-05,
+      "loss": 0.9604,
+      "step": 1228
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5875838926174497e-05,
+      "loss": 0.9917,
+      "step": 1229
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.587248322147651e-05,
+      "loss": 0.9741,
+      "step": 1230
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5869127516778525e-05,
+      "loss": 0.9043,
+      "step": 1231
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.586577181208054e-05,
+      "loss": 0.9453,
+      "step": 1232
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.586241610738255e-05,
+      "loss": 0.9741,
+      "step": 1233
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5859060402684565e-05,
+      "loss": 0.9712,
+      "step": 1234
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.585570469798658e-05,
+      "loss": 0.979,
+      "step": 1235
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.585234899328859e-05,
+      "loss": 0.978,
+      "step": 1236
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5848993288590608e-05,
+      "loss": 0.9648,
+      "step": 1237
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.584563758389262e-05,
+      "loss": 1.0312,
+      "step": 1238
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5842281879194633e-05,
+      "loss": 0.9902,
+      "step": 1239
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5838926174496644e-05,
+      "loss": 0.9873,
+      "step": 1240
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.583557046979866e-05,
+      "loss": 0.9419,
+      "step": 1241
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5832214765100672e-05,
+      "loss": 0.938,
+      "step": 1242
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5828859060402687e-05,
+      "loss": 1.0063,
+      "step": 1243
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5825503355704697e-05,
+      "loss": 0.9692,
+      "step": 1244
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5822147651006712e-05,
+      "loss": 0.9829,
+      "step": 1245
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5818791946308726e-05,
+      "loss": 0.9624,
+      "step": 1246
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.581543624161074e-05,
+      "loss": 0.9194,
+      "step": 1247
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.581208053691275e-05,
+      "loss": 0.9971,
+      "step": 1248
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5808724832214765e-05,
+      "loss": 1.0029,
+      "step": 1249
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.580536912751678e-05,
+      "loss": 0.9854,
+      "step": 1250
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5802013422818794e-05,
+      "loss": 1.0039,
+      "step": 1251
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5798657718120808e-05,
+      "loss": 0.998,
+      "step": 1252
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.579530201342282e-05,
+      "loss": 1.0024,
+      "step": 1253
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5791946308724833e-05,
+      "loss": 0.9668,
+      "step": 1254
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5788590604026848e-05,
+      "loss": 0.9634,
+      "step": 1255
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5785234899328862e-05,
+      "loss": 0.9595,
+      "step": 1256
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5781879194630873e-05,
+      "loss": 1.0068,
+      "step": 1257
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5778523489932887e-05,
+      "loss": 0.9927,
+      "step": 1258
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.57751677852349e-05,
+      "loss": 0.9741,
+      "step": 1259
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5771812080536916e-05,
+      "loss": 0.9902,
+      "step": 1260
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5768456375838927e-05,
+      "loss": 0.9888,
+      "step": 1261
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.576510067114094e-05,
+      "loss": 0.9575,
+      "step": 1262
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5761744966442952e-05,
+      "loss": 1.0083,
+      "step": 1263
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.575838926174497e-05,
+      "loss": 0.9463,
+      "step": 1264
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.575503355704698e-05,
+      "loss": 0.9375,
+      "step": 1265
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5751677852348995e-05,
+      "loss": 0.9688,
+      "step": 1266
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.574832214765101e-05,
+      "loss": 0.9614,
+      "step": 1267
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5744966442953023e-05,
+      "loss": 0.9634,
+      "step": 1268
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5741610738255034e-05,
+      "loss": 0.9883,
+      "step": 1269
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.5738255033557048e-05,
+      "loss": 0.9312,
+      "step": 1270
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.5734899328859063e-05,
+      "loss": 0.9722,
+      "step": 1271
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.5731543624161073e-05,
+      "loss": 0.9551,
+      "step": 1272
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.5728187919463088e-05,
+      "loss": 0.9702,
+      "step": 1273
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.5724832214765102e-05,
+      "loss": 0.9473,
+      "step": 1274
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.5721476510067116e-05,
+      "loss": 0.9678,
+      "step": 1275
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.5718120805369127e-05,
+      "loss": 0.9565,
+      "step": 1276
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.5714765100671145e-05,
+      "loss": 0.9565,
+      "step": 1277
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.5711409395973156e-05,
+      "loss": 0.9692,
+      "step": 1278
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.570805369127517e-05,
+      "loss": 1.0063,
+      "step": 1279
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.570469798657718e-05,
+      "loss": 0.9932,
+      "step": 1280
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.5701342281879195e-05,
+      "loss": 0.9707,
+      "step": 1281
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.569798657718121e-05,
+      "loss": 0.9868,
+      "step": 1282
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.5694630872483224e-05,
+      "loss": 1.021,
+      "step": 1283
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.5691275167785235e-05,
+      "loss": 0.9863,
+      "step": 1284
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.568791946308725e-05,
+      "loss": 0.9849,
+      "step": 1285
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.5684563758389263e-05,
+      "loss": 0.9722,
+      "step": 1286
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.5681208053691277e-05,
+      "loss": 0.9746,
+      "step": 1287
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.5677852348993288e-05,
+      "loss": 1.0005,
+      "step": 1288
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.5674496644295302e-05,
+      "loss": 0.9814,
+      "step": 1289
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.5671140939597317e-05,
+      "loss": 0.9619,
+      "step": 1290
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.566778523489933e-05,
+      "loss": 0.981,
+      "step": 1291
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.5664429530201345e-05,
+      "loss": 0.9844,
+      "step": 1292
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.5661073825503356e-05,
+      "loss": 0.9819,
+      "step": 1293
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.565771812080537e-05,
+      "loss": 0.9678,
+      "step": 1294
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.5654362416107385e-05,
+      "loss": 0.9795,
+      "step": 1295
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.56510067114094e-05,
+      "loss": 0.9468,
+      "step": 1296
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.564765100671141e-05,
+      "loss": 1.0,
+      "step": 1297
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.5644295302013424e-05,
+      "loss": 0.9453,
+      "step": 1298
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.5640939597315435e-05,
+      "loss": 0.9785,
+      "step": 1299
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.5637583892617453e-05,
+      "loss": 0.9497,
+      "step": 1300
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.5634228187919464e-05,
+      "loss": 0.9917,
+      "step": 1301
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.5630872483221478e-05,
+      "loss": 0.9907,
+      "step": 1302
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.5627516778523492e-05,
+      "loss": 0.9438,
+      "step": 1303
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.5624161073825506e-05,
+      "loss": 0.9619,
+      "step": 1304
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.5620805369127517e-05,
+      "loss": 0.9604,
+      "step": 1305
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.561744966442953e-05,
+      "loss": 1.0029,
+      "step": 1306
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.5614093959731546e-05,
+      "loss": 0.9712,
+      "step": 1307
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.5610738255033557e-05,
+      "loss": 1.002,
+      "step": 1308
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.560738255033557e-05,
+      "loss": 0.9663,
+      "step": 1309
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.5604026845637585e-05,
+      "loss": 0.9644,
+      "step": 1310
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.56006711409396e-05,
+      "loss": 0.9619,
+      "step": 1311
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.559731543624161e-05,
+      "loss": 0.9282,
+      "step": 1312
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.5593959731543625e-05,
+      "loss": 0.9966,
+      "step": 1313
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.559060402684564e-05,
+      "loss": 1.022,
+      "step": 1314
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.5587248322147653e-05,
+      "loss": 0.9502,
+      "step": 1315
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.5583892617449664e-05,
+      "loss": 0.9946,
+      "step": 1316
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.558053691275168e-05,
+      "loss": 1.0303,
+      "step": 1317
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.5577181208053693e-05,
+      "loss": 0.9883,
+      "step": 1318
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.5573825503355707e-05,
+      "loss": 0.9482,
+      "step": 1319
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.5570469798657718e-05,
+      "loss": 0.9814,
+      "step": 1320
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.5567114093959732e-05,
+      "loss": 0.9907,
+      "step": 1321
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.5563758389261746e-05,
+      "loss": 0.9634,
+      "step": 1322
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.556040268456376e-05,
+      "loss": 0.9409,
+      "step": 1323
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.555704697986577e-05,
+      "loss": 0.9531,
+      "step": 1324
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.5553691275167786e-05,
+      "loss": 1.0225,
+      "step": 1325
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.55503355704698e-05,
+      "loss": 0.9927,
+      "step": 1326
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.5546979865771814e-05,
+      "loss": 0.98,
+      "step": 1327
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.554362416107383e-05,
+      "loss": 0.9868,
+      "step": 1328
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.554026845637584e-05,
+      "loss": 0.9814,
+      "step": 1329
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.5536912751677854e-05,
+      "loss": 0.9653,
+      "step": 1330
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.5533557046979868e-05,
+      "loss": 0.9634,
+      "step": 1331
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.5530201342281882e-05,
+      "loss": 0.939,
+      "step": 1332
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.5526845637583893e-05,
+      "loss": 0.9712,
+      "step": 1333
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.5523489932885908e-05,
+      "loss": 0.9746,
+      "step": 1334
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.552013422818792e-05,
+      "loss": 0.978,
+      "step": 1335
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.5516778523489936e-05,
+      "loss": 0.9248,
+      "step": 1336
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.5513422818791947e-05,
+      "loss": 0.9775,
+      "step": 1337
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.551006711409396e-05,
+      "loss": 0.9902,
+      "step": 1338
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.5506711409395972e-05,
+      "loss": 0.9546,
+      "step": 1339
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.550335570469799e-05,
+      "loss": 0.9888,
+      "step": 1340
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.55e-05,
+      "loss": 0.9849,
+      "step": 1341
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.5496644295302015e-05,
+      "loss": 0.9795,
+      "step": 1342
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.549328859060403e-05,
+      "loss": 0.9575,
+      "step": 1343
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.548993288590604e-05,
+      "loss": 0.9473,
+      "step": 1344
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.5486577181208054e-05,
+      "loss": 0.9688,
+      "step": 1345
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.548322147651007e-05,
+      "loss": 0.9736,
+      "step": 1346
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.5479865771812083e-05,
+      "loss": 0.9673,
+      "step": 1347
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.5476510067114094e-05,
+      "loss": 0.9648,
+      "step": 1348
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.5473154362416108e-05,
+      "loss": 0.937,
+      "step": 1349
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.5469798657718122e-05,
+      "loss": 0.9248,
+      "step": 1350
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.5466442953020137e-05,
+      "loss": 0.9604,
+      "step": 1351
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.5463087248322148e-05,
+      "loss": 0.9785,
+      "step": 1352
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.5459731543624162e-05,
+      "loss": 0.9922,
+      "step": 1353
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.5456375838926176e-05,
+      "loss": 0.979,
+      "step": 1354
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.545302013422819e-05,
+      "loss": 0.9941,
+      "step": 1355
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.54496644295302e-05,
+      "loss": 0.9756,
+      "step": 1356
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.5446308724832215e-05,
+      "loss": 0.9722,
+      "step": 1357
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.544295302013423e-05,
+      "loss": 0.9966,
+      "step": 1358
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.5439597315436244e-05,
+      "loss": 0.9385,
+      "step": 1359
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.5436241610738255e-05,
+      "loss": 0.9551,
+      "step": 1360
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.543288590604027e-05,
+      "loss": 0.9507,
+      "step": 1361
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.5429530201342283e-05,
+      "loss": 0.9453,
+      "step": 1362
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.5426174496644298e-05,
+      "loss": 0.9341,
+      "step": 1363
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.542281879194631e-05,
+      "loss": 0.9761,
+      "step": 1364
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.5419463087248323e-05,
+      "loss": 0.9863,
+      "step": 1365
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.5416107382550337e-05,
+      "loss": 0.9077,
+      "step": 1366
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.541275167785235e-05,
+      "loss": 0.9521,
+      "step": 1367
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.5409395973154366e-05,
+      "loss": 0.9956,
+      "step": 1368
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.5406040268456377e-05,
+      "loss": 0.9639,
+      "step": 1369
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.540268456375839e-05,
+      "loss": 0.979,
+      "step": 1370
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.5399328859060402e-05,
+      "loss": 0.9053,
+      "step": 1371
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.539597315436242e-05,
+      "loss": 0.9214,
+      "step": 1372
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.539261744966443e-05,
+      "loss": 0.9282,
+      "step": 1373
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.5389261744966445e-05,
+      "loss": 0.9297,
+      "step": 1374
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.5385906040268455e-05,
+      "loss": 0.98,
+      "step": 1375
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.5382550335570473e-05,
+      "loss": 0.9561,
+      "step": 1376
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.5379194630872484e-05,
+      "loss": 0.9795,
+      "step": 1377
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.5375838926174498e-05,
+      "loss": 1.0464,
+      "step": 1378
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.537248322147651e-05,
+      "loss": 0.9521,
+      "step": 1379
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.5369127516778523e-05,
+      "loss": 0.9678,
+      "step": 1380
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.5365771812080538e-05,
+      "loss": 0.9639,
+      "step": 1381
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.5362416107382552e-05,
+      "loss": 0.958,
+      "step": 1382
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.5359060402684566e-05,
+      "loss": 0.939,
+      "step": 1383
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.5355704697986577e-05,
+      "loss": 0.9346,
+      "step": 1384
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.535234899328859e-05,
+      "loss": 0.9868,
+      "step": 1385
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.5348993288590606e-05,
+      "loss": 0.918,
+      "step": 1386
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.534563758389262e-05,
+      "loss": 0.9829,
+      "step": 1387
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.534228187919463e-05,
+      "loss": 0.9819,
+      "step": 1388
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.5338926174496645e-05,
+      "loss": 1.0142,
+      "step": 1389
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.533557046979866e-05,
+      "loss": 0.9849,
+      "step": 1390
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.5332214765100674e-05,
+      "loss": 0.9512,
+      "step": 1391
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.5328859060402685e-05,
+      "loss": 0.9404,
+      "step": 1392
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.53255033557047e-05,
+      "loss": 0.9614,
+      "step": 1393
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.5322147651006713e-05,
+      "loss": 1.0151,
+      "step": 1394
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.5318791946308727e-05,
+      "loss": 0.9858,
+      "step": 1395
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.5315436241610738e-05,
+      "loss": 0.9854,
+      "step": 1396
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.5312080536912753e-05,
+      "loss": 0.9795,
+      "step": 1397
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.5308724832214767e-05,
+      "loss": 1.0015,
+      "step": 1398
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.530536912751678e-05,
+      "loss": 0.9336,
+      "step": 1399
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.5302013422818792e-05,
+      "loss": 0.9639,
+      "step": 1400
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.5298657718120806e-05,
+      "loss": 0.9478,
+      "step": 1401
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.529530201342282e-05,
+      "loss": 0.978,
+      "step": 1402
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.5291946308724835e-05,
+      "loss": 0.9644,
+      "step": 1403
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.5288590604026846e-05,
+      "loss": 1.0195,
+      "step": 1404
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.528523489932886e-05,
+      "loss": 0.9468,
+      "step": 1405
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.5281879194630874e-05,
+      "loss": 0.9248,
+      "step": 1406
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.527852348993289e-05,
+      "loss": 0.9775,
+      "step": 1407
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.5275167785234903e-05,
+      "loss": 0.9678,
+      "step": 1408
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.5271812080536914e-05,
+      "loss": 0.9785,
+      "step": 1409
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.5268456375838928e-05,
+      "loss": 0.9351,
+      "step": 1410
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.526510067114094e-05,
+      "loss": 0.9839,
+      "step": 1411
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.5261744966442956e-05,
+      "loss": 0.9648,
+      "step": 1412
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.5258389261744967e-05,
+      "loss": 0.9897,
+      "step": 1413
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.5255033557046982e-05,
+      "loss": 0.9551,
+      "step": 1414
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.5251677852348994e-05,
+      "loss": 0.9858,
+      "step": 1415
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.5248322147651009e-05,
+      "loss": 0.9751,
+      "step": 1416
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.5244966442953023e-05,
+      "loss": 1.0005,
+      "step": 1417
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.5241610738255035e-05,
+      "loss": 0.9463,
+      "step": 1418
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.5238255033557048e-05,
+      "loss": 0.98,
+      "step": 1419
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.523489932885906e-05,
+      "loss": 0.9849,
+      "step": 1420
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.5231543624161076e-05,
+      "loss": 0.9541,
+      "step": 1421
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.5228187919463089e-05,
+      "loss": 0.9731,
+      "step": 1422
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.5224832214765102e-05,
+      "loss": 0.9824,
+      "step": 1423
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.5221476510067114e-05,
+      "loss": 0.959,
+      "step": 1424
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.521812080536913e-05,
+      "loss": 0.9561,
+      "step": 1425
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.5214765100671143e-05,
+      "loss": 0.9585,
+      "step": 1426
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.5211409395973155e-05,
+      "loss": 0.9941,
+      "step": 1427
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.5208053691275168e-05,
+      "loss": 1.0088,
+      "step": 1428
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.5204697986577182e-05,
+      "loss": 0.9785,
+      "step": 1429
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.5201342281879196e-05,
+      "loss": 0.9683,
+      "step": 1430
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.5197986577181209e-05,
+      "loss": 0.9668,
+      "step": 1431
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.5194630872483223e-05,
+      "loss": 0.9585,
+      "step": 1432
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.5191275167785236e-05,
+      "loss": 0.9731,
+      "step": 1433
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.518791946308725e-05,
+      "loss": 1.0024,
+      "step": 1434
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.5184563758389263e-05,
+      "loss": 0.9868,
+      "step": 1435
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.5181208053691277e-05,
+      "loss": 0.9956,
+      "step": 1436
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.517785234899329e-05,
+      "loss": 0.9546,
+      "step": 1437
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.5174496644295302e-05,
+      "loss": 0.9312,
+      "step": 1438
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.5171140939597318e-05,
+      "loss": 0.9595,
+      "step": 1439
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.516778523489933e-05,
+      "loss": 0.9546,
+      "step": 1440
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.5164429530201343e-05,
+      "loss": 0.9805,
+      "step": 1441
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.5161073825503356e-05,
+      "loss": 1.0381,
+      "step": 1442
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.5157718120805372e-05,
+      "loss": 0.9873,
+      "step": 1443
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.5154362416107384e-05,
+      "loss": 0.9541,
+      "step": 1444
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.5151006711409397e-05,
+      "loss": 0.9629,
+      "step": 1445
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.514765100671141e-05,
+      "loss": 0.9351,
+      "step": 1446
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.5144295302013424e-05,
+      "loss": 0.9668,
+      "step": 1447
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.5140939597315438e-05,
+      "loss": 1.0083,
+      "step": 1448
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.513758389261745e-05,
+      "loss": 0.9492,
+      "step": 1449
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.5134228187919463e-05,
+      "loss": 0.9775,
+      "step": 1450
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.5130872483221478e-05,
+      "loss": 0.9521,
+      "step": 1451
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.5127516778523492e-05,
+      "loss": 0.9683,
+      "step": 1452
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.5124161073825504e-05,
+      "loss": 0.9575,
+      "step": 1453
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.5120805369127519e-05,
+      "loss": 0.9854,
+      "step": 1454
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.5117449664429531e-05,
+      "loss": 0.9741,
+      "step": 1455
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.5114093959731544e-05,
+      "loss": 0.9707,
+      "step": 1456
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.511073825503356e-05,
+      "loss": 0.9873,
+      "step": 1457
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.5107382550335572e-05,
+      "loss": 0.9814,
+      "step": 1458
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.5104026845637585e-05,
+      "loss": 1.0034,
+      "step": 1459
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.5100671140939598e-05,
+      "loss": 0.9282,
+      "step": 1460
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.5097315436241614e-05,
+      "loss": 0.9634,
+      "step": 1461
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.5093959731543626e-05,
+      "loss": 0.979,
+      "step": 1462
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.5090604026845639e-05,
+      "loss": 0.9985,
+      "step": 1463
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.5087248322147651e-05,
+      "loss": 0.9717,
+      "step": 1464
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.5083892617449664e-05,
+      "loss": 0.9946,
+      "step": 1465
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.508053691275168e-05,
+      "loss": 0.9653,
+      "step": 1466
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.5077181208053692e-05,
+      "loss": 0.9604,
+      "step": 1467
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.5073825503355705e-05,
+      "loss": 0.9795,
+      "step": 1468
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.507046979865772e-05,
+      "loss": 0.981,
+      "step": 1469
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.5067114093959734e-05,
+      "loss": 0.9858,
+      "step": 1470
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.5063758389261746e-05,
+      "loss": 0.9551,
+      "step": 1471
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.506040268456376e-05,
+      "loss": 0.9795,
+      "step": 1472
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.5057046979865773e-05,
+      "loss": 0.9736,
+      "step": 1473
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.5053691275167786e-05,
+      "loss": 0.96,
+      "step": 1474
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.50503355704698e-05,
+      "loss": 0.9541,
+      "step": 1475
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.5046979865771814e-05,
+      "loss": 1.001,
+      "step": 1476
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.5043624161073827e-05,
+      "loss": 0.9468,
+      "step": 1477
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.504026845637584e-05,
+      "loss": 0.9839,
+      "step": 1478
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.5036912751677855e-05,
+      "loss": 0.9702,
+      "step": 1479
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.5033557046979868e-05,
+      "loss": 0.98,
+      "step": 1480
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.503020134228188e-05,
+      "loss": 0.9888,
+      "step": 1481
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.5026845637583893e-05,
+      "loss": 1.0234,
+      "step": 1482
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.5023489932885906e-05,
+      "loss": 0.9717,
+      "step": 1483
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.5020134228187922e-05,
+      "loss": 0.9448,
+      "step": 1484
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.5016778523489934e-05,
+      "loss": 0.9727,
+      "step": 1485
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.5013422818791947e-05,
+      "loss": 0.9585,
+      "step": 1486
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.5010067114093961e-05,
+      "loss": 0.9492,
+      "step": 1487
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.5006711409395975e-05,
+      "loss": 0.9355,
+      "step": 1488
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.5003355704697988e-05,
+      "loss": 0.9604,
+      "step": 1489
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 0.959,
+      "step": 1490
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.4996644295302015e-05,
+      "loss": 0.9946,
+      "step": 1491
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.4993288590604027e-05,
+      "loss": 0.9692,
+      "step": 1492
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.4989932885906042e-05,
+      "loss": 0.9385,
+      "step": 1493
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.4986577181208056e-05,
+      "loss": 1.0029,
+      "step": 1494
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.4983221476510068e-05,
+      "loss": 0.9761,
+      "step": 1495
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.4979865771812081e-05,
+      "loss": 1.0083,
+      "step": 1496
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.4976510067114097e-05,
+      "loss": 0.9683,
+      "step": 1497
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.497315436241611e-05,
+      "loss": 0.9741,
+      "step": 1498
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.4969798657718122e-05,
+      "loss": 1.0073,
+      "step": 1499
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.4966442953020135e-05,
+      "loss": 1.0312,
+      "step": 1500
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.4963087248322147e-05,
+      "loss": 0.9561,
+      "step": 1501
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.4959731543624163e-05,
+      "loss": 0.9834,
+      "step": 1502
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.4956375838926176e-05,
+      "loss": 1.0039,
+      "step": 1503
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.4953020134228188e-05,
+      "loss": 0.9941,
+      "step": 1504
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.4949664429530203e-05,
+      "loss": 0.9761,
+      "step": 1505
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.4946308724832217e-05,
+      "loss": 0.9741,
+      "step": 1506
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.494295302013423e-05,
+      "loss": 1.0029,
+      "step": 1507
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.4939597315436242e-05,
+      "loss": 0.9561,
+      "step": 1508
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.4936241610738256e-05,
+      "loss": 0.9458,
+      "step": 1509
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.4932885906040269e-05,
+      "loss": 0.9526,
+      "step": 1510
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.4929530201342283e-05,
+      "loss": 0.9414,
+      "step": 1511
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.4926174496644297e-05,
+      "loss": 0.9712,
+      "step": 1512
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.492281879194631e-05,
+      "loss": 0.9961,
+      "step": 1513
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.4919463087248323e-05,
+      "loss": 0.9541,
+      "step": 1514
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.4916107382550339e-05,
+      "loss": 0.9673,
+      "step": 1515
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.4912751677852351e-05,
+      "loss": 0.9404,
+      "step": 1516
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.4909395973154364e-05,
+      "loss": 0.9526,
+      "step": 1517
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.4906040268456376e-05,
+      "loss": 0.9561,
+      "step": 1518
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.4902684563758389e-05,
+      "loss": 0.9155,
+      "step": 1519
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.4899328859060405e-05,
+      "loss": 0.9551,
+      "step": 1520
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.4895973154362417e-05,
+      "loss": 0.9727,
+      "step": 1521
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.489261744966443e-05,
+      "loss": 0.936,
+      "step": 1522
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.4889261744966443e-05,
+      "loss": 0.9722,
+      "step": 1523
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.4885906040268459e-05,
+      "loss": 0.9956,
+      "step": 1524
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.4882550335570471e-05,
+      "loss": 0.9619,
+      "step": 1525
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.4879194630872484e-05,
+      "loss": 0.9971,
+      "step": 1526
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.4875838926174498e-05,
+      "loss": 0.9399,
+      "step": 1527
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.487248322147651e-05,
+      "loss": 0.9844,
+      "step": 1528
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.4869127516778525e-05,
+      "loss": 1.0322,
+      "step": 1529
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.4865771812080539e-05,
+      "loss": 0.9502,
+      "step": 1530
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.4862416107382552e-05,
+      "loss": 0.9712,
+      "step": 1531
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.4859060402684564e-05,
+      "loss": 0.9771,
+      "step": 1532
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.4855704697986579e-05,
+      "loss": 0.981,
+      "step": 1533
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.4852348993288593e-05,
+      "loss": 0.9814,
+      "step": 1534
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.4848993288590605e-05,
+      "loss": 0.9609,
+      "step": 1535
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.4845637583892618e-05,
+      "loss": 0.981,
+      "step": 1536
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.484228187919463e-05,
+      "loss": 0.9673,
+      "step": 1537
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.4838926174496647e-05,
+      "loss": 0.9487,
+      "step": 1538
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.4835570469798659e-05,
+      "loss": 0.9551,
+      "step": 1539
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.4832214765100672e-05,
+      "loss": 0.981,
+      "step": 1540
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.4828859060402684e-05,
+      "loss": 0.9575,
+      "step": 1541
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.48255033557047e-05,
+      "loss": 0.9434,
+      "step": 1542
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.4822147651006713e-05,
+      "loss": 0.9668,
+      "step": 1543
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.4818791946308725e-05,
+      "loss": 0.9648,
+      "step": 1544
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.481543624161074e-05,
+      "loss": 0.9512,
+      "step": 1545
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.4812080536912752e-05,
+      "loss": 0.9639,
+      "step": 1546
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.4808724832214767e-05,
+      "loss": 0.959,
+      "step": 1547
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.4805369127516779e-05,
+      "loss": 0.9575,
+      "step": 1548
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.4802013422818793e-05,
+      "loss": 1.0181,
+      "step": 1549
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.4798657718120806e-05,
+      "loss": 0.959,
+      "step": 1550
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.479530201342282e-05,
+      "loss": 0.9839,
+      "step": 1551
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.4791946308724835e-05,
+      "loss": 1.0195,
+      "step": 1552
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.4788590604026847e-05,
+      "loss": 0.9688,
+      "step": 1553
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.478523489932886e-05,
+      "loss": 0.9624,
+      "step": 1554
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.4781879194630872e-05,
+      "loss": 0.9697,
+      "step": 1555
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.4778523489932888e-05,
+      "loss": 0.9761,
+      "step": 1556
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.47751677852349e-05,
+      "loss": 0.9648,
+      "step": 1557
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.4771812080536913e-05,
+      "loss": 0.9331,
+      "step": 1558
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.4768456375838926e-05,
+      "loss": 0.9648,
+      "step": 1559
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.4765100671140942e-05,
+      "loss": 0.9624,
+      "step": 1560
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.4761744966442955e-05,
+      "loss": 0.9692,
+      "step": 1561
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.4758389261744967e-05,
+      "loss": 0.9585,
+      "step": 1562
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.475503355704698e-05,
+      "loss": 0.9609,
+      "step": 1563
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.4751677852348994e-05,
+      "loss": 0.9536,
+      "step": 1564
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.4748322147651008e-05,
+      "loss": 0.9883,
+      "step": 1565
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.474496644295302e-05,
+      "loss": 0.9927,
+      "step": 1566
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.4741610738255035e-05,
+      "loss": 0.9814,
+      "step": 1567
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.4738255033557048e-05,
+      "loss": 0.9971,
+      "step": 1568
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.4734899328859062e-05,
+      "loss": 0.9707,
+      "step": 1569
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.4731543624161076e-05,
+      "loss": 0.9678,
+      "step": 1570
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.4728187919463089e-05,
+      "loss": 0.9692,
+      "step": 1571
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.4724832214765101e-05,
+      "loss": 1.0093,
+      "step": 1572
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.4721476510067114e-05,
+      "loss": 0.9399,
+      "step": 1573
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.471812080536913e-05,
+      "loss": 0.9639,
+      "step": 1574
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.4714765100671142e-05,
+      "loss": 0.9624,
+      "step": 1575
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.4711409395973155e-05,
+      "loss": 0.9697,
+      "step": 1576
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.4708053691275168e-05,
+      "loss": 0.9663,
+      "step": 1577
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.4704697986577184e-05,
+      "loss": 0.9653,
+      "step": 1578
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.4701342281879196e-05,
+      "loss": 0.9438,
+      "step": 1579
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.4697986577181209e-05,
+      "loss": 0.9517,
+      "step": 1580
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.4694630872483221e-05,
+      "loss": 0.9824,
+      "step": 1581
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.4691275167785236e-05,
+      "loss": 0.9648,
+      "step": 1582
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.468791946308725e-05,
+      "loss": 1.0005,
+      "step": 1583
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.4684563758389262e-05,
+      "loss": 1.0293,
+      "step": 1584
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.4681208053691277e-05,
+      "loss": 0.9502,
+      "step": 1585
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.467785234899329e-05,
+      "loss": 0.9839,
+      "step": 1586
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.4674496644295304e-05,
+      "loss": 0.9595,
+      "step": 1587
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.4671140939597316e-05,
+      "loss": 0.9346,
+      "step": 1588
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.466778523489933e-05,
+      "loss": 0.998,
+      "step": 1589
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.4664429530201343e-05,
+      "loss": 0.9556,
+      "step": 1590
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.4661073825503357e-05,
+      "loss": 0.9268,
+      "step": 1591
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.4657718120805372e-05,
+      "loss": 0.9766,
+      "step": 1592
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.4654362416107384e-05,
+      "loss": 0.9634,
+      "step": 1593
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.4651006711409397e-05,
+      "loss": 0.9893,
+      "step": 1594
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.464765100671141e-05,
+      "loss": 1.0039,
+      "step": 1595
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.4644295302013425e-05,
+      "loss": 0.9697,
+      "step": 1596
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.4640939597315438e-05,
+      "loss": 0.9614,
+      "step": 1597
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.463758389261745e-05,
+      "loss": 0.9502,
+      "step": 1598
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.4634228187919463e-05,
+      "loss": 0.9458,
+      "step": 1599
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.4630872483221479e-05,
+      "loss": 0.9673,
+      "step": 1600
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.4627516778523492e-05,
+      "loss": 0.9883,
+      "step": 1601
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.4624161073825504e-05,
+      "loss": 0.9614,
+      "step": 1602
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.4620805369127518e-05,
+      "loss": 0.9854,
+      "step": 1603
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.4617449664429531e-05,
+      "loss": 0.9888,
+      "step": 1604
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.4614093959731545e-05,
+      "loss": 0.9639,
+      "step": 1605
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.4610738255033558e-05,
+      "loss": 0.9653,
+      "step": 1606
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.4607382550335572e-05,
+      "loss": 0.9653,
+      "step": 1607
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.4604026845637585e-05,
+      "loss": 0.9893,
+      "step": 1608
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.4600671140939599e-05,
+      "loss": 0.9268,
+      "step": 1609
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.4597315436241613e-05,
+      "loss": 0.9663,
+      "step": 1610
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.4593959731543626e-05,
+      "loss": 0.9595,
+      "step": 1611
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.4590604026845638e-05,
+      "loss": 0.9775,
+      "step": 1612
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.4587248322147651e-05,
+      "loss": 0.938,
+      "step": 1613
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.4583892617449667e-05,
+      "loss": 0.9443,
+      "step": 1614
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.458053691275168e-05,
+      "loss": 1.0083,
+      "step": 1615
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.4577181208053692e-05,
+      "loss": 0.9526,
+      "step": 1616
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.4573825503355705e-05,
+      "loss": 0.937,
+      "step": 1617
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.457046979865772e-05,
+      "loss": 0.9404,
+      "step": 1618
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.4567114093959733e-05,
+      "loss": 0.9629,
+      "step": 1619
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.4563758389261746e-05,
+      "loss": 0.918,
+      "step": 1620
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.4560402684563758e-05,
+      "loss": 0.9707,
+      "step": 1621
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.4557046979865773e-05,
+      "loss": 0.978,
+      "step": 1622
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.4553691275167787e-05,
+      "loss": 0.9727,
+      "step": 1623
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.45503355704698e-05,
+      "loss": 0.9717,
+      "step": 1624
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.4546979865771814e-05,
+      "loss": 0.9541,
+      "step": 1625
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.4543624161073826e-05,
+      "loss": 0.9683,
+      "step": 1626
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.454026845637584e-05,
+      "loss": 0.959,
+      "step": 1627
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.4536912751677855e-05,
+      "loss": 0.9443,
+      "step": 1628
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.4533557046979868e-05,
+      "loss": 0.9697,
+      "step": 1629
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.453020134228188e-05,
+      "loss": 0.9448,
+      "step": 1630
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.4526845637583893e-05,
+      "loss": 0.9688,
+      "step": 1631
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.4523489932885909e-05,
+      "loss": 0.9868,
+      "step": 1632
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.4520134228187921e-05,
+      "loss": 0.9829,
+      "step": 1633
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.4516778523489934e-05,
+      "loss": 1.0088,
+      "step": 1634
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.4513422818791946e-05,
+      "loss": 0.9409,
+      "step": 1635
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.4510067114093962e-05,
+      "loss": 0.9834,
+      "step": 1636
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.4506711409395975e-05,
+      "loss": 0.957,
+      "step": 1637
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.4503355704697988e-05,
+      "loss": 0.9546,
+      "step": 1638
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.45e-05,
+      "loss": 0.9312,
+      "step": 1639
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.4496644295302014e-05,
+      "loss": 0.9219,
+      "step": 1640
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.4493288590604029e-05,
+      "loss": 0.9614,
+      "step": 1641
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.4489932885906041e-05,
+      "loss": 0.9575,
+      "step": 1642
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.4486577181208055e-05,
+      "loss": 0.9727,
+      "step": 1643
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.4483221476510068e-05,
+      "loss": 0.9531,
+      "step": 1644
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.4479865771812082e-05,
+      "loss": 0.9814,
+      "step": 1645
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.4476510067114095e-05,
+      "loss": 0.9741,
+      "step": 1646
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.447315436241611e-05,
+      "loss": 0.9917,
+      "step": 1647
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.4469798657718122e-05,
+      "loss": 0.9663,
+      "step": 1648
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.4466442953020134e-05,
+      "loss": 0.9346,
+      "step": 1649
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.446308724832215e-05,
+      "loss": 0.978,
+      "step": 1650
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.4459731543624163e-05,
+      "loss": 0.96,
+      "step": 1651
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.4456375838926175e-05,
+      "loss": 0.9771,
+      "step": 1652
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.4453020134228188e-05,
+      "loss": 0.9536,
+      "step": 1653
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.4449664429530204e-05,
+      "loss": 0.959,
+      "step": 1654
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.4446308724832217e-05,
+      "loss": 0.9619,
+      "step": 1655
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.444295302013423e-05,
+      "loss": 0.9717,
+      "step": 1656
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.4439597315436242e-05,
+      "loss": 0.9355,
+      "step": 1657
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.4436241610738256e-05,
+      "loss": 0.9473,
+      "step": 1658
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.443288590604027e-05,
+      "loss": 0.9365,
+      "step": 1659
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.4429530201342283e-05,
+      "loss": 1.0161,
+      "step": 1660
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.4426174496644295e-05,
+      "loss": 0.9531,
+      "step": 1661
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.442281879194631e-05,
+      "loss": 0.9932,
+      "step": 1662
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.4419463087248324e-05,
+      "loss": 0.9736,
+      "step": 1663
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.4416107382550337e-05,
+      "loss": 1.0122,
+      "step": 1664
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.4412751677852351e-05,
+      "loss": 1.0063,
+      "step": 1665
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.4409395973154363e-05,
+      "loss": 0.9722,
+      "step": 1666
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.4406040268456376e-05,
+      "loss": 0.9604,
+      "step": 1667
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.4402684563758392e-05,
+      "loss": 0.957,
+      "step": 1668
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.4399328859060405e-05,
+      "loss": 0.978,
+      "step": 1669
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.4395973154362417e-05,
+      "loss": 0.9473,
+      "step": 1670
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.439261744966443e-05,
+      "loss": 0.9668,
+      "step": 1671
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.4389261744966446e-05,
+      "loss": 0.9561,
+      "step": 1672
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.4385906040268458e-05,
+      "loss": 0.9795,
+      "step": 1673
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.4382550335570471e-05,
+      "loss": 0.9263,
+      "step": 1674
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.4379194630872483e-05,
+      "loss": 0.9502,
+      "step": 1675
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.4375838926174496e-05,
+      "loss": 0.9634,
+      "step": 1676
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.4372483221476512e-05,
+      "loss": 0.96,
+      "step": 1677
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.4369127516778525e-05,
+      "loss": 0.9658,
+      "step": 1678
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.4365771812080537e-05,
+      "loss": 0.9541,
+      "step": 1679
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.4362416107382551e-05,
+      "loss": 0.9604,
+      "step": 1680
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.4359060402684566e-05,
+      "loss": 1.019,
+      "step": 1681
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.4355704697986578e-05,
+      "loss": 0.9463,
+      "step": 1682
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.4352348993288593e-05,
+      "loss": 0.9526,
+      "step": 1683
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.4348993288590605e-05,
+      "loss": 0.9565,
+      "step": 1684
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.4345637583892618e-05,
+      "loss": 1.0112,
+      "step": 1685
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.4342281879194632e-05,
+      "loss": 0.9707,
+      "step": 1686
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.4338926174496646e-05,
+      "loss": 0.9629,
+      "step": 1687
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.4335570469798659e-05,
+      "loss": 0.9658,
+      "step": 1688
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.4332214765100671e-05,
+      "loss": 0.9453,
+      "step": 1689
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.4328859060402687e-05,
+      "loss": 0.9458,
+      "step": 1690
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.43255033557047e-05,
+      "loss": 0.9531,
+      "step": 1691
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.4322147651006713e-05,
+      "loss": 0.9629,
+      "step": 1692
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.4318791946308725e-05,
+      "loss": 0.9644,
+      "step": 1693
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.4315436241610738e-05,
+      "loss": 0.9839,
+      "step": 1694
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.4312080536912754e-05,
+      "loss": 0.9316,
+      "step": 1695
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.4308724832214766e-05,
+      "loss": 0.9624,
+      "step": 1696
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.4305369127516779e-05,
+      "loss": 0.958,
+      "step": 1697
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.4302013422818793e-05,
+      "loss": 0.9839,
+      "step": 1698
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.4298657718120807e-05,
+      "loss": 0.9712,
+      "step": 1699
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.429530201342282e-05,
+      "loss": 0.9502,
+      "step": 1700
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.4291946308724833e-05,
+      "loss": 0.9619,
+      "step": 1701
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.4288590604026847e-05,
+      "loss": 0.959,
+      "step": 1702
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.428523489932886e-05,
+      "loss": 0.9624,
+      "step": 1703
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.4281879194630874e-05,
+      "loss": 0.9849,
+      "step": 1704
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.4278523489932888e-05,
+      "loss": 0.9937,
+      "step": 1705
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.42751677852349e-05,
+      "loss": 0.9707,
+      "step": 1706
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.4271812080536913e-05,
+      "loss": 0.9292,
+      "step": 1707
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.4268456375838929e-05,
+      "loss": 0.9771,
+      "step": 1708
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.4265100671140942e-05,
+      "loss": 0.9575,
+      "step": 1709
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.4261744966442954e-05,
+      "loss": 0.9639,
+      "step": 1710
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.4258389261744967e-05,
+      "loss": 0.9473,
+      "step": 1711
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.425503355704698e-05,
+      "loss": 0.9873,
+      "step": 1712
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.4251677852348995e-05,
+      "loss": 1.0059,
+      "step": 1713
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.4248322147651008e-05,
+      "loss": 0.9678,
+      "step": 1714
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.424496644295302e-05,
+      "loss": 0.9556,
+      "step": 1715
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.4241610738255035e-05,
+      "loss": 0.9766,
+      "step": 1716
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.4238255033557049e-05,
+      "loss": 0.981,
+      "step": 1717
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.4234899328859062e-05,
+      "loss": 0.9805,
+      "step": 1718
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.4231543624161074e-05,
+      "loss": 0.9609,
+      "step": 1719
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.4228187919463088e-05,
+      "loss": 0.9951,
+      "step": 1720
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.4224832214765101e-05,
+      "loss": 0.9614,
+      "step": 1721
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.4221476510067115e-05,
+      "loss": 0.9507,
+      "step": 1722
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.421812080536913e-05,
+      "loss": 0.9492,
+      "step": 1723
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.4214765100671142e-05,
+      "loss": 0.9575,
+      "step": 1724
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.4211409395973155e-05,
+      "loss": 0.9648,
+      "step": 1725
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.4208053691275169e-05,
+      "loss": 0.9868,
+      "step": 1726
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.4204697986577183e-05,
+      "loss": 0.9307,
+      "step": 1727
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.4201342281879196e-05,
+      "loss": 0.9785,
+      "step": 1728
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.4197986577181208e-05,
+      "loss": 0.9556,
+      "step": 1729
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.4194630872483221e-05,
+      "loss": 0.9458,
+      "step": 1730
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.4191275167785237e-05,
+      "loss": 0.9517,
+      "step": 1731
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.418791946308725e-05,
+      "loss": 0.9487,
+      "step": 1732
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.4184563758389262e-05,
+      "loss": 0.96,
+      "step": 1733
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.4181208053691275e-05,
+      "loss": 0.9604,
+      "step": 1734
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.417785234899329e-05,
+      "loss": 0.9849,
+      "step": 1735
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.4174496644295303e-05,
+      "loss": 0.9766,
+      "step": 1736
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.4171140939597316e-05,
+      "loss": 0.9443,
+      "step": 1737
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.416778523489933e-05,
+      "loss": 0.9316,
+      "step": 1738
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.4164429530201343e-05,
+      "loss": 0.9497,
+      "step": 1739
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.4161073825503357e-05,
+      "loss": 0.9341,
+      "step": 1740
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.4157718120805371e-05,
+      "loss": 0.9766,
+      "step": 1741
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.4154362416107384e-05,
+      "loss": 0.981,
+      "step": 1742
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.4151006711409396e-05,
+      "loss": 0.9624,
+      "step": 1743
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.414765100671141e-05,
+      "loss": 0.9751,
+      "step": 1744
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.4144295302013425e-05,
+      "loss": 0.9507,
+      "step": 1745
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.4140939597315438e-05,
+      "loss": 0.9697,
+      "step": 1746
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.413758389261745e-05,
+      "loss": 0.9639,
+      "step": 1747
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.4134228187919463e-05,
+      "loss": 0.959,
+      "step": 1748
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.4130872483221479e-05,
+      "loss": 0.9971,
+      "step": 1749
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.4127516778523491e-05,
+      "loss": 0.9785,
+      "step": 1750
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.4124161073825504e-05,
+      "loss": 0.9722,
+      "step": 1751
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.4120805369127516e-05,
+      "loss": 0.9883,
+      "step": 1752
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.4117449664429532e-05,
+      "loss": 0.9502,
+      "step": 1753
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.4114093959731545e-05,
+      "loss": 0.9487,
+      "step": 1754
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.4110738255033558e-05,
+      "loss": 0.9565,
+      "step": 1755
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.4107382550335572e-05,
+      "loss": 0.9688,
+      "step": 1756
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.4104026845637584e-05,
+      "loss": 0.9961,
+      "step": 1757
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.4100671140939599e-05,
+      "loss": 0.9775,
+      "step": 1758
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.4097315436241611e-05,
+      "loss": 0.9336,
+      "step": 1759
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.4093959731543626e-05,
+      "loss": 0.979,
+      "step": 1760
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.4090604026845638e-05,
+      "loss": 0.9795,
+      "step": 1761
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.4087248322147652e-05,
+      "loss": 0.9497,
+      "step": 1762
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.4083892617449667e-05,
+      "loss": 0.9536,
+      "step": 1763
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.408053691275168e-05,
+      "loss": 0.9375,
+      "step": 1764
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.4077181208053692e-05,
+      "loss": 0.9839,
+      "step": 1765
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.4073825503355704e-05,
+      "loss": 0.9717,
+      "step": 1766
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.407046979865772e-05,
+      "loss": 0.9429,
+      "step": 1767
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.4067114093959733e-05,
+      "loss": 0.9639,
+      "step": 1768
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.4063758389261746e-05,
+      "loss": 0.9487,
+      "step": 1769
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.4060402684563758e-05,
+      "loss": 0.9717,
+      "step": 1770
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.4057046979865774e-05,
+      "loss": 0.9604,
+      "step": 1771
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.4053691275167787e-05,
+      "loss": 0.9536,
+      "step": 1772
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.40503355704698e-05,
+      "loss": 0.9609,
+      "step": 1773
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.4046979865771812e-05,
+      "loss": 0.9521,
+      "step": 1774
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.4043624161073828e-05,
+      "loss": 0.9282,
+      "step": 1775
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.404026845637584e-05,
+      "loss": 0.9761,
+      "step": 1776
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.4036912751677853e-05,
+      "loss": 0.9521,
+      "step": 1777
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.4033557046979867e-05,
+      "loss": 0.9409,
+      "step": 1778
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.403020134228188e-05,
+      "loss": 0.9399,
+      "step": 1779
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.4026845637583894e-05,
+      "loss": 0.9604,
+      "step": 1780
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.4023489932885908e-05,
+      "loss": 0.98,
+      "step": 1781
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.4020134228187921e-05,
+      "loss": 0.9756,
+      "step": 1782
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.4016778523489934e-05,
+      "loss": 0.9429,
+      "step": 1783
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.4013422818791948e-05,
+      "loss": 0.9629,
+      "step": 1784
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.4010067114093962e-05,
+      "loss": 0.9683,
+      "step": 1785
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.4006711409395975e-05,
+      "loss": 0.9771,
+      "step": 1786
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.4003355704697987e-05,
+      "loss": 0.9766,
+      "step": 1787
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.4e-05,
+      "loss": 0.9521,
+      "step": 1788
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.3996644295302016e-05,
+      "loss": 0.9644,
+      "step": 1789
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.3993288590604028e-05,
+      "loss": 0.9561,
+      "step": 1790
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.3989932885906041e-05,
+      "loss": 0.9531,
+      "step": 1791
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.3986577181208053e-05,
+      "loss": 0.9956,
+      "step": 1792
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.398322147651007e-05,
+      "loss": 0.9312,
+      "step": 1793
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.3979865771812082e-05,
+      "loss": 0.9956,
+      "step": 1794
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.3976510067114095e-05,
+      "loss": 0.9736,
+      "step": 1795
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.3973154362416109e-05,
+      "loss": 0.9414,
+      "step": 1796
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.3969798657718121e-05,
+      "loss": 0.9541,
+      "step": 1797
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.3966442953020136e-05,
+      "loss": 0.9512,
+      "step": 1798
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.3963087248322148e-05,
+      "loss": 0.9551,
+      "step": 1799
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.3959731543624163e-05,
+      "loss": 0.9688,
+      "step": 1800
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.3956375838926175e-05,
+      "loss": 0.936,
+      "step": 1801
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.395302013422819e-05,
+      "loss": 0.9712,
+      "step": 1802
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.3949664429530204e-05,
+      "loss": 0.9634,
+      "step": 1803
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.3946308724832216e-05,
+      "loss": 0.9731,
+      "step": 1804
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.3942953020134229e-05,
+      "loss": 0.9443,
+      "step": 1805
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.3939597315436241e-05,
+      "loss": 0.9194,
+      "step": 1806
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.3936241610738257e-05,
+      "loss": 0.9292,
+      "step": 1807
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.393288590604027e-05,
+      "loss": 0.9214,
+      "step": 1808
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.3929530201342283e-05,
+      "loss": 0.9473,
+      "step": 1809
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.3926174496644295e-05,
+      "loss": 0.9478,
+      "step": 1810
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.3922818791946311e-05,
+      "loss": 0.9873,
+      "step": 1811
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.3919463087248324e-05,
+      "loss": 0.9717,
+      "step": 1812
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.3916107382550336e-05,
+      "loss": 0.9482,
+      "step": 1813
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.3912751677852349e-05,
+      "loss": 0.9526,
+      "step": 1814
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.3909395973154363e-05,
+      "loss": 0.9351,
+      "step": 1815
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.3906040268456377e-05,
+      "loss": 1.0078,
+      "step": 1816
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.390268456375839e-05,
+      "loss": 0.9575,
+      "step": 1817
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.3899328859060404e-05,
+      "loss": 0.9653,
+      "step": 1818
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.3895973154362417e-05,
+      "loss": 0.9629,
+      "step": 1819
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.3892617449664431e-05,
+      "loss": 0.9609,
+      "step": 1820
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.3889261744966445e-05,
+      "loss": 0.9351,
+      "step": 1821
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.3885906040268458e-05,
+      "loss": 0.9707,
+      "step": 1822
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.388255033557047e-05,
+      "loss": 0.936,
+      "step": 1823
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.3879194630872483e-05,
+      "loss": 0.9551,
+      "step": 1824
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.3875838926174499e-05,
+      "loss": 0.9404,
+      "step": 1825
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.3872483221476512e-05,
+      "loss": 0.9922,
+      "step": 1826
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.3869127516778524e-05,
+      "loss": 0.958,
+      "step": 1827
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.3865771812080537e-05,
+      "loss": 0.9414,
+      "step": 1828
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.3862416107382553e-05,
+      "loss": 0.9204,
+      "step": 1829
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.3859060402684565e-05,
+      "loss": 0.9766,
+      "step": 1830
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.3855704697986578e-05,
+      "loss": 0.9956,
+      "step": 1831
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.385234899328859e-05,
+      "loss": 0.9321,
+      "step": 1832
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.3848993288590605e-05,
+      "loss": 0.9722,
+      "step": 1833
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.3845637583892619e-05,
+      "loss": 0.9478,
+      "step": 1834
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.3842281879194632e-05,
+      "loss": 0.9619,
+      "step": 1835
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.3838926174496646e-05,
+      "loss": 0.9497,
+      "step": 1836
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.3835570469798659e-05,
+      "loss": 0.9829,
+      "step": 1837
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.3832214765100673e-05,
+      "loss": 0.9624,
+      "step": 1838
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.3828859060402685e-05,
+      "loss": 0.9878,
+      "step": 1839
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.38255033557047e-05,
+      "loss": 1.021,
+      "step": 1840
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.3822147651006712e-05,
+      "loss": 0.9722,
+      "step": 1841
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.3818791946308725e-05,
+      "loss": 0.9331,
+      "step": 1842
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.381543624161074e-05,
+      "loss": 0.9634,
+      "step": 1843
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.3812080536912753e-05,
+      "loss": 0.9385,
+      "step": 1844
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.3808724832214766e-05,
+      "loss": 0.9609,
+      "step": 1845
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.3805369127516779e-05,
+      "loss": 0.9385,
+      "step": 1846
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.3802013422818794e-05,
+      "loss": 0.9658,
+      "step": 1847
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.3798657718120807e-05,
+      "loss": 0.9673,
+      "step": 1848
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.379530201342282e-05,
+      "loss": 0.981,
+      "step": 1849
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.3791946308724832e-05,
+      "loss": 0.9507,
+      "step": 1850
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.3788590604026847e-05,
+      "loss": 0.9629,
+      "step": 1851
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.378523489932886e-05,
+      "loss": 0.9478,
+      "step": 1852
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1.3781879194630873e-05,
+      "loss": 0.9668,
+      "step": 1853
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.3778523489932888e-05,
+      "loss": 0.9785,
+      "step": 1854
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.37751677852349e-05,
+      "loss": 0.9355,
+      "step": 1855
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.3771812080536914e-05,
+      "loss": 0.9346,
+      "step": 1856
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.3768456375838927e-05,
+      "loss": 0.9805,
+      "step": 1857
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.3765100671140941e-05,
+      "loss": 0.9385,
+      "step": 1858
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.3761744966442954e-05,
+      "loss": 0.9502,
+      "step": 1859
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.3758389261744966e-05,
+      "loss": 0.9873,
+      "step": 1860
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.3755033557046982e-05,
+      "loss": 0.9688,
+      "step": 1861
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.3751677852348995e-05,
+      "loss": 0.9995,
+      "step": 1862
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.3748322147651008e-05,
+      "loss": 0.9463,
+      "step": 1863
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.374496644295302e-05,
+      "loss": 0.9243,
+      "step": 1864
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 1.3741610738255036e-05,
+      "loss": 0.9419,
+      "step": 1865
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.3738255033557049e-05,
+      "loss": 0.9443,
+      "step": 1866
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.3734899328859061e-05,
+      "loss": 0.9399,
+      "step": 1867
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.3731543624161074e-05,
+      "loss": 0.9551,
+      "step": 1868
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.3728187919463088e-05,
+      "loss": 0.9624,
+      "step": 1869
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.3724832214765102e-05,
+      "loss": 0.9536,
+      "step": 1870
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.3721476510067115e-05,
+      "loss": 0.9673,
+      "step": 1871
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.3718120805369128e-05,
+      "loss": 0.9697,
+      "step": 1872
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.3714765100671142e-05,
+      "loss": 0.96,
+      "step": 1873
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.3711409395973156e-05,
+      "loss": 1.0137,
+      "step": 1874
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.3708053691275169e-05,
+      "loss": 0.9736,
+      "step": 1875
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.3704697986577183e-05,
+      "loss": 0.9463,
+      "step": 1876
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 1.3701342281879196e-05,
+      "loss": 0.9614,
+      "step": 1877
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.3697986577181208e-05,
+      "loss": 0.9482,
+      "step": 1878
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.3694630872483224e-05,
+      "loss": 0.9595,
+      "step": 1879
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.3691275167785237e-05,
+      "loss": 0.9292,
+      "step": 1880
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.368791946308725e-05,
+      "loss": 0.9712,
+      "step": 1881
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.3684563758389262e-05,
+      "loss": 0.9556,
+      "step": 1882
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.3681208053691278e-05,
+      "loss": 0.9331,
+      "step": 1883
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.367785234899329e-05,
+      "loss": 0.98,
+      "step": 1884
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.3674496644295303e-05,
+      "loss": 0.9492,
+      "step": 1885
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.3671140939597316e-05,
+      "loss": 0.9771,
+      "step": 1886
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.3667785234899328e-05,
+      "loss": 0.9854,
+      "step": 1887
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.3664429530201344e-05,
+      "loss": 0.9697,
+      "step": 1888
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 1.3661073825503357e-05,
+      "loss": 0.9858,
+      "step": 1889
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.365771812080537e-05,
+      "loss": 0.9546,
+      "step": 1890
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.3654362416107384e-05,
+      "loss": 0.9736,
+      "step": 1891
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.3651006711409398e-05,
+      "loss": 0.9326,
+      "step": 1892
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.364765100671141e-05,
+      "loss": 0.9551,
+      "step": 1893
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.3644295302013425e-05,
+      "loss": 0.959,
+      "step": 1894
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.3640939597315437e-05,
+      "loss": 0.9795,
+      "step": 1895
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.363758389261745e-05,
+      "loss": 0.9731,
+      "step": 1896
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.3634228187919464e-05,
+      "loss": 0.9438,
+      "step": 1897
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.3630872483221478e-05,
+      "loss": 0.9287,
+      "step": 1898
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.3627516778523491e-05,
+      "loss": 0.9521,
+      "step": 1899
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.3624161073825504e-05,
+      "loss": 0.9448,
+      "step": 1900
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 1.362080536912752e-05,
+      "loss": 0.957,
+      "step": 1901
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.3617449664429532e-05,
+      "loss": 1.0005,
+      "step": 1902
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.3614093959731545e-05,
+      "loss": 0.9551,
+      "step": 1903
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.3610738255033557e-05,
+      "loss": 0.9604,
+      "step": 1904
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.360738255033557e-05,
+      "loss": 0.9771,
+      "step": 1905
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.3604026845637586e-05,
+      "loss": 0.9727,
+      "step": 1906
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.3600671140939598e-05,
+      "loss": 0.9438,
+      "step": 1907
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.3597315436241611e-05,
+      "loss": 0.9902,
+      "step": 1908
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.3593959731543625e-05,
+      "loss": 0.9434,
+      "step": 1909
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.359060402684564e-05,
+      "loss": 0.9644,
+      "step": 1910
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.3587248322147652e-05,
+      "loss": 0.9551,
+      "step": 1911
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.3583892617449665e-05,
+      "loss": 0.9775,
+      "step": 1912
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 1.3580536912751679e-05,
+      "loss": 0.9551,
+      "step": 1913
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.3577181208053692e-05,
+      "loss": 0.9668,
+      "step": 1914
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.3573825503355706e-05,
+      "loss": 0.9634,
+      "step": 1915
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.357046979865772e-05,
+      "loss": 0.9551,
+      "step": 1916
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.3567114093959733e-05,
+      "loss": 0.9844,
+      "step": 1917
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.3563758389261745e-05,
+      "loss": 0.9414,
+      "step": 1918
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.3560402684563761e-05,
+      "loss": 0.9932,
+      "step": 1919
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.3557046979865774e-05,
+      "loss": 0.9575,
+      "step": 1920
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.3553691275167786e-05,
+      "loss": 0.9619,
+      "step": 1921
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.3550335570469799e-05,
+      "loss": 0.9893,
+      "step": 1922
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.3546979865771812e-05,
+      "loss": 0.9614,
+      "step": 1923
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.3543624161073827e-05,
+      "loss": 0.9688,
+      "step": 1924
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 1.354026845637584e-05,
+      "loss": 0.9321,
+      "step": 1925
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.3536912751677853e-05,
+      "loss": 0.9263,
+      "step": 1926
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.3533557046979865e-05,
+      "loss": 0.9648,
+      "step": 1927
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.3530201342281881e-05,
+      "loss": 0.9458,
+      "step": 1928
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.3526845637583894e-05,
+      "loss": 0.9644,
+      "step": 1929
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.3523489932885906e-05,
+      "loss": 0.978,
+      "step": 1930
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.352013422818792e-05,
+      "loss": 0.9639,
+      "step": 1931
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.3516778523489933e-05,
+      "loss": 0.9297,
+      "step": 1932
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.3513422818791947e-05,
+      "loss": 0.9468,
+      "step": 1933
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.3510067114093962e-05,
+      "loss": 0.9243,
+      "step": 1934
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.3506711409395974e-05,
+      "loss": 0.9126,
+      "step": 1935
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.3503355704697987e-05,
+      "loss": 0.9614,
+      "step": 1936
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 1.3500000000000001e-05,
+      "loss": 0.9707,
+      "step": 1937
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.3496644295302015e-05,
+      "loss": 0.9878,
+      "step": 1938
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.3493288590604028e-05,
+      "loss": 0.9395,
+      "step": 1939
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.348993288590604e-05,
+      "loss": 0.958,
+      "step": 1940
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.3486577181208053e-05,
+      "loss": 0.9409,
+      "step": 1941
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.348322147651007e-05,
+      "loss": 0.9771,
+      "step": 1942
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.3479865771812082e-05,
+      "loss": 0.9712,
+      "step": 1943
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.3476510067114094e-05,
+      "loss": 0.9585,
+      "step": 1944
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.3473154362416107e-05,
+      "loss": 0.9673,
+      "step": 1945
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.3469798657718123e-05,
+      "loss": 0.9629,
+      "step": 1946
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.3466442953020135e-05,
+      "loss": 0.9292,
+      "step": 1947
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 1.3463087248322148e-05,
+      "loss": 0.9697,
+      "step": 1948
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.3459731543624162e-05,
+      "loss": 0.9824,
+      "step": 1949
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.3456375838926175e-05,
+      "loss": 0.9478,
+      "step": 1950
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.345302013422819e-05,
+      "loss": 0.9434,
+      "step": 1951
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.3449664429530202e-05,
+      "loss": 0.9961,
+      "step": 1952
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.3446308724832216e-05,
+      "loss": 0.9517,
+      "step": 1953
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.3442953020134229e-05,
+      "loss": 0.9751,
+      "step": 1954
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.3439597315436243e-05,
+      "loss": 1.0117,
+      "step": 1955
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.3436241610738257e-05,
+      "loss": 0.9272,
+      "step": 1956
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.343288590604027e-05,
+      "loss": 0.9458,
+      "step": 1957
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.3429530201342282e-05,
+      "loss": 0.9146,
+      "step": 1958
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.3426174496644298e-05,
+      "loss": 0.9663,
+      "step": 1959
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 1.342281879194631e-05,
+      "loss": 0.9111,
+      "step": 1960
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.3419463087248323e-05,
+      "loss": 0.9116,
+      "step": 1961
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.3416107382550336e-05,
+      "loss": 0.958,
+      "step": 1962
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.3412751677852349e-05,
+      "loss": 0.938,
+      "step": 1963
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.3409395973154365e-05,
+      "loss": 0.9883,
+      "step": 1964
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.3406040268456377e-05,
+      "loss": 0.98,
+      "step": 1965
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.340268456375839e-05,
+      "loss": 1.0005,
+      "step": 1966
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.3399328859060404e-05,
+      "loss": 0.938,
+      "step": 1967
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.3395973154362418e-05,
+      "loss": 0.9971,
+      "step": 1968
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.339261744966443e-05,
+      "loss": 0.9536,
+      "step": 1969
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.3389261744966443e-05,
+      "loss": 0.9849,
+      "step": 1970
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.3385906040268458e-05,
+      "loss": 0.9507,
+      "step": 1971
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 1.338255033557047e-05,
+      "loss": 0.9741,
+      "step": 1972
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.3379194630872485e-05,
+      "loss": 0.9478,
+      "step": 1973
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.3375838926174499e-05,
+      "loss": 0.9404,
+      "step": 1974
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.3372483221476511e-05,
+      "loss": 0.959,
+      "step": 1975
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.3369127516778524e-05,
+      "loss": 0.9814,
+      "step": 1976
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.3365771812080538e-05,
+      "loss": 0.9268,
+      "step": 1977
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.3362416107382553e-05,
+      "loss": 0.9507,
+      "step": 1978
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.3359060402684565e-05,
+      "loss": 0.9727,
+      "step": 1979
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.3355704697986578e-05,
+      "loss": 1.0005,
+      "step": 1980
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.335234899328859e-05,
+      "loss": 0.9629,
+      "step": 1981
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.3348993288590606e-05,
+      "loss": 0.9785,
+      "step": 1982
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.3345637583892619e-05,
+      "loss": 0.9541,
+      "step": 1983
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 1.3342281879194631e-05,
+      "loss": 0.9331,
+      "step": 1984
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3338926174496644e-05,
+      "loss": 0.998,
+      "step": 1985
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.333557046979866e-05,
+      "loss": 0.9766,
+      "step": 1986
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3332214765100673e-05,
+      "loss": 0.9385,
+      "step": 1987
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3328859060402685e-05,
+      "loss": 0.9429,
+      "step": 1988
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.33255033557047e-05,
+      "loss": 0.9634,
+      "step": 1989
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3322147651006712e-05,
+      "loss": 0.9648,
+      "step": 1990
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3318791946308726e-05,
+      "loss": 0.9233,
+      "step": 1991
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.331543624161074e-05,
+      "loss": 0.9468,
+      "step": 1992
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3312080536912753e-05,
+      "loss": 0.9565,
+      "step": 1993
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3308724832214766e-05,
+      "loss": 0.9487,
+      "step": 1994
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.330536912751678e-05,
+      "loss": 0.9541,
+      "step": 1995
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 1.3302013422818794e-05,
+      "loss": 0.9521,
+      "step": 1996
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3298657718120807e-05,
+      "loss": 0.9761,
+      "step": 1997
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.329530201342282e-05,
+      "loss": 0.9639,
+      "step": 1998
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3291946308724832e-05,
+      "loss": 0.9551,
+      "step": 1999
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3288590604026848e-05,
+      "loss": 0.9727,
+      "step": 2000
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.328523489932886e-05,
+      "loss": 0.9834,
+      "step": 2001
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3281879194630873e-05,
+      "loss": 0.9766,
+      "step": 2002
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3278523489932886e-05,
+      "loss": 0.981,
+      "step": 2003
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3275167785234902e-05,
+      "loss": 0.9771,
+      "step": 2004
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3271812080536914e-05,
+      "loss": 0.9712,
+      "step": 2005
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3268456375838927e-05,
+      "loss": 0.9619,
+      "step": 2006
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3265100671140941e-05,
+      "loss": 0.9541,
+      "step": 2007
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 1.3261744966442954e-05,
+      "loss": 0.9575,
+      "step": 2008
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.3258389261744968e-05,
+      "loss": 0.9336,
+      "step": 2009
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.325503355704698e-05,
+      "loss": 0.9229,
+      "step": 2010
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.3251677852348995e-05,
+      "loss": 0.9502,
+      "step": 2011
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.3248322147651007e-05,
+      "loss": 0.9917,
+      "step": 2012
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.3244966442953022e-05,
+      "loss": 0.8965,
+      "step": 2013
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.3241610738255036e-05,
+      "loss": 0.959,
+      "step": 2014
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.3238255033557048e-05,
+      "loss": 0.9434,
+      "step": 2015
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.3234899328859061e-05,
+      "loss": 0.9683,
+      "step": 2016
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.3231543624161074e-05,
+      "loss": 0.9927,
+      "step": 2017
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.322818791946309e-05,
+      "loss": 0.9575,
+      "step": 2018
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.3224832214765102e-05,
+      "loss": 0.9468,
+      "step": 2019
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 1.3221476510067115e-05,
+      "loss": 0.9487,
+      "step": 2020
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.3218120805369127e-05,
+      "loss": 0.9639,
+      "step": 2021
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.3214765100671143e-05,
+      "loss": 0.9346,
+      "step": 2022
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.3211409395973156e-05,
+      "loss": 0.9419,
+      "step": 2023
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.3208053691275168e-05,
+      "loss": 0.9175,
+      "step": 2024
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.3204697986577181e-05,
+      "loss": 0.9595,
+      "step": 2025
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.3201342281879195e-05,
+      "loss": 0.9048,
+      "step": 2026
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.319798657718121e-05,
+      "loss": 0.979,
+      "step": 2027
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.3194630872483222e-05,
+      "loss": 0.9614,
+      "step": 2028
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.3191275167785236e-05,
+      "loss": 0.9438,
+      "step": 2029
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.3187919463087249e-05,
+      "loss": 0.9194,
+      "step": 2030
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.3184563758389263e-05,
+      "loss": 0.9556,
+      "step": 2031
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 1.3181208053691278e-05,
+      "loss": 0.9741,
+      "step": 2032
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.317785234899329e-05,
+      "loss": 0.9287,
+      "step": 2033
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.3174496644295303e-05,
+      "loss": 0.9976,
+      "step": 2034
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.3171140939597315e-05,
+      "loss": 0.9507,
+      "step": 2035
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.3167785234899331e-05,
+      "loss": 0.9395,
+      "step": 2036
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.3164429530201344e-05,
+      "loss": 0.9453,
+      "step": 2037
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.3161073825503356e-05,
+      "loss": 0.9746,
+      "step": 2038
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.3157718120805369e-05,
+      "loss": 0.96,
+      "step": 2039
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.3154362416107385e-05,
+      "loss": 0.9824,
+      "step": 2040
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.3151006711409398e-05,
+      "loss": 0.9478,
+      "step": 2041
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.314765100671141e-05,
+      "loss": 0.9014,
+      "step": 2042
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.3144295302013423e-05,
+      "loss": 0.9668,
+      "step": 2043
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 1.3140939597315437e-05,
+      "loss": 0.9644,
+      "step": 2044
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.3137583892617451e-05,
+      "loss": 0.9243,
+      "step": 2045
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.3134228187919464e-05,
+      "loss": 0.9668,
+      "step": 2046
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.3130872483221478e-05,
+      "loss": 0.9321,
+      "step": 2047
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.312751677852349e-05,
+      "loss": 0.9155,
+      "step": 2048
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.3124161073825505e-05,
+      "loss": 0.9336,
+      "step": 2049
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.3120805369127518e-05,
+      "loss": 0.9404,
+      "step": 2050
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.3117449664429532e-05,
+      "loss": 0.9639,
+      "step": 2051
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.3114093959731544e-05,
+      "loss": 0.9434,
+      "step": 2052
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.3110738255033557e-05,
+      "loss": 0.981,
+      "step": 2053
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.3107382550335573e-05,
+      "loss": 0.9644,
+      "step": 2054
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.3104026845637586e-05,
+      "loss": 0.9707,
+      "step": 2055
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 1.3100671140939598e-05,
+      "loss": 0.9155,
+      "step": 2056
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.309731543624161e-05,
+      "loss": 0.9395,
+      "step": 2057
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3093959731543627e-05,
+      "loss": 0.9644,
+      "step": 2058
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.309060402684564e-05,
+      "loss": 0.9922,
+      "step": 2059
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3087248322147652e-05,
+      "loss": 0.9517,
+      "step": 2060
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3083892617449664e-05,
+      "loss": 0.917,
+      "step": 2061
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3080536912751679e-05,
+      "loss": 0.979,
+      "step": 2062
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3077181208053693e-05,
+      "loss": 0.9224,
+      "step": 2063
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3073825503355706e-05,
+      "loss": 0.9219,
+      "step": 2064
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3070469798657718e-05,
+      "loss": 0.9619,
+      "step": 2065
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3067114093959732e-05,
+      "loss": 0.9702,
+      "step": 2066
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.3063758389261747e-05,
+      "loss": 0.9453,
+      "step": 2067
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 1.306040268456376e-05,
+      "loss": 0.9258,
+      "step": 2068
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3057046979865773e-05,
+      "loss": 0.9316,
+      "step": 2069
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3053691275167786e-05,
+      "loss": 0.9482,
+      "step": 2070
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3050335570469799e-05,
+      "loss": 0.9492,
+      "step": 2071
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3046979865771815e-05,
+      "loss": 0.9814,
+      "step": 2072
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3043624161073827e-05,
+      "loss": 0.9468,
+      "step": 2073
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.304026845637584e-05,
+      "loss": 0.9546,
+      "step": 2074
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3036912751677852e-05,
+      "loss": 0.9751,
+      "step": 2075
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3033557046979868e-05,
+      "loss": 0.9536,
+      "step": 2076
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3030201342281881e-05,
+      "loss": 0.9761,
+      "step": 2077
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3026845637583893e-05,
+      "loss": 0.938,
+      "step": 2078
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.3023489932885906e-05,
+      "loss": 0.9204,
+      "step": 2079
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 1.302013422818792e-05,
+      "loss": 0.9517,
+      "step": 2080
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.3016778523489935e-05,
+      "loss": 0.9609,
+      "step": 2081
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.3013422818791947e-05,
+      "loss": 0.957,
+      "step": 2082
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.301006711409396e-05,
+      "loss": 0.9463,
+      "step": 2083
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.3006711409395974e-05,
+      "loss": 0.9458,
+      "step": 2084
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.3003355704697988e-05,
+      "loss": 1.002,
+      "step": 2085
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.3000000000000001e-05,
+      "loss": 0.981,
+      "step": 2086
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2996644295302015e-05,
+      "loss": 0.9536,
+      "step": 2087
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2993288590604028e-05,
+      "loss": 0.9546,
+      "step": 2088
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.298993288590604e-05,
+      "loss": 0.9849,
+      "step": 2089
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2986577181208055e-05,
+      "loss": 0.979,
+      "step": 2090
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 1.2983221476510069e-05,
+      "loss": 0.9072,
+      "step": 2091
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.2979865771812081e-05,
+      "loss": 0.9634,
+      "step": 2092
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.2976510067114094e-05,
+      "loss": 0.9097,
+      "step": 2093
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.297315436241611e-05,
+      "loss": 0.9722,
+      "step": 2094
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.2969798657718123e-05,
+      "loss": 0.9497,
+      "step": 2095
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.2966442953020135e-05,
+      "loss": 0.9185,
+      "step": 2096
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.2963087248322148e-05,
+      "loss": 0.9712,
+      "step": 2097
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.295973154362416e-05,
+      "loss": 0.9663,
+      "step": 2098
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.2956375838926176e-05,
+      "loss": 0.915,
+      "step": 2099
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.2953020134228189e-05,
+      "loss": 0.9653,
+      "step": 2100
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.2949664429530201e-05,
+      "loss": 0.957,
+      "step": 2101
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.2946308724832216e-05,
+      "loss": 0.9194,
+      "step": 2102
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 1.294295302013423e-05,
+      "loss": 0.9707,
+      "step": 2103
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.2939597315436243e-05,
+      "loss": 0.96,
+      "step": 2104
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.2936241610738257e-05,
+      "loss": 0.9336,
+      "step": 2105
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.293288590604027e-05,
+      "loss": 0.979,
+      "step": 2106
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.2929530201342282e-05,
+      "loss": 0.9204,
+      "step": 2107
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.2926174496644296e-05,
+      "loss": 0.9805,
+      "step": 2108
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.292281879194631e-05,
+      "loss": 0.9751,
+      "step": 2109
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.2919463087248323e-05,
+      "loss": 0.9565,
+      "step": 2110
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.2916107382550336e-05,
+      "loss": 0.9722,
+      "step": 2111
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.2912751677852352e-05,
+      "loss": 0.9844,
+      "step": 2112
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.2909395973154364e-05,
+      "loss": 0.9741,
+      "step": 2113
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.2906040268456377e-05,
+      "loss": 0.9272,
+      "step": 2114
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 1.290268456375839e-05,
+      "loss": 0.938,
+      "step": 2115
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.2899328859060402e-05,
+      "loss": 0.9785,
+      "step": 2116
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.2895973154362418e-05,
+      "loss": 0.9209,
+      "step": 2117
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.289261744966443e-05,
+      "loss": 0.9429,
+      "step": 2118
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.2889261744966443e-05,
+      "loss": 0.9604,
+      "step": 2119
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.2885906040268457e-05,
+      "loss": 0.9453,
+      "step": 2120
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.2882550335570472e-05,
+      "loss": 0.9473,
+      "step": 2121
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.2879194630872484e-05,
+      "loss": 0.9385,
+      "step": 2122
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.2875838926174497e-05,
+      "loss": 0.9473,
+      "step": 2123
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.2872483221476511e-05,
+      "loss": 0.917,
+      "step": 2124
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.2869127516778524e-05,
+      "loss": 0.9717,
+      "step": 2125
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.2865771812080538e-05,
+      "loss": 0.959,
+      "step": 2126
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 1.2862416107382552e-05,
+      "loss": 0.9277,
+      "step": 2127
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.2859060402684565e-05,
+      "loss": 0.9624,
+      "step": 2128
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.2855704697986577e-05,
+      "loss": 0.9507,
+      "step": 2129
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.2852348993288593e-05,
+      "loss": 0.9404,
+      "step": 2130
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.2848993288590606e-05,
+      "loss": 0.9194,
+      "step": 2131
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.2845637583892619e-05,
+      "loss": 0.9297,
+      "step": 2132
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.2842281879194631e-05,
+      "loss": 0.9858,
+      "step": 2133
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.2838926174496644e-05,
+      "loss": 0.9258,
+      "step": 2134
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.283557046979866e-05,
+      "loss": 0.9575,
+      "step": 2135
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.2832214765100672e-05,
+      "loss": 0.9561,
+      "step": 2136
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.2828859060402685e-05,
+      "loss": 0.9609,
+      "step": 2137
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.2825503355704697e-05,
+      "loss": 0.9072,
+      "step": 2138
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 1.2822147651006713e-05,
+      "loss": 0.9663,
+      "step": 2139
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.2818791946308726e-05,
+      "loss": 0.9829,
+      "step": 2140
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.2815436241610739e-05,
+      "loss": 0.9829,
+      "step": 2141
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.2812080536912753e-05,
+      "loss": 0.9497,
+      "step": 2142
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.2808724832214767e-05,
+      "loss": 0.9707,
+      "step": 2143
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.280536912751678e-05,
+      "loss": 0.9326,
+      "step": 2144
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.2802013422818794e-05,
+      "loss": 0.9697,
+      "step": 2145
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.2798657718120806e-05,
+      "loss": 0.9868,
+      "step": 2146
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.2795302013422819e-05,
+      "loss": 0.9775,
+      "step": 2147
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.2791946308724833e-05,
+      "loss": 0.9365,
+      "step": 2148
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.2788590604026848e-05,
+      "loss": 0.9346,
+      "step": 2149
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.278523489932886e-05,
+      "loss": 0.9766,
+      "step": 2150
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 1.2781879194630873e-05,
+      "loss": 0.9893,
+      "step": 2151
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.2778523489932889e-05,
+      "loss": 0.958,
+      "step": 2152
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.2775167785234901e-05,
+      "loss": 0.9326,
+      "step": 2153
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.2771812080536914e-05,
+      "loss": 0.9268,
+      "step": 2154
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.2768456375838926e-05,
+      "loss": 0.9248,
+      "step": 2155
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.2765100671140939e-05,
+      "loss": 0.9697,
+      "step": 2156
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.2761744966442955e-05,
+      "loss": 0.9595,
+      "step": 2157
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.2758389261744968e-05,
+      "loss": 0.9424,
+      "step": 2158
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.275503355704698e-05,
+      "loss": 0.9307,
+      "step": 2159
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.2751677852348994e-05,
+      "loss": 0.9717,
+      "step": 2160
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.2748322147651009e-05,
+      "loss": 0.9536,
+      "step": 2161
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.2744966442953021e-05,
+      "loss": 0.9951,
+      "step": 2162
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 1.2741610738255034e-05,
+      "loss": 1.021,
+      "step": 2163
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.2738255033557048e-05,
+      "loss": 0.9526,
+      "step": 2164
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.273489932885906e-05,
+      "loss": 0.9185,
+      "step": 2165
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.2731543624161075e-05,
+      "loss": 0.9917,
+      "step": 2166
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.272818791946309e-05,
+      "loss": 0.9673,
+      "step": 2167
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.2724832214765102e-05,
+      "loss": 0.9624,
+      "step": 2168
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.2721476510067114e-05,
+      "loss": 0.9399,
+      "step": 2169
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.271812080536913e-05,
+      "loss": 0.9404,
+      "step": 2170
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.2714765100671143e-05,
+      "loss": 0.9717,
+      "step": 2171
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.2711409395973156e-05,
+      "loss": 0.9243,
+      "step": 2172
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.2708053691275168e-05,
+      "loss": 0.9487,
+      "step": 2173
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.270469798657718e-05,
+      "loss": 0.9409,
+      "step": 2174
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 1.2701342281879197e-05,
+      "loss": 0.9175,
+      "step": 2175
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.269798657718121e-05,
+      "loss": 0.9468,
+      "step": 2176
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.2694630872483222e-05,
+      "loss": 0.9629,
+      "step": 2177
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.2691275167785234e-05,
+      "loss": 0.9829,
+      "step": 2178
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.268791946308725e-05,
+      "loss": 0.981,
+      "step": 2179
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.2684563758389263e-05,
+      "loss": 0.9507,
+      "step": 2180
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.2681208053691276e-05,
+      "loss": 0.958,
+      "step": 2181
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.267785234899329e-05,
+      "loss": 0.9556,
+      "step": 2182
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.2674496644295302e-05,
+      "loss": 0.9595,
+      "step": 2183
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.2671140939597317e-05,
+      "loss": 0.9404,
+      "step": 2184
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.2667785234899331e-05,
+      "loss": 0.9331,
+      "step": 2185
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.2664429530201344e-05,
+      "loss": 0.9385,
+      "step": 2186
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 1.2661073825503356e-05,
+      "loss": 0.9526,
+      "step": 2187
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.265771812080537e-05,
+      "loss": 0.9619,
+      "step": 2188
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.2654362416107385e-05,
+      "loss": 0.9448,
+      "step": 2189
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.2651006711409397e-05,
+      "loss": 0.9536,
+      "step": 2190
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.264765100671141e-05,
+      "loss": 0.9351,
+      "step": 2191
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.2644295302013422e-05,
+      "loss": 0.9712,
+      "step": 2192
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.2640939597315438e-05,
+      "loss": 0.9658,
+      "step": 2193
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.2637583892617451e-05,
+      "loss": 0.9585,
+      "step": 2194
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.2634228187919464e-05,
+      "loss": 0.9487,
+      "step": 2195
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.2630872483221476e-05,
+      "loss": 0.9478,
+      "step": 2196
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.2627516778523492e-05,
+      "loss": 0.9688,
+      "step": 2197
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.2624161073825505e-05,
+      "loss": 0.9463,
+      "step": 2198
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 1.2620805369127517e-05,
+      "loss": 1.0181,
+      "step": 2199
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.2617449664429532e-05,
+      "loss": 0.9385,
+      "step": 2200
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.2614093959731544e-05,
+      "loss": 0.981,
+      "step": 2201
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.2610738255033558e-05,
+      "loss": 0.9561,
+      "step": 2202
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.2607382550335571e-05,
+      "loss": 0.9248,
+      "step": 2203
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.2604026845637585e-05,
+      "loss": 0.9355,
+      "step": 2204
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.2600671140939598e-05,
+      "loss": 0.9351,
+      "step": 2205
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.2597315436241612e-05,
+      "loss": 0.9941,
+      "step": 2206
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.2593959731543626e-05,
+      "loss": 0.9805,
+      "step": 2207
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.2590604026845639e-05,
+      "loss": 0.9604,
+      "step": 2208
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.2587248322147652e-05,
+      "loss": 0.9624,
+      "step": 2209
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.2583892617449664e-05,
+      "loss": 0.9849,
+      "step": 2210
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 1.258053691275168e-05,
+      "loss": 0.9443,
+      "step": 2211
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.2577181208053693e-05,
+      "loss": 0.9565,
+      "step": 2212
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.2573825503355705e-05,
+      "loss": 0.9902,
+      "step": 2213
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.2570469798657718e-05,
+      "loss": 0.9727,
+      "step": 2214
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.2567114093959734e-05,
+      "loss": 0.9375,
+      "step": 2215
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.2563758389261746e-05,
+      "loss": 0.9146,
+      "step": 2216
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.2560402684563759e-05,
+      "loss": 0.9102,
+      "step": 2217
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.2557046979865773e-05,
+      "loss": 0.9653,
+      "step": 2218
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.2553691275167786e-05,
+      "loss": 0.9707,
+      "step": 2219
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.25503355704698e-05,
+      "loss": 0.9673,
+      "step": 2220
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.2546979865771813e-05,
+      "loss": 0.9512,
+      "step": 2221
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.2543624161073827e-05,
+      "loss": 0.9497,
+      "step": 2222
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 1.254026845637584e-05,
+      "loss": 0.9541,
+      "step": 2223
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.2536912751677854e-05,
+      "loss": 0.9326,
+      "step": 2224
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.2533557046979868e-05,
+      "loss": 0.936,
+      "step": 2225
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.253020134228188e-05,
+      "loss": 0.9604,
+      "step": 2226
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.2526845637583893e-05,
+      "loss": 0.96,
+      "step": 2227
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.2523489932885906e-05,
+      "loss": 0.9473,
+      "step": 2228
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.2520134228187922e-05,
+      "loss": 0.9331,
+      "step": 2229
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.2516778523489934e-05,
+      "loss": 0.9473,
+      "step": 2230
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.2513422818791947e-05,
+      "loss": 0.9639,
+      "step": 2231
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.251006711409396e-05,
+      "loss": 0.98,
+      "step": 2232
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.2506711409395975e-05,
+      "loss": 0.9658,
+      "step": 2233
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 1.2503355704697988e-05,
+      "loss": 0.9648,
+      "step": 2234
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.25e-05,
+      "loss": 0.9336,
+      "step": 2235
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.2496644295302013e-05,
+      "loss": 0.9741,
+      "step": 2236
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.2493288590604027e-05,
+      "loss": 0.9438,
+      "step": 2237
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.2489932885906042e-05,
+      "loss": 0.9512,
+      "step": 2238
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.2486577181208054e-05,
+      "loss": 0.9624,
+      "step": 2239
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.2483221476510069e-05,
+      "loss": 0.9414,
+      "step": 2240
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.2479865771812081e-05,
+      "loss": 0.9414,
+      "step": 2241
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.2476510067114095e-05,
+      "loss": 0.9604,
+      "step": 2242
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.247315436241611e-05,
+      "loss": 0.9575,
+      "step": 2243
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.2469798657718122e-05,
+      "loss": 0.9531,
+      "step": 2244
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.2466442953020135e-05,
+      "loss": 0.9829,
+      "step": 2245
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 1.2463087248322147e-05,
+      "loss": 0.9604,
+      "step": 2246
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.2459731543624163e-05,
+      "loss": 0.9126,
+      "step": 2247
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.2456375838926176e-05,
+      "loss": 0.9902,
+      "step": 2248
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.2453020134228189e-05,
+      "loss": 0.9297,
+      "step": 2249
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.2449664429530201e-05,
+      "loss": 0.9551,
+      "step": 2250
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.2446308724832217e-05,
+      "loss": 0.9189,
+      "step": 2251
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.244295302013423e-05,
+      "loss": 0.915,
+      "step": 2252
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.2439597315436242e-05,
+      "loss": 0.9355,
+      "step": 2253
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.2436241610738255e-05,
+      "loss": 0.9219,
+      "step": 2254
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.2432885906040269e-05,
+      "loss": 0.9487,
+      "step": 2255
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.2429530201342283e-05,
+      "loss": 0.9326,
+      "step": 2256
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.2426174496644296e-05,
+      "loss": 0.9663,
+      "step": 2257
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 1.242281879194631e-05,
+      "loss": 0.9204,
+      "step": 2258
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2419463087248323e-05,
+      "loss": 0.9512,
+      "step": 2259
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2416107382550337e-05,
+      "loss": 0.9341,
+      "step": 2260
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.241275167785235e-05,
+      "loss": 0.9619,
+      "step": 2261
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2409395973154364e-05,
+      "loss": 0.9771,
+      "step": 2262
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2406040268456377e-05,
+      "loss": 0.9707,
+      "step": 2263
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2402684563758389e-05,
+      "loss": 0.9263,
+      "step": 2264
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2399328859060405e-05,
+      "loss": 0.9043,
+      "step": 2265
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2395973154362418e-05,
+      "loss": 0.9761,
+      "step": 2266
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.239261744966443e-05,
+      "loss": 0.9517,
+      "step": 2267
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2389261744966443e-05,
+      "loss": 0.958,
+      "step": 2268
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2385906040268459e-05,
+      "loss": 0.9658,
+      "step": 2269
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 1.2382550335570471e-05,
+      "loss": 0.9429,
+      "step": 2270
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.2379194630872484e-05,
+      "loss": 0.9424,
+      "step": 2271
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.2375838926174497e-05,
+      "loss": 0.9561,
+      "step": 2272
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.237248322147651e-05,
+      "loss": 0.9131,
+      "step": 2273
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.2369127516778525e-05,
+      "loss": 0.9741,
+      "step": 2274
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.2365771812080538e-05,
+      "loss": 0.9619,
+      "step": 2275
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.236241610738255e-05,
+      "loss": 0.9224,
+      "step": 2276
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.2359060402684565e-05,
+      "loss": 0.895,
+      "step": 2277
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.2355704697986579e-05,
+      "loss": 0.9155,
+      "step": 2278
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.2352348993288591e-05,
+      "loss": 0.9531,
+      "step": 2279
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.2348993288590606e-05,
+      "loss": 0.9263,
+      "step": 2280
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.2345637583892618e-05,
+      "loss": 0.9429,
+      "step": 2281
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 1.234228187919463e-05,
+      "loss": 0.9414,
+      "step": 2282
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.2338926174496647e-05,
+      "loss": 0.957,
+      "step": 2283
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.233557046979866e-05,
+      "loss": 0.9634,
+      "step": 2284
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.2332214765100672e-05,
+      "loss": 0.9224,
+      "step": 2285
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.2328859060402685e-05,
+      "loss": 0.9746,
+      "step": 2286
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.23255033557047e-05,
+      "loss": 0.9419,
+      "step": 2287
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.2322147651006713e-05,
+      "loss": 0.9414,
+      "step": 2288
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.2318791946308726e-05,
+      "loss": 0.8999,
+      "step": 2289
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.2315436241610738e-05,
+      "loss": 0.9531,
+      "step": 2290
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.231208053691275e-05,
+      "loss": 0.9492,
+      "step": 2291
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.2308724832214767e-05,
+      "loss": 0.9634,
+      "step": 2292
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.230536912751678e-05,
+      "loss": 0.9502,
+      "step": 2293
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 1.2302013422818792e-05,
+      "loss": 0.9556,
+      "step": 2294
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.2298657718120806e-05,
+      "loss": 0.938,
+      "step": 2295
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.229530201342282e-05,
+      "loss": 0.9961,
+      "step": 2296
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.2291946308724833e-05,
+      "loss": 0.9434,
+      "step": 2297
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.2288590604026847e-05,
+      "loss": 0.9253,
+      "step": 2298
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.228523489932886e-05,
+      "loss": 0.9409,
+      "step": 2299
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.2281879194630872e-05,
+      "loss": 0.9507,
+      "step": 2300
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.2278523489932887e-05,
+      "loss": 0.9736,
+      "step": 2301
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.2275167785234901e-05,
+      "loss": 0.9355,
+      "step": 2302
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.2271812080536914e-05,
+      "loss": 0.9019,
+      "step": 2303
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.2268456375838926e-05,
+      "loss": 0.9609,
+      "step": 2304
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.2265100671140942e-05,
+      "loss": 0.9507,
+      "step": 2305
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 1.2261744966442955e-05,
+      "loss": 0.9409,
+      "step": 2306
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.2258389261744967e-05,
+      "loss": 0.9375,
+      "step": 2307
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.225503355704698e-05,
+      "loss": 0.9619,
+      "step": 2308
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.2251677852348992e-05,
+      "loss": 0.9741,
+      "step": 2309
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.2248322147651008e-05,
+      "loss": 0.9932,
+      "step": 2310
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.2244966442953021e-05,
+      "loss": 0.9351,
+      "step": 2311
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.2241610738255034e-05,
+      "loss": 0.9736,
+      "step": 2312
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.2238255033557048e-05,
+      "loss": 0.9507,
+      "step": 2313
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.2234899328859062e-05,
+      "loss": 0.9658,
+      "step": 2314
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.2231543624161075e-05,
+      "loss": 0.9727,
+      "step": 2315
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.2228187919463087e-05,
+      "loss": 0.9692,
+      "step": 2316
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.2224832214765102e-05,
+      "loss": 0.9351,
+      "step": 2317
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 1.2221476510067114e-05,
+      "loss": 0.9277,
+      "step": 2318
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.2218120805369128e-05,
+      "loss": 0.9565,
+      "step": 2319
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.2214765100671143e-05,
+      "loss": 0.9492,
+      "step": 2320
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.2211409395973155e-05,
+      "loss": 0.9814,
+      "step": 2321
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.2208053691275168e-05,
+      "loss": 0.938,
+      "step": 2322
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.2204697986577184e-05,
+      "loss": 0.9482,
+      "step": 2323
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.2201342281879196e-05,
+      "loss": 0.9546,
+      "step": 2324
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.2197986577181209e-05,
+      "loss": 0.9883,
+      "step": 2325
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.2194630872483222e-05,
+      "loss": 0.9385,
+      "step": 2326
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.2191275167785234e-05,
+      "loss": 0.9375,
+      "step": 2327
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.218791946308725e-05,
+      "loss": 0.9224,
+      "step": 2328
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.2184563758389263e-05,
+      "loss": 0.9502,
+      "step": 2329
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 1.2181208053691275e-05,
+      "loss": 0.9512,
+      "step": 2330
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.217785234899329e-05,
+      "loss": 0.9473,
+      "step": 2331
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.2174496644295304e-05,
+      "loss": 0.9702,
+      "step": 2332
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.2171140939597316e-05,
+      "loss": 0.8887,
+      "step": 2333
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.2167785234899329e-05,
+      "loss": 0.9346,
+      "step": 2334
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.2164429530201343e-05,
+      "loss": 0.9858,
+      "step": 2335
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.2161073825503358e-05,
+      "loss": 0.9248,
+      "step": 2336
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.215771812080537e-05,
+      "loss": 0.9556,
+      "step": 2337
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.2154362416107384e-05,
+      "loss": 0.9224,
+      "step": 2338
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.2151006711409397e-05,
+      "loss": 0.9688,
+      "step": 2339
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.214765100671141e-05,
+      "loss": 0.9497,
+      "step": 2340
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.2144295302013426e-05,
+      "loss": 1.0,
+      "step": 2341
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 1.2140939597315438e-05,
+      "loss": 0.959,
+      "step": 2342
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.213758389261745e-05,
+      "loss": 0.9126,
+      "step": 2343
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.2134228187919463e-05,
+      "loss": 0.9292,
+      "step": 2344
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.213087248322148e-05,
+      "loss": 0.98,
+      "step": 2345
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.2127516778523492e-05,
+      "loss": 0.9443,
+      "step": 2346
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.2124161073825504e-05,
+      "loss": 0.9639,
+      "step": 2347
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.2120805369127517e-05,
+      "loss": 0.9502,
+      "step": 2348
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.211744966442953e-05,
+      "loss": 0.9385,
+      "step": 2349
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.2114093959731545e-05,
+      "loss": 0.9199,
+      "step": 2350
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.2110738255033558e-05,
+      "loss": 0.9365,
+      "step": 2351
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.210738255033557e-05,
+      "loss": 0.9175,
+      "step": 2352
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.2104026845637585e-05,
+      "loss": 0.9746,
+      "step": 2353
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 1.21006711409396e-05,
+      "loss": 0.9766,
+      "step": 2354
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.2097315436241612e-05,
+      "loss": 0.9424,
+      "step": 2355
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.2093959731543626e-05,
+      "loss": 0.9805,
+      "step": 2356
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.2090604026845639e-05,
+      "loss": 0.9619,
+      "step": 2357
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.2087248322147651e-05,
+      "loss": 0.9766,
+      "step": 2358
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.2083892617449665e-05,
+      "loss": 0.9673,
+      "step": 2359
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.208053691275168e-05,
+      "loss": 0.978,
+      "step": 2360
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.2077181208053692e-05,
+      "loss": 0.9556,
+      "step": 2361
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.2073825503355705e-05,
+      "loss": 0.9595,
+      "step": 2362
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.2070469798657721e-05,
+      "loss": 0.9307,
+      "step": 2363
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.2067114093959733e-05,
+      "loss": 0.9624,
+      "step": 2364
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.2063758389261746e-05,
+      "loss": 0.9619,
+      "step": 2365
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 1.2060402684563759e-05,
+      "loss": 0.9668,
+      "step": 2366
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.2057046979865771e-05,
+      "loss": 0.9678,
+      "step": 2367
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.2053691275167787e-05,
+      "loss": 0.9043,
+      "step": 2368
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.20503355704698e-05,
+      "loss": 0.9282,
+      "step": 2369
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.2046979865771812e-05,
+      "loss": 0.9399,
+      "step": 2370
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.2043624161073827e-05,
+      "loss": 0.9697,
+      "step": 2371
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.2040268456375841e-05,
+      "loss": 0.9585,
+      "step": 2372
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.2036912751677853e-05,
+      "loss": 0.9561,
+      "step": 2373
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.2033557046979866e-05,
+      "loss": 0.9507,
+      "step": 2374
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.203020134228188e-05,
+      "loss": 0.9912,
+      "step": 2375
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.2026845637583893e-05,
+      "loss": 0.9048,
+      "step": 2376
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.2023489932885907e-05,
+      "loss": 0.9082,
+      "step": 2377
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 1.2020134228187921e-05,
+      "loss": 0.9062,
+      "step": 2378
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.2016778523489934e-05,
+      "loss": 0.96,
+      "step": 2379
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.2013422818791947e-05,
+      "loss": 0.9614,
+      "step": 2380
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.2010067114093963e-05,
+      "loss": 0.9399,
+      "step": 2381
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.2006711409395975e-05,
+      "loss": 0.9546,
+      "step": 2382
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.2003355704697988e-05,
+      "loss": 0.9121,
+      "step": 2383
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.2e-05,
+      "loss": 0.9619,
+      "step": 2384
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.1996644295302013e-05,
+      "loss": 0.9199,
+      "step": 2385
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.1993288590604029e-05,
+      "loss": 0.8784,
+      "step": 2386
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.1989932885906041e-05,
+      "loss": 0.9199,
+      "step": 2387
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.1986577181208054e-05,
+      "loss": 0.9209,
+      "step": 2388
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 1.1983221476510067e-05,
+      "loss": 0.9448,
+      "step": 2389
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.1979865771812083e-05,
+      "loss": 0.9224,
+      "step": 2390
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.1976510067114095e-05,
+      "loss": 0.8965,
+      "step": 2391
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.1973154362416108e-05,
+      "loss": 0.8853,
+      "step": 2392
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.1969798657718122e-05,
+      "loss": 0.915,
+      "step": 2393
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.1966442953020135e-05,
+      "loss": 0.8877,
+      "step": 2394
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.1963087248322149e-05,
+      "loss": 0.8608,
+      "step": 2395
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.1959731543624163e-05,
+      "loss": 0.9189,
+      "step": 2396
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.1956375838926176e-05,
+      "loss": 0.9336,
+      "step": 2397
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.1953020134228188e-05,
+      "loss": 0.9307,
+      "step": 2398
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.1949664429530203e-05,
+      "loss": 0.9087,
+      "step": 2399
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.1946308724832217e-05,
+      "loss": 0.9282,
+      "step": 2400
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 1.194295302013423e-05,
+      "loss": 0.8706,
+      "step": 2401
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.1939597315436242e-05,
+      "loss": 0.9189,
+      "step": 2402
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.1936241610738255e-05,
+      "loss": 0.8774,
+      "step": 2403
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.193288590604027e-05,
+      "loss": 0.8999,
+      "step": 2404
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.1929530201342283e-05,
+      "loss": 0.8823,
+      "step": 2405
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.1926174496644296e-05,
+      "loss": 0.939,
+      "step": 2406
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.1922818791946308e-05,
+      "loss": 0.8882,
+      "step": 2407
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.1919463087248324e-05,
+      "loss": 0.9292,
+      "step": 2408
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.1916107382550337e-05,
+      "loss": 0.8857,
+      "step": 2409
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.191275167785235e-05,
+      "loss": 0.8984,
+      "step": 2410
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.1909395973154364e-05,
+      "loss": 0.9106,
+      "step": 2411
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.1906040268456376e-05,
+      "loss": 0.9194,
+      "step": 2412
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 1.190268456375839e-05,
+      "loss": 0.8975,
+      "step": 2413
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.1899328859060403e-05,
+      "loss": 0.8945,
+      "step": 2414
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.1895973154362417e-05,
+      "loss": 0.8921,
+      "step": 2415
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.189261744966443e-05,
+      "loss": 0.9121,
+      "step": 2416
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.1889261744966444e-05,
+      "loss": 0.958,
+      "step": 2417
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.1885906040268458e-05,
+      "loss": 0.9014,
+      "step": 2418
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.1882550335570471e-05,
+      "loss": 0.8945,
+      "step": 2419
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.1879194630872484e-05,
+      "loss": 0.8516,
+      "step": 2420
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.1875838926174496e-05,
+      "loss": 0.9087,
+      "step": 2421
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.1872483221476512e-05,
+      "loss": 0.8901,
+      "step": 2422
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.1869127516778525e-05,
+      "loss": 0.9067,
+      "step": 2423
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.1865771812080537e-05,
+      "loss": 0.8979,
+      "step": 2424
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 1.186241610738255e-05,
+      "loss": 0.9263,
+      "step": 2425
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.1859060402684566e-05,
+      "loss": 0.9092,
+      "step": 2426
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.1855704697986578e-05,
+      "loss": 0.9077,
+      "step": 2427
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.1852348993288591e-05,
+      "loss": 0.9072,
+      "step": 2428
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.1848993288590604e-05,
+      "loss": 0.876,
+      "step": 2429
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.1845637583892618e-05,
+      "loss": 0.9268,
+      "step": 2430
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.1842281879194632e-05,
+      "loss": 0.9019,
+      "step": 2431
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.1838926174496645e-05,
+      "loss": 0.9082,
+      "step": 2432
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.1835570469798659e-05,
+      "loss": 0.8862,
+      "step": 2433
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.1832214765100672e-05,
+      "loss": 0.9023,
+      "step": 2434
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.1828859060402686e-05,
+      "loss": 0.9038,
+      "step": 2435
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.18255033557047e-05,
+      "loss": 0.9287,
+      "step": 2436
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 1.1822147651006713e-05,
+      "loss": 0.9209,
+      "step": 2437
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.1818791946308725e-05,
+      "loss": 0.895,
+      "step": 2438
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.1815436241610738e-05,
+      "loss": 0.9243,
+      "step": 2439
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.1812080536912754e-05,
+      "loss": 0.9185,
+      "step": 2440
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.1808724832214766e-05,
+      "loss": 0.9038,
+      "step": 2441
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.1805369127516779e-05,
+      "loss": 0.8481,
+      "step": 2442
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.1802013422818792e-05,
+      "loss": 0.9053,
+      "step": 2443
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.1798657718120808e-05,
+      "loss": 0.9209,
+      "step": 2444
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.179530201342282e-05,
+      "loss": 0.8652,
+      "step": 2445
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.1791946308724833e-05,
+      "loss": 0.8955,
+      "step": 2446
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.1788590604026845e-05,
+      "loss": 0.9165,
+      "step": 2447
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.178523489932886e-05,
+      "loss": 0.9077,
+      "step": 2448
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 1.1781879194630874e-05,
+      "loss": 0.8877,
+      "step": 2449
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.1778523489932886e-05,
+      "loss": 0.9258,
+      "step": 2450
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.17751677852349e-05,
+      "loss": 0.8828,
+      "step": 2451
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.1771812080536913e-05,
+      "loss": 0.8877,
+      "step": 2452
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.1768456375838928e-05,
+      "loss": 0.8882,
+      "step": 2453
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.1765100671140942e-05,
+      "loss": 0.9473,
+      "step": 2454
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.1761744966442954e-05,
+      "loss": 0.917,
+      "step": 2455
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.1758389261744967e-05,
+      "loss": 0.9253,
+      "step": 2456
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.175503355704698e-05,
+      "loss": 0.8892,
+      "step": 2457
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.1751677852348996e-05,
+      "loss": 0.915,
+      "step": 2458
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.1748322147651008e-05,
+      "loss": 0.8916,
+      "step": 2459
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.174496644295302e-05,
+      "loss": 0.8848,
+      "step": 2460
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 1.1741610738255033e-05,
+      "loss": 0.894,
+      "step": 2461
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.173825503355705e-05,
+      "loss": 0.8579,
+      "step": 2462
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.1734899328859062e-05,
+      "loss": 0.9141,
+      "step": 2463
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.1731543624161074e-05,
+      "loss": 0.9277,
+      "step": 2464
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.1728187919463087e-05,
+      "loss": 0.8828,
+      "step": 2465
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.1724832214765101e-05,
+      "loss": 0.9062,
+      "step": 2466
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.1721476510067116e-05,
+      "loss": 0.9131,
+      "step": 2467
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.1718120805369128e-05,
+      "loss": 0.8877,
+      "step": 2468
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.1714765100671142e-05,
+      "loss": 0.8994,
+      "step": 2469
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.1711409395973155e-05,
+      "loss": 0.8999,
+      "step": 2470
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.170805369127517e-05,
+      "loss": 0.8779,
+      "step": 2471
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.1704697986577182e-05,
+      "loss": 0.9048,
+      "step": 2472
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 1.1701342281879196e-05,
+      "loss": 0.8931,
+      "step": 2473
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.1697986577181209e-05,
+      "loss": 0.9468,
+      "step": 2474
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.1694630872483221e-05,
+      "loss": 0.8945,
+      "step": 2475
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.1691275167785237e-05,
+      "loss": 0.9028,
+      "step": 2476
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.168791946308725e-05,
+      "loss": 0.9082,
+      "step": 2477
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.1684563758389262e-05,
+      "loss": 0.9302,
+      "step": 2478
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.1681208053691275e-05,
+      "loss": 0.8999,
+      "step": 2479
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.1677852348993291e-05,
+      "loss": 0.9258,
+      "step": 2480
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.1674496644295304e-05,
+      "loss": 0.8618,
+      "step": 2481
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.1671140939597316e-05,
+      "loss": 0.8599,
+      "step": 2482
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.1667785234899329e-05,
+      "loss": 0.9126,
+      "step": 2483
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.1664429530201343e-05,
+      "loss": 0.9482,
+      "step": 2484
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 1.1661073825503357e-05,
+      "loss": 0.9019,
+      "step": 2485
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.165771812080537e-05,
+      "loss": 0.8833,
+      "step": 2486
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.1654362416107382e-05,
+      "loss": 0.9438,
+      "step": 2487
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.1651006711409397e-05,
+      "loss": 0.9238,
+      "step": 2488
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.1647651006711411e-05,
+      "loss": 0.9043,
+      "step": 2489
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.1644295302013424e-05,
+      "loss": 0.8774,
+      "step": 2490
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.1640939597315438e-05,
+      "loss": 0.9185,
+      "step": 2491
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.163758389261745e-05,
+      "loss": 0.9297,
+      "step": 2492
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.1634228187919463e-05,
+      "loss": 0.8931,
+      "step": 2493
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.1630872483221479e-05,
+      "loss": 0.9038,
+      "step": 2494
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.1627516778523491e-05,
+      "loss": 0.9199,
+      "step": 2495
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.1624161073825504e-05,
+      "loss": 0.9629,
+      "step": 2496
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 1.1620805369127517e-05,
+      "loss": 0.916,
+      "step": 2497
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.1617449664429533e-05,
+      "loss": 0.96,
+      "step": 2498
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.1614093959731545e-05,
+      "loss": 0.9106,
+      "step": 2499
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.1610738255033558e-05,
+      "loss": 0.8306,
+      "step": 2500
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.160738255033557e-05,
+      "loss": 0.9116,
+      "step": 2501
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.1604026845637583e-05,
+      "loss": 0.917,
+      "step": 2502
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.1600671140939599e-05,
+      "loss": 0.8994,
+      "step": 2503
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.1597315436241611e-05,
+      "loss": 0.8965,
+      "step": 2504
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.1593959731543624e-05,
+      "loss": 0.8999,
+      "step": 2505
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.1590604026845638e-05,
+      "loss": 0.8892,
+      "step": 2506
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.1587248322147653e-05,
+      "loss": 0.8921,
+      "step": 2507
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.1583892617449665e-05,
+      "loss": 0.9653,
+      "step": 2508
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 1.158053691275168e-05,
+      "loss": 0.9355,
+      "step": 2509
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.1577181208053692e-05,
+      "loss": 0.9165,
+      "step": 2510
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.1573825503355705e-05,
+      "loss": 0.9302,
+      "step": 2511
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.1570469798657719e-05,
+      "loss": 0.8867,
+      "step": 2512
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.1567114093959733e-05,
+      "loss": 0.9121,
+      "step": 2513
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.1563758389261746e-05,
+      "loss": 0.937,
+      "step": 2514
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.1560402684563758e-05,
+      "loss": 0.9307,
+      "step": 2515
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.1557046979865774e-05,
+      "loss": 0.9194,
+      "step": 2516
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.1553691275167787e-05,
+      "loss": 0.9116,
+      "step": 2517
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.15503355704698e-05,
+      "loss": 0.9053,
+      "step": 2518
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.1546979865771812e-05,
+      "loss": 0.9336,
+      "step": 2519
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.1543624161073828e-05,
+      "loss": 0.8735,
+      "step": 2520
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 1.154026845637584e-05,
+      "loss": 0.8862,
+      "step": 2521
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.1536912751677853e-05,
+      "loss": 0.9233,
+      "step": 2522
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.1533557046979866e-05,
+      "loss": 0.8872,
+      "step": 2523
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.153020134228188e-05,
+      "loss": 0.9258,
+      "step": 2524
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.1526845637583894e-05,
+      "loss": 0.856,
+      "step": 2525
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.1523489932885907e-05,
+      "loss": 0.9126,
+      "step": 2526
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.152013422818792e-05,
+      "loss": 0.9146,
+      "step": 2527
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.1516778523489934e-05,
+      "loss": 0.9009,
+      "step": 2528
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.1513422818791948e-05,
+      "loss": 0.8716,
+      "step": 2529
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.151006711409396e-05,
+      "loss": 0.8857,
+      "step": 2530
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.1506711409395975e-05,
+      "loss": 0.9092,
+      "step": 2531
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.1503355704697987e-05,
+      "loss": 0.9155,
+      "step": 2532
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 1.15e-05,
+      "loss": 0.9014,
+      "step": 2533
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.1496644295302016e-05,
+      "loss": 0.9263,
+      "step": 2534
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.1493288590604029e-05,
+      "loss": 0.8882,
+      "step": 2535
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.1489932885906041e-05,
+      "loss": 0.9009,
+      "step": 2536
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.1486577181208054e-05,
+      "loss": 0.9189,
+      "step": 2537
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.148322147651007e-05,
+      "loss": 0.9189,
+      "step": 2538
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.1479865771812082e-05,
+      "loss": 0.9004,
+      "step": 2539
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.1476510067114095e-05,
+      "loss": 0.9028,
+      "step": 2540
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.1473154362416107e-05,
+      "loss": 0.9116,
+      "step": 2541
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.146979865771812e-05,
+      "loss": 0.8999,
+      "step": 2542
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.1466442953020136e-05,
+      "loss": 0.9009,
+      "step": 2543
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 1.1463087248322149e-05,
+      "loss": 0.8848,
+      "step": 2544
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.1459731543624161e-05,
+      "loss": 0.876,
+      "step": 2545
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.1456375838926175e-05,
+      "loss": 0.8926,
+      "step": 2546
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.145302013422819e-05,
+      "loss": 0.9292,
+      "step": 2547
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.1449664429530202e-05,
+      "loss": 0.8979,
+      "step": 2548
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.1446308724832217e-05,
+      "loss": 0.9043,
+      "step": 2549
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.1442953020134229e-05,
+      "loss": 0.8945,
+      "step": 2550
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.1439597315436242e-05,
+      "loss": 0.937,
+      "step": 2551
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.1436241610738256e-05,
+      "loss": 0.895,
+      "step": 2552
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.143288590604027e-05,
+      "loss": 0.9111,
+      "step": 2553
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.1429530201342283e-05,
+      "loss": 0.8652,
+      "step": 2554
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.1426174496644295e-05,
+      "loss": 0.9082,
+      "step": 2555
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 1.1422818791946311e-05,
+      "loss": 0.896,
+      "step": 2556
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.1419463087248324e-05,
+      "loss": 0.9033,
+      "step": 2557
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.1416107382550337e-05,
+      "loss": 0.9233,
+      "step": 2558
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.1412751677852349e-05,
+      "loss": 0.9077,
+      "step": 2559
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.1409395973154362e-05,
+      "loss": 0.9336,
+      "step": 2560
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.1406040268456378e-05,
+      "loss": 0.8882,
+      "step": 2561
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.140268456375839e-05,
+      "loss": 0.897,
+      "step": 2562
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.1399328859060403e-05,
+      "loss": 0.9009,
+      "step": 2563
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.1395973154362417e-05,
+      "loss": 0.8828,
+      "step": 2564
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.1392617449664431e-05,
+      "loss": 0.9097,
+      "step": 2565
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.1389261744966444e-05,
+      "loss": 0.9126,
+      "step": 2566
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.1385906040268458e-05,
+      "loss": 0.9526,
+      "step": 2567
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 1.138255033557047e-05,
+      "loss": 0.9131,
+      "step": 2568
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.1379194630872483e-05,
+      "loss": 0.8945,
+      "step": 2569
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.1375838926174498e-05,
+      "loss": 0.876,
+      "step": 2570
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.1372483221476512e-05,
+      "loss": 0.8677,
+      "step": 2571
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.1369127516778524e-05,
+      "loss": 0.9458,
+      "step": 2572
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.1365771812080537e-05,
+      "loss": 0.915,
+      "step": 2573
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.1362416107382553e-05,
+      "loss": 0.917,
+      "step": 2574
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.1359060402684566e-05,
+      "loss": 0.8784,
+      "step": 2575
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.1355704697986578e-05,
+      "loss": 0.897,
+      "step": 2576
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.135234899328859e-05,
+      "loss": 0.8892,
+      "step": 2577
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.1348993288590603e-05,
+      "loss": 0.9233,
+      "step": 2578
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.134563758389262e-05,
+      "loss": 0.9414,
+      "step": 2579
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 1.1342281879194632e-05,
+      "loss": 0.8999,
+      "step": 2580
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.1338926174496644e-05,
+      "loss": 0.9023,
+      "step": 2581
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.1335570469798659e-05,
+      "loss": 0.9478,
+      "step": 2582
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.1332214765100673e-05,
+      "loss": 0.9243,
+      "step": 2583
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.1328859060402686e-05,
+      "loss": 0.9214,
+      "step": 2584
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.1325503355704698e-05,
+      "loss": 0.9111,
+      "step": 2585
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.1322147651006712e-05,
+      "loss": 0.9023,
+      "step": 2586
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.1318791946308725e-05,
+      "loss": 0.8926,
+      "step": 2587
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.131543624161074e-05,
+      "loss": 0.9536,
+      "step": 2588
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.1312080536912754e-05,
+      "loss": 0.9106,
+      "step": 2589
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.1308724832214766e-05,
+      "loss": 0.9146,
+      "step": 2590
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.1305369127516779e-05,
+      "loss": 0.9263,
+      "step": 2591
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 1.1302013422818795e-05,
+      "loss": 0.897,
+      "step": 2592
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.1298657718120807e-05,
+      "loss": 0.9282,
+      "step": 2593
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.129530201342282e-05,
+      "loss": 0.9268,
+      "step": 2594
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.1291946308724832e-05,
+      "loss": 0.9121,
+      "step": 2595
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.1288590604026845e-05,
+      "loss": 0.9077,
+      "step": 2596
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.1285234899328861e-05,
+      "loss": 0.918,
+      "step": 2597
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.1281879194630874e-05,
+      "loss": 0.8848,
+      "step": 2598
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.1278523489932886e-05,
+      "loss": 0.9321,
+      "step": 2599
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.1275167785234899e-05,
+      "loss": 0.8921,
+      "step": 2600
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.1271812080536915e-05,
+      "loss": 0.9067,
+      "step": 2601
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.1268456375838927e-05,
+      "loss": 0.895,
+      "step": 2602
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.126510067114094e-05,
+      "loss": 0.8652,
+      "step": 2603
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 1.1261744966442954e-05,
+      "loss": 0.9111,
+      "step": 2604
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.1258389261744967e-05,
+      "loss": 0.8965,
+      "step": 2605
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.1255033557046981e-05,
+      "loss": 0.8867,
+      "step": 2606
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.1251677852348995e-05,
+      "loss": 0.897,
+      "step": 2607
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.1248322147651008e-05,
+      "loss": 0.8706,
+      "step": 2608
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.124496644295302e-05,
+      "loss": 0.9268,
+      "step": 2609
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.1241610738255035e-05,
+      "loss": 0.9111,
+      "step": 2610
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.1238255033557049e-05,
+      "loss": 0.8955,
+      "step": 2611
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.1234899328859062e-05,
+      "loss": 0.9014,
+      "step": 2612
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.1231543624161074e-05,
+      "loss": 0.9326,
+      "step": 2613
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.1228187919463087e-05,
+      "loss": 0.896,
+      "step": 2614
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.1224832214765103e-05,
+      "loss": 0.8979,
+      "step": 2615
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 1.1221476510067115e-05,
+      "loss": 0.8804,
+      "step": 2616
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.1218120805369128e-05,
+      "loss": 0.8975,
+      "step": 2617
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.121476510067114e-05,
+      "loss": 0.9194,
+      "step": 2618
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.1211409395973156e-05,
+      "loss": 0.9058,
+      "step": 2619
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.1208053691275169e-05,
+      "loss": 0.9219,
+      "step": 2620
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.1204697986577182e-05,
+      "loss": 0.917,
+      "step": 2621
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.1201342281879196e-05,
+      "loss": 0.918,
+      "step": 2622
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.1197986577181208e-05,
+      "loss": 0.9395,
+      "step": 2623
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.1194630872483223e-05,
+      "loss": 0.9419,
+      "step": 2624
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.1191275167785235e-05,
+      "loss": 0.896,
+      "step": 2625
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.118791946308725e-05,
+      "loss": 0.8945,
+      "step": 2626
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.1184563758389262e-05,
+      "loss": 0.9038,
+      "step": 2627
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 1.1181208053691276e-05,
+      "loss": 0.8965,
+      "step": 2628
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.117785234899329e-05,
+      "loss": 0.9077,
+      "step": 2629
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.1174496644295303e-05,
+      "loss": 0.8911,
+      "step": 2630
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.1171140939597316e-05,
+      "loss": 0.9326,
+      "step": 2631
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.1167785234899328e-05,
+      "loss": 0.9072,
+      "step": 2632
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.1164429530201344e-05,
+      "loss": 0.8853,
+      "step": 2633
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.1161073825503357e-05,
+      "loss": 0.8647,
+      "step": 2634
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.115771812080537e-05,
+      "loss": 0.9175,
+      "step": 2635
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.1154362416107382e-05,
+      "loss": 0.8882,
+      "step": 2636
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.1151006711409398e-05,
+      "loss": 0.8813,
+      "step": 2637
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.114765100671141e-05,
+      "loss": 0.894,
+      "step": 2638
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.1144295302013423e-05,
+      "loss": 0.8706,
+      "step": 2639
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 1.1140939597315436e-05,
+      "loss": 0.9038,
+      "step": 2640
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.113758389261745e-05,
+      "loss": 0.9067,
+      "step": 2641
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.1134228187919464e-05,
+      "loss": 0.8989,
+      "step": 2642
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.1130872483221477e-05,
+      "loss": 0.9053,
+      "step": 2643
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.1127516778523491e-05,
+      "loss": 0.9053,
+      "step": 2644
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.1124161073825504e-05,
+      "loss": 0.9258,
+      "step": 2645
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.1120805369127518e-05,
+      "loss": 0.8945,
+      "step": 2646
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.1117449664429532e-05,
+      "loss": 0.9331,
+      "step": 2647
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.1114093959731545e-05,
+      "loss": 0.9512,
+      "step": 2648
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.1110738255033557e-05,
+      "loss": 0.8867,
+      "step": 2649
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.110738255033557e-05,
+      "loss": 0.8843,
+      "step": 2650
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.1104026845637586e-05,
+      "loss": 0.8672,
+      "step": 2651
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 1.1100671140939599e-05,
+      "loss": 0.9199,
+      "step": 2652
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.1097315436241611e-05,
+      "loss": 0.9336,
+      "step": 2653
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.1093959731543624e-05,
+      "loss": 0.8955,
+      "step": 2654
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.109060402684564e-05,
+      "loss": 0.8911,
+      "step": 2655
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.1087248322147652e-05,
+      "loss": 0.9126,
+      "step": 2656
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.1083892617449665e-05,
+      "loss": 0.9087,
+      "step": 2657
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.1080536912751677e-05,
+      "loss": 0.8989,
+      "step": 2658
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.1077181208053692e-05,
+      "loss": 0.9316,
+      "step": 2659
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.1073825503355706e-05,
+      "loss": 0.9248,
+      "step": 2660
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.1070469798657719e-05,
+      "loss": 0.8989,
+      "step": 2661
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.1067114093959733e-05,
+      "loss": 0.9365,
+      "step": 2662
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.1063758389261745e-05,
+      "loss": 0.9048,
+      "step": 2663
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 1.106040268456376e-05,
+      "loss": 0.8931,
+      "step": 2664
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.1057046979865772e-05,
+      "loss": 0.8608,
+      "step": 2665
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.1053691275167787e-05,
+      "loss": 0.8901,
+      "step": 2666
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.10503355704698e-05,
+      "loss": 0.8745,
+      "step": 2667
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.1046979865771812e-05,
+      "loss": 0.9609,
+      "step": 2668
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.1043624161073828e-05,
+      "loss": 0.874,
+      "step": 2669
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.104026845637584e-05,
+      "loss": 0.8857,
+      "step": 2670
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.1036912751677853e-05,
+      "loss": 0.8774,
+      "step": 2671
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.1033557046979865e-05,
+      "loss": 0.8872,
+      "step": 2672
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.1030201342281881e-05,
+      "loss": 0.9009,
+      "step": 2673
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.1026845637583894e-05,
+      "loss": 0.8569,
+      "step": 2674
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.1023489932885907e-05,
+      "loss": 0.9126,
+      "step": 2675
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 1.102013422818792e-05,
+      "loss": 0.8848,
+      "step": 2676
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.1016778523489933e-05,
+      "loss": 0.8975,
+      "step": 2677
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.1013422818791948e-05,
+      "loss": 0.9312,
+      "step": 2678
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.101006711409396e-05,
+      "loss": 0.9355,
+      "step": 2679
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.1006711409395975e-05,
+      "loss": 0.9473,
+      "step": 2680
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.1003355704697987e-05,
+      "loss": 0.895,
+      "step": 2681
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.1000000000000001e-05,
+      "loss": 0.8696,
+      "step": 2682
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.0996644295302014e-05,
+      "loss": 0.8872,
+      "step": 2683
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.0993288590604028e-05,
+      "loss": 0.8843,
+      "step": 2684
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.098993288590604e-05,
+      "loss": 0.8901,
+      "step": 2685
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.0986577181208053e-05,
+      "loss": 0.8452,
+      "step": 2686
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 1.098322147651007e-05,
+      "loss": 0.9297,
+      "step": 2687
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.0979865771812082e-05,
+      "loss": 0.9248,
+      "step": 2688
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.0976510067114095e-05,
+      "loss": 0.938,
+      "step": 2689
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.0973154362416107e-05,
+      "loss": 0.9009,
+      "step": 2690
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.0969798657718123e-05,
+      "loss": 0.9043,
+      "step": 2691
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.0966442953020136e-05,
+      "loss": 0.8828,
+      "step": 2692
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.0963087248322148e-05,
+      "loss": 0.9136,
+      "step": 2693
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.095973154362416e-05,
+      "loss": 0.9282,
+      "step": 2694
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.0956375838926175e-05,
+      "loss": 0.8975,
+      "step": 2695
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.095302013422819e-05,
+      "loss": 0.9048,
+      "step": 2696
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.0949664429530202e-05,
+      "loss": 0.9087,
+      "step": 2697
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.0946308724832215e-05,
+      "loss": 0.8716,
+      "step": 2698
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 1.0942953020134229e-05,
+      "loss": 0.9243,
+      "step": 2699
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.0939597315436243e-05,
+      "loss": 0.9111,
+      "step": 2700
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.0936241610738256e-05,
+      "loss": 0.936,
+      "step": 2701
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.093288590604027e-05,
+      "loss": 0.873,
+      "step": 2702
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.0929530201342283e-05,
+      "loss": 0.875,
+      "step": 2703
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.0926174496644297e-05,
+      "loss": 0.9438,
+      "step": 2704
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.0922818791946311e-05,
+      "loss": 0.8838,
+      "step": 2705
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.0919463087248324e-05,
+      "loss": 0.8901,
+      "step": 2706
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.0916107382550336e-05,
+      "loss": 0.9277,
+      "step": 2707
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.0912751677852349e-05,
+      "loss": 0.9448,
+      "step": 2708
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.0909395973154365e-05,
+      "loss": 0.8906,
+      "step": 2709
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.0906040268456377e-05,
+      "loss": 0.9097,
+      "step": 2710
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 1.090268456375839e-05,
+      "loss": 0.8916,
+      "step": 2711
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.0899328859060403e-05,
+      "loss": 0.8911,
+      "step": 2712
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.0895973154362418e-05,
+      "loss": 0.9214,
+      "step": 2713
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.0892617449664431e-05,
+      "loss": 0.9082,
+      "step": 2714
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.0889261744966444e-05,
+      "loss": 0.8809,
+      "step": 2715
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.0885906040268456e-05,
+      "loss": 0.9126,
+      "step": 2716
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.088255033557047e-05,
+      "loss": 0.8999,
+      "step": 2717
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.0879194630872485e-05,
+      "loss": 0.8901,
+      "step": 2718
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.0875838926174497e-05,
+      "loss": 0.9312,
+      "step": 2719
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.0872483221476512e-05,
+      "loss": 0.8809,
+      "step": 2720
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.0869127516778524e-05,
+      "loss": 0.8989,
+      "step": 2721
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.0865771812080538e-05,
+      "loss": 0.9102,
+      "step": 2722
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 1.0862416107382551e-05,
+      "loss": 0.9292,
+      "step": 2723
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.0859060402684565e-05,
+      "loss": 0.897,
+      "step": 2724
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.0855704697986578e-05,
+      "loss": 0.8936,
+      "step": 2725
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.085234899328859e-05,
+      "loss": 0.9399,
+      "step": 2726
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.0848993288590606e-05,
+      "loss": 0.9126,
+      "step": 2727
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.0845637583892619e-05,
+      "loss": 0.9619,
+      "step": 2728
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.0842281879194632e-05,
+      "loss": 0.8892,
+      "step": 2729
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.0838926174496644e-05,
+      "loss": 0.8877,
+      "step": 2730
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.083557046979866e-05,
+      "loss": 0.8647,
+      "step": 2731
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.0832214765100673e-05,
+      "loss": 0.8828,
+      "step": 2732
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.0828859060402685e-05,
+      "loss": 0.8848,
+      "step": 2733
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.0825503355704698e-05,
+      "loss": 0.8755,
+      "step": 2734
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 1.0822147651006712e-05,
+      "loss": 0.938,
+      "step": 2735
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.0818791946308726e-05,
+      "loss": 0.9043,
+      "step": 2736
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.0815436241610739e-05,
+      "loss": 0.8989,
+      "step": 2737
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.0812080536912752e-05,
+      "loss": 0.8901,
+      "step": 2738
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.0808724832214766e-05,
+      "loss": 0.9072,
+      "step": 2739
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.080536912751678e-05,
+      "loss": 0.8975,
+      "step": 2740
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.0802013422818793e-05,
+      "loss": 0.9048,
+      "step": 2741
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.0798657718120807e-05,
+      "loss": 0.9316,
+      "step": 2742
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.079530201342282e-05,
+      "loss": 0.9092,
+      "step": 2743
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.0791946308724832e-05,
+      "loss": 0.9092,
+      "step": 2744
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.0788590604026848e-05,
+      "loss": 0.9121,
+      "step": 2745
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.078523489932886e-05,
+      "loss": 0.9116,
+      "step": 2746
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 1.0781879194630873e-05,
+      "loss": 0.9263,
+      "step": 2747
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.0778523489932886e-05,
+      "loss": 0.936,
+      "step": 2748
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.0775167785234902e-05,
+      "loss": 0.894,
+      "step": 2749
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.0771812080536914e-05,
+      "loss": 0.9214,
+      "step": 2750
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.0768456375838927e-05,
+      "loss": 0.9038,
+      "step": 2751
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.076510067114094e-05,
+      "loss": 0.8677,
+      "step": 2752
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.0761744966442952e-05,
+      "loss": 0.8872,
+      "step": 2753
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.0758389261744968e-05,
+      "loss": 0.8999,
+      "step": 2754
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.075503355704698e-05,
+      "loss": 0.9077,
+      "step": 2755
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.0751677852348993e-05,
+      "loss": 0.9033,
+      "step": 2756
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.0748322147651008e-05,
+      "loss": 0.8989,
+      "step": 2757
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.0744966442953022e-05,
+      "loss": 0.8701,
+      "step": 2758
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 1.0741610738255034e-05,
+      "loss": 0.8633,
+      "step": 2759
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.0738255033557049e-05,
+      "loss": 0.915,
+      "step": 2760
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.0734899328859061e-05,
+      "loss": 0.917,
+      "step": 2761
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.0731543624161074e-05,
+      "loss": 0.9019,
+      "step": 2762
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.0728187919463088e-05,
+      "loss": 0.9287,
+      "step": 2763
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.0724832214765102e-05,
+      "loss": 0.9312,
+      "step": 2764
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.0721476510067115e-05,
+      "loss": 0.8579,
+      "step": 2765
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.0718120805369128e-05,
+      "loss": 0.9199,
+      "step": 2766
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.0714765100671144e-05,
+      "loss": 0.8892,
+      "step": 2767
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.0711409395973156e-05,
+      "loss": 0.8779,
+      "step": 2768
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.0708053691275169e-05,
+      "loss": 0.8647,
+      "step": 2769
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.0704697986577181e-05,
+      "loss": 0.9019,
+      "step": 2770
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 1.0701342281879194e-05,
+      "loss": 0.9092,
+      "step": 2771
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.069798657718121e-05,
+      "loss": 0.9092,
+      "step": 2772
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.0694630872483222e-05,
+      "loss": 0.9404,
+      "step": 2773
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.0691275167785235e-05,
+      "loss": 0.8989,
+      "step": 2774
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.068791946308725e-05,
+      "loss": 0.8906,
+      "step": 2775
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.0684563758389264e-05,
+      "loss": 0.9136,
+      "step": 2776
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.0681208053691276e-05,
+      "loss": 0.9224,
+      "step": 2777
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.0677852348993289e-05,
+      "loss": 0.8994,
+      "step": 2778
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.0674496644295303e-05,
+      "loss": 0.8813,
+      "step": 2779
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.0671140939597316e-05,
+      "loss": 0.9146,
+      "step": 2780
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.066778523489933e-05,
+      "loss": 0.8921,
+      "step": 2781
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.0664429530201344e-05,
+      "loss": 0.876,
+      "step": 2782
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 1.0661073825503357e-05,
+      "loss": 0.9131,
+      "step": 2783
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.065771812080537e-05,
+      "loss": 0.9189,
+      "step": 2784
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.0654362416107385e-05,
+      "loss": 0.9248,
+      "step": 2785
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.0651006711409398e-05,
+      "loss": 0.9019,
+      "step": 2786
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.064765100671141e-05,
+      "loss": 0.8999,
+      "step": 2787
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.0644295302013423e-05,
+      "loss": 0.8984,
+      "step": 2788
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.0640939597315436e-05,
+      "loss": 0.8984,
+      "step": 2789
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.0637583892617451e-05,
+      "loss": 0.8911,
+      "step": 2790
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.0634228187919464e-05,
+      "loss": 0.9092,
+      "step": 2791
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.0630872483221477e-05,
+      "loss": 0.9028,
+      "step": 2792
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.0627516778523491e-05,
+      "loss": 0.877,
+      "step": 2793
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.0624161073825505e-05,
+      "loss": 0.9155,
+      "step": 2794
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 1.0620805369127518e-05,
+      "loss": 0.8882,
+      "step": 2795
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.061744966442953e-05,
+      "loss": 0.9077,
+      "step": 2796
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0614093959731545e-05,
+      "loss": 0.9214,
+      "step": 2797
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0610738255033557e-05,
+      "loss": 0.9058,
+      "step": 2798
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0607382550335571e-05,
+      "loss": 0.8955,
+      "step": 2799
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0604026845637586e-05,
+      "loss": 0.8906,
+      "step": 2800
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0600671140939598e-05,
+      "loss": 0.9043,
+      "step": 2801
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0597315436241611e-05,
+      "loss": 0.9009,
+      "step": 2802
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0593959731543625e-05,
+      "loss": 0.8857,
+      "step": 2803
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.059060402684564e-05,
+      "loss": 0.9258,
+      "step": 2804
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0587248322147652e-05,
+      "loss": 0.8682,
+      "step": 2805
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0583892617449665e-05,
+      "loss": 0.8965,
+      "step": 2806
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 1.0580536912751677e-05,
+      "loss": 0.8975,
+      "step": 2807
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0577181208053693e-05,
+      "loss": 0.9048,
+      "step": 2808
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0573825503355706e-05,
+      "loss": 0.8657,
+      "step": 2809
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0570469798657718e-05,
+      "loss": 0.8784,
+      "step": 2810
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0567114093959731e-05,
+      "loss": 0.8887,
+      "step": 2811
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0563758389261747e-05,
+      "loss": 0.8999,
+      "step": 2812
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.056040268456376e-05,
+      "loss": 0.8779,
+      "step": 2813
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0557046979865772e-05,
+      "loss": 0.9082,
+      "step": 2814
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0553691275167786e-05,
+      "loss": 0.8828,
+      "step": 2815
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0550335570469799e-05,
+      "loss": 0.8901,
+      "step": 2816
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0546979865771813e-05,
+      "loss": 0.9375,
+      "step": 2817
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.0543624161073827e-05,
+      "loss": 0.8901,
+      "step": 2818
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 1.054026845637584e-05,
+      "loss": 0.8975,
+      "step": 2819
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0536912751677853e-05,
+      "loss": 0.8906,
+      "step": 2820
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0533557046979867e-05,
+      "loss": 0.8872,
+      "step": 2821
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0530201342281881e-05,
+      "loss": 0.8994,
+      "step": 2822
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0526845637583894e-05,
+      "loss": 0.9102,
+      "step": 2823
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0523489932885906e-05,
+      "loss": 0.9116,
+      "step": 2824
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0520134228187919e-05,
+      "loss": 0.8843,
+      "step": 2825
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0516778523489935e-05,
+      "loss": 0.8906,
+      "step": 2826
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0513422818791947e-05,
+      "loss": 0.9229,
+      "step": 2827
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.051006711409396e-05,
+      "loss": 0.8916,
+      "step": 2828
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0506711409395973e-05,
+      "loss": 0.9326,
+      "step": 2829
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 1.0503355704697989e-05,
+      "loss": 0.8896,
+      "step": 2830
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0500000000000001e-05,
+      "loss": 0.9258,
+      "step": 2831
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0496644295302014e-05,
+      "loss": 0.8896,
+      "step": 2832
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0493288590604028e-05,
+      "loss": 0.8945,
+      "step": 2833
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.048993288590604e-05,
+      "loss": 0.9331,
+      "step": 2834
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0486577181208055e-05,
+      "loss": 0.9238,
+      "step": 2835
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0483221476510067e-05,
+      "loss": 0.8994,
+      "step": 2836
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0479865771812082e-05,
+      "loss": 0.8804,
+      "step": 2837
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0476510067114094e-05,
+      "loss": 0.9106,
+      "step": 2838
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0473154362416109e-05,
+      "loss": 0.9565,
+      "step": 2839
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0469798657718123e-05,
+      "loss": 0.9014,
+      "step": 2840
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0466442953020135e-05,
+      "loss": 0.9141,
+      "step": 2841
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 1.0463087248322148e-05,
+      "loss": 0.9126,
+      "step": 2842
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.045973154362416e-05,
+      "loss": 0.8975,
+      "step": 2843
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0456375838926177e-05,
+      "loss": 0.896,
+      "step": 2844
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0453020134228189e-05,
+      "loss": 0.9399,
+      "step": 2845
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0449664429530202e-05,
+      "loss": 0.936,
+      "step": 2846
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0446308724832214e-05,
+      "loss": 0.8867,
+      "step": 2847
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.044295302013423e-05,
+      "loss": 0.9004,
+      "step": 2848
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0439597315436243e-05,
+      "loss": 0.8901,
+      "step": 2849
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0436241610738255e-05,
+      "loss": 0.8765,
+      "step": 2850
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0432885906040268e-05,
+      "loss": 0.9106,
+      "step": 2851
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0429530201342282e-05,
+      "loss": 0.9082,
+      "step": 2852
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0426174496644296e-05,
+      "loss": 0.9136,
+      "step": 2853
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 1.0422818791946309e-05,
+      "loss": 0.8862,
+      "step": 2854
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0419463087248323e-05,
+      "loss": 0.8887,
+      "step": 2855
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0416107382550336e-05,
+      "loss": 0.9126,
+      "step": 2856
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.041275167785235e-05,
+      "loss": 0.9229,
+      "step": 2857
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0409395973154364e-05,
+      "loss": 0.9111,
+      "step": 2858
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0406040268456377e-05,
+      "loss": 0.9023,
+      "step": 2859
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.040268456375839e-05,
+      "loss": 0.9019,
+      "step": 2860
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0399328859060402e-05,
+      "loss": 0.8887,
+      "step": 2861
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0395973154362418e-05,
+      "loss": 0.9302,
+      "step": 2862
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.039261744966443e-05,
+      "loss": 0.9058,
+      "step": 2863
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0389261744966443e-05,
+      "loss": 0.8706,
+      "step": 2864
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0385906040268456e-05,
+      "loss": 0.8701,
+      "step": 2865
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.0382550335570472e-05,
+      "loss": 0.9126,
+      "step": 2866
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.0379194630872484e-05,
+      "loss": 0.8882,
+      "step": 2867
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.0375838926174497e-05,
+      "loss": 0.918,
+      "step": 2868
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.037248322147651e-05,
+      "loss": 0.9038,
+      "step": 2869
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.0369127516778524e-05,
+      "loss": 0.9346,
+      "step": 2870
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.0365771812080538e-05,
+      "loss": 0.8979,
+      "step": 2871
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.036241610738255e-05,
+      "loss": 0.8804,
+      "step": 2872
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.0359060402684565e-05,
+      "loss": 0.8911,
+      "step": 2873
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.0355704697986578e-05,
+      "loss": 0.9116,
+      "step": 2874
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.0352348993288592e-05,
+      "loss": 0.9136,
+      "step": 2875
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.0348993288590604e-05,
+      "loss": 0.9224,
+      "step": 2876
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.0345637583892619e-05,
+      "loss": 0.8809,
+      "step": 2877
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.0342281879194631e-05,
+      "loss": 0.9038,
+      "step": 2878
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0338926174496644e-05,
+      "loss": 0.9014,
+      "step": 2879
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.033557046979866e-05,
+      "loss": 0.9009,
+      "step": 2880
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0332214765100672e-05,
+      "loss": 0.8965,
+      "step": 2881
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0328859060402685e-05,
+      "loss": 0.9282,
+      "step": 2882
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0325503355704698e-05,
+      "loss": 0.8965,
+      "step": 2883
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0322147651006714e-05,
+      "loss": 0.9033,
+      "step": 2884
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0318791946308726e-05,
+      "loss": 0.8916,
+      "step": 2885
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0315436241610739e-05,
+      "loss": 0.9155,
+      "step": 2886
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0312080536912751e-05,
+      "loss": 0.8711,
+      "step": 2887
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0308724832214767e-05,
+      "loss": 0.9136,
+      "step": 2888
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.030536912751678e-05,
+      "loss": 0.9204,
+      "step": 2889
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.0302013422818792e-05,
+      "loss": 0.9048,
+      "step": 2890
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0298657718120805e-05,
+      "loss": 0.9038,
+      "step": 2891
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.029530201342282e-05,
+      "loss": 0.897,
+      "step": 2892
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0291946308724834e-05,
+      "loss": 0.8721,
+      "step": 2893
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0288590604026846e-05,
+      "loss": 0.9268,
+      "step": 2894
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.028523489932886e-05,
+      "loss": 0.9272,
+      "step": 2895
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0281879194630873e-05,
+      "loss": 0.9077,
+      "step": 2896
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0278523489932887e-05,
+      "loss": 0.9092,
+      "step": 2897
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0275167785234902e-05,
+      "loss": 0.916,
+      "step": 2898
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0271812080536914e-05,
+      "loss": 0.9268,
+      "step": 2899
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0268456375838927e-05,
+      "loss": 0.917,
+      "step": 2900
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.026510067114094e-05,
+      "loss": 0.9097,
+      "step": 2901
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.0261744966442955e-05,
+      "loss": 0.9185,
+      "step": 2902
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0258389261744968e-05,
+      "loss": 0.8911,
+      "step": 2903
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.025503355704698e-05,
+      "loss": 0.8989,
+      "step": 2904
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0251677852348993e-05,
+      "loss": 0.9072,
+      "step": 2905
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0248322147651009e-05,
+      "loss": 0.8955,
+      "step": 2906
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0244966442953022e-05,
+      "loss": 0.9004,
+      "step": 2907
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0241610738255034e-05,
+      "loss": 0.9443,
+      "step": 2908
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0238255033557047e-05,
+      "loss": 0.9053,
+      "step": 2909
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0234899328859061e-05,
+      "loss": 0.9355,
+      "step": 2910
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0231543624161075e-05,
+      "loss": 0.8926,
+      "step": 2911
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0228187919463088e-05,
+      "loss": 0.8999,
+      "step": 2912
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0224832214765102e-05,
+      "loss": 0.9097,
+      "step": 2913
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.0221476510067115e-05,
+      "loss": 0.8613,
+      "step": 2914
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0218120805369129e-05,
+      "loss": 0.9014,
+      "step": 2915
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0214765100671142e-05,
+      "loss": 0.9106,
+      "step": 2916
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0211409395973156e-05,
+      "loss": 0.936,
+      "step": 2917
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0208053691275168e-05,
+      "loss": 0.9043,
+      "step": 2918
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0204697986577181e-05,
+      "loss": 0.8726,
+      "step": 2919
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0201342281879197e-05,
+      "loss": 0.9268,
+      "step": 2920
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.019798657718121e-05,
+      "loss": 0.9204,
+      "step": 2921
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0194630872483222e-05,
+      "loss": 0.896,
+      "step": 2922
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0191275167785235e-05,
+      "loss": 0.896,
+      "step": 2923
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.018791946308725e-05,
+      "loss": 0.8892,
+      "step": 2924
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0184563758389263e-05,
+      "loss": 0.8867,
+      "step": 2925
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.0181208053691276e-05,
+      "loss": 0.8984,
+      "step": 2926
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0177852348993288e-05,
+      "loss": 0.9131,
+      "step": 2927
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0174496644295303e-05,
+      "loss": 0.894,
+      "step": 2928
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0171140939597317e-05,
+      "loss": 0.9199,
+      "step": 2929
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.016778523489933e-05,
+      "loss": 0.9185,
+      "step": 2930
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0164429530201344e-05,
+      "loss": 0.9082,
+      "step": 2931
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0161073825503356e-05,
+      "loss": 0.9106,
+      "step": 2932
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.015771812080537e-05,
+      "loss": 0.9102,
+      "step": 2933
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0154362416107383e-05,
+      "loss": 0.9204,
+      "step": 2934
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0151006711409397e-05,
+      "loss": 0.8613,
+      "step": 2935
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.014765100671141e-05,
+      "loss": 0.9287,
+      "step": 2936
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0144295302013423e-05,
+      "loss": 0.8813,
+      "step": 2937
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.0140939597315439e-05,
+      "loss": 0.9092,
+      "step": 2938
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0137583892617451e-05,
+      "loss": 0.916,
+      "step": 2939
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0134228187919464e-05,
+      "loss": 0.8896,
+      "step": 2940
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0130872483221476e-05,
+      "loss": 0.8936,
+      "step": 2941
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0127516778523492e-05,
+      "loss": 0.8711,
+      "step": 2942
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0124161073825505e-05,
+      "loss": 0.9023,
+      "step": 2943
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0120805369127517e-05,
+      "loss": 0.9243,
+      "step": 2944
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.011744966442953e-05,
+      "loss": 0.9307,
+      "step": 2945
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0114093959731544e-05,
+      "loss": 0.9121,
+      "step": 2946
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0110738255033559e-05,
+      "loss": 0.8921,
+      "step": 2947
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0107382550335571e-05,
+      "loss": 0.8931,
+      "step": 2948
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0104026845637584e-05,
+      "loss": 0.8931,
+      "step": 2949
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.0100671140939598e-05,
+      "loss": 0.8887,
+      "step": 2950
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0097315436241612e-05,
+      "loss": 0.9233,
+      "step": 2951
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0093959731543625e-05,
+      "loss": 0.9092,
+      "step": 2952
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0090604026845639e-05,
+      "loss": 0.9092,
+      "step": 2953
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0087248322147652e-05,
+      "loss": 0.8936,
+      "step": 2954
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0083892617449664e-05,
+      "loss": 0.9185,
+      "step": 2955
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.008053691275168e-05,
+      "loss": 0.9258,
+      "step": 2956
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0077181208053693e-05,
+      "loss": 0.918,
+      "step": 2957
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0073825503355705e-05,
+      "loss": 0.9033,
+      "step": 2958
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0070469798657718e-05,
+      "loss": 0.9053,
+      "step": 2959
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0067114093959734e-05,
+      "loss": 0.9028,
+      "step": 2960
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0063758389261747e-05,
+      "loss": 0.8511,
+      "step": 2961
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.0060402684563759e-05,
+      "loss": 0.8765,
+      "step": 2962
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0057046979865772e-05,
+      "loss": 0.8818,
+      "step": 2963
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0053691275167784e-05,
+      "loss": 0.918,
+      "step": 2964
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.00503355704698e-05,
+      "loss": 0.8901,
+      "step": 2965
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0046979865771813e-05,
+      "loss": 0.9004,
+      "step": 2966
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0043624161073825e-05,
+      "loss": 0.9282,
+      "step": 2967
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.004026845637584e-05,
+      "loss": 0.9102,
+      "step": 2968
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0036912751677854e-05,
+      "loss": 0.8989,
+      "step": 2969
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0033557046979867e-05,
+      "loss": 0.9043,
+      "step": 2970
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.003020134228188e-05,
+      "loss": 0.8613,
+      "step": 2971
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0026845637583893e-05,
+      "loss": 0.9106,
+      "step": 2972
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.0023489932885906e-05,
+      "loss": 0.9375,
+      "step": 2973
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.002013422818792e-05,
+      "loss": 0.9385,
+      "step": 2974
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0016778523489935e-05,
+      "loss": 0.9028,
+      "step": 2975
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0013422818791947e-05,
+      "loss": 0.9116,
+      "step": 2976
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.001006711409396e-05,
+      "loss": 0.8501,
+      "step": 2977
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0006711409395976e-05,
+      "loss": 0.9014,
+      "step": 2978
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.0003355704697988e-05,
+      "loss": 0.8979,
+      "step": 2979
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1e-05,
+      "loss": 0.8896,
+      "step": 2980
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 9.996644295302013e-06,
+      "loss": 0.9375,
+      "step": 2981
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 9.993288590604028e-06,
+      "loss": 0.9053,
+      "step": 2982
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 9.98993288590604e-06,
+      "loss": 0.8936,
+      "step": 2983
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 9.986577181208055e-06,
+      "loss": 0.9087,
+      "step": 2984
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 9.983221476510067e-06,
+      "loss": 0.9028,
+      "step": 2985
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 9.979865771812081e-06,
+      "loss": 0.9395,
+      "step": 2986
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 9.976510067114094e-06,
+      "loss": 0.9248,
+      "step": 2987
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 9.973154362416108e-06,
+      "loss": 0.9019,
+      "step": 2988
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 9.96979865771812e-06,
+      "loss": 0.896,
+      "step": 2989
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 9.966442953020135e-06,
+      "loss": 0.8652,
+      "step": 2990
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 9.96308724832215e-06,
+      "loss": 0.9067,
+      "step": 2991
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 9.959731543624162e-06,
+      "loss": 0.877,
+      "step": 2992
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 9.956375838926176e-06,
+      "loss": 0.9043,
+      "step": 2993
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 9.953020134228189e-06,
+      "loss": 0.8926,
+      "step": 2994
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 9.949664429530203e-06,
+      "loss": 0.9082,
+      "step": 2995
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 9.946308724832216e-06,
+      "loss": 0.9092,
+      "step": 2996
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 9.94295302013423e-06,
+      "loss": 0.8711,
+      "step": 2997
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 9.939597315436242e-06,
+      "loss": 0.9146,
+      "step": 2998
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 9.936241610738255e-06,
+      "loss": 0.9194,
+      "step": 2999
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 9.93288590604027e-06,
+      "loss": 0.9248,
+      "step": 3000
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 9.929530201342282e-06,
+      "loss": 0.8716,
+      "step": 3001
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 9.926174496644296e-06,
+      "loss": 0.8789,
+      "step": 3002
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 9.922818791946309e-06,
+      "loss": 0.9102,
+      "step": 3003
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 9.919463087248323e-06,
+      "loss": 0.9243,
+      "step": 3004
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 9.916107382550336e-06,
+      "loss": 0.9219,
+      "step": 3005
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 9.91275167785235e-06,
+      "loss": 0.9229,
+      "step": 3006
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 9.909395973154362e-06,
+      "loss": 0.9111,
+      "step": 3007
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 9.906040268456377e-06,
+      "loss": 0.9189,
+      "step": 3008
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 9.90268456375839e-06,
+      "loss": 0.9033,
+      "step": 3009
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.899328859060404e-06,
+      "loss": 0.8706,
+      "step": 3010
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.895973154362418e-06,
+      "loss": 0.8862,
+      "step": 3011
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.89261744966443e-06,
+      "loss": 0.8882,
+      "step": 3012
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.889261744966445e-06,
+      "loss": 0.9019,
+      "step": 3013
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.885906040268457e-06,
+      "loss": 0.8853,
+      "step": 3014
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.882550335570472e-06,
+      "loss": 0.917,
+      "step": 3015
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.879194630872484e-06,
+      "loss": 0.8706,
+      "step": 3016
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.875838926174497e-06,
+      "loss": 0.8911,
+      "step": 3017
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.872483221476511e-06,
+      "loss": 0.895,
+      "step": 3018
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.869127516778524e-06,
+      "loss": 0.8662,
+      "step": 3019
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.865771812080538e-06,
+      "loss": 0.897,
+      "step": 3020
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 9.86241610738255e-06,
+      "loss": 0.9214,
+      "step": 3021
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.859060402684565e-06,
+      "loss": 0.9062,
+      "step": 3022
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.855704697986577e-06,
+      "loss": 0.8784,
+      "step": 3023
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.852348993288592e-06,
+      "loss": 0.917,
+      "step": 3024
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.848993288590604e-06,
+      "loss": 0.9087,
+      "step": 3025
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.845637583892618e-06,
+      "loss": 0.8916,
+      "step": 3026
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.842281879194631e-06,
+      "loss": 0.9028,
+      "step": 3027
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.838926174496645e-06,
+      "loss": 0.9043,
+      "step": 3028
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.835570469798658e-06,
+      "loss": 0.8984,
+      "step": 3029
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.832214765100672e-06,
+      "loss": 0.8896,
+      "step": 3030
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.828859060402686e-06,
+      "loss": 0.894,
+      "step": 3031
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.825503355704699e-06,
+      "loss": 0.8862,
+      "step": 3032
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 9.822147651006713e-06,
+      "loss": 0.9033,
+      "step": 3033
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 9.818791946308726e-06,
+      "loss": 0.9067,
+      "step": 3034
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 9.81543624161074e-06,
+      "loss": 0.8931,
+      "step": 3035
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 9.812080536912753e-06,
+      "loss": 0.8965,
+      "step": 3036
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 9.808724832214765e-06,
+      "loss": 0.8696,
+      "step": 3037
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 9.80536912751678e-06,
+      "loss": 0.9551,
+      "step": 3038
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 9.802013422818792e-06,
+      "loss": 0.8931,
+      "step": 3039
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 9.798657718120806e-06,
+      "loss": 0.9082,
+      "step": 3040
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 9.795302013422819e-06,
+      "loss": 0.8999,
+      "step": 3041
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 9.791946308724833e-06,
+      "loss": 0.8931,
+      "step": 3042
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 9.788590604026846e-06,
+      "loss": 0.9014,
+      "step": 3043
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 9.78523489932886e-06,
+      "loss": 0.895,
+      "step": 3044
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 9.781879194630873e-06,
+      "loss": 0.9141,
+      "step": 3045
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 9.778523489932887e-06,
+      "loss": 0.9038,
+      "step": 3046
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 9.7751677852349e-06,
+      "loss": 0.917,
+      "step": 3047
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 9.771812080536914e-06,
+      "loss": 0.9072,
+      "step": 3048
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 9.768456375838928e-06,
+      "loss": 0.8877,
+      "step": 3049
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 9.76510067114094e-06,
+      "loss": 0.8999,
+      "step": 3050
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 9.761744966442955e-06,
+      "loss": 0.8926,
+      "step": 3051
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 9.758389261744968e-06,
+      "loss": 0.8779,
+      "step": 3052
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 9.755033557046982e-06,
+      "loss": 0.9224,
+      "step": 3053
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 9.751677852348994e-06,
+      "loss": 0.8975,
+      "step": 3054
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 9.748322147651007e-06,
+      "loss": 0.897,
+      "step": 3055
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 9.744966442953021e-06,
+      "loss": 0.8843,
+      "step": 3056
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 9.741610738255034e-06,
+      "loss": 0.9155,
+      "step": 3057
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 9.738255033557048e-06,
+      "loss": 0.8843,
+      "step": 3058
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 9.73489932885906e-06,
+      "loss": 0.915,
+      "step": 3059
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 9.731543624161075e-06,
+      "loss": 0.9038,
+      "step": 3060
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 9.728187919463088e-06,
+      "loss": 0.9492,
+      "step": 3061
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 9.724832214765102e-06,
+      "loss": 0.8975,
+      "step": 3062
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 9.721476510067114e-06,
+      "loss": 0.9121,
+      "step": 3063
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 9.718120805369129e-06,
+      "loss": 0.8994,
+      "step": 3064
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 9.714765100671141e-06,
+      "loss": 0.9253,
+      "step": 3065
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 9.711409395973155e-06,
+      "loss": 0.895,
+      "step": 3066
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 9.708053691275168e-06,
+      "loss": 0.8687,
+      "step": 3067
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 9.704697986577182e-06,
+      "loss": 0.8887,
+      "step": 3068
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 9.701342281879197e-06,
+      "loss": 0.9155,
+      "step": 3069
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.69798657718121e-06,
+      "loss": 0.9019,
+      "step": 3070
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.694630872483223e-06,
+      "loss": 0.9307,
+      "step": 3071
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.691275167785236e-06,
+      "loss": 0.8672,
+      "step": 3072
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.687919463087249e-06,
+      "loss": 0.8486,
+      "step": 3073
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.684563758389263e-06,
+      "loss": 0.9189,
+      "step": 3074
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.681208053691275e-06,
+      "loss": 0.9233,
+      "step": 3075
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.67785234899329e-06,
+      "loss": 0.9072,
+      "step": 3076
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.674496644295302e-06,
+      "loss": 0.9263,
+      "step": 3077
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.671140939597317e-06,
+      "loss": 0.9365,
+      "step": 3078
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.66778523489933e-06,
+      "loss": 0.894,
+      "step": 3079
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.664429530201343e-06,
+      "loss": 0.8984,
+      "step": 3080
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 9.661073825503356e-06,
+      "loss": 0.9111,
+      "step": 3081
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.657718120805369e-06,
+      "loss": 0.8872,
+      "step": 3082
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.654362416107383e-06,
+      "loss": 0.9233,
+      "step": 3083
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.651006711409397e-06,
+      "loss": 0.9316,
+      "step": 3084
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.64765100671141e-06,
+      "loss": 0.8628,
+      "step": 3085
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.644295302013424e-06,
+      "loss": 0.8926,
+      "step": 3086
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.640939597315437e-06,
+      "loss": 0.9121,
+      "step": 3087
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.637583892617451e-06,
+      "loss": 0.8887,
+      "step": 3088
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.634228187919465e-06,
+      "loss": 0.9019,
+      "step": 3089
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.630872483221478e-06,
+      "loss": 0.9194,
+      "step": 3090
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.62751677852349e-06,
+      "loss": 0.8872,
+      "step": 3091
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.624161073825505e-06,
+      "loss": 0.8809,
+      "step": 3092
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 9.620805369127517e-06,
+      "loss": 0.9526,
+      "step": 3093
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.617449664429531e-06,
+      "loss": 0.9165,
+      "step": 3094
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.614093959731544e-06,
+      "loss": 0.8789,
+      "step": 3095
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.610738255033558e-06,
+      "loss": 0.8916,
+      "step": 3096
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.607382550335571e-06,
+      "loss": 0.8765,
+      "step": 3097
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.604026845637585e-06,
+      "loss": 0.9331,
+      "step": 3098
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.600671140939598e-06,
+      "loss": 0.8989,
+      "step": 3099
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.59731543624161e-06,
+      "loss": 0.9043,
+      "step": 3100
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.593959731543625e-06,
+      "loss": 0.9014,
+      "step": 3101
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.590604026845637e-06,
+      "loss": 0.9092,
+      "step": 3102
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.587248322147651e-06,
+      "loss": 0.9033,
+      "step": 3103
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.583892617449666e-06,
+      "loss": 0.8979,
+      "step": 3104
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 9.580536912751678e-06,
+      "loss": 0.9072,
+      "step": 3105
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.577181208053693e-06,
+      "loss": 0.8838,
+      "step": 3106
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.573825503355705e-06,
+      "loss": 0.8809,
+      "step": 3107
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.57046979865772e-06,
+      "loss": 0.8838,
+      "step": 3108
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.567114093959732e-06,
+      "loss": 0.9116,
+      "step": 3109
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.563758389261746e-06,
+      "loss": 0.8843,
+      "step": 3110
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.560402684563759e-06,
+      "loss": 0.9009,
+      "step": 3111
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.557046979865773e-06,
+      "loss": 0.9072,
+      "step": 3112
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.553691275167786e-06,
+      "loss": 0.9326,
+      "step": 3113
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.5503355704698e-06,
+      "loss": 0.8813,
+      "step": 3114
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.546979865771813e-06,
+      "loss": 0.9126,
+      "step": 3115
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.543624161073827e-06,
+      "loss": 0.8921,
+      "step": 3116
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 9.54026845637584e-06,
+      "loss": 0.877,
+      "step": 3117
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.536912751677852e-06,
+      "loss": 0.916,
+      "step": 3118
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.533557046979866e-06,
+      "loss": 0.8857,
+      "step": 3119
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.530201342281879e-06,
+      "loss": 0.9019,
+      "step": 3120
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.526845637583893e-06,
+      "loss": 0.8965,
+      "step": 3121
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.523489932885906e-06,
+      "loss": 0.9248,
+      "step": 3122
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.52013422818792e-06,
+      "loss": 0.8638,
+      "step": 3123
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.516778523489934e-06,
+      "loss": 0.9268,
+      "step": 3124
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.513422818791947e-06,
+      "loss": 0.8794,
+      "step": 3125
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.510067114093961e-06,
+      "loss": 0.9214,
+      "step": 3126
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.506711409395974e-06,
+      "loss": 0.9141,
+      "step": 3127
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.503355704697988e-06,
+      "loss": 0.8975,
+      "step": 3128
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 9.5e-06,
+      "loss": 0.9092,
+      "step": 3129
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.496644295302015e-06,
+      "loss": 0.9541,
+      "step": 3130
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.493288590604027e-06,
+      "loss": 0.9033,
+      "step": 3131
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.489932885906042e-06,
+      "loss": 0.8833,
+      "step": 3132
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.486577181208054e-06,
+      "loss": 0.8726,
+      "step": 3133
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.483221476510069e-06,
+      "loss": 0.8755,
+      "step": 3134
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.479865771812081e-06,
+      "loss": 0.9121,
+      "step": 3135
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.476510067114095e-06,
+      "loss": 0.8838,
+      "step": 3136
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.473154362416108e-06,
+      "loss": 0.9146,
+      "step": 3137
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.46979865771812e-06,
+      "loss": 0.8809,
+      "step": 3138
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.466442953020135e-06,
+      "loss": 0.8569,
+      "step": 3139
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 9.463087248322147e-06,
+      "loss": 0.8838,
+      "step": 3140
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.459731543624162e-06,
+      "loss": 0.8931,
+      "step": 3141
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.456375838926174e-06,
+      "loss": 0.8984,
+      "step": 3142
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.453020134228188e-06,
+      "loss": 0.936,
+      "step": 3143
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.449664429530203e-06,
+      "loss": 0.8906,
+      "step": 3144
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.446308724832215e-06,
+      "loss": 0.9102,
+      "step": 3145
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.44295302013423e-06,
+      "loss": 0.8906,
+      "step": 3146
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.439597315436242e-06,
+      "loss": 0.8774,
+      "step": 3147
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.436241610738256e-06,
+      "loss": 0.8872,
+      "step": 3148
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.432885906040269e-06,
+      "loss": 0.8594,
+      "step": 3149
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.429530201342283e-06,
+      "loss": 0.8652,
+      "step": 3150
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.426174496644296e-06,
+      "loss": 0.9438,
+      "step": 3151
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 9.42281879194631e-06,
+      "loss": 0.8955,
+      "step": 3152
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.419463087248323e-06,
+      "loss": 0.8887,
+      "step": 3153
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.416107382550337e-06,
+      "loss": 0.8936,
+      "step": 3154
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.41275167785235e-06,
+      "loss": 0.9214,
+      "step": 3155
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.409395973154362e-06,
+      "loss": 0.9111,
+      "step": 3156
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.406040268456376e-06,
+      "loss": 0.9194,
+      "step": 3157
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.402684563758389e-06,
+      "loss": 0.8877,
+      "step": 3158
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.399328859060403e-06,
+      "loss": 0.9097,
+      "step": 3159
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.395973154362416e-06,
+      "loss": 0.9126,
+      "step": 3160
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.39261744966443e-06,
+      "loss": 0.9004,
+      "step": 3161
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.389261744966444e-06,
+      "loss": 0.9102,
+      "step": 3162
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.385906040268457e-06,
+      "loss": 0.9209,
+      "step": 3163
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 9.382550335570471e-06,
+      "loss": 0.8994,
+      "step": 3164
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.379194630872484e-06,
+      "loss": 0.9004,
+      "step": 3165
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.375838926174498e-06,
+      "loss": 0.9023,
+      "step": 3166
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.37248322147651e-06,
+      "loss": 0.8809,
+      "step": 3167
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.369127516778525e-06,
+      "loss": 0.9146,
+      "step": 3168
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.365771812080538e-06,
+      "loss": 0.9048,
+      "step": 3169
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.362416107382552e-06,
+      "loss": 0.8999,
+      "step": 3170
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.359060402684564e-06,
+      "loss": 0.9033,
+      "step": 3171
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.355704697986579e-06,
+      "loss": 0.9106,
+      "step": 3172
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.352348993288591e-06,
+      "loss": 0.9087,
+      "step": 3173
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.348993288590604e-06,
+      "loss": 0.8848,
+      "step": 3174
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.345637583892618e-06,
+      "loss": 0.9185,
+      "step": 3175
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 9.34228187919463e-06,
+      "loss": 0.8892,
+      "step": 3176
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.338926174496645e-06,
+      "loss": 0.9214,
+      "step": 3177
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.335570469798658e-06,
+      "loss": 0.9214,
+      "step": 3178
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.332214765100672e-06,
+      "loss": 0.9185,
+      "step": 3179
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.328859060402684e-06,
+      "loss": 0.9028,
+      "step": 3180
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.325503355704699e-06,
+      "loss": 0.9028,
+      "step": 3181
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.322147651006713e-06,
+      "loss": 0.9028,
+      "step": 3182
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.318791946308726e-06,
+      "loss": 0.8711,
+      "step": 3183
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.31543624161074e-06,
+      "loss": 0.8823,
+      "step": 3184
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.312080536912752e-06,
+      "loss": 0.8682,
+      "step": 3185
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.308724832214767e-06,
+      "loss": 0.8887,
+      "step": 3186
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.30536912751678e-06,
+      "loss": 0.8794,
+      "step": 3187
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 9.302013422818794e-06,
+      "loss": 0.9229,
+      "step": 3188
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.298657718120806e-06,
+      "loss": 0.876,
+      "step": 3189
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.29530201342282e-06,
+      "loss": 0.8857,
+      "step": 3190
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.291946308724833e-06,
+      "loss": 0.916,
+      "step": 3191
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.288590604026846e-06,
+      "loss": 0.9282,
+      "step": 3192
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.28523489932886e-06,
+      "loss": 0.9053,
+      "step": 3193
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.281879194630872e-06,
+      "loss": 0.9077,
+      "step": 3194
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.278523489932887e-06,
+      "loss": 0.9102,
+      "step": 3195
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.2751677852349e-06,
+      "loss": 0.877,
+      "step": 3196
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.271812080536914e-06,
+      "loss": 0.9019,
+      "step": 3197
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.268456375838926e-06,
+      "loss": 0.8984,
+      "step": 3198
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.26510067114094e-06,
+      "loss": 0.8984,
+      "step": 3199
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 9.261744966442953e-06,
+      "loss": 0.8696,
+      "step": 3200
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.258389261744967e-06,
+      "loss": 0.8745,
+      "step": 3201
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.255033557046982e-06,
+      "loss": 0.9375,
+      "step": 3202
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.251677852348994e-06,
+      "loss": 0.8901,
+      "step": 3203
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.248322147651008e-06,
+      "loss": 0.8687,
+      "step": 3204
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.244966442953021e-06,
+      "loss": 0.8848,
+      "step": 3205
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.241610738255035e-06,
+      "loss": 0.9419,
+      "step": 3206
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.238255033557048e-06,
+      "loss": 0.916,
+      "step": 3207
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.234899328859062e-06,
+      "loss": 0.9067,
+      "step": 3208
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.231543624161075e-06,
+      "loss": 0.8721,
+      "step": 3209
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.228187919463087e-06,
+      "loss": 0.8857,
+      "step": 3210
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.224832214765101e-06,
+      "loss": 0.894,
+      "step": 3211
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 9.221476510067114e-06,
+      "loss": 0.9106,
+      "step": 3212
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.218120805369128e-06,
+      "loss": 0.8994,
+      "step": 3213
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.214765100671141e-06,
+      "loss": 0.8999,
+      "step": 3214
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.211409395973155e-06,
+      "loss": 0.8979,
+      "step": 3215
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.208053691275168e-06,
+      "loss": 0.9048,
+      "step": 3216
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.204697986577182e-06,
+      "loss": 0.8911,
+      "step": 3217
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.201342281879195e-06,
+      "loss": 0.8931,
+      "step": 3218
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.197986577181209e-06,
+      "loss": 0.876,
+      "step": 3219
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.194630872483221e-06,
+      "loss": 0.9072,
+      "step": 3220
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.191275167785236e-06,
+      "loss": 0.9092,
+      "step": 3221
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.18791946308725e-06,
+      "loss": 0.8926,
+      "step": 3222
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.184563758389263e-06,
+      "loss": 0.8633,
+      "step": 3223
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.181208053691277e-06,
+      "loss": 0.8979,
+      "step": 3224
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.17785234899329e-06,
+      "loss": 0.8633,
+      "step": 3225
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.174496644295304e-06,
+      "loss": 0.9302,
+      "step": 3226
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.171140939597316e-06,
+      "loss": 0.9341,
+      "step": 3227
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.16778523489933e-06,
+      "loss": 0.9272,
+      "step": 3228
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.164429530201343e-06,
+      "loss": 0.9214,
+      "step": 3229
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.161073825503356e-06,
+      "loss": 0.9175,
+      "step": 3230
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.15771812080537e-06,
+      "loss": 0.873,
+      "step": 3231
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.154362416107383e-06,
+      "loss": 0.9014,
+      "step": 3232
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.151006711409397e-06,
+      "loss": 0.8691,
+      "step": 3233
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.14765100671141e-06,
+      "loss": 0.8481,
+      "step": 3234
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.144295302013424e-06,
+      "loss": 0.9009,
+      "step": 3235
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.140939597315436e-06,
+      "loss": 0.9121,
+      "step": 3236
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.13758389261745e-06,
+      "loss": 0.876,
+      "step": 3237
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.134228187919463e-06,
+      "loss": 0.9307,
+      "step": 3238
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.130872483221477e-06,
+      "loss": 0.8989,
+      "step": 3239
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.12751677852349e-06,
+      "loss": 0.8804,
+      "step": 3240
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.124161073825504e-06,
+      "loss": 0.9165,
+      "step": 3241
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.120805369127519e-06,
+      "loss": 0.8672,
+      "step": 3242
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.117449664429531e-06,
+      "loss": 0.8755,
+      "step": 3243
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.114093959731545e-06,
+      "loss": 0.8784,
+      "step": 3244
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.110738255033558e-06,
+      "loss": 0.9043,
+      "step": 3245
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.107382550335572e-06,
+      "loss": 0.8691,
+      "step": 3246
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.104026845637585e-06,
+      "loss": 0.8462,
+      "step": 3247
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.100671140939597e-06,
+      "loss": 0.8662,
+      "step": 3248
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.097315436241612e-06,
+      "loss": 0.8833,
+      "step": 3249
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.093959731543624e-06,
+      "loss": 0.8896,
+      "step": 3250
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.090604026845639e-06,
+      "loss": 0.9019,
+      "step": 3251
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.087248322147651e-06,
+      "loss": 0.9072,
+      "step": 3252
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.083892617449665e-06,
+      "loss": 0.8921,
+      "step": 3253
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.080536912751678e-06,
+      "loss": 0.8594,
+      "step": 3254
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.077181208053692e-06,
+      "loss": 0.9082,
+      "step": 3255
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.073825503355705e-06,
+      "loss": 0.8486,
+      "step": 3256
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.070469798657719e-06,
+      "loss": 0.894,
+      "step": 3257
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.067114093959732e-06,
+      "loss": 0.8809,
+      "step": 3258
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.063758389261746e-06,
+      "loss": 0.8882,
+      "step": 3259
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.060402684563759e-06,
+      "loss": 0.8696,
+      "step": 3260
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.057046979865773e-06,
+      "loss": 0.8979,
+      "step": 3261
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.053691275167787e-06,
+      "loss": 0.8931,
+      "step": 3262
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.0503355704698e-06,
+      "loss": 0.8848,
+      "step": 3263
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.046979865771814e-06,
+      "loss": 0.9263,
+      "step": 3264
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.043624161073827e-06,
+      "loss": 0.9023,
+      "step": 3265
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.040268456375839e-06,
+      "loss": 0.9131,
+      "step": 3266
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.036912751677853e-06,
+      "loss": 0.876,
+      "step": 3267
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.033557046979866e-06,
+      "loss": 0.875,
+      "step": 3268
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.03020134228188e-06,
+      "loss": 0.8965,
+      "step": 3269
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.026845637583893e-06,
+      "loss": 0.8955,
+      "step": 3270
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.023489932885907e-06,
+      "loss": 0.8667,
+      "step": 3271
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 9.02013422818792e-06,
+      "loss": 0.9048,
+      "step": 3272
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 9.016778523489934e-06,
+      "loss": 0.8774,
+      "step": 3273
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 9.013422818791947e-06,
+      "loss": 0.9189,
+      "step": 3274
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 9.01006711409396e-06,
+      "loss": 0.8989,
+      "step": 3275
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 9.006711409395973e-06,
+      "loss": 0.918,
+      "step": 3276
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 9.003355704697988e-06,
+      "loss": 0.9194,
+      "step": 3277
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 9e-06,
+      "loss": 0.8926,
+      "step": 3278
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.996644295302015e-06,
+      "loss": 0.8794,
+      "step": 3279
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.993288590604027e-06,
+      "loss": 0.9082,
+      "step": 3280
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.989932885906041e-06,
+      "loss": 0.9043,
+      "step": 3281
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.986577181208056e-06,
+      "loss": 0.9048,
+      "step": 3282
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.983221476510068e-06,
+      "loss": 0.8989,
+      "step": 3283
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.97986577181208e-06,
+      "loss": 0.9204,
+      "step": 3284
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.976510067114095e-06,
+      "loss": 0.894,
+      "step": 3285
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.973154362416108e-06,
+      "loss": 0.8896,
+      "step": 3286
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.969798657718122e-06,
+      "loss": 0.9058,
+      "step": 3287
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.966442953020134e-06,
+      "loss": 0.8901,
+      "step": 3288
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.963087248322149e-06,
+      "loss": 0.897,
+      "step": 3289
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.959731543624161e-06,
+      "loss": 0.9165,
+      "step": 3290
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.956375838926176e-06,
+      "loss": 0.9282,
+      "step": 3291
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.953020134228188e-06,
+      "loss": 0.8892,
+      "step": 3292
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.9496644295302e-06,
+      "loss": 0.8696,
+      "step": 3293
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.946308724832215e-06,
+      "loss": 0.9043,
+      "step": 3294
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.94295302013423e-06,
+      "loss": 0.9106,
+      "step": 3295
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.939597315436242e-06,
+      "loss": 0.9106,
+      "step": 3296
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.936241610738256e-06,
+      "loss": 0.9185,
+      "step": 3297
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.932885906040269e-06,
+      "loss": 0.9082,
+      "step": 3298
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.929530201342283e-06,
+      "loss": 0.895,
+      "step": 3299
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.926174496644297e-06,
+      "loss": 0.9082,
+      "step": 3300
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.92281879194631e-06,
+      "loss": 0.8979,
+      "step": 3301
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.919463087248322e-06,
+      "loss": 0.8809,
+      "step": 3302
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.916107382550337e-06,
+      "loss": 0.8589,
+      "step": 3303
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.91275167785235e-06,
+      "loss": 0.8979,
+      "step": 3304
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.909395973154364e-06,
+      "loss": 0.9331,
+      "step": 3305
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.906040268456376e-06,
+      "loss": 0.9067,
+      "step": 3306
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 8.90268456375839e-06,
+      "loss": 0.8696,
+      "step": 3307
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.899328859060403e-06,
+      "loss": 0.8921,
+      "step": 3308
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.895973154362417e-06,
+      "loss": 0.916,
+      "step": 3309
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.89261744966443e-06,
+      "loss": 0.9321,
+      "step": 3310
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.889261744966444e-06,
+      "loss": 0.9248,
+      "step": 3311
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.885906040268457e-06,
+      "loss": 0.8926,
+      "step": 3312
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.88255033557047e-06,
+      "loss": 0.8867,
+      "step": 3313
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.879194630872484e-06,
+      "loss": 0.9082,
+      "step": 3314
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.875838926174498e-06,
+      "loss": 0.9268,
+      "step": 3315
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.87248322147651e-06,
+      "loss": 0.8857,
+      "step": 3316
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.869127516778525e-06,
+      "loss": 0.9199,
+      "step": 3317
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.865771812080537e-06,
+      "loss": 0.8887,
+      "step": 3318
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 8.862416107382552e-06,
+      "loss": 0.9058,
+      "step": 3319
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.859060402684566e-06,
+      "loss": 0.8828,
+      "step": 3320
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.855704697986578e-06,
+      "loss": 0.9243,
+      "step": 3321
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.852348993288591e-06,
+      "loss": 0.875,
+      "step": 3322
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.848993288590605e-06,
+      "loss": 0.8989,
+      "step": 3323
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.845637583892618e-06,
+      "loss": 0.8877,
+      "step": 3324
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.842281879194632e-06,
+      "loss": 0.9268,
+      "step": 3325
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.838926174496645e-06,
+      "loss": 0.9238,
+      "step": 3326
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.835570469798659e-06,
+      "loss": 0.9023,
+      "step": 3327
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.832214765100672e-06,
+      "loss": 0.9487,
+      "step": 3328
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.828859060402686e-06,
+      "loss": 0.8921,
+      "step": 3329
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.825503355704698e-06,
+      "loss": 0.8989,
+      "step": 3330
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 8.822147651006711e-06,
+      "loss": 0.8687,
+      "step": 3331
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.818791946308725e-06,
+      "loss": 0.8828,
+      "step": 3332
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.815436241610738e-06,
+      "loss": 0.8872,
+      "step": 3333
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.812080536912752e-06,
+      "loss": 0.8872,
+      "step": 3334
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.808724832214766e-06,
+      "loss": 0.9087,
+      "step": 3335
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.805369127516779e-06,
+      "loss": 0.9067,
+      "step": 3336
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.802013422818793e-06,
+      "loss": 0.8818,
+      "step": 3337
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.798657718120806e-06,
+      "loss": 0.9272,
+      "step": 3338
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.79530201342282e-06,
+      "loss": 0.9004,
+      "step": 3339
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.791946308724833e-06,
+      "loss": 0.9219,
+      "step": 3340
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.788590604026847e-06,
+      "loss": 0.8784,
+      "step": 3341
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.78523489932886e-06,
+      "loss": 0.9116,
+      "step": 3342
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 8.781879194630874e-06,
+      "loss": 0.9243,
+      "step": 3343
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.778523489932886e-06,
+      "loss": 0.9106,
+      "step": 3344
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.7751677852349e-06,
+      "loss": 0.8931,
+      "step": 3345
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.771812080536913e-06,
+      "loss": 0.873,
+      "step": 3346
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.768456375838928e-06,
+      "loss": 0.8691,
+      "step": 3347
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.76510067114094e-06,
+      "loss": 0.9263,
+      "step": 3348
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.761744966442953e-06,
+      "loss": 0.9214,
+      "step": 3349
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.758389261744967e-06,
+      "loss": 0.9268,
+      "step": 3350
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.75503355704698e-06,
+      "loss": 0.8574,
+      "step": 3351
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.751677852348994e-06,
+      "loss": 0.8477,
+      "step": 3352
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.748322147651006e-06,
+      "loss": 0.9072,
+      "step": 3353
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.74496644295302e-06,
+      "loss": 0.9111,
+      "step": 3354
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 8.741610738255035e-06,
+      "loss": 0.8945,
+      "step": 3355
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.738255033557047e-06,
+      "loss": 0.9072,
+      "step": 3356
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.734899328859062e-06,
+      "loss": 0.916,
+      "step": 3357
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.731543624161074e-06,
+      "loss": 0.8726,
+      "step": 3358
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.728187919463089e-06,
+      "loss": 0.8574,
+      "step": 3359
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.724832214765101e-06,
+      "loss": 0.9058,
+      "step": 3360
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.721476510067115e-06,
+      "loss": 0.8818,
+      "step": 3361
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.718120805369128e-06,
+      "loss": 0.9141,
+      "step": 3362
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.714765100671142e-06,
+      "loss": 0.8857,
+      "step": 3363
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.711409395973155e-06,
+      "loss": 0.8848,
+      "step": 3364
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.70805369127517e-06,
+      "loss": 0.9106,
+      "step": 3365
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.704697986577182e-06,
+      "loss": 0.9194,
+      "step": 3366
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 8.701342281879194e-06,
+      "loss": 0.8853,
+      "step": 3367
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.697986577181209e-06,
+      "loss": 0.9292,
+      "step": 3368
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.694630872483221e-06,
+      "loss": 0.9307,
+      "step": 3369
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.691275167785235e-06,
+      "loss": 0.9033,
+      "step": 3370
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.687919463087248e-06,
+      "loss": 0.8887,
+      "step": 3371
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.684563758389262e-06,
+      "loss": 0.9175,
+      "step": 3372
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.681208053691275e-06,
+      "loss": 0.9238,
+      "step": 3373
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.67785234899329e-06,
+      "loss": 0.8838,
+      "step": 3374
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.674496644295303e-06,
+      "loss": 0.8779,
+      "step": 3375
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.671140939597316e-06,
+      "loss": 0.8618,
+      "step": 3376
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.66778523489933e-06,
+      "loss": 0.9204,
+      "step": 3377
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.664429530201343e-06,
+      "loss": 0.9194,
+      "step": 3378
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 8.661073825503357e-06,
+      "loss": 0.9106,
+      "step": 3379
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.65771812080537e-06,
+      "loss": 0.8975,
+      "step": 3380
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.654362416107384e-06,
+      "loss": 0.9204,
+      "step": 3381
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.651006711409397e-06,
+      "loss": 0.894,
+      "step": 3382
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.647651006711411e-06,
+      "loss": 0.9233,
+      "step": 3383
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.644295302013423e-06,
+      "loss": 0.9072,
+      "step": 3384
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.640939597315436e-06,
+      "loss": 0.8804,
+      "step": 3385
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.63758389261745e-06,
+      "loss": 0.9067,
+      "step": 3386
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.634228187919463e-06,
+      "loss": 0.9048,
+      "step": 3387
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.630872483221477e-06,
+      "loss": 0.9092,
+      "step": 3388
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.62751677852349e-06,
+      "loss": 0.9302,
+      "step": 3389
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.624161073825504e-06,
+      "loss": 0.8574,
+      "step": 3390
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 8.620805369127517e-06,
+      "loss": 0.8784,
+      "step": 3391
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.617449664429531e-06,
+      "loss": 0.8633,
+      "step": 3392
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.614093959731543e-06,
+      "loss": 0.9023,
+      "step": 3393
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.610738255033558e-06,
+      "loss": 0.8662,
+      "step": 3394
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.607382550335572e-06,
+      "loss": 0.9048,
+      "step": 3395
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.604026845637585e-06,
+      "loss": 0.895,
+      "step": 3396
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.600671140939599e-06,
+      "loss": 0.9043,
+      "step": 3397
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.597315436241611e-06,
+      "loss": 0.9126,
+      "step": 3398
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.593959731543626e-06,
+      "loss": 0.8896,
+      "step": 3399
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.590604026845638e-06,
+      "loss": 0.9268,
+      "step": 3400
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.587248322147653e-06,
+      "loss": 0.9238,
+      "step": 3401
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.583892617449665e-06,
+      "loss": 0.916,
+      "step": 3402
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 8.58053691275168e-06,
+      "loss": 0.8848,
+      "step": 3403
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.577181208053692e-06,
+      "loss": 0.9058,
+      "step": 3404
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.573825503355705e-06,
+      "loss": 0.9028,
+      "step": 3405
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.570469798657719e-06,
+      "loss": 0.9043,
+      "step": 3406
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.567114093959731e-06,
+      "loss": 0.8691,
+      "step": 3407
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.563758389261746e-06,
+      "loss": 0.8931,
+      "step": 3408
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.560402684563758e-06,
+      "loss": 0.917,
+      "step": 3409
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.557046979865773e-06,
+      "loss": 0.8804,
+      "step": 3410
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.553691275167785e-06,
+      "loss": 0.9253,
+      "step": 3411
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.5503355704698e-06,
+      "loss": 0.8921,
+      "step": 3412
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.546979865771814e-06,
+      "loss": 0.8896,
+      "step": 3413
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.543624161073826e-06,
+      "loss": 0.8687,
+      "step": 3414
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 8.54026845637584e-06,
+      "loss": 0.8916,
+      "step": 3415
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.536912751677853e-06,
+      "loss": 0.8901,
+      "step": 3416
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.533557046979867e-06,
+      "loss": 0.895,
+      "step": 3417
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.53020134228188e-06,
+      "loss": 0.8882,
+      "step": 3418
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.526845637583894e-06,
+      "loss": 0.895,
+      "step": 3419
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.523489932885907e-06,
+      "loss": 0.8965,
+      "step": 3420
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.520134228187921e-06,
+      "loss": 0.8765,
+      "step": 3421
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.516778523489934e-06,
+      "loss": 0.9111,
+      "step": 3422
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.513422818791946e-06,
+      "loss": 0.9014,
+      "step": 3423
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.51006711409396e-06,
+      "loss": 0.8887,
+      "step": 3424
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.506711409395973e-06,
+      "loss": 0.9058,
+      "step": 3425
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 8.503355704697987e-06,
+      "loss": 0.8926,
+      "step": 3426
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.5e-06,
+      "loss": 0.8906,
+      "step": 3427
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.496644295302014e-06,
+      "loss": 0.9023,
+      "step": 3428
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.493288590604027e-06,
+      "loss": 0.8936,
+      "step": 3429
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.489932885906041e-06,
+      "loss": 0.9116,
+      "step": 3430
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.486577181208054e-06,
+      "loss": 0.877,
+      "step": 3431
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.483221476510068e-06,
+      "loss": 0.8979,
+      "step": 3432
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.479865771812082e-06,
+      "loss": 0.8711,
+      "step": 3433
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.476510067114095e-06,
+      "loss": 0.8984,
+      "step": 3434
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.473154362416109e-06,
+      "loss": 0.9121,
+      "step": 3435
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.469798657718122e-06,
+      "loss": 0.8804,
+      "step": 3436
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.466442953020136e-06,
+      "loss": 0.8945,
+      "step": 3437
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 8.463087248322148e-06,
+      "loss": 0.9023,
+      "step": 3438
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 8.459731543624163e-06,
+      "loss": 0.916,
+      "step": 3439
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 8.456375838926175e-06,
+      "loss": 0.9155,
+      "step": 3440
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 8.453020134228188e-06,
+      "loss": 0.9077,
+      "step": 3441
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 8.449664429530202e-06,
+      "loss": 0.9038,
+      "step": 3442
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 8.446308724832215e-06,
+      "loss": 0.8867,
+      "step": 3443
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 8.442953020134229e-06,
+      "loss": 0.8706,
+      "step": 3444
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 8.439597315436242e-06,
+      "loss": 0.8882,
+      "step": 3445
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 8.436241610738256e-06,
+      "loss": 0.8726,
+      "step": 3446
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 8.432885906040268e-06,
+      "loss": 0.8867,
+      "step": 3447
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 8.429530201342283e-06,
+      "loss": 0.8994,
+      "step": 3448
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 8.426174496644295e-06,
+      "loss": 0.9053,
+      "step": 3449
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 8.42281879194631e-06,
+      "loss": 0.8823,
+      "step": 3450
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 8.419463087248322e-06,
+      "loss": 0.9048,
+      "step": 3451
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 8.416107382550336e-06,
+      "loss": 0.8618,
+      "step": 3452
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 8.41275167785235e-06,
+      "loss": 0.8936,
+      "step": 3453
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 8.409395973154363e-06,
+      "loss": 0.9082,
+      "step": 3454
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 8.406040268456378e-06,
+      "loss": 0.8672,
+      "step": 3455
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 8.40268456375839e-06,
+      "loss": 0.9111,
+      "step": 3456
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 8.399328859060404e-06,
+      "loss": 0.8643,
+      "step": 3457
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 8.395973154362417e-06,
+      "loss": 0.9077,
+      "step": 3458
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 8.39261744966443e-06,
+      "loss": 0.9023,
+      "step": 3459
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 8.389261744966444e-06,
+      "loss": 0.877,
+      "step": 3460
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 8.385906040268456e-06,
+      "loss": 0.8765,
+      "step": 3461
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 8.38255033557047e-06,
+      "loss": 0.9053,
+      "step": 3462
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 8.379194630872483e-06,
+      "loss": 0.876,
+      "step": 3463
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 8.375838926174498e-06,
+      "loss": 0.8872,
+      "step": 3464
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 8.37248322147651e-06,
+      "loss": 0.8926,
+      "step": 3465
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 8.369127516778524e-06,
+      "loss": 0.8735,
+      "step": 3466
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 8.365771812080537e-06,
+      "loss": 0.9272,
+      "step": 3467
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 8.362416107382551e-06,
+      "loss": 0.9209,
+      "step": 3468
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 8.359060402684564e-06,
+      "loss": 0.9082,
+      "step": 3469
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 8.355704697986578e-06,
+      "loss": 0.8901,
+      "step": 3470
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 8.35234899328859e-06,
+      "loss": 0.9067,
+      "step": 3471
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 8.348993288590605e-06,
+      "loss": 0.9014,
+      "step": 3472
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 8.34563758389262e-06,
+      "loss": 0.8647,
+      "step": 3473
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 8.342281879194632e-06,
+      "loss": 0.9409,
+      "step": 3474
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.338926174496646e-06,
+      "loss": 0.8862,
+      "step": 3475
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.335570469798659e-06,
+      "loss": 0.8872,
+      "step": 3476
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.332214765100671e-06,
+      "loss": 0.915,
+      "step": 3477
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.328859060402686e-06,
+      "loss": 0.8813,
+      "step": 3478
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.325503355704698e-06,
+      "loss": 0.9141,
+      "step": 3479
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.322147651006712e-06,
+      "loss": 0.8989,
+      "step": 3480
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.318791946308725e-06,
+      "loss": 0.9233,
+      "step": 3481
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.31543624161074e-06,
+      "loss": 0.9204,
+      "step": 3482
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.312080536912752e-06,
+      "loss": 0.9258,
+      "step": 3483
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.308724832214766e-06,
+      "loss": 0.8711,
+      "step": 3484
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.305369127516779e-06,
+      "loss": 0.9121,
+      "step": 3485
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 8.302013422818791e-06,
+      "loss": 0.8965,
+      "step": 3486
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 8.298657718120806e-06,
+      "loss": 0.873,
+      "step": 3487
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 8.29530201342282e-06,
+      "loss": 0.8677,
+      "step": 3488
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 8.291946308724832e-06,
+      "loss": 0.9033,
+      "step": 3489
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 8.288590604026847e-06,
+      "loss": 0.9199,
+      "step": 3490
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 8.28523489932886e-06,
+      "loss": 0.9126,
+      "step": 3491
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 8.281879194630874e-06,
+      "loss": 0.8843,
+      "step": 3492
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 8.278523489932888e-06,
+      "loss": 0.8696,
+      "step": 3493
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 8.2751677852349e-06,
+      "loss": 0.936,
+      "step": 3494
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 8.271812080536915e-06,
+      "loss": 0.9019,
+      "step": 3495
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 8.268456375838927e-06,
+      "loss": 0.9214,
+      "step": 3496
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 8.26510067114094e-06,
+      "loss": 0.9004,
+      "step": 3497
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 8.261744966442954e-06,
+      "loss": 0.8467,
+      "step": 3498
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 8.258389261744967e-06,
+      "loss": 0.8643,
+      "step": 3499
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 8.255033557046981e-06,
+      "loss": 0.8701,
+      "step": 3500
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 8.251677852348993e-06,
+      "loss": 0.916,
+      "step": 3501
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 8.248322147651008e-06,
+      "loss": 0.8911,
+      "step": 3502
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 8.24496644295302e-06,
+      "loss": 0.873,
+      "step": 3503
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 8.241610738255035e-06,
+      "loss": 0.9248,
+      "step": 3504
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 8.238255033557047e-06,
+      "loss": 0.8594,
+      "step": 3505
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 8.23489932885906e-06,
+      "loss": 0.8716,
+      "step": 3506
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 8.231543624161074e-06,
+      "loss": 0.8569,
+      "step": 3507
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 8.228187919463088e-06,
+      "loss": 0.9038,
+      "step": 3508
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 8.224832214765101e-06,
+      "loss": 0.8955,
+      "step": 3509
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 8.221476510067115e-06,
+      "loss": 0.8955,
+      "step": 3510
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.218120805369128e-06,
+      "loss": 0.9077,
+      "step": 3511
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.214765100671142e-06,
+      "loss": 0.958,
+      "step": 3512
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.211409395973156e-06,
+      "loss": 0.9019,
+      "step": 3513
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.208053691275169e-06,
+      "loss": 0.8926,
+      "step": 3514
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.204697986577181e-06,
+      "loss": 0.8896,
+      "step": 3515
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.201342281879196e-06,
+      "loss": 0.9019,
+      "step": 3516
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.197986577181208e-06,
+      "loss": 0.9106,
+      "step": 3517
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.194630872483223e-06,
+      "loss": 0.9126,
+      "step": 3518
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.191275167785235e-06,
+      "loss": 0.9316,
+      "step": 3519
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.18791946308725e-06,
+      "loss": 0.9155,
+      "step": 3520
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.184563758389262e-06,
+      "loss": 0.896,
+      "step": 3521
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 8.181208053691276e-06,
+      "loss": 0.9058,
+      "step": 3522
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.177852348993289e-06,
+      "loss": 0.8823,
+      "step": 3523
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.174496644295301e-06,
+      "loss": 0.8867,
+      "step": 3524
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.171140939597316e-06,
+      "loss": 0.9082,
+      "step": 3525
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.16778523489933e-06,
+      "loss": 0.8862,
+      "step": 3526
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.164429530201343e-06,
+      "loss": 0.9146,
+      "step": 3527
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.161073825503357e-06,
+      "loss": 0.8936,
+      "step": 3528
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.15771812080537e-06,
+      "loss": 0.8945,
+      "step": 3529
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.154362416107384e-06,
+      "loss": 0.9072,
+      "step": 3530
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.151006711409398e-06,
+      "loss": 0.8813,
+      "step": 3531
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.14765100671141e-06,
+      "loss": 0.9062,
+      "step": 3532
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.144295302013423e-06,
+      "loss": 0.8413,
+      "step": 3533
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 8.140939597315437e-06,
+      "loss": 0.8892,
+      "step": 3534
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 8.13758389261745e-06,
+      "loss": 0.9546,
+      "step": 3535
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 8.134228187919464e-06,
+      "loss": 0.9346,
+      "step": 3536
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 8.130872483221477e-06,
+      "loss": 0.8931,
+      "step": 3537
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 8.127516778523491e-06,
+      "loss": 0.896,
+      "step": 3538
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 8.124161073825504e-06,
+      "loss": 0.877,
+      "step": 3539
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 8.120805369127518e-06,
+      "loss": 0.8853,
+      "step": 3540
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 8.11744966442953e-06,
+      "loss": 0.9214,
+      "step": 3541
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 8.114093959731543e-06,
+      "loss": 0.9014,
+      "step": 3542
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 8.110738255033557e-06,
+      "loss": 0.8652,
+      "step": 3543
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 8.10738255033557e-06,
+      "loss": 0.9004,
+      "step": 3544
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 8.104026845637584e-06,
+      "loss": 0.9102,
+      "step": 3545
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 8.100671140939599e-06,
+      "loss": 0.8765,
+      "step": 3546
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 8.097315436241611e-06,
+      "loss": 0.8955,
+      "step": 3547
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 8.093959731543625e-06,
+      "loss": 0.8999,
+      "step": 3548
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 8.090604026845638e-06,
+      "loss": 0.897,
+      "step": 3549
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 8.087248322147652e-06,
+      "loss": 0.8862,
+      "step": 3550
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 8.083892617449665e-06,
+      "loss": 0.9331,
+      "step": 3551
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 8.080536912751679e-06,
+      "loss": 0.9028,
+      "step": 3552
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 8.077181208053692e-06,
+      "loss": 0.9302,
+      "step": 3553
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 8.073825503355706e-06,
+      "loss": 0.8867,
+      "step": 3554
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 8.070469798657719e-06,
+      "loss": 0.8848,
+      "step": 3555
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 8.067114093959733e-06,
+      "loss": 0.9014,
+      "step": 3556
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 8.063758389261745e-06,
+      "loss": 0.916,
+      "step": 3557
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 8.06040268456376e-06,
+      "loss": 0.9194,
+      "step": 3558
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.057046979865772e-06,
+      "loss": 0.874,
+      "step": 3559
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.053691275167785e-06,
+      "loss": 0.9028,
+      "step": 3560
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.050335570469799e-06,
+      "loss": 0.9038,
+      "step": 3561
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.046979865771812e-06,
+      "loss": 0.8975,
+      "step": 3562
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.043624161073826e-06,
+      "loss": 0.9209,
+      "step": 3563
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.040268456375839e-06,
+      "loss": 0.8633,
+      "step": 3564
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.036912751677853e-06,
+      "loss": 0.8735,
+      "step": 3565
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.033557046979867e-06,
+      "loss": 0.8867,
+      "step": 3566
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.03020134228188e-06,
+      "loss": 0.8701,
+      "step": 3567
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.026845637583894e-06,
+      "loss": 0.8833,
+      "step": 3568
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.023489932885906e-06,
+      "loss": 0.8652,
+      "step": 3569
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 8.02013422818792e-06,
+      "loss": 0.9263,
+      "step": 3570
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 8.016778523489933e-06,
+      "loss": 0.873,
+      "step": 3571
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 8.013422818791948e-06,
+      "loss": 0.8477,
+      "step": 3572
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 8.01006711409396e-06,
+      "loss": 0.8608,
+      "step": 3573
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 8.006711409395974e-06,
+      "loss": 0.9116,
+      "step": 3574
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 8.003355704697987e-06,
+      "loss": 0.8862,
+      "step": 3575
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 0.8281,
+      "step": 3576
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.996644295302014e-06,
+      "loss": 0.8521,
+      "step": 3577
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.993288590604026e-06,
+      "loss": 0.8657,
+      "step": 3578
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.98993288590604e-06,
+      "loss": 0.8672,
+      "step": 3579
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.986577181208053e-06,
+      "loss": 0.8745,
+      "step": 3580
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 7.983221476510068e-06,
+      "loss": 0.8843,
+      "step": 3581
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.97986577181208e-06,
+      "loss": 0.8481,
+      "step": 3582
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.976510067114094e-06,
+      "loss": 0.8711,
+      "step": 3583
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.973154362416107e-06,
+      "loss": 0.8442,
+      "step": 3584
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.969798657718121e-06,
+      "loss": 0.8823,
+      "step": 3585
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.966442953020136e-06,
+      "loss": 0.8589,
+      "step": 3586
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.963087248322148e-06,
+      "loss": 0.8638,
+      "step": 3587
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.959731543624162e-06,
+      "loss": 0.8926,
+      "step": 3588
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.956375838926175e-06,
+      "loss": 0.813,
+      "step": 3589
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.95302013422819e-06,
+      "loss": 0.8394,
+      "step": 3590
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.949664429530202e-06,
+      "loss": 0.8701,
+      "step": 3591
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.946308724832216e-06,
+      "loss": 0.8154,
+      "step": 3592
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 7.942953020134229e-06,
+      "loss": 0.8101,
+      "step": 3593
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.939597315436243e-06,
+      "loss": 0.8901,
+      "step": 3594
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.936241610738256e-06,
+      "loss": 0.8403,
+      "step": 3595
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.93288590604027e-06,
+      "loss": 0.8555,
+      "step": 3596
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.929530201342282e-06,
+      "loss": 0.8604,
+      "step": 3597
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.926174496644295e-06,
+      "loss": 0.8413,
+      "step": 3598
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.92281879194631e-06,
+      "loss": 0.8677,
+      "step": 3599
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.919463087248322e-06,
+      "loss": 0.8789,
+      "step": 3600
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.916107382550336e-06,
+      "loss": 0.8477,
+      "step": 3601
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.912751677852349e-06,
+      "loss": 0.8887,
+      "step": 3602
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.909395973154363e-06,
+      "loss": 0.8843,
+      "step": 3603
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.906040268456376e-06,
+      "loss": 0.8525,
+      "step": 3604
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 7.90268456375839e-06,
+      "loss": 0.8696,
+      "step": 3605
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.899328859060404e-06,
+      "loss": 0.8657,
+      "step": 3606
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.895973154362417e-06,
+      "loss": 0.8496,
+      "step": 3607
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.892617449664431e-06,
+      "loss": 0.8799,
+      "step": 3608
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.889261744966444e-06,
+      "loss": 0.8677,
+      "step": 3609
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.885906040268458e-06,
+      "loss": 0.8394,
+      "step": 3610
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.88255033557047e-06,
+      "loss": 0.8623,
+      "step": 3611
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.879194630872485e-06,
+      "loss": 0.8667,
+      "step": 3612
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.875838926174497e-06,
+      "loss": 0.877,
+      "step": 3613
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.872483221476512e-06,
+      "loss": 0.8667,
+      "step": 3614
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.869127516778524e-06,
+      "loss": 0.876,
+      "step": 3615
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.865771812080537e-06,
+      "loss": 0.8677,
+      "step": 3616
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 7.862416107382551e-06,
+      "loss": 0.8672,
+      "step": 3617
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.859060402684564e-06,
+      "loss": 0.8564,
+      "step": 3618
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.855704697986578e-06,
+      "loss": 0.8501,
+      "step": 3619
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.85234899328859e-06,
+      "loss": 0.8384,
+      "step": 3620
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.848993288590605e-06,
+      "loss": 0.8672,
+      "step": 3621
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.845637583892617e-06,
+      "loss": 0.8394,
+      "step": 3622
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.842281879194632e-06,
+      "loss": 0.8467,
+      "step": 3623
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.838926174496644e-06,
+      "loss": 0.8452,
+      "step": 3624
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.835570469798658e-06,
+      "loss": 0.8594,
+      "step": 3625
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.832214765100673e-06,
+      "loss": 0.8887,
+      "step": 3626
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.828859060402685e-06,
+      "loss": 0.8784,
+      "step": 3627
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.8255033557047e-06,
+      "loss": 0.8613,
+      "step": 3628
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 7.822147651006712e-06,
+      "loss": 0.8667,
+      "step": 3629
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 7.818791946308726e-06,
+      "loss": 0.8535,
+      "step": 3630
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 7.815436241610739e-06,
+      "loss": 0.8594,
+      "step": 3631
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 7.812080536912753e-06,
+      "loss": 0.8784,
+      "step": 3632
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 7.808724832214766e-06,
+      "loss": 0.8452,
+      "step": 3633
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 7.805369127516778e-06,
+      "loss": 0.8584,
+      "step": 3634
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 7.802013422818793e-06,
+      "loss": 0.8403,
+      "step": 3635
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 7.798657718120805e-06,
+      "loss": 0.8677,
+      "step": 3636
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 7.79530201342282e-06,
+      "loss": 0.8755,
+      "step": 3637
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 7.791946308724832e-06,
+      "loss": 0.835,
+      "step": 3638
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 7.788590604026846e-06,
+      "loss": 0.873,
+      "step": 3639
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 7.785234899328859e-06,
+      "loss": 0.8638,
+      "step": 3640
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 7.781879194630873e-06,
+      "loss": 0.8535,
+      "step": 3641
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 7.778523489932886e-06,
+      "loss": 0.8281,
+      "step": 3642
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 7.7751677852349e-06,
+      "loss": 0.8818,
+      "step": 3643
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 7.771812080536914e-06,
+      "loss": 0.8379,
+      "step": 3644
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 7.768456375838927e-06,
+      "loss": 0.8604,
+      "step": 3645
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 7.765100671140941e-06,
+      "loss": 0.876,
+      "step": 3646
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 7.761744966442954e-06,
+      "loss": 0.8462,
+      "step": 3647
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 7.758389261744968e-06,
+      "loss": 0.8335,
+      "step": 3648
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 7.75503355704698e-06,
+      "loss": 0.8232,
+      "step": 3649
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 7.751677852348995e-06,
+      "loss": 0.8887,
+      "step": 3650
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 7.748322147651007e-06,
+      "loss": 0.8481,
+      "step": 3651
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 7.74496644295302e-06,
+      "loss": 0.8481,
+      "step": 3652
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 7.741610738255034e-06,
+      "loss": 0.8384,
+      "step": 3653
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 7.738255033557047e-06,
+      "loss": 0.8848,
+      "step": 3654
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 7.734899328859061e-06,
+      "loss": 0.876,
+      "step": 3655
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 7.731543624161074e-06,
+      "loss": 0.8403,
+      "step": 3656
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 7.728187919463088e-06,
+      "loss": 0.8491,
+      "step": 3657
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 7.7248322147651e-06,
+      "loss": 0.8511,
+      "step": 3658
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 7.721476510067115e-06,
+      "loss": 0.8828,
+      "step": 3659
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 7.718120805369127e-06,
+      "loss": 0.8169,
+      "step": 3660
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 7.714765100671142e-06,
+      "loss": 0.8491,
+      "step": 3661
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 7.711409395973154e-06,
+      "loss": 0.8677,
+      "step": 3662
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 7.708053691275169e-06,
+      "loss": 0.8457,
+      "step": 3663
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 7.704697986577183e-06,
+      "loss": 0.875,
+      "step": 3664
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 7.701342281879195e-06,
+      "loss": 0.8359,
+      "step": 3665
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 7.69798657718121e-06,
+      "loss": 0.856,
+      "step": 3666
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 7.694630872483222e-06,
+      "loss": 0.8521,
+      "step": 3667
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 7.691275167785237e-06,
+      "loss": 0.875,
+      "step": 3668
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 7.687919463087249e-06,
+      "loss": 0.8604,
+      "step": 3669
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 7.684563758389262e-06,
+      "loss": 0.8604,
+      "step": 3670
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 7.681208053691276e-06,
+      "loss": 0.8931,
+      "step": 3671
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 7.677852348993289e-06,
+      "loss": 0.8374,
+      "step": 3672
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 7.674496644295303e-06,
+      "loss": 0.8496,
+      "step": 3673
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 7.671140939597315e-06,
+      "loss": 0.8735,
+      "step": 3674
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 7.66778523489933e-06,
+      "loss": 0.8638,
+      "step": 3675
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 7.664429530201342e-06,
+      "loss": 0.8794,
+      "step": 3676
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 7.661073825503357e-06,
+      "loss": 0.8511,
+      "step": 3677
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 7.657718120805369e-06,
+      "loss": 0.8467,
+      "step": 3678
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 7.654362416107383e-06,
+      "loss": 0.8433,
+      "step": 3679
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 7.651006711409396e-06,
+      "loss": 0.8691,
+      "step": 3680
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 7.64765100671141e-06,
+      "loss": 0.8438,
+      "step": 3681
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 7.644295302013423e-06,
+      "loss": 0.8477,
+      "step": 3682
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 7.640939597315437e-06,
+      "loss": 0.8589,
+      "step": 3683
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 7.637583892617451e-06,
+      "loss": 0.8657,
+      "step": 3684
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 7.634228187919464e-06,
+      "loss": 0.8237,
+      "step": 3685
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 7.630872483221478e-06,
+      "loss": 0.8735,
+      "step": 3686
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 7.627516778523491e-06,
+      "loss": 0.8306,
+      "step": 3687
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 7.624161073825504e-06,
+      "loss": 0.8638,
+      "step": 3688
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 7.620805369127518e-06,
+      "loss": 0.8618,
+      "step": 3689
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 7.61744966442953e-06,
+      "loss": 0.8604,
+      "step": 3690
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 7.6140939597315445e-06,
+      "loss": 0.8755,
+      "step": 3691
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 7.610738255033557e-06,
+      "loss": 0.8379,
+      "step": 3692
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 7.607382550335571e-06,
+      "loss": 0.8706,
+      "step": 3693
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 7.604026845637584e-06,
+      "loss": 0.8774,
+      "step": 3694
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 7.600671140939598e-06,
+      "loss": 0.8555,
+      "step": 3695
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 7.597315436241612e-06,
+      "loss": 0.8667,
+      "step": 3696
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 7.593959731543625e-06,
+      "loss": 0.8735,
+      "step": 3697
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 7.5906040268456385e-06,
+      "loss": 0.853,
+      "step": 3698
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 7.587248322147651e-06,
+      "loss": 0.8584,
+      "step": 3699
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 7.583892617449665e-06,
+      "loss": 0.877,
+      "step": 3700
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 7.580536912751678e-06,
+      "loss": 0.8457,
+      "step": 3701
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 7.577181208053692e-06,
+      "loss": 0.8501,
+      "step": 3702
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 7.573825503355705e-06,
+      "loss": 0.8389,
+      "step": 3703
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 7.570469798657719e-06,
+      "loss": 0.8438,
+      "step": 3704
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 7.567114093959732e-06,
+      "loss": 0.8701,
+      "step": 3705
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 7.563758389261746e-06,
+      "loss": 0.8525,
+      "step": 3706
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 7.560402684563759e-06,
+      "loss": 0.8569,
+      "step": 3707
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 7.557046979865772e-06,
+      "loss": 0.8589,
+      "step": 3708
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 7.553691275167786e-06,
+      "loss": 0.8511,
+      "step": 3709
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 7.550335570469799e-06,
+      "loss": 0.8735,
+      "step": 3710
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 7.546979865771813e-06,
+      "loss": 0.8735,
+      "step": 3711
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 7.543624161073826e-06,
+      "loss": 0.8872,
+      "step": 3712
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 7.54026845637584e-06,
+      "loss": 0.8682,
+      "step": 3713
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 7.5369127516778525e-06,
+      "loss": 0.8691,
+      "step": 3714
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 7.533557046979867e-06,
+      "loss": 0.8726,
+      "step": 3715
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 7.53020134228188e-06,
+      "loss": 0.8638,
+      "step": 3716
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 7.526845637583893e-06,
+      "loss": 0.8608,
+      "step": 3717
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 7.523489932885907e-06,
+      "loss": 0.8237,
+      "step": 3718
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 7.52013422818792e-06,
+      "loss": 0.8545,
+      "step": 3719
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 7.516778523489934e-06,
+      "loss": 0.8506,
+      "step": 3720
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 7.5134228187919465e-06,
+      "loss": 0.8462,
+      "step": 3721
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 7.510067114093961e-06,
+      "loss": 0.8584,
+      "step": 3722
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 7.506711409395973e-06,
+      "loss": 0.8579,
+      "step": 3723
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 7.503355704697988e-06,
+      "loss": 0.8779,
+      "step": 3724
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 7.500000000000001e-06,
+      "loss": 0.8467,
+      "step": 3725
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 7.496644295302014e-06,
+      "loss": 0.8667,
+      "step": 3726
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 7.493288590604028e-06,
+      "loss": 0.8408,
+      "step": 3727
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 7.4899328859060405e-06,
+      "loss": 0.8818,
+      "step": 3728
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 7.486577181208055e-06,
+      "loss": 0.8237,
+      "step": 3729
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 7.483221476510067e-06,
+      "loss": 0.8442,
+      "step": 3730
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 7.479865771812082e-06,
+      "loss": 0.8569,
+      "step": 3731
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 7.476510067114094e-06,
+      "loss": 0.832,
+      "step": 3732
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 7.4731543624161084e-06,
+      "loss": 0.8423,
+      "step": 3733
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 7.469798657718121e-06,
+      "loss": 0.8838,
+      "step": 3734
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 7.4664429530201345e-06,
+      "loss": 0.8218,
+      "step": 3735
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 7.463087248322149e-06,
+      "loss": 0.897,
+      "step": 3736
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 7.459731543624161e-06,
+      "loss": 0.8789,
+      "step": 3737
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 7.456375838926176e-06,
+      "loss": 0.8325,
+      "step": 3738
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 7.453020134228188e-06,
+      "loss": 0.8535,
+      "step": 3739
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 7.4496644295302024e-06,
+      "loss": 0.873,
+      "step": 3740
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 7.446308724832215e-06,
+      "loss": 0.8794,
+      "step": 3741
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 7.442953020134229e-06,
+      "loss": 0.8618,
+      "step": 3742
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 7.439597315436242e-06,
+      "loss": 0.8774,
+      "step": 3743
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 7.436241610738255e-06,
+      "loss": 0.8643,
+      "step": 3744
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 7.4328859060402696e-06,
+      "loss": 0.8833,
+      "step": 3745
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 7.429530201342282e-06,
+      "loss": 0.8384,
+      "step": 3746
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 7.426174496644296e-06,
+      "loss": 0.8809,
+      "step": 3747
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 7.422818791946309e-06,
+      "loss": 0.8711,
+      "step": 3748
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 7.419463087248323e-06,
+      "loss": 0.8047,
+      "step": 3749
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 7.416107382550336e-06,
+      "loss": 0.8433,
+      "step": 3750
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 7.41275167785235e-06,
+      "loss": 0.8691,
+      "step": 3751
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 7.409395973154363e-06,
+      "loss": 0.8765,
+      "step": 3752
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 7.406040268456376e-06,
+      "loss": 0.8823,
+      "step": 3753
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 7.4026845637583896e-06,
+      "loss": 0.8594,
+      "step": 3754
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 7.399328859060403e-06,
+      "loss": 0.8623,
+      "step": 3755
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 7.395973154362417e-06,
+      "loss": 0.8643,
+      "step": 3756
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 7.39261744966443e-06,
+      "loss": 0.8569,
+      "step": 3757
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 7.389261744966444e-06,
+      "loss": 0.8672,
+      "step": 3758
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 7.385906040268457e-06,
+      "loss": 0.8818,
+      "step": 3759
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 7.382550335570471e-06,
+      "loss": 0.8755,
+      "step": 3760
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 7.3791946308724835e-06,
+      "loss": 0.8438,
+      "step": 3761
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 7.375838926174497e-06,
+      "loss": 0.8682,
+      "step": 3762
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 7.37248322147651e-06,
+      "loss": 0.8623,
+      "step": 3763
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 7.369127516778524e-06,
+      "loss": 0.8281,
+      "step": 3764
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 7.365771812080538e-06,
+      "loss": 0.8862,
+      "step": 3765
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 7.362416107382551e-06,
+      "loss": 0.8682,
+      "step": 3766
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 7.359060402684565e-06,
+      "loss": 0.8428,
+      "step": 3767
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 7.3557046979865775e-06,
+      "loss": 0.8784,
+      "step": 3768
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 7.352348993288592e-06,
+      "loss": 0.8813,
+      "step": 3769
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 7.348993288590604e-06,
+      "loss": 0.9038,
+      "step": 3770
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 7.345637583892618e-06,
+      "loss": 0.856,
+      "step": 3771
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 7.342281879194631e-06,
+      "loss": 0.8838,
+      "step": 3772
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 7.338926174496645e-06,
+      "loss": 0.8628,
+      "step": 3773
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 7.335570469798658e-06,
+      "loss": 0.8784,
+      "step": 3774
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 7.3322147651006715e-06,
+      "loss": 0.8701,
+      "step": 3775
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 7.328859060402686e-06,
+      "loss": 0.8579,
+      "step": 3776
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 7.325503355704698e-06,
+      "loss": 0.8462,
+      "step": 3777
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 7.322147651006713e-06,
+      "loss": 0.8667,
+      "step": 3778
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 7.318791946308725e-06,
+      "loss": 0.8877,
+      "step": 3779
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 7.3154362416107395e-06,
+      "loss": 0.8828,
+      "step": 3780
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 7.312080536912752e-06,
+      "loss": 0.8633,
+      "step": 3781
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 7.3087248322147655e-06,
+      "loss": 0.8506,
+      "step": 3782
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 7.305369127516779e-06,
+      "loss": 0.8892,
+      "step": 3783
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 7.302013422818792e-06,
+      "loss": 0.853,
+      "step": 3784
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 7.298657718120807e-06,
+      "loss": 0.8525,
+      "step": 3785
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 7.295302013422819e-06,
+      "loss": 0.8662,
+      "step": 3786
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 7.2919463087248335e-06,
+      "loss": 0.8564,
+      "step": 3787
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 7.288590604026846e-06,
+      "loss": 0.8696,
+      "step": 3788
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 7.28523489932886e-06,
+      "loss": 0.8472,
+      "step": 3789
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 7.281879194630873e-06,
+      "loss": 0.8408,
+      "step": 3790
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 7.278523489932886e-06,
+      "loss": 0.8066,
+      "step": 3791
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 7.2751677852349e-06,
+      "loss": 0.8491,
+      "step": 3792
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 7.271812080536913e-06,
+      "loss": 0.8223,
+      "step": 3793
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 7.2684563758389275e-06,
+      "loss": 0.8452,
+      "step": 3794
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 7.26510067114094e-06,
+      "loss": 0.8628,
+      "step": 3795
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 7.261744966442954e-06,
+      "loss": 0.8691,
+      "step": 3796
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 7.258389261744967e-06,
+      "loss": 0.8843,
+      "step": 3797
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 7.255033557046981e-06,
+      "loss": 0.8848,
+      "step": 3798
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 7.251677852348994e-06,
+      "loss": 0.8774,
+      "step": 3799
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 7.248322147651007e-06,
+      "loss": 0.8457,
+      "step": 3800
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 7.244966442953021e-06,
+      "loss": 0.8638,
+      "step": 3801
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 7.241610738255034e-06,
+      "loss": 0.8599,
+      "step": 3802
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 7.2382550335570475e-06,
+      "loss": 0.8726,
+      "step": 3803
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 7.234899328859061e-06,
+      "loss": 0.8398,
+      "step": 3804
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 7.231543624161075e-06,
+      "loss": 0.8574,
+      "step": 3805
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 7.228187919463088e-06,
+      "loss": 0.854,
+      "step": 3806
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 7.224832214765102e-06,
+      "loss": 0.8462,
+      "step": 3807
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 7.221476510067115e-06,
+      "loss": 0.8638,
+      "step": 3808
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 7.218120805369128e-06,
+      "loss": 0.8462,
+      "step": 3809
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 7.2147651006711414e-06,
+      "loss": 0.8652,
+      "step": 3810
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 7.211409395973155e-06,
+      "loss": 0.8569,
+      "step": 3811
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 7.208053691275168e-06,
+      "loss": 0.8501,
+      "step": 3812
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 7.204697986577182e-06,
+      "loss": 0.8794,
+      "step": 3813
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 7.201342281879196e-06,
+      "loss": 0.8755,
+      "step": 3814
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 7.197986577181209e-06,
+      "loss": 0.8594,
+      "step": 3815
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 7.194630872483223e-06,
+      "loss": 0.856,
+      "step": 3816
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 7.1912751677852354e-06,
+      "loss": 0.8896,
+      "step": 3817
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 7.187919463087248e-06,
+      "loss": 0.8232,
+      "step": 3818
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 7.184563758389262e-06,
+      "loss": 0.8438,
+      "step": 3819
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 7.181208053691276e-06,
+      "loss": 0.8589,
+      "step": 3820
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 7.177852348993289e-06,
+      "loss": 0.8569,
+      "step": 3821
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 7.1744966442953026e-06,
+      "loss": 0.8667,
+      "step": 3822
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 7.171140939597316e-06,
+      "loss": 0.833,
+      "step": 3823
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 7.167785234899329e-06,
+      "loss": 0.8652,
+      "step": 3824
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 7.164429530201344e-06,
+      "loss": 0.8838,
+      "step": 3825
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 7.161073825503356e-06,
+      "loss": 0.8677,
+      "step": 3826
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 7.157718120805369e-06,
+      "loss": 0.8696,
+      "step": 3827
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 7.154362416107383e-06,
+      "loss": 0.8818,
+      "step": 3828
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 7.1510067114093965e-06,
+      "loss": 0.8486,
+      "step": 3829
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 7.14765100671141e-06,
+      "loss": 0.8594,
+      "step": 3830
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 7.144295302013423e-06,
+      "loss": 0.8447,
+      "step": 3831
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 7.140939597315437e-06,
+      "loss": 0.8584,
+      "step": 3832
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 7.13758389261745e-06,
+      "loss": 0.8628,
+      "step": 3833
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 7.1342281879194645e-06,
+      "loss": 0.8599,
+      "step": 3834
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 7.130872483221477e-06,
+      "loss": 0.8379,
+      "step": 3835
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 7.12751677852349e-06,
+      "loss": 0.8486,
+      "step": 3836
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 7.124161073825504e-06,
+      "loss": 0.8315,
+      "step": 3837
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 7.120805369127517e-06,
+      "loss": 0.8491,
+      "step": 3838
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 7.117449664429531e-06,
+      "loss": 0.8608,
+      "step": 3839
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 7.114093959731544e-06,
+      "loss": 0.8618,
+      "step": 3840
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 7.110738255033558e-06,
+      "loss": 0.854,
+      "step": 3841
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 7.107382550335571e-06,
+      "loss": 0.832,
+      "step": 3842
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 7.1040268456375845e-06,
+      "loss": 0.8481,
+      "step": 3843
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 7.100671140939598e-06,
+      "loss": 0.8389,
+      "step": 3844
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 7.0973154362416105e-06,
+      "loss": 0.8706,
+      "step": 3845
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 7.093959731543625e-06,
+      "loss": 0.8589,
+      "step": 3846
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 7.090604026845637e-06,
+      "loss": 0.8447,
+      "step": 3847
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 7.087248322147652e-06,
+      "loss": 0.853,
+      "step": 3848
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 7.083892617449665e-06,
+      "loss": 0.8647,
+      "step": 3849
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 7.0805369127516785e-06,
+      "loss": 0.8169,
+      "step": 3850
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 7.077181208053692e-06,
+      "loss": 0.8311,
+      "step": 3851
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 7.073825503355705e-06,
+      "loss": 0.8374,
+      "step": 3852
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 7.070469798657719e-06,
+      "loss": 0.8472,
+      "step": 3853
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 7.067114093959731e-06,
+      "loss": 0.8228,
+      "step": 3854
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 7.063758389261746e-06,
+      "loss": 0.8701,
+      "step": 3855
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 7.060402684563758e-06,
+      "loss": 0.832,
+      "step": 3856
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 7.0570469798657725e-06,
+      "loss": 0.853,
+      "step": 3857
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 7.053691275167786e-06,
+      "loss": 0.8818,
+      "step": 3858
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 7.050335570469799e-06,
+      "loss": 0.8711,
+      "step": 3859
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 7.046979865771813e-06,
+      "loss": 0.8604,
+      "step": 3860
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 7.043624161073826e-06,
+      "loss": 0.8491,
+      "step": 3861
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 7.04026845637584e-06,
+      "loss": 0.8428,
+      "step": 3862
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 7.036912751677852e-06,
+      "loss": 0.8413,
+      "step": 3863
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 7.0335570469798665e-06,
+      "loss": 0.8618,
+      "step": 3864
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 7.030201342281879e-06,
+      "loss": 0.8481,
+      "step": 3865
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 7.026845637583893e-06,
+      "loss": 0.8604,
+      "step": 3866
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 7.023489932885906e-06,
+      "loss": 0.8267,
+      "step": 3867
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 7.02013422818792e-06,
+      "loss": 0.8174,
+      "step": 3868
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 7.016778523489934e-06,
+      "loss": 0.7974,
+      "step": 3869
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 7.013422818791947e-06,
+      "loss": 0.8496,
+      "step": 3870
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 7.0100671140939605e-06,
+      "loss": 0.835,
+      "step": 3871
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 7.006711409395974e-06,
+      "loss": 0.8545,
+      "step": 3872
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 7.003355704697987e-06,
+      "loss": 0.8584,
+      "step": 3873
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 7e-06,
+      "loss": 0.8623,
+      "step": 3874
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 6.996644295302014e-06,
+      "loss": 0.8735,
+      "step": 3875
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 6.993288590604027e-06,
+      "loss": 0.8345,
+      "step": 3876
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 6.989932885906041e-06,
+      "loss": 0.8765,
+      "step": 3877
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 6.9865771812080544e-06,
+      "loss": 0.8765,
+      "step": 3878
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 6.983221476510068e-06,
+      "loss": 0.8408,
+      "step": 3879
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 6.979865771812081e-06,
+      "loss": 0.8247,
+      "step": 3880
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 6.976510067114095e-06,
+      "loss": 0.8706,
+      "step": 3881
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 6.973154362416108e-06,
+      "loss": 0.8179,
+      "step": 3882
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 6.969798657718121e-06,
+      "loss": 0.8447,
+      "step": 3883
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 6.966442953020135e-06,
+      "loss": 0.8628,
+      "step": 3884
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 6.963087248322148e-06,
+      "loss": 0.8281,
+      "step": 3885
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 6.959731543624162e-06,
+      "loss": 0.8569,
+      "step": 3886
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 6.9563758389261744e-06,
+      "loss": 0.8438,
+      "step": 3887
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 6.953020134228189e-06,
+      "loss": 0.8438,
+      "step": 3888
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 6.949664429530202e-06,
+      "loss": 0.8428,
+      "step": 3889
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 6.9463087248322156e-06,
+      "loss": 0.8662,
+      "step": 3890
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 6.942953020134229e-06,
+      "loss": 0.8716,
+      "step": 3891
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 6.9395973154362416e-06,
+      "loss": 0.8628,
+      "step": 3892
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 6.936241610738256e-06,
+      "loss": 0.8394,
+      "step": 3893
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 6.9328859060402684e-06,
+      "loss": 0.8345,
+      "step": 3894
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 6.929530201342283e-06,
+      "loss": 0.894,
+      "step": 3895
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 6.926174496644295e-06,
+      "loss": 0.8706,
+      "step": 3896
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 6.9228187919463096e-06,
+      "loss": 0.8594,
+      "step": 3897
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 6.919463087248323e-06,
+      "loss": 0.8892,
+      "step": 3898
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 6.916107382550336e-06,
+      "loss": 0.876,
+      "step": 3899
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 6.91275167785235e-06,
+      "loss": 0.8765,
+      "step": 3900
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 6.909395973154362e-06,
+      "loss": 0.8379,
+      "step": 3901
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 6.906040268456377e-06,
+      "loss": 0.8779,
+      "step": 3902
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 6.902684563758389e-06,
+      "loss": 0.8872,
+      "step": 3903
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 6.8993288590604035e-06,
+      "loss": 0.8408,
+      "step": 3904
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 6.895973154362416e-06,
+      "loss": 0.8335,
+      "step": 3905
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 6.89261744966443e-06,
+      "loss": 0.8213,
+      "step": 3906
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 6.889261744966444e-06,
+      "loss": 0.8491,
+      "step": 3907
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 6.885906040268457e-06,
+      "loss": 0.8506,
+      "step": 3908
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 6.882550335570471e-06,
+      "loss": 0.916,
+      "step": 3909
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 6.879194630872483e-06,
+      "loss": 0.8696,
+      "step": 3910
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 6.8758389261744975e-06,
+      "loss": 0.8799,
+      "step": 3911
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 6.87248322147651e-06,
+      "loss": 0.8467,
+      "step": 3912
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 6.869127516778524e-06,
+      "loss": 0.8613,
+      "step": 3913
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 6.865771812080537e-06,
+      "loss": 0.8413,
+      "step": 3914
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 6.862416107382551e-06,
+      "loss": 0.8501,
+      "step": 3915
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 6.859060402684564e-06,
+      "loss": 0.8652,
+      "step": 3916
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 6.855704697986578e-06,
+      "loss": 0.8604,
+      "step": 3917
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 6.8523489932885915e-06,
+      "loss": 0.8345,
+      "step": 3918
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 6.848993288590604e-06,
+      "loss": 0.833,
+      "step": 3919
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 6.845637583892618e-06,
+      "loss": 0.855,
+      "step": 3920
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 6.842281879194631e-06,
+      "loss": 0.8521,
+      "step": 3921
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 6.838926174496645e-06,
+      "loss": 0.8535,
+      "step": 3922
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 6.835570469798658e-06,
+      "loss": 0.8398,
+      "step": 3923
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 6.832214765100672e-06,
+      "loss": 0.8857,
+      "step": 3924
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 6.828859060402685e-06,
+      "loss": 0.8501,
+      "step": 3925
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 6.825503355704699e-06,
+      "loss": 0.8901,
+      "step": 3926
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 6.822147651006712e-06,
+      "loss": 0.8599,
+      "step": 3927
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 6.818791946308725e-06,
+      "loss": 0.8262,
+      "step": 3928
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 6.815436241610739e-06,
+      "loss": 0.8467,
+      "step": 3929
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 6.812080536912752e-06,
+      "loss": 0.8535,
+      "step": 3930
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 6.808724832214766e-06,
+      "loss": 0.8584,
+      "step": 3931
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 6.805369127516779e-06,
+      "loss": 0.875,
+      "step": 3932
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 6.802013422818793e-06,
+      "loss": 0.8491,
+      "step": 3933
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 6.7986577181208055e-06,
+      "loss": 0.8677,
+      "step": 3934
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 6.79530201342282e-06,
+      "loss": 0.8496,
+      "step": 3935
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 6.791946308724832e-06,
+      "loss": 0.8604,
+      "step": 3936
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 6.788590604026846e-06,
+      "loss": 0.856,
+      "step": 3937
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 6.78523489932886e-06,
+      "loss": 0.8521,
+      "step": 3938
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 6.781879194630873e-06,
+      "loss": 0.897,
+      "step": 3939
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 6.778523489932887e-06,
+      "loss": 0.8442,
+      "step": 3940
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 6.7751677852348995e-06,
+      "loss": 0.8262,
+      "step": 3941
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 6.771812080536914e-06,
+      "loss": 0.9097,
+      "step": 3942
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 6.768456375838926e-06,
+      "loss": 0.8462,
+      "step": 3943
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 6.765100671140941e-06,
+      "loss": 0.8447,
+      "step": 3944
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 6.761744966442953e-06,
+      "loss": 0.8638,
+      "step": 3945
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 6.758389261744967e-06,
+      "loss": 0.8867,
+      "step": 3946
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 6.755033557046981e-06,
+      "loss": 0.8604,
+      "step": 3947
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 6.7516778523489935e-06,
+      "loss": 0.8545,
+      "step": 3948
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 6.748322147651008e-06,
+      "loss": 0.894,
+      "step": 3949
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 6.74496644295302e-06,
+      "loss": 0.856,
+      "step": 3950
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 6.741610738255035e-06,
+      "loss": 0.8779,
+      "step": 3951
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 6.738255033557047e-06,
+      "loss": 0.8555,
+      "step": 3952
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 6.7348993288590614e-06,
+      "loss": 0.832,
+      "step": 3953
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 6.731543624161074e-06,
+      "loss": 0.8599,
+      "step": 3954
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 6.7281879194630874e-06,
+      "loss": 0.8516,
+      "step": 3955
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 6.724832214765101e-06,
+      "loss": 0.8193,
+      "step": 3956
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 6.721476510067114e-06,
+      "loss": 0.8628,
+      "step": 3957
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 6.7181208053691286e-06,
+      "loss": 0.8774,
+      "step": 3958
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 6.714765100671141e-06,
+      "loss": 0.833,
+      "step": 3959
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 6.711409395973155e-06,
+      "loss": 0.8818,
+      "step": 3960
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 6.708053691275168e-06,
+      "loss": 0.8628,
+      "step": 3961
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 6.704697986577182e-06,
+      "loss": 0.8457,
+      "step": 3962
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 6.701342281879195e-06,
+      "loss": 0.8398,
+      "step": 3963
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 6.697986577181209e-06,
+      "loss": 0.8271,
+      "step": 3964
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 6.694630872483222e-06,
+      "loss": 0.8525,
+      "step": 3965
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 6.691275167785235e-06,
+      "loss": 0.8594,
+      "step": 3966
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 6.687919463087249e-06,
+      "loss": 0.8281,
+      "step": 3967
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 6.684563758389262e-06,
+      "loss": 0.8584,
+      "step": 3968
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 6.681208053691276e-06,
+      "loss": 0.8257,
+      "step": 3969
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 6.677852348993289e-06,
+      "loss": 0.8818,
+      "step": 3970
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 6.674496644295303e-06,
+      "loss": 0.8384,
+      "step": 3971
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 6.671140939597316e-06,
+      "loss": 0.8491,
+      "step": 3972
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 6.66778523489933e-06,
+      "loss": 0.8721,
+      "step": 3973
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 6.6644295302013425e-06,
+      "loss": 0.856,
+      "step": 3974
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 6.661073825503356e-06,
+      "loss": 0.8677,
+      "step": 3975
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 6.65771812080537e-06,
+      "loss": 0.8584,
+      "step": 3976
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 6.654362416107383e-06,
+      "loss": 0.8354,
+      "step": 3977
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 6.651006711409397e-06,
+      "loss": 0.8516,
+      "step": 3978
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 6.64765100671141e-06,
+      "loss": 0.8608,
+      "step": 3979
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 6.644295302013424e-06,
+      "loss": 0.8853,
+      "step": 3980
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 6.6409395973154365e-06,
+      "loss": 0.8418,
+      "step": 3981
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 6.637583892617451e-06,
+      "loss": 0.8711,
+      "step": 3982
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 6.634228187919463e-06,
+      "loss": 0.8931,
+      "step": 3983
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 6.630872483221477e-06,
+      "loss": 0.8618,
+      "step": 3984
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 6.62751677852349e-06,
+      "loss": 0.8721,
+      "step": 3985
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 6.624161073825504e-06,
+      "loss": 0.8545,
+      "step": 3986
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 6.620805369127518e-06,
+      "loss": 0.8423,
+      "step": 3987
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 6.6174496644295305e-06,
+      "loss": 0.8203,
+      "step": 3988
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 6.614093959731545e-06,
+      "loss": 0.8711,
+      "step": 3989
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 6.610738255033557e-06,
+      "loss": 0.8555,
+      "step": 3990
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 6.607382550335572e-06,
+      "loss": 0.8276,
+      "step": 3991
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 6.604026845637584e-06,
+      "loss": 0.8369,
+      "step": 3992
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 6.600671140939598e-06,
+      "loss": 0.8242,
+      "step": 3993
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 6.597315436241611e-06,
+      "loss": 0.8809,
+      "step": 3994
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 6.5939597315436245e-06,
+      "loss": 0.8184,
+      "step": 3995
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 6.590604026845639e-06,
+      "loss": 0.8721,
+      "step": 3996
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 6.587248322147651e-06,
+      "loss": 0.8628,
+      "step": 3997
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 6.583892617449666e-06,
+      "loss": 0.853,
+      "step": 3998
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 6.580536912751678e-06,
+      "loss": 0.8442,
+      "step": 3999
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 6.5771812080536925e-06,
+      "loss": 0.8555,
+      "step": 4000
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 6.573825503355705e-06,
+      "loss": 0.8516,
+      "step": 4001
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 6.5704697986577185e-06,
+      "loss": 0.8462,
+      "step": 4002
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 6.567114093959732e-06,
+      "loss": 0.8794,
+      "step": 4003
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 6.563758389261745e-06,
+      "loss": 0.8638,
+      "step": 4004
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 6.560402684563759e-06,
+      "loss": 0.8477,
+      "step": 4005
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 6.557046979865772e-06,
+      "loss": 0.8828,
+      "step": 4006
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 6.5536912751677865e-06,
+      "loss": 0.8799,
+      "step": 4007
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 6.550335570469799e-06,
+      "loss": 0.8442,
+      "step": 4008
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 6.546979865771813e-06,
+      "loss": 0.8667,
+      "step": 4009
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 6.543624161073826e-06,
+      "loss": 0.8691,
+      "step": 4010
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 6.540268456375839e-06,
+      "loss": 0.8657,
+      "step": 4011
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 6.536912751677853e-06,
+      "loss": 0.8345,
+      "step": 4012
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 6.533557046979866e-06,
+      "loss": 0.8442,
+      "step": 4013
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 6.53020134228188e-06,
+      "loss": 0.8574,
+      "step": 4014
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 6.526845637583893e-06,
+      "loss": 0.8525,
+      "step": 4015
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 6.523489932885907e-06,
+      "loss": 0.8374,
+      "step": 4016
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 6.52013422818792e-06,
+      "loss": 0.832,
+      "step": 4017
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 6.516778523489934e-06,
+      "loss": 0.855,
+      "step": 4018
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 6.513422818791947e-06,
+      "loss": 0.8403,
+      "step": 4019
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 6.51006711409396e-06,
+      "loss": 0.8555,
+      "step": 4020
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 6.506711409395974e-06,
+      "loss": 0.8496,
+      "step": 4021
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 6.503355704697987e-06,
+      "loss": 0.8525,
+      "step": 4022
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 6.5000000000000004e-06,
+      "loss": 0.8892,
+      "step": 4023
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 6.496644295302014e-06,
+      "loss": 0.8721,
+      "step": 4024
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 6.493288590604027e-06,
+      "loss": 0.8486,
+      "step": 4025
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 6.489932885906041e-06,
+      "loss": 0.8271,
+      "step": 4026
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 6.486577181208055e-06,
+      "loss": 0.8413,
+      "step": 4027
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 6.483221476510068e-06,
+      "loss": 0.8574,
+      "step": 4028
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 6.47986577181208e-06,
+      "loss": 0.8745,
+      "step": 4029
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 6.4765100671140944e-06,
+      "loss": 0.8281,
+      "step": 4030
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 6.473154362416108e-06,
+      "loss": 0.8467,
+      "step": 4031
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 6.469798657718121e-06,
+      "loss": 0.856,
+      "step": 4032
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 6.466442953020135e-06,
+      "loss": 0.8535,
+      "step": 4033
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 6.463087248322148e-06,
+      "loss": 0.8657,
+      "step": 4034
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 6.4597315436241616e-06,
+      "loss": 0.8291,
+      "step": 4035
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 6.456375838926176e-06,
+      "loss": 0.8291,
+      "step": 4036
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 6.453020134228188e-06,
+      "loss": 0.8628,
+      "step": 4037
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 6.449664429530201e-06,
+      "loss": 0.8467,
+      "step": 4038
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 6.446308724832215e-06,
+      "loss": 0.8545,
+      "step": 4039
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 6.442953020134229e-06,
+      "loss": 0.8413,
+      "step": 4040
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 6.439597315436242e-06,
+      "loss": 0.8657,
+      "step": 4041
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 6.4362416107382556e-06,
+      "loss": 0.8486,
+      "step": 4042
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 6.432885906040269e-06,
+      "loss": 0.8379,
+      "step": 4043
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 6.429530201342282e-06,
+      "loss": 0.8394,
+      "step": 4044
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 6.426174496644297e-06,
+      "loss": 0.835,
+      "step": 4045
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 6.422818791946309e-06,
+      "loss": 0.877,
+      "step": 4046
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 6.419463087248322e-06,
+      "loss": 0.8496,
+      "step": 4047
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 6.416107382550336e-06,
+      "loss": 0.8628,
+      "step": 4048
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 6.412751677852349e-06,
+      "loss": 0.8472,
+      "step": 4049
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 6.409395973154363e-06,
+      "loss": 0.8818,
+      "step": 4050
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 6.406040268456376e-06,
+      "loss": 0.8613,
+      "step": 4051
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 6.40268456375839e-06,
+      "loss": 0.8428,
+      "step": 4052
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 6.399328859060403e-06,
+      "loss": 0.8711,
+      "step": 4053
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 6.395973154362417e-06,
+      "loss": 0.8311,
+      "step": 4054
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 6.39261744966443e-06,
+      "loss": 0.8638,
+      "step": 4055
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 6.389261744966444e-06,
+      "loss": 0.853,
+      "step": 4056
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 6.385906040268457e-06,
+      "loss": 0.8906,
+      "step": 4057
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 6.3825503355704695e-06,
+      "loss": 0.8984,
+      "step": 4058
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 6.379194630872484e-06,
+      "loss": 0.8169,
+      "step": 4059
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 6.375838926174497e-06,
+      "loss": 0.8672,
+      "step": 4060
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 6.372483221476511e-06,
+      "loss": 0.8628,
+      "step": 4061
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 6.369127516778524e-06,
+      "loss": 0.855,
+      "step": 4062
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 6.3657718120805375e-06,
+      "loss": 0.8711,
+      "step": 4063
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 6.362416107382551e-06,
+      "loss": 0.8857,
+      "step": 4064
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 6.359060402684565e-06,
+      "loss": 0.8623,
+      "step": 4065
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 6.355704697986578e-06,
+      "loss": 0.8115,
+      "step": 4066
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 6.35234899328859e-06,
+      "loss": 0.8618,
+      "step": 4067
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 6.348993288590605e-06,
+      "loss": 0.8633,
+      "step": 4068
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 6.345637583892617e-06,
+      "loss": 0.8784,
+      "step": 4069
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 6.3422818791946315e-06,
+      "loss": 0.8379,
+      "step": 4070
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.338926174496645e-06,
+      "loss": 0.854,
+      "step": 4071
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.335570469798658e-06,
+      "loss": 0.8462,
+      "step": 4072
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.332214765100672e-06,
+      "loss": 0.7686,
+      "step": 4073
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.328859060402685e-06,
+      "loss": 0.8545,
+      "step": 4074
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.325503355704699e-06,
+      "loss": 0.874,
+      "step": 4075
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.322147651006711e-06,
+      "loss": 0.877,
+      "step": 4076
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.3187919463087255e-06,
+      "loss": 0.8638,
+      "step": 4077
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.315436241610738e-06,
+      "loss": 0.874,
+      "step": 4078
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.312080536912752e-06,
+      "loss": 0.8555,
+      "step": 4079
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.308724832214766e-06,
+      "loss": 0.8579,
+      "step": 4080
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.305369127516779e-06,
+      "loss": 0.874,
+      "step": 4081
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 6.302013422818793e-06,
+      "loss": 0.8633,
+      "step": 4082
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.298657718120806e-06,
+      "loss": 0.8711,
+      "step": 4083
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.2953020134228195e-06,
+      "loss": 0.8657,
+      "step": 4084
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.291946308724832e-06,
+      "loss": 0.8291,
+      "step": 4085
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.288590604026846e-06,
+      "loss": 0.8643,
+      "step": 4086
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.285234899328859e-06,
+      "loss": 0.8579,
+      "step": 4087
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.281879194630873e-06,
+      "loss": 0.8687,
+      "step": 4088
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.278523489932887e-06,
+      "loss": 0.8335,
+      "step": 4089
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.2751677852349e-06,
+      "loss": 0.876,
+      "step": 4090
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.2718120805369135e-06,
+      "loss": 0.8813,
+      "step": 4091
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.268456375838927e-06,
+      "loss": 0.8647,
+      "step": 4092
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.26510067114094e-06,
+      "loss": 0.8687,
+      "step": 4093
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 6.261744966442953e-06,
+      "loss": 0.8652,
+      "step": 4094
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.258389261744967e-06,
+      "loss": 0.8364,
+      "step": 4095
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.25503355704698e-06,
+      "loss": 0.8608,
+      "step": 4096
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.251677852348994e-06,
+      "loss": 0.8589,
+      "step": 4097
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.248322147651007e-06,
+      "loss": 0.8765,
+      "step": 4098
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.244966442953021e-06,
+      "loss": 0.8579,
+      "step": 4099
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.241610738255034e-06,
+      "loss": 0.8877,
+      "step": 4100
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.238255033557048e-06,
+      "loss": 0.8716,
+      "step": 4101
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.234899328859061e-06,
+      "loss": 0.8687,
+      "step": 4102
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.231543624161074e-06,
+      "loss": 0.8608,
+      "step": 4103
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.228187919463088e-06,
+      "loss": 0.877,
+      "step": 4104
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.2248322147651006e-06,
+      "loss": 0.8418,
+      "step": 4105
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 6.221476510067115e-06,
+      "loss": 0.8398,
+      "step": 4106
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.2181208053691274e-06,
+      "loss": 0.8232,
+      "step": 4107
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.214765100671142e-06,
+      "loss": 0.8506,
+      "step": 4108
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.211409395973155e-06,
+      "loss": 0.8252,
+      "step": 4109
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.2080536912751686e-06,
+      "loss": 0.8706,
+      "step": 4110
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.204697986577182e-06,
+      "loss": 0.8545,
+      "step": 4111
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.2013422818791946e-06,
+      "loss": 0.875,
+      "step": 4112
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.197986577181209e-06,
+      "loss": 0.8589,
+      "step": 4113
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.194630872483221e-06,
+      "loss": 0.8931,
+      "step": 4114
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.191275167785236e-06,
+      "loss": 0.8652,
+      "step": 4115
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.187919463087248e-06,
+      "loss": 0.8589,
+      "step": 4116
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.1845637583892625e-06,
+      "loss": 0.8472,
+      "step": 4117
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 6.181208053691275e-06,
+      "loss": 0.8882,
+      "step": 4118
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.177852348993289e-06,
+      "loss": 0.854,
+      "step": 4119
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.174496644295303e-06,
+      "loss": 0.8716,
+      "step": 4120
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.171140939597315e-06,
+      "loss": 0.8853,
+      "step": 4121
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.16778523489933e-06,
+      "loss": 0.8872,
+      "step": 4122
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.164429530201342e-06,
+      "loss": 0.8345,
+      "step": 4123
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.1610738255033565e-06,
+      "loss": 0.8682,
+      "step": 4124
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.157718120805369e-06,
+      "loss": 0.8765,
+      "step": 4125
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.154362416107383e-06,
+      "loss": 0.8452,
+      "step": 4126
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.151006711409396e-06,
+      "loss": 0.8154,
+      "step": 4127
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.14765100671141e-06,
+      "loss": 0.8457,
+      "step": 4128
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.144295302013424e-06,
+      "loss": 0.8403,
+      "step": 4129
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 6.140939597315436e-06,
+      "loss": 0.8438,
+      "step": 4130
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.1375838926174505e-06,
+      "loss": 0.8887,
+      "step": 4131
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.134228187919463e-06,
+      "loss": 0.8379,
+      "step": 4132
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.130872483221477e-06,
+      "loss": 0.8652,
+      "step": 4133
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.12751677852349e-06,
+      "loss": 0.854,
+      "step": 4134
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.124161073825504e-06,
+      "loss": 0.8486,
+      "step": 4135
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.120805369127517e-06,
+      "loss": 0.8481,
+      "step": 4136
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.117449664429531e-06,
+      "loss": 0.8774,
+      "step": 4137
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.114093959731544e-06,
+      "loss": 0.8442,
+      "step": 4138
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.110738255033557e-06,
+      "loss": 0.8496,
+      "step": 4139
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.107382550335571e-06,
+      "loss": 0.8228,
+      "step": 4140
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.104026845637584e-06,
+      "loss": 0.8647,
+      "step": 4141
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 6.100671140939598e-06,
+      "loss": 0.8213,
+      "step": 4142
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.097315436241611e-06,
+      "loss": 0.8765,
+      "step": 4143
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.093959731543625e-06,
+      "loss": 0.8608,
+      "step": 4144
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.090604026845638e-06,
+      "loss": 0.856,
+      "step": 4145
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.087248322147652e-06,
+      "loss": 0.8506,
+      "step": 4146
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.0838926174496645e-06,
+      "loss": 0.8594,
+      "step": 4147
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.080536912751679e-06,
+      "loss": 0.8325,
+      "step": 4148
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.077181208053692e-06,
+      "loss": 0.8555,
+      "step": 4149
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.073825503355705e-06,
+      "loss": 0.835,
+      "step": 4150
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.070469798657719e-06,
+      "loss": 0.8491,
+      "step": 4151
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.067114093959732e-06,
+      "loss": 0.876,
+      "step": 4152
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.063758389261746e-06,
+      "loss": 0.8765,
+      "step": 4153
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 6.0604026845637585e-06,
+      "loss": 0.8599,
+      "step": 4154
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.057046979865773e-06,
+      "loss": 0.8213,
+      "step": 4155
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.053691275167785e-06,
+      "loss": 0.8433,
+      "step": 4156
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.0503355704698e-06,
+      "loss": 0.855,
+      "step": 4157
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.046979865771813e-06,
+      "loss": 0.856,
+      "step": 4158
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.043624161073826e-06,
+      "loss": 0.8584,
+      "step": 4159
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.04026845637584e-06,
+      "loss": 0.8521,
+      "step": 4160
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.0369127516778525e-06,
+      "loss": 0.8315,
+      "step": 4161
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.033557046979867e-06,
+      "loss": 0.8599,
+      "step": 4162
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.030201342281879e-06,
+      "loss": 0.8579,
+      "step": 4163
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.026845637583894e-06,
+      "loss": 0.8477,
+      "step": 4164
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.023489932885906e-06,
+      "loss": 0.8184,
+      "step": 4165
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 6.0201342281879204e-06,
+      "loss": 0.8066,
+      "step": 4166
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.016778523489933e-06,
+      "loss": 0.8516,
+      "step": 4167
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.0134228187919464e-06,
+      "loss": 0.8589,
+      "step": 4168
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.010067114093961e-06,
+      "loss": 0.8569,
+      "step": 4169
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.006711409395973e-06,
+      "loss": 0.8696,
+      "step": 4170
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6.0033557046979876e-06,
+      "loss": 0.835,
+      "step": 4171
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 6e-06,
+      "loss": 0.8359,
+      "step": 4172
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 5.9966442953020144e-06,
+      "loss": 0.8628,
+      "step": 4173
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 5.993288590604027e-06,
+      "loss": 0.8716,
+      "step": 4174
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 5.989932885906041e-06,
+      "loss": 0.8667,
+      "step": 4175
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 5.986577181208054e-06,
+      "loss": 0.8696,
+      "step": 4176
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 5.983221476510067e-06,
+      "loss": 0.8193,
+      "step": 4177
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 5.9798657718120816e-06,
+      "loss": 0.8628,
+      "step": 4178
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 5.976510067114094e-06,
+      "loss": 0.8784,
+      "step": 4179
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 5.973154362416108e-06,
+      "loss": 0.8667,
+      "step": 4180
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 5.969798657718121e-06,
+      "loss": 0.8496,
+      "step": 4181
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 5.966442953020135e-06,
+      "loss": 0.8608,
+      "step": 4182
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 5.963087248322148e-06,
+      "loss": 0.8335,
+      "step": 4183
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 5.959731543624162e-06,
+      "loss": 0.8511,
+      "step": 4184
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 5.956375838926175e-06,
+      "loss": 0.8652,
+      "step": 4185
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 5.953020134228188e-06,
+      "loss": 0.8721,
+      "step": 4186
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 5.9496644295302015e-06,
+      "loss": 0.8345,
+      "step": 4187
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 5.946308724832215e-06,
+      "loss": 0.8892,
+      "step": 4188
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 5.942953020134229e-06,
+      "loss": 0.8628,
+      "step": 4189
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 5.939597315436242e-06,
+      "loss": 0.8315,
+      "step": 4190
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 5.936241610738256e-06,
+      "loss": 0.8325,
+      "step": 4191
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 5.932885906040269e-06,
+      "loss": 0.853,
+      "step": 4192
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 5.929530201342283e-06,
+      "loss": 0.8931,
+      "step": 4193
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 5.9261744966442955e-06,
+      "loss": 0.8809,
+      "step": 4194
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 5.922818791946309e-06,
+      "loss": 0.8438,
+      "step": 4195
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 5.919463087248322e-06,
+      "loss": 0.8667,
+      "step": 4196
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 5.916107382550336e-06,
+      "loss": 0.8398,
+      "step": 4197
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 5.91275167785235e-06,
+      "loss": 0.832,
+      "step": 4198
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 5.909395973154363e-06,
+      "loss": 0.8394,
+      "step": 4199
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 5.906040268456377e-06,
+      "loss": 0.8696,
+      "step": 4200
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 5.9026845637583895e-06,
+      "loss": 0.8398,
+      "step": 4201
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 5.899328859060404e-06,
+      "loss": 0.873,
+      "step": 4202
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 5.895973154362416e-06,
+      "loss": 0.8223,
+      "step": 4203
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 5.89261744966443e-06,
+      "loss": 0.8228,
+      "step": 4204
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 5.889261744966443e-06,
+      "loss": 0.855,
+      "step": 4205
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 5.885906040268457e-06,
+      "loss": 0.8721,
+      "step": 4206
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 5.882550335570471e-06,
+      "loss": 0.8496,
+      "step": 4207
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 5.8791946308724835e-06,
+      "loss": 0.8491,
+      "step": 4208
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 5.875838926174498e-06,
+      "loss": 0.8594,
+      "step": 4209
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 5.87248322147651e-06,
+      "loss": 0.8735,
+      "step": 4210
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 5.869127516778525e-06,
+      "loss": 0.8242,
+      "step": 4211
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 5.865771812080537e-06,
+      "loss": 0.8525,
+      "step": 4212
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 5.862416107382551e-06,
+      "loss": 0.854,
+      "step": 4213
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.859060402684564e-06,
+      "loss": 0.8389,
+      "step": 4214
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.8557046979865775e-06,
+      "loss": 0.854,
+      "step": 4215
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.852348993288591e-06,
+      "loss": 0.8911,
+      "step": 4216
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.848993288590604e-06,
+      "loss": 0.8867,
+      "step": 4217
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.845637583892619e-06,
+      "loss": 0.8604,
+      "step": 4218
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.842281879194631e-06,
+      "loss": 0.8857,
+      "step": 4219
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.8389261744966455e-06,
+      "loss": 0.8682,
+      "step": 4220
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.835570469798658e-06,
+      "loss": 0.835,
+      "step": 4221
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.8322147651006715e-06,
+      "loss": 0.8184,
+      "step": 4222
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.828859060402685e-06,
+      "loss": 0.8462,
+      "step": 4223
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.825503355704698e-06,
+      "loss": 0.8525,
+      "step": 4224
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 5.822147651006712e-06,
+      "loss": 0.8467,
+      "step": 4225
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.818791946308725e-06,
+      "loss": 0.8726,
+      "step": 4226
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.8154362416107395e-06,
+      "loss": 0.8584,
+      "step": 4227
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.812080536912752e-06,
+      "loss": 0.8511,
+      "step": 4228
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.808724832214766e-06,
+      "loss": 0.8496,
+      "step": 4229
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.805369127516779e-06,
+      "loss": 0.8462,
+      "step": 4230
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.8020134228187915e-06,
+      "loss": 0.855,
+      "step": 4231
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.798657718120806e-06,
+      "loss": 0.8745,
+      "step": 4232
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.795302013422819e-06,
+      "loss": 0.8774,
+      "step": 4233
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.791946308724833e-06,
+      "loss": 0.8789,
+      "step": 4234
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.788590604026846e-06,
+      "loss": 0.8169,
+      "step": 4235
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.7852348993288594e-06,
+      "loss": 0.8638,
+      "step": 4236
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 5.781879194630873e-06,
+      "loss": 0.8892,
+      "step": 4237
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.778523489932887e-06,
+      "loss": 0.8794,
+      "step": 4238
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.7751677852349e-06,
+      "loss": 0.8555,
+      "step": 4239
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.771812080536914e-06,
+      "loss": 0.8633,
+      "step": 4240
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.768456375838927e-06,
+      "loss": 0.855,
+      "step": 4241
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.76510067114094e-06,
+      "loss": 0.8613,
+      "step": 4242
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.7617449664429534e-06,
+      "loss": 0.8579,
+      "step": 4243
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.758389261744967e-06,
+      "loss": 0.8276,
+      "step": 4244
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.75503355704698e-06,
+      "loss": 0.8452,
+      "step": 4245
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.751677852348994e-06,
+      "loss": 0.8506,
+      "step": 4246
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.748322147651008e-06,
+      "loss": 0.8472,
+      "step": 4247
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.7449664429530206e-06,
+      "loss": 0.8813,
+      "step": 4248
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 5.741610738255035e-06,
+      "loss": 0.8506,
+      "step": 4249
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.738255033557047e-06,
+      "loss": 0.8618,
+      "step": 4250
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.73489932885906e-06,
+      "loss": 0.8247,
+      "step": 4251
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.731543624161074e-06,
+      "loss": 0.8535,
+      "step": 4252
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.728187919463088e-06,
+      "loss": 0.8389,
+      "step": 4253
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.724832214765101e-06,
+      "loss": 0.8706,
+      "step": 4254
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.7214765100671146e-06,
+      "loss": 0.8262,
+      "step": 4255
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.718120805369128e-06,
+      "loss": 0.8535,
+      "step": 4256
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.714765100671141e-06,
+      "loss": 0.8857,
+      "step": 4257
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.711409395973156e-06,
+      "loss": 0.8408,
+      "step": 4258
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.708053691275168e-06,
+      "loss": 0.8467,
+      "step": 4259
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.704697986577181e-06,
+      "loss": 0.8813,
+      "step": 4260
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 5.701342281879195e-06,
+      "loss": 0.8789,
+      "step": 4261
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.6979865771812085e-06,
+      "loss": 0.8887,
+      "step": 4262
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.694630872483222e-06,
+      "loss": 0.8237,
+      "step": 4263
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.691275167785235e-06,
+      "loss": 0.8589,
+      "step": 4264
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.687919463087249e-06,
+      "loss": 0.873,
+      "step": 4265
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.684563758389262e-06,
+      "loss": 0.8408,
+      "step": 4266
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.6812080536912765e-06,
+      "loss": 0.8506,
+      "step": 4267
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.677852348993289e-06,
+      "loss": 0.8804,
+      "step": 4268
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.674496644295302e-06,
+      "loss": 0.8716,
+      "step": 4269
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.671140939597316e-06,
+      "loss": 0.8813,
+      "step": 4270
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.667785234899329e-06,
+      "loss": 0.8506,
+      "step": 4271
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.664429530201343e-06,
+      "loss": 0.8838,
+      "step": 4272
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 5.661073825503356e-06,
+      "loss": 0.856,
+      "step": 4273
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.65771812080537e-06,
+      "loss": 0.8491,
+      "step": 4274
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.654362416107383e-06,
+      "loss": 0.8506,
+      "step": 4275
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.651006711409397e-06,
+      "loss": 0.8467,
+      "step": 4276
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.64765100671141e-06,
+      "loss": 0.8516,
+      "step": 4277
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.6442953020134225e-06,
+      "loss": 0.8672,
+      "step": 4278
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.640939597315437e-06,
+      "loss": 0.8281,
+      "step": 4279
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.637583892617449e-06,
+      "loss": 0.8608,
+      "step": 4280
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.634228187919464e-06,
+      "loss": 0.8442,
+      "step": 4281
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.630872483221477e-06,
+      "loss": 0.8467,
+      "step": 4282
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.6275167785234905e-06,
+      "loss": 0.8813,
+      "step": 4283
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.624161073825504e-06,
+      "loss": 0.8745,
+      "step": 4284
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 5.620805369127517e-06,
+      "loss": 0.8579,
+      "step": 4285
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.617449664429531e-06,
+      "loss": 0.8398,
+      "step": 4286
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.614093959731543e-06,
+      "loss": 0.8765,
+      "step": 4287
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.610738255033558e-06,
+      "loss": 0.8672,
+      "step": 4288
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.60738255033557e-06,
+      "loss": 0.8306,
+      "step": 4289
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.6040268456375845e-06,
+      "loss": 0.8481,
+      "step": 4290
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.600671140939598e-06,
+      "loss": 0.8535,
+      "step": 4291
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.597315436241611e-06,
+      "loss": 0.853,
+      "step": 4292
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.593959731543625e-06,
+      "loss": 0.8579,
+      "step": 4293
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.590604026845638e-06,
+      "loss": 0.856,
+      "step": 4294
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.587248322147652e-06,
+      "loss": 0.8486,
+      "step": 4295
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.583892617449664e-06,
+      "loss": 0.8564,
+      "step": 4296
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 5.5805369127516785e-06,
+      "loss": 0.8247,
+      "step": 4297
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.577181208053691e-06,
+      "loss": 0.8447,
+      "step": 4298
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.573825503355705e-06,
+      "loss": 0.8589,
+      "step": 4299
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.570469798657718e-06,
+      "loss": 0.8716,
+      "step": 4300
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.567114093959732e-06,
+      "loss": 0.8511,
+      "step": 4301
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.563758389261746e-06,
+      "loss": 0.8716,
+      "step": 4302
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.560402684563759e-06,
+      "loss": 0.8389,
+      "step": 4303
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.5570469798657725e-06,
+      "loss": 0.854,
+      "step": 4304
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.553691275167785e-06,
+      "loss": 0.876,
+      "step": 4305
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.550335570469799e-06,
+      "loss": 0.8301,
+      "step": 4306
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.546979865771812e-06,
+      "loss": 0.835,
+      "step": 4307
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.543624161073826e-06,
+      "loss": 0.8052,
+      "step": 4308
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 5.540268456375839e-06,
+      "loss": 0.8564,
+      "step": 4309
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.536912751677853e-06,
+      "loss": 0.8418,
+      "step": 4310
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.5335570469798664e-06,
+      "loss": 0.8584,
+      "step": 4311
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.53020134228188e-06,
+      "loss": 0.8306,
+      "step": 4312
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.526845637583893e-06,
+      "loss": 0.8584,
+      "step": 4313
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.523489932885906e-06,
+      "loss": 0.8701,
+      "step": 4314
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.52013422818792e-06,
+      "loss": 0.8696,
+      "step": 4315
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.516778523489933e-06,
+      "loss": 0.8711,
+      "step": 4316
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.513422818791947e-06,
+      "loss": 0.8262,
+      "step": 4317
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.51006711409396e-06,
+      "loss": 0.8804,
+      "step": 4318
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.506711409395974e-06,
+      "loss": 0.8394,
+      "step": 4319
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.503355704697987e-06,
+      "loss": 0.8608,
+      "step": 4320
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 5.500000000000001e-06,
+      "loss": 0.8369,
+      "step": 4321
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.496644295302014e-06,
+      "loss": 0.856,
+      "step": 4322
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.493288590604027e-06,
+      "loss": 0.8604,
+      "step": 4323
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.489932885906041e-06,
+      "loss": 0.835,
+      "step": 4324
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.4865771812080536e-06,
+      "loss": 0.8486,
+      "step": 4325
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.483221476510068e-06,
+      "loss": 0.8184,
+      "step": 4326
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.47986577181208e-06,
+      "loss": 0.8521,
+      "step": 4327
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.476510067114095e-06,
+      "loss": 0.8433,
+      "step": 4328
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.473154362416107e-06,
+      "loss": 0.8774,
+      "step": 4329
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.4697986577181215e-06,
+      "loss": 0.8623,
+      "step": 4330
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.466442953020135e-06,
+      "loss": 0.8486,
+      "step": 4331
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 5.463087248322148e-06,
+      "loss": 0.8794,
+      "step": 4332
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.459731543624162e-06,
+      "loss": 0.8462,
+      "step": 4333
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.456375838926174e-06,
+      "loss": 0.875,
+      "step": 4334
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.453020134228189e-06,
+      "loss": 0.8491,
+      "step": 4335
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.449664429530201e-06,
+      "loss": 0.835,
+      "step": 4336
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.4463087248322155e-06,
+      "loss": 0.8384,
+      "step": 4337
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.442953020134228e-06,
+      "loss": 0.8579,
+      "step": 4338
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.439597315436242e-06,
+      "loss": 0.8862,
+      "step": 4339
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.436241610738256e-06,
+      "loss": 0.8643,
+      "step": 4340
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.432885906040269e-06,
+      "loss": 0.8638,
+      "step": 4341
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.429530201342283e-06,
+      "loss": 0.8418,
+      "step": 4342
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.426174496644295e-06,
+      "loss": 0.8433,
+      "step": 4343
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 5.4228187919463095e-06,
+      "loss": 0.8745,
+      "step": 4344
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.419463087248322e-06,
+      "loss": 0.853,
+      "step": 4345
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.416107382550336e-06,
+      "loss": 0.8491,
+      "step": 4346
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.412751677852349e-06,
+      "loss": 0.8618,
+      "step": 4347
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.409395973154363e-06,
+      "loss": 0.8789,
+      "step": 4348
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.406040268456376e-06,
+      "loss": 0.8242,
+      "step": 4349
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.40268456375839e-06,
+      "loss": 0.8711,
+      "step": 4350
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.3993288590604035e-06,
+      "loss": 0.874,
+      "step": 4351
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.395973154362416e-06,
+      "loss": 0.8696,
+      "step": 4352
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.39261744966443e-06,
+      "loss": 0.856,
+      "step": 4353
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.389261744966443e-06,
+      "loss": 0.8706,
+      "step": 4354
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.385906040268457e-06,
+      "loss": 0.8462,
+      "step": 4355
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 5.38255033557047e-06,
+      "loss": 0.8545,
+      "step": 4356
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.379194630872484e-06,
+      "loss": 0.8604,
+      "step": 4357
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.375838926174497e-06,
+      "loss": 0.8345,
+      "step": 4358
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.372483221476511e-06,
+      "loss": 0.8657,
+      "step": 4359
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.369127516778524e-06,
+      "loss": 0.8379,
+      "step": 4360
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.365771812080537e-06,
+      "loss": 0.8638,
+      "step": 4361
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.362416107382551e-06,
+      "loss": 0.8394,
+      "step": 4362
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.359060402684564e-06,
+      "loss": 0.832,
+      "step": 4363
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.355704697986578e-06,
+      "loss": 0.876,
+      "step": 4364
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.352348993288591e-06,
+      "loss": 0.8179,
+      "step": 4365
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.348993288590605e-06,
+      "loss": 0.8398,
+      "step": 4366
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.3456375838926175e-06,
+      "loss": 0.8408,
+      "step": 4367
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 5.342281879194632e-06,
+      "loss": 0.8784,
+      "step": 4368
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.338926174496644e-06,
+      "loss": 0.8511,
+      "step": 4369
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.335570469798658e-06,
+      "loss": 0.8506,
+      "step": 4370
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.332214765100672e-06,
+      "loss": 0.8809,
+      "step": 4371
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.328859060402685e-06,
+      "loss": 0.8594,
+      "step": 4372
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.325503355704699e-06,
+      "loss": 0.8652,
+      "step": 4373
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.3221476510067115e-06,
+      "loss": 0.8662,
+      "step": 4374
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.318791946308726e-06,
+      "loss": 0.8379,
+      "step": 4375
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.315436241610738e-06,
+      "loss": 0.8477,
+      "step": 4376
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.312080536912753e-06,
+      "loss": 0.8662,
+      "step": 4377
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.308724832214765e-06,
+      "loss": 0.8359,
+      "step": 4378
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.305369127516779e-06,
+      "loss": 0.855,
+      "step": 4379
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 5.302013422818793e-06,
+      "loss": 0.8682,
+      "step": 4380
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 5.2986577181208054e-06,
+      "loss": 0.9053,
+      "step": 4381
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 5.29530201342282e-06,
+      "loss": 0.8774,
+      "step": 4382
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 5.291946308724832e-06,
+      "loss": 0.8608,
+      "step": 4383
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 5.2885906040268466e-06,
+      "loss": 0.8584,
+      "step": 4384
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 5.285234899328859e-06,
+      "loss": 0.8286,
+      "step": 4385
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 5.2818791946308734e-06,
+      "loss": 0.8589,
+      "step": 4386
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 5.278523489932886e-06,
+      "loss": 0.8784,
+      "step": 4387
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 5.2751677852348994e-06,
+      "loss": 0.8394,
+      "step": 4388
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 5.271812080536914e-06,
+      "loss": 0.8574,
+      "step": 4389
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 5.268456375838926e-06,
+      "loss": 0.8716,
+      "step": 4390
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 5.2651006711409406e-06,
+      "loss": 0.8516,
+      "step": 4391
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 5.261744966442953e-06,
+      "loss": 0.8677,
+      "step": 4392
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 5.258389261744967e-06,
+      "loss": 0.8564,
+      "step": 4393
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 5.25503355704698e-06,
+      "loss": 0.8613,
+      "step": 4394
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 5.251677852348994e-06,
+      "loss": 0.8462,
+      "step": 4395
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 5.248322147651007e-06,
+      "loss": 0.8794,
+      "step": 4396
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 5.24496644295302e-06,
+      "loss": 0.8667,
+      "step": 4397
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 5.241610738255034e-06,
+      "loss": 0.8398,
+      "step": 4398
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 5.238255033557047e-06,
+      "loss": 0.8633,
+      "step": 4399
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 5.234899328859061e-06,
+      "loss": 0.8408,
+      "step": 4400
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 5.231543624161074e-06,
+      "loss": 0.8911,
+      "step": 4401
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 5.228187919463088e-06,
+      "loss": 0.8564,
+      "step": 4402
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 5.224832214765101e-06,
+      "loss": 0.8315,
+      "step": 4403
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 5.221476510067115e-06,
+      "loss": 0.8955,
+      "step": 4404
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 5.218120805369128e-06,
+      "loss": 0.8779,
+      "step": 4405
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 5.214765100671141e-06,
+      "loss": 0.854,
+      "step": 4406
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 5.2114093959731545e-06,
+      "loss": 0.8369,
+      "step": 4407
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 5.208053691275168e-06,
+      "loss": 0.8286,
+      "step": 4408
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 5.204697986577182e-06,
+      "loss": 0.8511,
+      "step": 4409
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 5.201342281879195e-06,
+      "loss": 0.8608,
+      "step": 4410
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 5.197986577181209e-06,
+      "loss": 0.8345,
+      "step": 4411
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 5.194630872483222e-06,
+      "loss": 0.8413,
+      "step": 4412
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 5.191275167785236e-06,
+      "loss": 0.8374,
+      "step": 4413
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 5.1879194630872485e-06,
+      "loss": 0.8652,
+      "step": 4414
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 5.184563758389262e-06,
+      "loss": 0.8618,
+      "step": 4415
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 5.181208053691275e-06,
+      "loss": 0.8628,
+      "step": 4416
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 5.177852348993289e-06,
+      "loss": 0.8711,
+      "step": 4417
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 5.174496644295302e-06,
+      "loss": 0.873,
+      "step": 4418
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 5.171140939597316e-06,
+      "loss": 0.8696,
+      "step": 4419
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 5.16778523489933e-06,
+      "loss": 0.8838,
+      "step": 4420
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 5.1644295302013425e-06,
+      "loss": 0.8789,
+      "step": 4421
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 5.161073825503357e-06,
+      "loss": 0.8564,
+      "step": 4422
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 5.157718120805369e-06,
+      "loss": 0.8711,
+      "step": 4423
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 5.154362416107384e-06,
+      "loss": 0.8735,
+      "step": 4424
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 5.151006711409396e-06,
+      "loss": 0.8589,
+      "step": 4425
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 5.14765100671141e-06,
+      "loss": 0.8564,
+      "step": 4426
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 5.144295302013423e-06,
+      "loss": 0.8838,
+      "step": 4427
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 5.1409395973154365e-06,
+      "loss": 0.8926,
+      "step": 4428
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 5.137583892617451e-06,
+      "loss": 0.8862,
+      "step": 4429
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 5.134228187919463e-06,
+      "loss": 0.8188,
+      "step": 4430
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 5.130872483221478e-06,
+      "loss": 0.832,
+      "step": 4431
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 5.12751677852349e-06,
+      "loss": 0.8281,
+      "step": 4432
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 5.1241610738255045e-06,
+      "loss": 0.8252,
+      "step": 4433
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 5.120805369127517e-06,
+      "loss": 0.8442,
+      "step": 4434
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 5.1174496644295305e-06,
+      "loss": 0.8115,
+      "step": 4435
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 5.114093959731544e-06,
+      "loss": 0.8711,
+      "step": 4436
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 5.110738255033557e-06,
+      "loss": 0.835,
+      "step": 4437
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 5.107382550335571e-06,
+      "loss": 0.8354,
+      "step": 4438
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 5.104026845637584e-06,
+      "loss": 0.853,
+      "step": 4439
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 5.1006711409395985e-06,
+      "loss": 0.8643,
+      "step": 4440
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 5.097315436241611e-06,
+      "loss": 0.8848,
+      "step": 4441
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 5.093959731543625e-06,
+      "loss": 0.8418,
+      "step": 4442
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 5.090604026845638e-06,
+      "loss": 0.8369,
+      "step": 4443
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 5.087248322147651e-06,
+      "loss": 0.8379,
+      "step": 4444
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 5.083892617449665e-06,
+      "loss": 0.8687,
+      "step": 4445
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 5.080536912751678e-06,
+      "loss": 0.8428,
+      "step": 4446
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 5.077181208053692e-06,
+      "loss": 0.8652,
+      "step": 4447
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 5.073825503355705e-06,
+      "loss": 0.8716,
+      "step": 4448
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 5.070469798657719e-06,
+      "loss": 0.8315,
+      "step": 4449
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 5.067114093959732e-06,
+      "loss": 0.8618,
+      "step": 4450
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 5.063758389261746e-06,
+      "loss": 0.8242,
+      "step": 4451
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 5.060402684563759e-06,
+      "loss": 0.8389,
+      "step": 4452
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 5.057046979865772e-06,
+      "loss": 0.8047,
+      "step": 4453
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 5.053691275167786e-06,
+      "loss": 0.8633,
+      "step": 4454
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 5.050335570469799e-06,
+      "loss": 0.8652,
+      "step": 4455
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 5.0469798657718124e-06,
+      "loss": 0.8569,
+      "step": 4456
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 5.043624161073826e-06,
+      "loss": 0.8862,
+      "step": 4457
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 5.04026845637584e-06,
+      "loss": 0.8467,
+      "step": 4458
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 5.036912751677853e-06,
+      "loss": 0.8638,
+      "step": 4459
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 5.033557046979867e-06,
+      "loss": 0.8774,
+      "step": 4460
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 5.0302013422818796e-06,
+      "loss": 0.8584,
+      "step": 4461
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 5.026845637583892e-06,
+      "loss": 0.8228,
+      "step": 4462
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 5.023489932885906e-06,
+      "loss": 0.8618,
+      "step": 4463
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 5.02013422818792e-06,
+      "loss": 0.8584,
+      "step": 4464
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 5.016778523489933e-06,
+      "loss": 0.8574,
+      "step": 4465
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 5.013422818791947e-06,
+      "loss": 0.8613,
+      "step": 4466
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 5.01006711409396e-06,
+      "loss": 0.8696,
+      "step": 4467
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 5.0067114093959736e-06,
+      "loss": 0.8755,
+      "step": 4468
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 5.003355704697988e-06,
+      "loss": 0.8755,
+      "step": 4469
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 5e-06,
+      "loss": 0.8623,
+      "step": 4470
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.996644295302014e-06,
+      "loss": 0.8389,
+      "step": 4471
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.993288590604027e-06,
+      "loss": 0.8687,
+      "step": 4472
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.989932885906041e-06,
+      "loss": 0.8755,
+      "step": 4473
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.986577181208054e-06,
+      "loss": 0.8828,
+      "step": 4474
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 4.9832214765100675e-06,
+      "loss": 0.8569,
+      "step": 4475
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.979865771812081e-06,
+      "loss": 0.8672,
+      "step": 4476
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.976510067114094e-06,
+      "loss": 0.8179,
+      "step": 4477
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.973154362416108e-06,
+      "loss": 0.8281,
+      "step": 4478
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.969798657718121e-06,
+      "loss": 0.8379,
+      "step": 4479
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.966442953020135e-06,
+      "loss": 0.8848,
+      "step": 4480
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.963087248322148e-06,
+      "loss": 0.8555,
+      "step": 4481
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.9597315436241615e-06,
+      "loss": 0.8379,
+      "step": 4482
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.956375838926175e-06,
+      "loss": 0.8545,
+      "step": 4483
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.953020134228188e-06,
+      "loss": 0.8979,
+      "step": 4484
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.949664429530202e-06,
+      "loss": 0.8906,
+      "step": 4485
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.946308724832215e-06,
+      "loss": 0.8706,
+      "step": 4486
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 4.942953020134229e-06,
+      "loss": 0.8594,
+      "step": 4487
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.939597315436242e-06,
+      "loss": 0.8354,
+      "step": 4488
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.9362416107382555e-06,
+      "loss": 0.833,
+      "step": 4489
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.932885906040269e-06,
+      "loss": 0.8789,
+      "step": 4490
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.929530201342282e-06,
+      "loss": 0.8442,
+      "step": 4491
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.926174496644296e-06,
+      "loss": 0.8516,
+      "step": 4492
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.922818791946309e-06,
+      "loss": 0.8374,
+      "step": 4493
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.919463087248323e-06,
+      "loss": 0.8804,
+      "step": 4494
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.916107382550336e-06,
+      "loss": 0.8281,
+      "step": 4495
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.9127516778523495e-06,
+      "loss": 0.8643,
+      "step": 4496
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.909395973154363e-06,
+      "loss": 0.8267,
+      "step": 4497
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.906040268456376e-06,
+      "loss": 0.8394,
+      "step": 4498
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 4.90268456375839e-06,
+      "loss": 0.8643,
+      "step": 4499
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.899328859060403e-06,
+      "loss": 0.8594,
+      "step": 4500
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.895973154362417e-06,
+      "loss": 0.8828,
+      "step": 4501
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.89261744966443e-06,
+      "loss": 0.8379,
+      "step": 4502
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.8892617449664435e-06,
+      "loss": 0.8574,
+      "step": 4503
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.885906040268457e-06,
+      "loss": 0.8403,
+      "step": 4504
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.88255033557047e-06,
+      "loss": 0.8677,
+      "step": 4505
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.879194630872484e-06,
+      "loss": 0.8599,
+      "step": 4506
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.875838926174497e-06,
+      "loss": 0.8628,
+      "step": 4507
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.872483221476511e-06,
+      "loss": 0.8462,
+      "step": 4508
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.869127516778524e-06,
+      "loss": 0.8491,
+      "step": 4509
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.8657718120805375e-06,
+      "loss": 0.8579,
+      "step": 4510
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 4.862416107382551e-06,
+      "loss": 0.8638,
+      "step": 4511
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.859060402684564e-06,
+      "loss": 0.8623,
+      "step": 4512
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.855704697986578e-06,
+      "loss": 0.8701,
+      "step": 4513
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.852348993288591e-06,
+      "loss": 0.8281,
+      "step": 4514
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.848993288590605e-06,
+      "loss": 0.8462,
+      "step": 4515
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.845637583892618e-06,
+      "loss": 0.8843,
+      "step": 4516
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.8422818791946315e-06,
+      "loss": 0.8765,
+      "step": 4517
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.838926174496645e-06,
+      "loss": 0.8359,
+      "step": 4518
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.835570469798658e-06,
+      "loss": 0.8799,
+      "step": 4519
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.832214765100672e-06,
+      "loss": 0.8765,
+      "step": 4520
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.828859060402684e-06,
+      "loss": 0.8589,
+      "step": 4521
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.825503355704699e-06,
+      "loss": 0.8516,
+      "step": 4522
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 4.822147651006712e-06,
+      "loss": 0.8276,
+      "step": 4523
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.8187919463087254e-06,
+      "loss": 0.8584,
+      "step": 4524
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.815436241610739e-06,
+      "loss": 0.8286,
+      "step": 4525
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.812080536912752e-06,
+      "loss": 0.853,
+      "step": 4526
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.808724832214766e-06,
+      "loss": 0.8647,
+      "step": 4527
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.805369127516779e-06,
+      "loss": 0.8726,
+      "step": 4528
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.8020134228187926e-06,
+      "loss": 0.8667,
+      "step": 4529
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.798657718120805e-06,
+      "loss": 0.853,
+      "step": 4530
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.795302013422819e-06,
+      "loss": 0.8301,
+      "step": 4531
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.791946308724833e-06,
+      "loss": 0.8794,
+      "step": 4532
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.788590604026846e-06,
+      "loss": 0.8481,
+      "step": 4533
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.78523489932886e-06,
+      "loss": 0.8735,
+      "step": 4534
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 4.781879194630873e-06,
+      "loss": 0.8618,
+      "step": 4535
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.7785234899328866e-06,
+      "loss": 0.8364,
+      "step": 4536
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.7751677852349e-06,
+      "loss": 0.8516,
+      "step": 4537
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.771812080536913e-06,
+      "loss": 0.8579,
+      "step": 4538
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.768456375838926e-06,
+      "loss": 0.8506,
+      "step": 4539
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.765100671140939e-06,
+      "loss": 0.8677,
+      "step": 4540
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.761744966442953e-06,
+      "loss": 0.8872,
+      "step": 4541
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.758389261744967e-06,
+      "loss": 0.8511,
+      "step": 4542
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.7550335570469805e-06,
+      "loss": 0.8628,
+      "step": 4543
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.751677852348994e-06,
+      "loss": 0.8276,
+      "step": 4544
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.748322147651007e-06,
+      "loss": 0.8306,
+      "step": 4545
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.744966442953021e-06,
+      "loss": 0.8423,
+      "step": 4546
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 4.741610738255034e-06,
+      "loss": 0.8521,
+      "step": 4547
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.738255033557048e-06,
+      "loss": 0.8701,
+      "step": 4548
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.73489932885906e-06,
+      "loss": 0.8364,
+      "step": 4549
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.731543624161074e-06,
+      "loss": 0.8638,
+      "step": 4550
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.728187919463087e-06,
+      "loss": 0.8555,
+      "step": 4551
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.724832214765101e-06,
+      "loss": 0.8511,
+      "step": 4552
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.721476510067115e-06,
+      "loss": 0.8906,
+      "step": 4553
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.718120805369128e-06,
+      "loss": 0.854,
+      "step": 4554
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.714765100671142e-06,
+      "loss": 0.8389,
+      "step": 4555
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.711409395973155e-06,
+      "loss": 0.8633,
+      "step": 4556
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.7080536912751685e-06,
+      "loss": 0.8511,
+      "step": 4557
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.704697986577181e-06,
+      "loss": 0.8491,
+      "step": 4558
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 4.7013422818791945e-06,
+      "loss": 0.8506,
+      "step": 4559
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 4.697986577181208e-06,
+      "loss": 0.8423,
+      "step": 4560
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 4.694630872483222e-06,
+      "loss": 0.8647,
+      "step": 4561
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 4.691275167785236e-06,
+      "loss": 0.8442,
+      "step": 4562
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 4.687919463087249e-06,
+      "loss": 0.8521,
+      "step": 4563
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 4.6845637583892625e-06,
+      "loss": 0.8657,
+      "step": 4564
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 4.681208053691276e-06,
+      "loss": 0.8472,
+      "step": 4565
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 4.677852348993289e-06,
+      "loss": 0.8364,
+      "step": 4566
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 4.674496644295302e-06,
+      "loss": 0.8633,
+      "step": 4567
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 4.671140939597315e-06,
+      "loss": 0.8604,
+      "step": 4568
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 4.667785234899329e-06,
+      "loss": 0.8745,
+      "step": 4569
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 4.664429530201342e-06,
+      "loss": 0.8389,
+      "step": 4570
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 4.6610738255033565e-06,
+      "loss": 0.896,
+      "step": 4571
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.65771812080537e-06,
+      "loss": 0.856,
+      "step": 4572
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.654362416107383e-06,
+      "loss": 0.855,
+      "step": 4573
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.651006711409397e-06,
+      "loss": 0.8574,
+      "step": 4574
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.64765100671141e-06,
+      "loss": 0.8784,
+      "step": 4575
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.644295302013423e-06,
+      "loss": 0.8486,
+      "step": 4576
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.640939597315436e-06,
+      "loss": 0.8262,
+      "step": 4577
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.63758389261745e-06,
+      "loss": 0.8403,
+      "step": 4578
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.634228187919463e-06,
+      "loss": 0.8833,
+      "step": 4579
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.6308724832214765e-06,
+      "loss": 0.8501,
+      "step": 4580
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.627516778523491e-06,
+      "loss": 0.8428,
+      "step": 4581
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.624161073825504e-06,
+      "loss": 0.8555,
+      "step": 4582
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 4.620805369127518e-06,
+      "loss": 0.8257,
+      "step": 4583
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 4.617449664429531e-06,
+      "loss": 0.8638,
+      "step": 4584
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 4.614093959731544e-06,
+      "loss": 0.8276,
+      "step": 4585
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 4.610738255033557e-06,
+      "loss": 0.855,
+      "step": 4586
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 4.6073825503355705e-06,
+      "loss": 0.8545,
+      "step": 4587
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 4.604026845637584e-06,
+      "loss": 0.8403,
+      "step": 4588
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 4.600671140939597e-06,
+      "loss": 0.8359,
+      "step": 4589
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 4.597315436241611e-06,
+      "loss": 0.8569,
+      "step": 4590
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 4.593959731543625e-06,
+      "loss": 0.8657,
+      "step": 4591
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 4.5906040268456384e-06,
+      "loss": 0.8516,
+      "step": 4592
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 4.587248322147652e-06,
+      "loss": 0.8599,
+      "step": 4593
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 4.583892617449665e-06,
+      "loss": 0.8433,
+      "step": 4594
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 4.580536912751678e-06,
+      "loss": 0.8521,
+      "step": 4595
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 4.577181208053691e-06,
+      "loss": 0.8281,
+      "step": 4596
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 4.573825503355705e-06,
+      "loss": 0.8618,
+      "step": 4597
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 4.570469798657718e-06,
+      "loss": 0.8433,
+      "step": 4598
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 4.567114093959732e-06,
+      "loss": 0.8267,
+      "step": 4599
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 4.563758389261745e-06,
+      "loss": 0.8691,
+      "step": 4600
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 4.560402684563759e-06,
+      "loss": 0.8726,
+      "step": 4601
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 4.557046979865773e-06,
+      "loss": 0.8584,
+      "step": 4602
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 4.553691275167786e-06,
+      "loss": 0.8667,
+      "step": 4603
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 4.550335570469799e-06,
+      "loss": 0.8486,
+      "step": 4604
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 4.546979865771812e-06,
+      "loss": 0.8628,
+      "step": 4605
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 4.5436241610738256e-06,
+      "loss": 0.8447,
+      "step": 4606
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 4.540268456375839e-06,
+      "loss": 0.8608,
+      "step": 4607
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 4.536912751677852e-06,
+      "loss": 0.8203,
+      "step": 4608
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 4.533557046979866e-06,
+      "loss": 0.8374,
+      "step": 4609
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 4.530201342281879e-06,
+      "loss": 0.8652,
+      "step": 4610
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 4.5268456375838935e-06,
+      "loss": 0.8647,
+      "step": 4611
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 4.523489932885907e-06,
+      "loss": 0.8604,
+      "step": 4612
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 4.5201342281879196e-06,
+      "loss": 0.8442,
+      "step": 4613
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 4.516778523489933e-06,
+      "loss": 0.8662,
+      "step": 4614
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 4.513422818791946e-06,
+      "loss": 0.8286,
+      "step": 4615
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 4.51006711409396e-06,
+      "loss": 0.8735,
+      "step": 4616
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 4.506711409395973e-06,
+      "loss": 0.894,
+      "step": 4617
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 4.503355704697987e-06,
+      "loss": 0.8647,
+      "step": 4618
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.5e-06,
+      "loss": 0.8501,
+      "step": 4619
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.4966442953020135e-06,
+      "loss": 0.8345,
+      "step": 4620
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.493288590604028e-06,
+      "loss": 0.8599,
+      "step": 4621
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.48993288590604e-06,
+      "loss": 0.8765,
+      "step": 4622
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.486577181208054e-06,
+      "loss": 0.8604,
+      "step": 4623
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.483221476510067e-06,
+      "loss": 0.9009,
+      "step": 4624
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.479865771812081e-06,
+      "loss": 0.8486,
+      "step": 4625
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.476510067114094e-06,
+      "loss": 0.8687,
+      "step": 4626
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.4731543624161075e-06,
+      "loss": 0.8828,
+      "step": 4627
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.469798657718121e-06,
+      "loss": 0.8818,
+      "step": 4628
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.466442953020134e-06,
+      "loss": 0.897,
+      "step": 4629
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 4.463087248322149e-06,
+      "loss": 0.8174,
+      "step": 4630
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 4.459731543624161e-06,
+      "loss": 0.8774,
+      "step": 4631
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 4.456375838926175e-06,
+      "loss": 0.8623,
+      "step": 4632
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 4.453020134228188e-06,
+      "loss": 0.8833,
+      "step": 4633
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 4.4496644295302015e-06,
+      "loss": 0.854,
+      "step": 4634
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 4.446308724832215e-06,
+      "loss": 0.8408,
+      "step": 4635
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 4.442953020134228e-06,
+      "loss": 0.8638,
+      "step": 4636
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 4.439597315436242e-06,
+      "loss": 0.8271,
+      "step": 4637
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 4.436241610738255e-06,
+      "loss": 0.8638,
+      "step": 4638
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 4.432885906040269e-06,
+      "loss": 0.8364,
+      "step": 4639
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 4.429530201342283e-06,
+      "loss": 0.8291,
+      "step": 4640
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 4.4261744966442955e-06,
+      "loss": 0.8638,
+      "step": 4641
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 4.422818791946309e-06,
+      "loss": 0.8657,
+      "step": 4642
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 4.419463087248322e-06,
+      "loss": 0.8589,
+      "step": 4643
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 4.416107382550336e-06,
+      "loss": 0.8066,
+      "step": 4644
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 4.412751677852349e-06,
+      "loss": 0.8711,
+      "step": 4645
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 4.409395973154363e-06,
+      "loss": 0.8481,
+      "step": 4646
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 4.406040268456376e-06,
+      "loss": 0.8667,
+      "step": 4647
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 4.4026845637583895e-06,
+      "loss": 0.8394,
+      "step": 4648
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 4.399328859060403e-06,
+      "loss": 0.8794,
+      "step": 4649
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 4.395973154362416e-06,
+      "loss": 0.8369,
+      "step": 4650
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 4.39261744966443e-06,
+      "loss": 0.8794,
+      "step": 4651
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 4.389261744966443e-06,
+      "loss": 0.8286,
+      "step": 4652
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 4.385906040268457e-06,
+      "loss": 0.8647,
+      "step": 4653
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 4.38255033557047e-06,
+      "loss": 0.856,
+      "step": 4654
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 4.3791946308724835e-06,
+      "loss": 0.876,
+      "step": 4655
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 4.375838926174497e-06,
+      "loss": 0.8267,
+      "step": 4656
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 4.37248322147651e-06,
+      "loss": 0.8306,
+      "step": 4657
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 4.369127516778524e-06,
+      "loss": 0.8599,
+      "step": 4658
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 4.365771812080537e-06,
+      "loss": 0.8223,
+      "step": 4659
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 4.362416107382551e-06,
+      "loss": 0.834,
+      "step": 4660
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 4.359060402684564e-06,
+      "loss": 0.8525,
+      "step": 4661
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 4.3557046979865775e-06,
+      "loss": 0.8696,
+      "step": 4662
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 4.352348993288591e-06,
+      "loss": 0.8413,
+      "step": 4663
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 4.348993288590604e-06,
+      "loss": 0.895,
+      "step": 4664
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 4.345637583892618e-06,
+      "loss": 0.8584,
+      "step": 4665
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 4.342281879194631e-06,
+      "loss": 0.8892,
+      "step": 4666
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 4.338926174496645e-06,
+      "loss": 0.8555,
+      "step": 4667
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 4.335570469798658e-06,
+      "loss": 0.8325,
+      "step": 4668
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 4.3322147651006714e-06,
+      "loss": 0.8271,
+      "step": 4669
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 4.328859060402685e-06,
+      "loss": 0.8354,
+      "step": 4670
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 4.325503355704698e-06,
+      "loss": 0.875,
+      "step": 4671
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 4.322147651006712e-06,
+      "loss": 0.8379,
+      "step": 4672
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 4.318791946308725e-06,
+      "loss": 0.8804,
+      "step": 4673
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 4.3154362416107386e-06,
+      "loss": 0.8408,
+      "step": 4674
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 4.312080536912752e-06,
+      "loss": 0.8696,
+      "step": 4675
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 4.3087248322147654e-06,
+      "loss": 0.8271,
+      "step": 4676
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 4.305369127516779e-06,
+      "loss": 0.835,
+      "step": 4677
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 4.302013422818792e-06,
+      "loss": 0.8516,
+      "step": 4678
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 4.298657718120806e-06,
+      "loss": 0.8599,
+      "step": 4679
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 4.295302013422819e-06,
+      "loss": 0.8467,
+      "step": 4680
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 4.2919463087248326e-06,
+      "loss": 0.8589,
+      "step": 4681
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 4.288590604026846e-06,
+      "loss": 0.8589,
+      "step": 4682
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 4.285234899328859e-06,
+      "loss": 0.8257,
+      "step": 4683
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 4.281879194630873e-06,
+      "loss": 0.8379,
+      "step": 4684
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 4.278523489932886e-06,
+      "loss": 0.8379,
+      "step": 4685
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 4.2751677852349e-06,
+      "loss": 0.8687,
+      "step": 4686
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 4.271812080536913e-06,
+      "loss": 0.8433,
+      "step": 4687
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 4.2684563758389265e-06,
+      "loss": 0.8755,
+      "step": 4688
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 4.26510067114094e-06,
+      "loss": 0.8447,
+      "step": 4689
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 4.261744966442953e-06,
+      "loss": 0.8511,
+      "step": 4690
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 4.258389261744967e-06,
+      "loss": 0.873,
+      "step": 4691
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 4.25503355704698e-06,
+      "loss": 0.8564,
+      "step": 4692
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 4.251677852348994e-06,
+      "loss": 0.8647,
+      "step": 4693
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 4.248322147651007e-06,
+      "loss": 0.8594,
+      "step": 4694
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 4.2449664429530205e-06,
+      "loss": 0.8413,
+      "step": 4695
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 4.241610738255034e-06,
+      "loss": 0.8403,
+      "step": 4696
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 4.238255033557047e-06,
+      "loss": 0.8701,
+      "step": 4697
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 4.234899328859061e-06,
+      "loss": 0.8599,
+      "step": 4698
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 4.231543624161074e-06,
+      "loss": 0.8853,
+      "step": 4699
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 4.228187919463088e-06,
+      "loss": 0.8423,
+      "step": 4700
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 4.224832214765101e-06,
+      "loss": 0.8188,
+      "step": 4701
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 4.2214765100671145e-06,
+      "loss": 0.8657,
+      "step": 4702
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 4.218120805369128e-06,
+      "loss": 0.8672,
+      "step": 4703
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 4.214765100671141e-06,
+      "loss": 0.856,
+      "step": 4704
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 4.211409395973155e-06,
+      "loss": 0.8374,
+      "step": 4705
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 4.208053691275168e-06,
+      "loss": 0.8569,
+      "step": 4706
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 4.204697986577182e-06,
+      "loss": 0.8105,
+      "step": 4707
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 4.201342281879195e-06,
+      "loss": 0.8223,
+      "step": 4708
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 4.1979865771812085e-06,
+      "loss": 0.8545,
+      "step": 4709
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 4.194630872483222e-06,
+      "loss": 0.8237,
+      "step": 4710
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 4.191275167785235e-06,
+      "loss": 0.8765,
+      "step": 4711
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 4.187919463087249e-06,
+      "loss": 0.854,
+      "step": 4712
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 4.184563758389262e-06,
+      "loss": 0.8721,
+      "step": 4713
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 4.181208053691276e-06,
+      "loss": 0.8608,
+      "step": 4714
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.177852348993289e-06,
+      "loss": 0.811,
+      "step": 4715
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.1744966442953025e-06,
+      "loss": 0.8491,
+      "step": 4716
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.171140939597316e-06,
+      "loss": 0.8423,
+      "step": 4717
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.167785234899329e-06,
+      "loss": 0.8506,
+      "step": 4718
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.164429530201343e-06,
+      "loss": 0.8599,
+      "step": 4719
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.161073825503356e-06,
+      "loss": 0.8442,
+      "step": 4720
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.15771812080537e-06,
+      "loss": 0.8486,
+      "step": 4721
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.154362416107383e-06,
+      "loss": 0.8579,
+      "step": 4722
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.151006711409396e-06,
+      "loss": 0.8726,
+      "step": 4723
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.14765100671141e-06,
+      "loss": 0.8545,
+      "step": 4724
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.144295302013423e-06,
+      "loss": 0.8701,
+      "step": 4725
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 4.140939597315437e-06,
+      "loss": 0.8682,
+      "step": 4726
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.13758389261745e-06,
+      "loss": 0.8574,
+      "step": 4727
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.134228187919464e-06,
+      "loss": 0.8296,
+      "step": 4728
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.130872483221477e-06,
+      "loss": 0.8628,
+      "step": 4729
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.1275167785234905e-06,
+      "loss": 0.853,
+      "step": 4730
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.124161073825504e-06,
+      "loss": 0.8569,
+      "step": 4731
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.120805369127517e-06,
+      "loss": 0.832,
+      "step": 4732
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.11744966442953e-06,
+      "loss": 0.8403,
+      "step": 4733
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.114093959731544e-06,
+      "loss": 0.8604,
+      "step": 4734
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.110738255033558e-06,
+      "loss": 0.856,
+      "step": 4735
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.107382550335571e-06,
+      "loss": 0.8345,
+      "step": 4736
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.1040268456375844e-06,
+      "loss": 0.8545,
+      "step": 4737
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 4.100671140939598e-06,
+      "loss": 0.8735,
+      "step": 4738
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 4.097315436241611e-06,
+      "loss": 0.8687,
+      "step": 4739
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 4.093959731543625e-06,
+      "loss": 0.854,
+      "step": 4740
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 4.090604026845638e-06,
+      "loss": 0.8535,
+      "step": 4741
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 4.087248322147651e-06,
+      "loss": 0.8735,
+      "step": 4742
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 4.083892617449665e-06,
+      "loss": 0.856,
+      "step": 4743
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 4.0805369127516784e-06,
+      "loss": 0.8682,
+      "step": 4744
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 4.077181208053692e-06,
+      "loss": 0.8481,
+      "step": 4745
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 4.073825503355705e-06,
+      "loss": 0.8149,
+      "step": 4746
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 4.070469798657719e-06,
+      "loss": 0.8633,
+      "step": 4747
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 4.067114093959732e-06,
+      "loss": 0.8726,
+      "step": 4748
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 4.0637583892617456e-06,
+      "loss": 0.8501,
+      "step": 4749
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 4.060402684563759e-06,
+      "loss": 0.8779,
+      "step": 4750
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 4.0570469798657716e-06,
+      "loss": 0.8433,
+      "step": 4751
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 4.053691275167785e-06,
+      "loss": 0.8276,
+      "step": 4752
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 4.050335570469799e-06,
+      "loss": 0.8774,
+      "step": 4753
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 4.046979865771813e-06,
+      "loss": 0.874,
+      "step": 4754
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 4.043624161073826e-06,
+      "loss": 0.8438,
+      "step": 4755
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 4.0402684563758395e-06,
+      "loss": 0.8281,
+      "step": 4756
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 4.036912751677853e-06,
+      "loss": 0.894,
+      "step": 4757
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 4.033557046979866e-06,
+      "loss": 0.8438,
+      "step": 4758
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 4.03020134228188e-06,
+      "loss": 0.8579,
+      "step": 4759
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 4.026845637583892e-06,
+      "loss": 0.8384,
+      "step": 4760
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 4.023489932885906e-06,
+      "loss": 0.8618,
+      "step": 4761
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 4.020134228187919e-06,
+      "loss": 0.8569,
+      "step": 4762
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 4.0167785234899335e-06,
+      "loss": 0.8354,
+      "step": 4763
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 4.013422818791947e-06,
+      "loss": 0.8833,
+      "step": 4764
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 4.01006711409396e-06,
+      "loss": 0.8569,
+      "step": 4765
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 4.006711409395974e-06,
+      "loss": 0.8701,
+      "step": 4766
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 4.003355704697987e-06,
+      "loss": 0.8677,
+      "step": 4767
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 0.8779,
+      "step": 4768
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.996644295302013e-06,
+      "loss": 0.8052,
+      "step": 4769
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.993288590604027e-06,
+      "loss": 0.8354,
+      "step": 4770
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.98993288590604e-06,
+      "loss": 0.8472,
+      "step": 4771
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.9865771812080535e-06,
+      "loss": 0.8271,
+      "step": 4772
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 3.983221476510068e-06,
+      "loss": 0.8052,
+      "step": 4773
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 3.979865771812081e-06,
+      "loss": 0.8354,
+      "step": 4774
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 3.976510067114095e-06,
+      "loss": 0.834,
+      "step": 4775
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 3.973154362416108e-06,
+      "loss": 0.7979,
+      "step": 4776
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 3.9697986577181215e-06,
+      "loss": 0.7798,
+      "step": 4777
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 3.966442953020135e-06,
+      "loss": 0.8076,
+      "step": 4778
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 3.9630872483221475e-06,
+      "loss": 0.8193,
+      "step": 4779
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 3.959731543624161e-06,
+      "loss": 0.8037,
+      "step": 4780
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 3.956375838926174e-06,
+      "loss": 0.8447,
+      "step": 4781
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 3.953020134228188e-06,
+      "loss": 0.8472,
+      "step": 4782
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 3.949664429530202e-06,
+      "loss": 0.8066,
+      "step": 4783
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 3.9463087248322155e-06,
+      "loss": 0.834,
+      "step": 4784
+    },
+    {
+      "epoch": 4.01,
+      "learning_rate": 3.942953020134229e-06,
+      "loss": 0.8535,
+      "step": 4785
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 3.939597315436242e-06,
+      "loss": 0.8555,
+      "step": 4786
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 3.936241610738256e-06,
+      "loss": 0.8047,
+      "step": 4787
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 3.932885906040268e-06,
+      "loss": 0.8442,
+      "step": 4788
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 3.929530201342282e-06,
+      "loss": 0.855,
+      "step": 4789
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 3.926174496644295e-06,
+      "loss": 0.8311,
+      "step": 4790
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 3.922818791946309e-06,
+      "loss": 0.7954,
+      "step": 4791
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 3.919463087248322e-06,
+      "loss": 0.8379,
+      "step": 4792
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 3.916107382550336e-06,
+      "loss": 0.832,
+      "step": 4793
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 3.91275167785235e-06,
+      "loss": 0.8335,
+      "step": 4794
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 3.909395973154363e-06,
+      "loss": 0.8452,
+      "step": 4795
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 3.906040268456377e-06,
+      "loss": 0.8003,
+      "step": 4796
+    },
+    {
+      "epoch": 4.02,
+      "learning_rate": 3.902684563758389e-06,
+      "loss": 0.8149,
+      "step": 4797
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 3.899328859060403e-06,
+      "loss": 0.8296,
+      "step": 4798
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 3.895973154362416e-06,
+      "loss": 0.8003,
+      "step": 4799
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 3.8926174496644295e-06,
+      "loss": 0.8071,
+      "step": 4800
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 3.889261744966443e-06,
+      "loss": 0.8271,
+      "step": 4801
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 3.885906040268457e-06,
+      "loss": 0.8145,
+      "step": 4802
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 3.882550335570471e-06,
+      "loss": 0.8345,
+      "step": 4803
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 3.879194630872484e-06,
+      "loss": 0.812,
+      "step": 4804
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 3.8758389261744974e-06,
+      "loss": 0.8228,
+      "step": 4805
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 3.87248322147651e-06,
+      "loss": 0.8218,
+      "step": 4806
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 3.8691275167785235e-06,
+      "loss": 0.8022,
+      "step": 4807
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 3.865771812080537e-06,
+      "loss": 0.811,
+      "step": 4808
+    },
+    {
+      "epoch": 4.03,
+      "learning_rate": 3.86241610738255e-06,
+      "loss": 0.8286,
+      "step": 4809
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 3.859060402684564e-06,
+      "loss": 0.8091,
+      "step": 4810
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 3.855704697986577e-06,
+      "loss": 0.8389,
+      "step": 4811
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 3.8523489932885914e-06,
+      "loss": 0.8452,
+      "step": 4812
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 3.848993288590605e-06,
+      "loss": 0.8052,
+      "step": 4813
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 3.845637583892618e-06,
+      "loss": 0.8242,
+      "step": 4814
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 3.842281879194631e-06,
+      "loss": 0.8188,
+      "step": 4815
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 3.838926174496644e-06,
+      "loss": 0.8398,
+      "step": 4816
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 3.835570469798658e-06,
+      "loss": 0.8579,
+      "step": 4817
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 3.832214765100671e-06,
+      "loss": 0.8345,
+      "step": 4818
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 3.8288590604026846e-06,
+      "loss": 0.8257,
+      "step": 4819
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 3.825503355704698e-06,
+      "loss": 0.8638,
+      "step": 4820
+    },
+    {
+      "epoch": 4.04,
+      "learning_rate": 3.8221476510067114e-06,
+      "loss": 0.8354,
+      "step": 4821
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 3.818791946308726e-06,
+      "loss": 0.8154,
+      "step": 4822
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 3.815436241610739e-06,
+      "loss": 0.8232,
+      "step": 4823
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 3.812080536912752e-06,
+      "loss": 0.8516,
+      "step": 4824
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 3.808724832214765e-06,
+      "loss": 0.8052,
+      "step": 4825
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 3.8053691275167786e-06,
+      "loss": 0.8394,
+      "step": 4826
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 3.802013422818792e-06,
+      "loss": 0.8115,
+      "step": 4827
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 3.798657718120806e-06,
+      "loss": 0.8335,
+      "step": 4828
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 3.7953020134228193e-06,
+      "loss": 0.8105,
+      "step": 4829
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 3.7919463087248327e-06,
+      "loss": 0.8276,
+      "step": 4830
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 3.788590604026846e-06,
+      "loss": 0.8164,
+      "step": 4831
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 3.7852348993288595e-06,
+      "loss": 0.8071,
+      "step": 4832
+    },
+    {
+      "epoch": 4.05,
+      "learning_rate": 3.781879194630873e-06,
+      "loss": 0.8315,
+      "step": 4833
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 3.778523489932886e-06,
+      "loss": 0.8433,
+      "step": 4834
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 3.7751677852348994e-06,
+      "loss": 0.8232,
+      "step": 4835
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 3.771812080536913e-06,
+      "loss": 0.8267,
+      "step": 4836
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 3.7684563758389262e-06,
+      "loss": 0.8145,
+      "step": 4837
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 3.76510067114094e-06,
+      "loss": 0.8096,
+      "step": 4838
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 3.7617449664429535e-06,
+      "loss": 0.8354,
+      "step": 4839
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 3.758389261744967e-06,
+      "loss": 0.812,
+      "step": 4840
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 3.7550335570469804e-06,
+      "loss": 0.811,
+      "step": 4841
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 3.751677852348994e-06,
+      "loss": 0.8369,
+      "step": 4842
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 3.748322147651007e-06,
+      "loss": 0.8335,
+      "step": 4843
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 3.7449664429530202e-06,
+      "loss": 0.8291,
+      "step": 4844
+    },
+    {
+      "epoch": 4.06,
+      "learning_rate": 3.7416107382550337e-06,
+      "loss": 0.811,
+      "step": 4845
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 3.738255033557047e-06,
+      "loss": 0.856,
+      "step": 4846
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 3.7348993288590605e-06,
+      "loss": 0.8345,
+      "step": 4847
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 3.7315436241610744e-06,
+      "loss": 0.8242,
+      "step": 4848
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 3.728187919463088e-06,
+      "loss": 0.8213,
+      "step": 4849
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 3.7248322147651012e-06,
+      "loss": 0.8203,
+      "step": 4850
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 3.7214765100671146e-06,
+      "loss": 0.812,
+      "step": 4851
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 3.7181208053691276e-06,
+      "loss": 0.8159,
+      "step": 4852
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 3.714765100671141e-06,
+      "loss": 0.7988,
+      "step": 4853
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 3.7114093959731545e-06,
+      "loss": 0.8306,
+      "step": 4854
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 3.708053691275168e-06,
+      "loss": 0.8096,
+      "step": 4855
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 3.7046979865771814e-06,
+      "loss": 0.854,
+      "step": 4856
+    },
+    {
+      "epoch": 4.07,
+      "learning_rate": 3.7013422818791948e-06,
+      "loss": 0.8081,
+      "step": 4857
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 3.6979865771812086e-06,
+      "loss": 0.8091,
+      "step": 4858
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 3.694630872483222e-06,
+      "loss": 0.8042,
+      "step": 4859
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 3.6912751677852355e-06,
+      "loss": 0.791,
+      "step": 4860
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 3.6879194630872485e-06,
+      "loss": 0.8384,
+      "step": 4861
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 3.684563758389262e-06,
+      "loss": 0.8242,
+      "step": 4862
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 3.6812080536912753e-06,
+      "loss": 0.8213,
+      "step": 4863
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 3.6778523489932888e-06,
+      "loss": 0.8296,
+      "step": 4864
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 3.674496644295302e-06,
+      "loss": 0.8364,
+      "step": 4865
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 3.6711409395973156e-06,
+      "loss": 0.8213,
+      "step": 4866
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 3.667785234899329e-06,
+      "loss": 0.8198,
+      "step": 4867
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 3.664429530201343e-06,
+      "loss": 0.8315,
+      "step": 4868
+    },
+    {
+      "epoch": 4.08,
+      "learning_rate": 3.6610738255033563e-06,
+      "loss": 0.7974,
+      "step": 4869
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 3.6577181208053697e-06,
+      "loss": 0.833,
+      "step": 4870
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 3.6543624161073828e-06,
+      "loss": 0.8389,
+      "step": 4871
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 3.651006711409396e-06,
+      "loss": 0.8423,
+      "step": 4872
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 3.6476510067114096e-06,
+      "loss": 0.8237,
+      "step": 4873
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 3.644295302013423e-06,
+      "loss": 0.8066,
+      "step": 4874
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 3.6409395973154365e-06,
+      "loss": 0.7944,
+      "step": 4875
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 3.63758389261745e-06,
+      "loss": 0.8623,
+      "step": 4876
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 3.6342281879194637e-06,
+      "loss": 0.8438,
+      "step": 4877
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 3.630872483221477e-06,
+      "loss": 0.8291,
+      "step": 4878
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 3.6275167785234906e-06,
+      "loss": 0.8062,
+      "step": 4879
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 3.6241610738255036e-06,
+      "loss": 0.8076,
+      "step": 4880
+    },
+    {
+      "epoch": 4.09,
+      "learning_rate": 3.620805369127517e-06,
+      "loss": 0.8237,
+      "step": 4881
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 3.6174496644295304e-06,
+      "loss": 0.8369,
+      "step": 4882
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 3.614093959731544e-06,
+      "loss": 0.8354,
+      "step": 4883
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 3.6107382550335573e-06,
+      "loss": 0.8223,
+      "step": 4884
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 3.6073825503355707e-06,
+      "loss": 0.8018,
+      "step": 4885
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 3.604026845637584e-06,
+      "loss": 0.8184,
+      "step": 4886
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 3.600671140939598e-06,
+      "loss": 0.8193,
+      "step": 4887
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 3.5973154362416114e-06,
+      "loss": 0.8203,
+      "step": 4888
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 3.593959731543624e-06,
+      "loss": 0.8447,
+      "step": 4889
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 3.590604026845638e-06,
+      "loss": 0.7798,
+      "step": 4890
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 3.5872483221476513e-06,
+      "loss": 0.7993,
+      "step": 4891
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 3.5838926174496647e-06,
+      "loss": 0.8125,
+      "step": 4892
+    },
+    {
+      "epoch": 4.1,
+      "learning_rate": 3.580536912751678e-06,
+      "loss": 0.8364,
+      "step": 4893
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 3.5771812080536916e-06,
+      "loss": 0.8179,
+      "step": 4894
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 3.573825503355705e-06,
+      "loss": 0.8086,
+      "step": 4895
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 3.5704697986577184e-06,
+      "loss": 0.8076,
+      "step": 4896
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 3.5671140939597323e-06,
+      "loss": 0.8271,
+      "step": 4897
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 3.563758389261745e-06,
+      "loss": 0.834,
+      "step": 4898
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 3.5604026845637587e-06,
+      "loss": 0.8262,
+      "step": 4899
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 3.557046979865772e-06,
+      "loss": 0.8125,
+      "step": 4900
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 3.5536912751677855e-06,
+      "loss": 0.8301,
+      "step": 4901
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 3.550335570469799e-06,
+      "loss": 0.8447,
+      "step": 4902
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 3.5469798657718124e-06,
+      "loss": 0.8301,
+      "step": 4903
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 3.543624161073826e-06,
+      "loss": 0.8608,
+      "step": 4904
+    },
+    {
+      "epoch": 4.11,
+      "learning_rate": 3.5402684563758393e-06,
+      "loss": 0.7959,
+      "step": 4905
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.5369127516778527e-06,
+      "loss": 0.8379,
+      "step": 4906
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.5335570469798657e-06,
+      "loss": 0.8213,
+      "step": 4907
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.530201342281879e-06,
+      "loss": 0.8232,
+      "step": 4908
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.526845637583893e-06,
+      "loss": 0.8149,
+      "step": 4909
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.5234899328859064e-06,
+      "loss": 0.8335,
+      "step": 4910
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.52013422818792e-06,
+      "loss": 0.8027,
+      "step": 4911
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.5167785234899332e-06,
+      "loss": 0.8389,
+      "step": 4912
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.5134228187919467e-06,
+      "loss": 0.8521,
+      "step": 4913
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.51006711409396e-06,
+      "loss": 0.7969,
+      "step": 4914
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.5067114093959735e-06,
+      "loss": 0.8257,
+      "step": 4915
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.503355704697987e-06,
+      "loss": 0.8589,
+      "step": 4916
+    },
+    {
+      "epoch": 4.12,
+      "learning_rate": 3.5e-06,
+      "loss": 0.8481,
+      "step": 4917
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.4966442953020134e-06,
+      "loss": 0.8345,
+      "step": 4918
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.4932885906040272e-06,
+      "loss": 0.814,
+      "step": 4919
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.4899328859060407e-06,
+      "loss": 0.8628,
+      "step": 4920
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.486577181208054e-06,
+      "loss": 0.8716,
+      "step": 4921
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.4832214765100675e-06,
+      "loss": 0.8755,
+      "step": 4922
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.479865771812081e-06,
+      "loss": 0.8335,
+      "step": 4923
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.4765100671140944e-06,
+      "loss": 0.8335,
+      "step": 4924
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.4731543624161078e-06,
+      "loss": 0.8032,
+      "step": 4925
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.4697986577181208e-06,
+      "loss": 0.8457,
+      "step": 4926
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.4664429530201342e-06,
+      "loss": 0.8237,
+      "step": 4927
+    },
+    {
+      "epoch": 4.13,
+      "learning_rate": 3.4630872483221476e-06,
+      "loss": 0.8022,
+      "step": 4928
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.4597315436241615e-06,
+      "loss": 0.8145,
+      "step": 4929
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.456375838926175e-06,
+      "loss": 0.8252,
+      "step": 4930
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.4530201342281883e-06,
+      "loss": 0.8159,
+      "step": 4931
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.4496644295302018e-06,
+      "loss": 0.8423,
+      "step": 4932
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.446308724832215e-06,
+      "loss": 0.8271,
+      "step": 4933
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.4429530201342286e-06,
+      "loss": 0.833,
+      "step": 4934
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.4395973154362416e-06,
+      "loss": 0.8198,
+      "step": 4935
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.436241610738255e-06,
+      "loss": 0.8447,
+      "step": 4936
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.4328859060402685e-06,
+      "loss": 0.8262,
+      "step": 4937
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.429530201342282e-06,
+      "loss": 0.8511,
+      "step": 4938
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.4261744966442958e-06,
+      "loss": 0.8271,
+      "step": 4939
+    },
+    {
+      "epoch": 4.14,
+      "learning_rate": 3.422818791946309e-06,
+      "loss": 0.8286,
+      "step": 4940
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.4194630872483226e-06,
+      "loss": 0.8364,
+      "step": 4941
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.416107382550336e-06,
+      "loss": 0.8184,
+      "step": 4942
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.4127516778523495e-06,
+      "loss": 0.833,
+      "step": 4943
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.4093959731543625e-06,
+      "loss": 0.8145,
+      "step": 4944
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.406040268456376e-06,
+      "loss": 0.8467,
+      "step": 4945
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.4026845637583893e-06,
+      "loss": 0.8184,
+      "step": 4946
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.3993288590604027e-06,
+      "loss": 0.8389,
+      "step": 4947
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.395973154362416e-06,
+      "loss": 0.8291,
+      "step": 4948
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.39261744966443e-06,
+      "loss": 0.8237,
+      "step": 4949
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.3892617449664434e-06,
+      "loss": 0.8174,
+      "step": 4950
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.385906040268457e-06,
+      "loss": 0.8154,
+      "step": 4951
+    },
+    {
+      "epoch": 4.15,
+      "learning_rate": 3.3825503355704703e-06,
+      "loss": 0.8491,
+      "step": 4952
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.3791946308724833e-06,
+      "loss": 0.8096,
+      "step": 4953
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.3758389261744967e-06,
+      "loss": 0.8232,
+      "step": 4954
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.37248322147651e-06,
+      "loss": 0.8135,
+      "step": 4955
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.3691275167785236e-06,
+      "loss": 0.8442,
+      "step": 4956
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.365771812080537e-06,
+      "loss": 0.8135,
+      "step": 4957
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.3624161073825504e-06,
+      "loss": 0.8091,
+      "step": 4958
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.3590604026845643e-06,
+      "loss": 0.8301,
+      "step": 4959
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.3557046979865777e-06,
+      "loss": 0.7944,
+      "step": 4960
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.352348993288591e-06,
+      "loss": 0.8145,
+      "step": 4961
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.3489932885906046e-06,
+      "loss": 0.8438,
+      "step": 4962
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.3456375838926176e-06,
+      "loss": 0.8115,
+      "step": 4963
+    },
+    {
+      "epoch": 4.16,
+      "learning_rate": 3.342281879194631e-06,
+      "loss": 0.8057,
+      "step": 4964
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.3389261744966444e-06,
+      "loss": 0.8369,
+      "step": 4965
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.335570469798658e-06,
+      "loss": 0.8306,
+      "step": 4966
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.3322147651006713e-06,
+      "loss": 0.8066,
+      "step": 4967
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.328859060402685e-06,
+      "loss": 0.834,
+      "step": 4968
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.3255033557046986e-06,
+      "loss": 0.8516,
+      "step": 4969
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.322147651006712e-06,
+      "loss": 0.8267,
+      "step": 4970
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.3187919463087254e-06,
+      "loss": 0.8364,
+      "step": 4971
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.3154362416107384e-06,
+      "loss": 0.8018,
+      "step": 4972
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.312080536912752e-06,
+      "loss": 0.8335,
+      "step": 4973
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.3087248322147653e-06,
+      "loss": 0.812,
+      "step": 4974
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.3053691275167787e-06,
+      "loss": 0.8257,
+      "step": 4975
+    },
+    {
+      "epoch": 4.17,
+      "learning_rate": 3.302013422818792e-06,
+      "loss": 0.8042,
+      "step": 4976
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.2986577181208055e-06,
+      "loss": 0.8208,
+      "step": 4977
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.2953020134228194e-06,
+      "loss": 0.8184,
+      "step": 4978
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.291946308724833e-06,
+      "loss": 0.7842,
+      "step": 4979
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.2885906040268462e-06,
+      "loss": 0.8398,
+      "step": 4980
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.2852348993288592e-06,
+      "loss": 0.8286,
+      "step": 4981
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.2818791946308727e-06,
+      "loss": 0.8267,
+      "step": 4982
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.278523489932886e-06,
+      "loss": 0.8218,
+      "step": 4983
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.2751677852348995e-06,
+      "loss": 0.8281,
+      "step": 4984
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.271812080536913e-06,
+      "loss": 0.8379,
+      "step": 4985
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.2684563758389264e-06,
+      "loss": 0.8154,
+      "step": 4986
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.26510067114094e-06,
+      "loss": 0.8296,
+      "step": 4987
+    },
+    {
+      "epoch": 4.18,
+      "learning_rate": 3.2617449664429537e-06,
+      "loss": 0.854,
+      "step": 4988
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.258389261744967e-06,
+      "loss": 0.8042,
+      "step": 4989
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.25503355704698e-06,
+      "loss": 0.8184,
+      "step": 4990
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.2516778523489935e-06,
+      "loss": 0.8403,
+      "step": 4991
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.248322147651007e-06,
+      "loss": 0.8315,
+      "step": 4992
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.2449664429530204e-06,
+      "loss": 0.8271,
+      "step": 4993
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.241610738255034e-06,
+      "loss": 0.8335,
+      "step": 4994
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.2382550335570472e-06,
+      "loss": 0.8477,
+      "step": 4995
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.2348993288590606e-06,
+      "loss": 0.813,
+      "step": 4996
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.231543624161074e-06,
+      "loss": 0.8423,
+      "step": 4997
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.228187919463088e-06,
+      "loss": 0.8276,
+      "step": 4998
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.2248322147651005e-06,
+      "loss": 0.8022,
+      "step": 4999
+    },
+    {
+      "epoch": 4.19,
+      "learning_rate": 3.2214765100671143e-06,
+      "loss": 0.8223,
+      "step": 5000
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.2181208053691278e-06,
+      "loss": 0.8369,
+      "step": 5001
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.214765100671141e-06,
+      "loss": 0.8071,
+      "step": 5002
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.2114093959731546e-06,
+      "loss": 0.811,
+      "step": 5003
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.208053691275168e-06,
+      "loss": 0.8286,
+      "step": 5004
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.2046979865771815e-06,
+      "loss": 0.8013,
+      "step": 5005
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.201342281879195e-06,
+      "loss": 0.8071,
+      "step": 5006
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.1979865771812083e-06,
+      "loss": 0.8447,
+      "step": 5007
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.194630872483222e-06,
+      "loss": 0.8467,
+      "step": 5008
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.1912751677852348e-06,
+      "loss": 0.8252,
+      "step": 5009
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.1879194630872486e-06,
+      "loss": 0.8545,
+      "step": 5010
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.184563758389262e-06,
+      "loss": 0.8184,
+      "step": 5011
+    },
+    {
+      "epoch": 4.2,
+      "learning_rate": 3.1812080536912755e-06,
+      "loss": 0.8154,
+      "step": 5012
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.177852348993289e-06,
+      "loss": 0.8218,
+      "step": 5013
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.1744966442953023e-06,
+      "loss": 0.8066,
+      "step": 5014
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.1711409395973157e-06,
+      "loss": 0.812,
+      "step": 5015
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.167785234899329e-06,
+      "loss": 0.8081,
+      "step": 5016
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.1644295302013426e-06,
+      "loss": 0.8574,
+      "step": 5017
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.1610738255033556e-06,
+      "loss": 0.8369,
+      "step": 5018
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.157718120805369e-06,
+      "loss": 0.8135,
+      "step": 5019
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.154362416107383e-06,
+      "loss": 0.8213,
+      "step": 5020
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.1510067114093963e-06,
+      "loss": 0.8198,
+      "step": 5021
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.1476510067114097e-06,
+      "loss": 0.7949,
+      "step": 5022
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.144295302013423e-06,
+      "loss": 0.8086,
+      "step": 5023
+    },
+    {
+      "epoch": 4.21,
+      "learning_rate": 3.1409395973154366e-06,
+      "loss": 0.8345,
+      "step": 5024
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.13758389261745e-06,
+      "loss": 0.8394,
+      "step": 5025
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.1342281879194634e-06,
+      "loss": 0.8398,
+      "step": 5026
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.1308724832214764e-06,
+      "loss": 0.811,
+      "step": 5027
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.12751677852349e-06,
+      "loss": 0.8379,
+      "step": 5028
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.1241610738255033e-06,
+      "loss": 0.8589,
+      "step": 5029
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.120805369127517e-06,
+      "loss": 0.8286,
+      "step": 5030
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.1174496644295306e-06,
+      "loss": 0.834,
+      "step": 5031
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.114093959731544e-06,
+      "loss": 0.8545,
+      "step": 5032
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.1107382550335574e-06,
+      "loss": 0.832,
+      "step": 5033
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.107382550335571e-06,
+      "loss": 0.8228,
+      "step": 5034
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.1040268456375843e-06,
+      "loss": 0.8359,
+      "step": 5035
+    },
+    {
+      "epoch": 4.22,
+      "learning_rate": 3.1006711409395973e-06,
+      "loss": 0.8057,
+      "step": 5036
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.0973154362416107e-06,
+      "loss": 0.8188,
+      "step": 5037
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.093959731543624e-06,
+      "loss": 0.8018,
+      "step": 5038
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.0906040268456376e-06,
+      "loss": 0.7842,
+      "step": 5039
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.0872483221476514e-06,
+      "loss": 0.8374,
+      "step": 5040
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.083892617449665e-06,
+      "loss": 0.8345,
+      "step": 5041
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.0805369127516783e-06,
+      "loss": 0.8135,
+      "step": 5042
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.0771812080536917e-06,
+      "loss": 0.8389,
+      "step": 5043
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.073825503355705e-06,
+      "loss": 0.8325,
+      "step": 5044
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.070469798657718e-06,
+      "loss": 0.812,
+      "step": 5045
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.0671140939597315e-06,
+      "loss": 0.853,
+      "step": 5046
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.063758389261745e-06,
+      "loss": 0.8315,
+      "step": 5047
+    },
+    {
+      "epoch": 4.23,
+      "learning_rate": 3.0604026845637584e-06,
+      "loss": 0.8228,
+      "step": 5048
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 3.057046979865772e-06,
+      "loss": 0.8394,
+      "step": 5049
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 3.0536912751677857e-06,
+      "loss": 0.8184,
+      "step": 5050
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 3.050335570469799e-06,
+      "loss": 0.8452,
+      "step": 5051
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 3.0469798657718125e-06,
+      "loss": 0.8398,
+      "step": 5052
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 3.043624161073826e-06,
+      "loss": 0.8525,
+      "step": 5053
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 3.0402684563758394e-06,
+      "loss": 0.8188,
+      "step": 5054
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 3.0369127516778524e-06,
+      "loss": 0.8149,
+      "step": 5055
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 3.033557046979866e-06,
+      "loss": 0.8472,
+      "step": 5056
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 3.0302013422818792e-06,
+      "loss": 0.835,
+      "step": 5057
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 3.0268456375838927e-06,
+      "loss": 0.834,
+      "step": 5058
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 3.0234899328859065e-06,
+      "loss": 0.8086,
+      "step": 5059
+    },
+    {
+      "epoch": 4.24,
+      "learning_rate": 3.02013422818792e-06,
+      "loss": 0.7988,
+      "step": 5060
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 3.0167785234899334e-06,
+      "loss": 0.8271,
+      "step": 5061
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 3.013422818791947e-06,
+      "loss": 0.8408,
+      "step": 5062
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 3.0100671140939602e-06,
+      "loss": 0.8262,
+      "step": 5063
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 3.0067114093959732e-06,
+      "loss": 0.8281,
+      "step": 5064
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 3.0033557046979866e-06,
+      "loss": 0.8188,
+      "step": 5065
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 3e-06,
+      "loss": 0.8237,
+      "step": 5066
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.9966442953020135e-06,
+      "loss": 0.8286,
+      "step": 5067
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.993288590604027e-06,
+      "loss": 0.8384,
+      "step": 5068
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.9899328859060408e-06,
+      "loss": 0.7866,
+      "step": 5069
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.986577181208054e-06,
+      "loss": 0.8188,
+      "step": 5070
+    },
+    {
+      "epoch": 4.25,
+      "learning_rate": 2.9832214765100676e-06,
+      "loss": 0.8384,
+      "step": 5071
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.979865771812081e-06,
+      "loss": 0.8154,
+      "step": 5072
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.976510067114094e-06,
+      "loss": 0.8633,
+      "step": 5073
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.9731543624161075e-06,
+      "loss": 0.8286,
+      "step": 5074
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.969798657718121e-06,
+      "loss": 0.8394,
+      "step": 5075
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.9664429530201343e-06,
+      "loss": 0.8228,
+      "step": 5076
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.9630872483221478e-06,
+      "loss": 0.8584,
+      "step": 5077
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.959731543624161e-06,
+      "loss": 0.812,
+      "step": 5078
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.956375838926175e-06,
+      "loss": 0.8145,
+      "step": 5079
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.9530201342281885e-06,
+      "loss": 0.8184,
+      "step": 5080
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.949664429530202e-06,
+      "loss": 0.8188,
+      "step": 5081
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.946308724832215e-06,
+      "loss": 0.8213,
+      "step": 5082
+    },
+    {
+      "epoch": 4.26,
+      "learning_rate": 2.9429530201342283e-06,
+      "loss": 0.8394,
+      "step": 5083
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.9395973154362418e-06,
+      "loss": 0.8315,
+      "step": 5084
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.936241610738255e-06,
+      "loss": 0.8589,
+      "step": 5085
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.9328859060402686e-06,
+      "loss": 0.8374,
+      "step": 5086
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.929530201342282e-06,
+      "loss": 0.8086,
+      "step": 5087
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.9261744966442955e-06,
+      "loss": 0.8267,
+      "step": 5088
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.9228187919463093e-06,
+      "loss": 0.8198,
+      "step": 5089
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.9194630872483227e-06,
+      "loss": 0.834,
+      "step": 5090
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.9161073825503357e-06,
+      "loss": 0.8091,
+      "step": 5091
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.912751677852349e-06,
+      "loss": 0.8232,
+      "step": 5092
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.9093959731543626e-06,
+      "loss": 0.8325,
+      "step": 5093
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.906040268456376e-06,
+      "loss": 0.8008,
+      "step": 5094
+    },
+    {
+      "epoch": 4.27,
+      "learning_rate": 2.9026845637583894e-06,
+      "loss": 0.8057,
+      "step": 5095
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.899328859060403e-06,
+      "loss": 0.8242,
+      "step": 5096
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.8959731543624163e-06,
+      "loss": 0.8374,
+      "step": 5097
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.8926174496644297e-06,
+      "loss": 0.7974,
+      "step": 5098
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.8892617449664436e-06,
+      "loss": 0.8169,
+      "step": 5099
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.885906040268457e-06,
+      "loss": 0.8198,
+      "step": 5100
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.88255033557047e-06,
+      "loss": 0.8398,
+      "step": 5101
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.8791946308724834e-06,
+      "loss": 0.8184,
+      "step": 5102
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.875838926174497e-06,
+      "loss": 0.8174,
+      "step": 5103
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.8724832214765103e-06,
+      "loss": 0.8413,
+      "step": 5104
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.8691275167785237e-06,
+      "loss": 0.8169,
+      "step": 5105
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.865771812080537e-06,
+      "loss": 0.8003,
+      "step": 5106
+    },
+    {
+      "epoch": 4.28,
+      "learning_rate": 2.8624161073825506e-06,
+      "loss": 0.8267,
+      "step": 5107
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.859060402684564e-06,
+      "loss": 0.8555,
+      "step": 5108
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.855704697986578e-06,
+      "loss": 0.7993,
+      "step": 5109
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.8523489932885904e-06,
+      "loss": 0.8613,
+      "step": 5110
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.8489932885906043e-06,
+      "loss": 0.8369,
+      "step": 5111
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.8456375838926177e-06,
+      "loss": 0.8257,
+      "step": 5112
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.842281879194631e-06,
+      "loss": 0.8398,
+      "step": 5113
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.8389261744966445e-06,
+      "loss": 0.8433,
+      "step": 5114
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.835570469798658e-06,
+      "loss": 0.8101,
+      "step": 5115
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.8322147651006714e-06,
+      "loss": 0.8291,
+      "step": 5116
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.828859060402685e-06,
+      "loss": 0.8164,
+      "step": 5117
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.8255033557046987e-06,
+      "loss": 0.8179,
+      "step": 5118
+    },
+    {
+      "epoch": 4.29,
+      "learning_rate": 2.8221476510067113e-06,
+      "loss": 0.8247,
+      "step": 5119
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.8187919463087247e-06,
+      "loss": 0.8374,
+      "step": 5120
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.8154362416107385e-06,
+      "loss": 0.8188,
+      "step": 5121
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.812080536912752e-06,
+      "loss": 0.8062,
+      "step": 5122
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.8087248322147654e-06,
+      "loss": 0.8467,
+      "step": 5123
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.805369127516779e-06,
+      "loss": 0.8154,
+      "step": 5124
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.8020134228187922e-06,
+      "loss": 0.8325,
+      "step": 5125
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.7986577181208057e-06,
+      "loss": 0.8228,
+      "step": 5126
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.795302013422819e-06,
+      "loss": 0.8149,
+      "step": 5127
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.791946308724832e-06,
+      "loss": 0.7983,
+      "step": 5128
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.7885906040268455e-06,
+      "loss": 0.8193,
+      "step": 5129
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.785234899328859e-06,
+      "loss": 0.8154,
+      "step": 5130
+    },
+    {
+      "epoch": 4.3,
+      "learning_rate": 2.781879194630873e-06,
+      "loss": 0.8242,
+      "step": 5131
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.7785234899328862e-06,
+      "loss": 0.8252,
+      "step": 5132
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.7751677852348997e-06,
+      "loss": 0.8682,
+      "step": 5133
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.771812080536913e-06,
+      "loss": 0.8003,
+      "step": 5134
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.7684563758389265e-06,
+      "loss": 0.8262,
+      "step": 5135
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.76510067114094e-06,
+      "loss": 0.8228,
+      "step": 5136
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.761744966442953e-06,
+      "loss": 0.8208,
+      "step": 5137
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.7583892617449664e-06,
+      "loss": 0.8457,
+      "step": 5138
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.75503355704698e-06,
+      "loss": 0.8096,
+      "step": 5139
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.7516778523489936e-06,
+      "loss": 0.8032,
+      "step": 5140
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.748322147651007e-06,
+      "loss": 0.8037,
+      "step": 5141
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.7449664429530205e-06,
+      "loss": 0.8232,
+      "step": 5142
+    },
+    {
+      "epoch": 4.31,
+      "learning_rate": 2.741610738255034e-06,
+      "loss": 0.8472,
+      "step": 5143
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.7382550335570473e-06,
+      "loss": 0.8188,
+      "step": 5144
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.7348993288590608e-06,
+      "loss": 0.8071,
+      "step": 5145
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.731543624161074e-06,
+      "loss": 0.8398,
+      "step": 5146
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.728187919463087e-06,
+      "loss": 0.8354,
+      "step": 5147
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.7248322147651006e-06,
+      "loss": 0.832,
+      "step": 5148
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.721476510067114e-06,
+      "loss": 0.8115,
+      "step": 5149
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.718120805369128e-06,
+      "loss": 0.8345,
+      "step": 5150
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.7147651006711413e-06,
+      "loss": 0.8008,
+      "step": 5151
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.7114093959731548e-06,
+      "loss": 0.7988,
+      "step": 5152
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.708053691275168e-06,
+      "loss": 0.8462,
+      "step": 5153
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.7046979865771816e-06,
+      "loss": 0.8154,
+      "step": 5154
+    },
+    {
+      "epoch": 4.32,
+      "learning_rate": 2.701342281879195e-06,
+      "loss": 0.854,
+      "step": 5155
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.697986577181208e-06,
+      "loss": 0.8384,
+      "step": 5156
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.6946308724832215e-06,
+      "loss": 0.8145,
+      "step": 5157
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.691275167785235e-06,
+      "loss": 0.8164,
+      "step": 5158
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.6879194630872483e-06,
+      "loss": 0.8198,
+      "step": 5159
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.684563758389262e-06,
+      "loss": 0.8135,
+      "step": 5160
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.6812080536912756e-06,
+      "loss": 0.7969,
+      "step": 5161
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.677852348993289e-06,
+      "loss": 0.8086,
+      "step": 5162
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.6744966442953024e-06,
+      "loss": 0.8486,
+      "step": 5163
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.671140939597316e-06,
+      "loss": 0.8247,
+      "step": 5164
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.667785234899329e-06,
+      "loss": 0.8501,
+      "step": 5165
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.6644295302013423e-06,
+      "loss": 0.792,
+      "step": 5166
+    },
+    {
+      "epoch": 4.33,
+      "learning_rate": 2.6610738255033557e-06,
+      "loss": 0.8506,
+      "step": 5167
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.657718120805369e-06,
+      "loss": 0.8555,
+      "step": 5168
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.6543624161073826e-06,
+      "loss": 0.8081,
+      "step": 5169
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.6510067114093964e-06,
+      "loss": 0.8545,
+      "step": 5170
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.64765100671141e-06,
+      "loss": 0.7998,
+      "step": 5171
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.6442953020134233e-06,
+      "loss": 0.834,
+      "step": 5172
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.6409395973154367e-06,
+      "loss": 0.8125,
+      "step": 5173
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.6375838926174497e-06,
+      "loss": 0.8569,
+      "step": 5174
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.634228187919463e-06,
+      "loss": 0.7944,
+      "step": 5175
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.6308724832214766e-06,
+      "loss": 0.8105,
+      "step": 5176
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.62751677852349e-06,
+      "loss": 0.8076,
+      "step": 5177
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.6241610738255034e-06,
+      "loss": 0.8384,
+      "step": 5178
+    },
+    {
+      "epoch": 4.34,
+      "learning_rate": 2.620805369127517e-06,
+      "loss": 0.8384,
+      "step": 5179
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.6174496644295307e-06,
+      "loss": 0.8369,
+      "step": 5180
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.614093959731544e-06,
+      "loss": 0.8311,
+      "step": 5181
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.6107382550335576e-06,
+      "loss": 0.8496,
+      "step": 5182
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.6073825503355706e-06,
+      "loss": 0.8486,
+      "step": 5183
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.604026845637584e-06,
+      "loss": 0.8267,
+      "step": 5184
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.6006711409395974e-06,
+      "loss": 0.8389,
+      "step": 5185
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.597315436241611e-06,
+      "loss": 0.8179,
+      "step": 5186
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.5939597315436243e-06,
+      "loss": 0.7866,
+      "step": 5187
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.5906040268456377e-06,
+      "loss": 0.7993,
+      "step": 5188
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.587248322147651e-06,
+      "loss": 0.8442,
+      "step": 5189
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.583892617449665e-06,
+      "loss": 0.8027,
+      "step": 5190
+    },
+    {
+      "epoch": 4.35,
+      "learning_rate": 2.5805369127516784e-06,
+      "loss": 0.8145,
+      "step": 5191
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.577181208053692e-06,
+      "loss": 0.812,
+      "step": 5192
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.573825503355705e-06,
+      "loss": 0.7949,
+      "step": 5193
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.5704697986577182e-06,
+      "loss": 0.8208,
+      "step": 5194
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.5671140939597317e-06,
+      "loss": 0.834,
+      "step": 5195
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.563758389261745e-06,
+      "loss": 0.8223,
+      "step": 5196
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.5604026845637585e-06,
+      "loss": 0.8091,
+      "step": 5197
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.557046979865772e-06,
+      "loss": 0.8154,
+      "step": 5198
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.5536912751677854e-06,
+      "loss": 0.813,
+      "step": 5199
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.5503355704697992e-06,
+      "loss": 0.8042,
+      "step": 5200
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.5469798657718127e-06,
+      "loss": 0.8481,
+      "step": 5201
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.5436241610738257e-06,
+      "loss": 0.8672,
+      "step": 5202
+    },
+    {
+      "epoch": 4.36,
+      "learning_rate": 2.540268456375839e-06,
+      "loss": 0.7949,
+      "step": 5203
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.5369127516778525e-06,
+      "loss": 0.8193,
+      "step": 5204
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.533557046979866e-06,
+      "loss": 0.7998,
+      "step": 5205
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.5302013422818794e-06,
+      "loss": 0.8467,
+      "step": 5206
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.526845637583893e-06,
+      "loss": 0.8452,
+      "step": 5207
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.5234899328859062e-06,
+      "loss": 0.8159,
+      "step": 5208
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.52013422818792e-06,
+      "loss": 0.8086,
+      "step": 5209
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.5167785234899335e-06,
+      "loss": 0.8535,
+      "step": 5210
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.513422818791946e-06,
+      "loss": 0.7954,
+      "step": 5211
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.51006711409396e-06,
+      "loss": 0.8022,
+      "step": 5212
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.5067114093959734e-06,
+      "loss": 0.8438,
+      "step": 5213
+    },
+    {
+      "epoch": 4.37,
+      "learning_rate": 2.5033557046979868e-06,
+      "loss": 0.8076,
+      "step": 5214
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.5e-06,
+      "loss": 0.8091,
+      "step": 5215
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.4966442953020136e-06,
+      "loss": 0.8149,
+      "step": 5216
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.493288590604027e-06,
+      "loss": 0.8096,
+      "step": 5217
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.4899328859060405e-06,
+      "loss": 0.7998,
+      "step": 5218
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.486577181208054e-06,
+      "loss": 0.8125,
+      "step": 5219
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.4832214765100673e-06,
+      "loss": 0.8398,
+      "step": 5220
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.4798657718120808e-06,
+      "loss": 0.8228,
+      "step": 5221
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.476510067114094e-06,
+      "loss": 0.8354,
+      "step": 5222
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.4731543624161076e-06,
+      "loss": 0.8813,
+      "step": 5223
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.469798657718121e-06,
+      "loss": 0.8213,
+      "step": 5224
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.4664429530201345e-06,
+      "loss": 0.8486,
+      "step": 5225
+    },
+    {
+      "epoch": 4.38,
+      "learning_rate": 2.463087248322148e-06,
+      "loss": 0.8081,
+      "step": 5226
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.4597315436241613e-06,
+      "loss": 0.832,
+      "step": 5227
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.4563758389261747e-06,
+      "loss": 0.8335,
+      "step": 5228
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.453020134228188e-06,
+      "loss": 0.8208,
+      "step": 5229
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.4496644295302016e-06,
+      "loss": 0.8018,
+      "step": 5230
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.446308724832215e-06,
+      "loss": 0.79,
+      "step": 5231
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.4429530201342285e-06,
+      "loss": 0.7964,
+      "step": 5232
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.439597315436242e-06,
+      "loss": 0.8188,
+      "step": 5233
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.4362416107382553e-06,
+      "loss": 0.8096,
+      "step": 5234
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.4328859060402687e-06,
+      "loss": 0.8564,
+      "step": 5235
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.429530201342282e-06,
+      "loss": 0.7974,
+      "step": 5236
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.4261744966442956e-06,
+      "loss": 0.8276,
+      "step": 5237
+    },
+    {
+      "epoch": 4.39,
+      "learning_rate": 2.422818791946309e-06,
+      "loss": 0.855,
+      "step": 5238
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.4194630872483224e-06,
+      "loss": 0.8218,
+      "step": 5239
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.416107382550336e-06,
+      "loss": 0.833,
+      "step": 5240
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.4127516778523493e-06,
+      "loss": 0.8062,
+      "step": 5241
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.4093959731543627e-06,
+      "loss": 0.8413,
+      "step": 5242
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.406040268456376e-06,
+      "loss": 0.8423,
+      "step": 5243
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.4026845637583896e-06,
+      "loss": 0.853,
+      "step": 5244
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.3993288590604026e-06,
+      "loss": 0.7939,
+      "step": 5245
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.3959731543624164e-06,
+      "loss": 0.8354,
+      "step": 5246
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.39261744966443e-06,
+      "loss": 0.8008,
+      "step": 5247
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.3892617449664433e-06,
+      "loss": 0.8442,
+      "step": 5248
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.3859060402684567e-06,
+      "loss": 0.8369,
+      "step": 5249
+    },
+    {
+      "epoch": 4.4,
+      "learning_rate": 2.3825503355704697e-06,
+      "loss": 0.8359,
+      "step": 5250
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 2.3791946308724836e-06,
+      "loss": 0.8545,
+      "step": 5251
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 2.375838926174497e-06,
+      "loss": 0.8433,
+      "step": 5252
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 2.3724832214765104e-06,
+      "loss": 0.834,
+      "step": 5253
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 2.369127516778524e-06,
+      "loss": 0.8345,
+      "step": 5254
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 2.365771812080537e-06,
+      "loss": 0.8262,
+      "step": 5255
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 2.3624161073825507e-06,
+      "loss": 0.8145,
+      "step": 5256
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 2.359060402684564e-06,
+      "loss": 0.8218,
+      "step": 5257
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 2.3557046979865775e-06,
+      "loss": 0.8281,
+      "step": 5258
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 2.3523489932885905e-06,
+      "loss": 0.8081,
+      "step": 5259
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 2.348993288590604e-06,
+      "loss": 0.8301,
+      "step": 5260
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 2.345637583892618e-06,
+      "loss": 0.8574,
+      "step": 5261
+    },
+    {
+      "epoch": 4.41,
+      "learning_rate": 2.3422818791946313e-06,
+      "loss": 0.8271,
+      "step": 5262
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.3389261744966447e-06,
+      "loss": 0.8574,
+      "step": 5263
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.3355704697986577e-06,
+      "loss": 0.8374,
+      "step": 5264
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.332214765100671e-06,
+      "loss": 0.832,
+      "step": 5265
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.328859060402685e-06,
+      "loss": 0.7803,
+      "step": 5266
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.3255033557046984e-06,
+      "loss": 0.8423,
+      "step": 5267
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.3221476510067114e-06,
+      "loss": 0.8423,
+      "step": 5268
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.318791946308725e-06,
+      "loss": 0.8125,
+      "step": 5269
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.3154362416107382e-06,
+      "loss": 0.8389,
+      "step": 5270
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.312080536912752e-06,
+      "loss": 0.835,
+      "step": 5271
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.3087248322147655e-06,
+      "loss": 0.835,
+      "step": 5272
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.3053691275167785e-06,
+      "loss": 0.812,
+      "step": 5273
+    },
+    {
+      "epoch": 4.42,
+      "learning_rate": 2.302013422818792e-06,
+      "loss": 0.8174,
+      "step": 5274
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 2.2986577181208054e-06,
+      "loss": 0.8228,
+      "step": 5275
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 2.2953020134228192e-06,
+      "loss": 0.8203,
+      "step": 5276
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 2.2919463087248326e-06,
+      "loss": 0.8257,
+      "step": 5277
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 2.2885906040268457e-06,
+      "loss": 0.8579,
+      "step": 5278
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 2.285234899328859e-06,
+      "loss": 0.8071,
+      "step": 5279
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 2.2818791946308725e-06,
+      "loss": 0.8096,
+      "step": 5280
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 2.2785234899328864e-06,
+      "loss": 0.8535,
+      "step": 5281
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 2.2751677852348994e-06,
+      "loss": 0.854,
+      "step": 5282
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 2.2718120805369128e-06,
+      "loss": 0.8384,
+      "step": 5283
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 2.268456375838926e-06,
+      "loss": 0.8267,
+      "step": 5284
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 2.2651006711409396e-06,
+      "loss": 0.8086,
+      "step": 5285
+    },
+    {
+      "epoch": 4.43,
+      "learning_rate": 2.2617449664429535e-06,
+      "loss": 0.8135,
+      "step": 5286
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 2.2583892617449665e-06,
+      "loss": 0.7974,
+      "step": 5287
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 2.25503355704698e-06,
+      "loss": 0.7959,
+      "step": 5288
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 2.2516778523489933e-06,
+      "loss": 0.8169,
+      "step": 5289
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 2.2483221476510068e-06,
+      "loss": 0.8257,
+      "step": 5290
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 2.24496644295302e-06,
+      "loss": 0.7798,
+      "step": 5291
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 2.2416107382550336e-06,
+      "loss": 0.8115,
+      "step": 5292
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 2.238255033557047e-06,
+      "loss": 0.8423,
+      "step": 5293
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 2.2348993288590605e-06,
+      "loss": 0.8271,
+      "step": 5294
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 2.2315436241610743e-06,
+      "loss": 0.814,
+      "step": 5295
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 2.2281879194630873e-06,
+      "loss": 0.8096,
+      "step": 5296
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 2.2248322147651008e-06,
+      "loss": 0.8296,
+      "step": 5297
+    },
+    {
+      "epoch": 4.44,
+      "learning_rate": 2.221476510067114e-06,
+      "loss": 0.8228,
+      "step": 5298
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 2.2181208053691276e-06,
+      "loss": 0.8286,
+      "step": 5299
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 2.2147651006711415e-06,
+      "loss": 0.8086,
+      "step": 5300
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 2.2114093959731545e-06,
+      "loss": 0.8101,
+      "step": 5301
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 2.208053691275168e-06,
+      "loss": 0.8262,
+      "step": 5302
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 2.2046979865771813e-06,
+      "loss": 0.8296,
+      "step": 5303
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 2.2013422818791947e-06,
+      "loss": 0.8447,
+      "step": 5304
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 2.197986577181208e-06,
+      "loss": 0.8506,
+      "step": 5305
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 2.1946308724832216e-06,
+      "loss": 0.8071,
+      "step": 5306
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 2.191275167785235e-06,
+      "loss": 0.8579,
+      "step": 5307
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 2.1879194630872484e-06,
+      "loss": 0.8276,
+      "step": 5308
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 2.184563758389262e-06,
+      "loss": 0.8135,
+      "step": 5309
+    },
+    {
+      "epoch": 4.45,
+      "learning_rate": 2.1812080536912753e-06,
+      "loss": 0.8584,
+      "step": 5310
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.1778523489932887e-06,
+      "loss": 0.8452,
+      "step": 5311
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.174496644295302e-06,
+      "loss": 0.8467,
+      "step": 5312
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.1711409395973156e-06,
+      "loss": 0.7896,
+      "step": 5313
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.167785234899329e-06,
+      "loss": 0.8262,
+      "step": 5314
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.1644295302013424e-06,
+      "loss": 0.8232,
+      "step": 5315
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.161073825503356e-06,
+      "loss": 0.8525,
+      "step": 5316
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.1577181208053693e-06,
+      "loss": 0.8516,
+      "step": 5317
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.1543624161073827e-06,
+      "loss": 0.8071,
+      "step": 5318
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.151006711409396e-06,
+      "loss": 0.8159,
+      "step": 5319
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.1476510067114096e-06,
+      "loss": 0.8184,
+      "step": 5320
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.144295302013423e-06,
+      "loss": 0.8247,
+      "step": 5321
+    },
+    {
+      "epoch": 4.46,
+      "learning_rate": 2.1409395973154364e-06,
+      "loss": 0.8271,
+      "step": 5322
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.13758389261745e-06,
+      "loss": 0.8423,
+      "step": 5323
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.1342281879194633e-06,
+      "loss": 0.8447,
+      "step": 5324
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.1308724832214767e-06,
+      "loss": 0.812,
+      "step": 5325
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.12751677852349e-06,
+      "loss": 0.8042,
+      "step": 5326
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.1241610738255036e-06,
+      "loss": 0.8374,
+      "step": 5327
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.120805369127517e-06,
+      "loss": 0.8247,
+      "step": 5328
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.1174496644295304e-06,
+      "loss": 0.8081,
+      "step": 5329
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.114093959731544e-06,
+      "loss": 0.854,
+      "step": 5330
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.1107382550335573e-06,
+      "loss": 0.8335,
+      "step": 5331
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.1073825503355707e-06,
+      "loss": 0.8208,
+      "step": 5332
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.104026845637584e-06,
+      "loss": 0.8354,
+      "step": 5333
+    },
+    {
+      "epoch": 4.47,
+      "learning_rate": 2.1006711409395975e-06,
+      "loss": 0.8472,
+      "step": 5334
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.097315436241611e-06,
+      "loss": 0.7964,
+      "step": 5335
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.0939597315436244e-06,
+      "loss": 0.8574,
+      "step": 5336
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.090604026845638e-06,
+      "loss": 0.811,
+      "step": 5337
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.0872483221476512e-06,
+      "loss": 0.8101,
+      "step": 5338
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.0838926174496647e-06,
+      "loss": 0.8379,
+      "step": 5339
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.080536912751678e-06,
+      "loss": 0.8389,
+      "step": 5340
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.0771812080536915e-06,
+      "loss": 0.7988,
+      "step": 5341
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.073825503355705e-06,
+      "loss": 0.8379,
+      "step": 5342
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.0704697986577184e-06,
+      "loss": 0.8262,
+      "step": 5343
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.067114093959732e-06,
+      "loss": 0.769,
+      "step": 5344
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.0637583892617452e-06,
+      "loss": 0.7979,
+      "step": 5345
+    },
+    {
+      "epoch": 4.48,
+      "learning_rate": 2.0604026845637587e-06,
+      "loss": 0.8369,
+      "step": 5346
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.057046979865772e-06,
+      "loss": 0.8379,
+      "step": 5347
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.0536912751677855e-06,
+      "loss": 0.8784,
+      "step": 5348
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.050335570469799e-06,
+      "loss": 0.8276,
+      "step": 5349
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.0469798657718124e-06,
+      "loss": 0.811,
+      "step": 5350
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.0436241610738254e-06,
+      "loss": 0.8232,
+      "step": 5351
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.0402684563758392e-06,
+      "loss": 0.832,
+      "step": 5352
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.0369127516778526e-06,
+      "loss": 0.8276,
+      "step": 5353
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.033557046979866e-06,
+      "loss": 0.814,
+      "step": 5354
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.0302013422818795e-06,
+      "loss": 0.8257,
+      "step": 5355
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.0268456375838925e-06,
+      "loss": 0.832,
+      "step": 5356
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.0234899328859063e-06,
+      "loss": 0.8008,
+      "step": 5357
+    },
+    {
+      "epoch": 4.49,
+      "learning_rate": 2.0201342281879198e-06,
+      "loss": 0.8213,
+      "step": 5358
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 2.016778523489933e-06,
+      "loss": 0.7886,
+      "step": 5359
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 2.013422818791946e-06,
+      "loss": 0.8135,
+      "step": 5360
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 2.0100671140939596e-06,
+      "loss": 0.8037,
+      "step": 5361
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 2.0067114093959735e-06,
+      "loss": 0.7959,
+      "step": 5362
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 2.003355704697987e-06,
+      "loss": 0.811,
+      "step": 5363
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 0.8325,
+      "step": 5364
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.9966442953020133e-06,
+      "loss": 0.8359,
+      "step": 5365
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.9932885906040268e-06,
+      "loss": 0.835,
+      "step": 5366
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.9899328859060406e-06,
+      "loss": 0.8296,
+      "step": 5367
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.986577181208054e-06,
+      "loss": 0.8184,
+      "step": 5368
+    },
+    {
+      "epoch": 4.5,
+      "learning_rate": 1.9832214765100675e-06,
+      "loss": 0.8555,
+      "step": 5369
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.9798657718120805e-06,
+      "loss": 0.8042,
+      "step": 5370
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.976510067114094e-06,
+      "loss": 0.8457,
+      "step": 5371
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.9731543624161077e-06,
+      "loss": 0.8149,
+      "step": 5372
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.969798657718121e-06,
+      "loss": 0.8267,
+      "step": 5373
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.966442953020134e-06,
+      "loss": 0.813,
+      "step": 5374
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.9630872483221476e-06,
+      "loss": 0.8091,
+      "step": 5375
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.959731543624161e-06,
+      "loss": 0.7734,
+      "step": 5376
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.956375838926175e-06,
+      "loss": 0.8262,
+      "step": 5377
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.9530201342281883e-06,
+      "loss": 0.8252,
+      "step": 5378
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.9496644295302013e-06,
+      "loss": 0.8096,
+      "step": 5379
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.9463087248322147e-06,
+      "loss": 0.8467,
+      "step": 5380
+    },
+    {
+      "epoch": 4.51,
+      "learning_rate": 1.9429530201342286e-06,
+      "loss": 0.8164,
+      "step": 5381
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.939597315436242e-06,
+      "loss": 0.8193,
+      "step": 5382
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.936241610738255e-06,
+      "loss": 0.8262,
+      "step": 5383
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.9328859060402684e-06,
+      "loss": 0.812,
+      "step": 5384
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.929530201342282e-06,
+      "loss": 0.8027,
+      "step": 5385
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.9261744966442957e-06,
+      "loss": 0.8076,
+      "step": 5386
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.922818791946309e-06,
+      "loss": 0.8467,
+      "step": 5387
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.919463087248322e-06,
+      "loss": 0.8354,
+      "step": 5388
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.9161073825503356e-06,
+      "loss": 0.8247,
+      "step": 5389
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.912751677852349e-06,
+      "loss": 0.8325,
+      "step": 5390
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.909395973154363e-06,
+      "loss": 0.8291,
+      "step": 5391
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.906040268456376e-06,
+      "loss": 0.8247,
+      "step": 5392
+    },
+    {
+      "epoch": 4.52,
+      "learning_rate": 1.9026845637583893e-06,
+      "loss": 0.812,
+      "step": 5393
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.899328859060403e-06,
+      "loss": 0.8071,
+      "step": 5394
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.8959731543624163e-06,
+      "loss": 0.8184,
+      "step": 5395
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.8926174496644298e-06,
+      "loss": 0.8125,
+      "step": 5396
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.889261744966443e-06,
+      "loss": 0.8379,
+      "step": 5397
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.8859060402684564e-06,
+      "loss": 0.8135,
+      "step": 5398
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.88255033557047e-06,
+      "loss": 0.8457,
+      "step": 5399
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.8791946308724835e-06,
+      "loss": 0.8208,
+      "step": 5400
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.875838926174497e-06,
+      "loss": 0.8306,
+      "step": 5401
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.8724832214765101e-06,
+      "loss": 0.8027,
+      "step": 5402
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.8691275167785235e-06,
+      "loss": 0.8115,
+      "step": 5403
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.8657718120805372e-06,
+      "loss": 0.835,
+      "step": 5404
+    },
+    {
+      "epoch": 4.53,
+      "learning_rate": 1.8624161073825506e-06,
+      "loss": 0.8096,
+      "step": 5405
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.8590604026845638e-06,
+      "loss": 0.8286,
+      "step": 5406
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.8557046979865773e-06,
+      "loss": 0.8462,
+      "step": 5407
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.8523489932885907e-06,
+      "loss": 0.7969,
+      "step": 5408
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.8489932885906043e-06,
+      "loss": 0.8174,
+      "step": 5409
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.8456375838926177e-06,
+      "loss": 0.8315,
+      "step": 5410
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.842281879194631e-06,
+      "loss": 0.7979,
+      "step": 5411
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.8389261744966444e-06,
+      "loss": 0.8525,
+      "step": 5412
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.8355704697986578e-06,
+      "loss": 0.832,
+      "step": 5413
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.8322147651006714e-06,
+      "loss": 0.8354,
+      "step": 5414
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.8288590604026849e-06,
+      "loss": 0.8218,
+      "step": 5415
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.825503355704698e-06,
+      "loss": 0.8096,
+      "step": 5416
+    },
+    {
+      "epoch": 4.54,
+      "learning_rate": 1.8221476510067115e-06,
+      "loss": 0.8301,
+      "step": 5417
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.818791946308725e-06,
+      "loss": 0.8076,
+      "step": 5418
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.8154362416107386e-06,
+      "loss": 0.855,
+      "step": 5419
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.8120805369127518e-06,
+      "loss": 0.8218,
+      "step": 5420
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.8087248322147652e-06,
+      "loss": 0.8481,
+      "step": 5421
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.8053691275167786e-06,
+      "loss": 0.8193,
+      "step": 5422
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.802013422818792e-06,
+      "loss": 0.8105,
+      "step": 5423
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.7986577181208057e-06,
+      "loss": 0.8408,
+      "step": 5424
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.795302013422819e-06,
+      "loss": 0.8462,
+      "step": 5425
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.7919463087248324e-06,
+      "loss": 0.8198,
+      "step": 5426
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.7885906040268458e-06,
+      "loss": 0.8291,
+      "step": 5427
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.7852348993288592e-06,
+      "loss": 0.8159,
+      "step": 5428
+    },
+    {
+      "epoch": 4.55,
+      "learning_rate": 1.7818791946308724e-06,
+      "loss": 0.8604,
+      "step": 5429
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.778523489932886e-06,
+      "loss": 0.8174,
+      "step": 5430
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.7751677852348995e-06,
+      "loss": 0.8604,
+      "step": 5431
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.771812080536913e-06,
+      "loss": 0.812,
+      "step": 5432
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.7684563758389263e-06,
+      "loss": 0.8359,
+      "step": 5433
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.7651006711409396e-06,
+      "loss": 0.7979,
+      "step": 5434
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.7617449664429532e-06,
+      "loss": 0.8208,
+      "step": 5435
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.7583892617449666e-06,
+      "loss": 0.8237,
+      "step": 5436
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.75503355704698e-06,
+      "loss": 0.8027,
+      "step": 5437
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.7516778523489935e-06,
+      "loss": 0.8081,
+      "step": 5438
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.7483221476510067e-06,
+      "loss": 0.812,
+      "step": 5439
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.7449664429530203e-06,
+      "loss": 0.8125,
+      "step": 5440
+    },
+    {
+      "epoch": 4.56,
+      "learning_rate": 1.7416107382550338e-06,
+      "loss": 0.7939,
+      "step": 5441
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.7382550335570472e-06,
+      "loss": 0.833,
+      "step": 5442
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.7348993288590604e-06,
+      "loss": 0.8286,
+      "step": 5443
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.7315436241610738e-06,
+      "loss": 0.8271,
+      "step": 5444
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.7281879194630875e-06,
+      "loss": 0.8164,
+      "step": 5445
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.7248322147651009e-06,
+      "loss": 0.8047,
+      "step": 5446
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.7214765100671143e-06,
+      "loss": 0.835,
+      "step": 5447
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.7181208053691275e-06,
+      "loss": 0.8438,
+      "step": 5448
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.714765100671141e-06,
+      "loss": 0.8218,
+      "step": 5449
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.7114093959731546e-06,
+      "loss": 0.8477,
+      "step": 5450
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.708053691275168e-06,
+      "loss": 0.8081,
+      "step": 5451
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.7046979865771812e-06,
+      "loss": 0.8589,
+      "step": 5452
+    },
+    {
+      "epoch": 4.57,
+      "learning_rate": 1.7013422818791947e-06,
+      "loss": 0.8379,
+      "step": 5453
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.697986577181208e-06,
+      "loss": 0.8276,
+      "step": 5454
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.6946308724832217e-06,
+      "loss": 0.856,
+      "step": 5455
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.6912751677852351e-06,
+      "loss": 0.8018,
+      "step": 5456
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.6879194630872484e-06,
+      "loss": 0.8281,
+      "step": 5457
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.6845637583892618e-06,
+      "loss": 0.8418,
+      "step": 5458
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.6812080536912752e-06,
+      "loss": 0.8096,
+      "step": 5459
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.6778523489932889e-06,
+      "loss": 0.8242,
+      "step": 5460
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.6744966442953023e-06,
+      "loss": 0.7993,
+      "step": 5461
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.6711409395973155e-06,
+      "loss": 0.8271,
+      "step": 5462
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.667785234899329e-06,
+      "loss": 0.8486,
+      "step": 5463
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.6644295302013426e-06,
+      "loss": 0.8218,
+      "step": 5464
+    },
+    {
+      "epoch": 4.58,
+      "learning_rate": 1.661073825503356e-06,
+      "loss": 0.7881,
+      "step": 5465
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.6577181208053692e-06,
+      "loss": 0.8228,
+      "step": 5466
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.6543624161073826e-06,
+      "loss": 0.8564,
+      "step": 5467
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.651006711409396e-06,
+      "loss": 0.8169,
+      "step": 5468
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.6476510067114097e-06,
+      "loss": 0.7993,
+      "step": 5469
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.6442953020134231e-06,
+      "loss": 0.7842,
+      "step": 5470
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.6409395973154363e-06,
+      "loss": 0.856,
+      "step": 5471
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.6375838926174498e-06,
+      "loss": 0.8257,
+      "step": 5472
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.6342281879194632e-06,
+      "loss": 0.833,
+      "step": 5473
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.6308724832214768e-06,
+      "loss": 0.8335,
+      "step": 5474
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.62751677852349e-06,
+      "loss": 0.8242,
+      "step": 5475
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.6241610738255035e-06,
+      "loss": 0.8198,
+      "step": 5476
+    },
+    {
+      "epoch": 4.59,
+      "learning_rate": 1.620805369127517e-06,
+      "loss": 0.833,
+      "step": 5477
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.6174496644295303e-06,
+      "loss": 0.8135,
+      "step": 5478
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.614093959731544e-06,
+      "loss": 0.7822,
+      "step": 5479
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.6107382550335572e-06,
+      "loss": 0.8193,
+      "step": 5480
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.6073825503355706e-06,
+      "loss": 0.8384,
+      "step": 5481
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.604026845637584e-06,
+      "loss": 0.8296,
+      "step": 5482
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.6006711409395975e-06,
+      "loss": 0.8311,
+      "step": 5483
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.597315436241611e-06,
+      "loss": 0.8657,
+      "step": 5484
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.5939597315436243e-06,
+      "loss": 0.8198,
+      "step": 5485
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.5906040268456377e-06,
+      "loss": 0.8237,
+      "step": 5486
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.5872483221476512e-06,
+      "loss": 0.8511,
+      "step": 5487
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.5838926174496646e-06,
+      "loss": 0.8315,
+      "step": 5488
+    },
+    {
+      "epoch": 4.6,
+      "learning_rate": 1.5805369127516778e-06,
+      "loss": 0.8013,
+      "step": 5489
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.5771812080536914e-06,
+      "loss": 0.853,
+      "step": 5490
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.5738255033557049e-06,
+      "loss": 0.8105,
+      "step": 5491
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.5704697986577183e-06,
+      "loss": 0.8584,
+      "step": 5492
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.5671140939597317e-06,
+      "loss": 0.8066,
+      "step": 5493
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.563758389261745e-06,
+      "loss": 0.8667,
+      "step": 5494
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.5604026845637586e-06,
+      "loss": 0.8252,
+      "step": 5495
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.557046979865772e-06,
+      "loss": 0.8481,
+      "step": 5496
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.5536912751677854e-06,
+      "loss": 0.8481,
+      "step": 5497
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.5503355704697986e-06,
+      "loss": 0.8481,
+      "step": 5498
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.546979865771812e-06,
+      "loss": 0.8242,
+      "step": 5499
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.5436241610738257e-06,
+      "loss": 0.8296,
+      "step": 5500
+    },
+    {
+      "epoch": 4.61,
+      "learning_rate": 1.5402684563758391e-06,
+      "loss": 0.8599,
+      "step": 5501
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.5369127516778526e-06,
+      "loss": 0.7954,
+      "step": 5502
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.5335570469798658e-06,
+      "loss": 0.8247,
+      "step": 5503
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.5302013422818792e-06,
+      "loss": 0.7954,
+      "step": 5504
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.5268456375838928e-06,
+      "loss": 0.8115,
+      "step": 5505
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.5234899328859063e-06,
+      "loss": 0.8145,
+      "step": 5506
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.5201342281879197e-06,
+      "loss": 0.8223,
+      "step": 5507
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.516778523489933e-06,
+      "loss": 0.8081,
+      "step": 5508
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.5134228187919463e-06,
+      "loss": 0.8364,
+      "step": 5509
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.51006711409396e-06,
+      "loss": 0.8296,
+      "step": 5510
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.5067114093959734e-06,
+      "loss": 0.7915,
+      "step": 5511
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.5033557046979866e-06,
+      "loss": 0.7954,
+      "step": 5512
+    },
+    {
+      "epoch": 4.62,
+      "learning_rate": 1.5e-06,
+      "loss": 0.8267,
+      "step": 5513
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.4966442953020135e-06,
+      "loss": 0.8223,
+      "step": 5514
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.493288590604027e-06,
+      "loss": 0.8145,
+      "step": 5515
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.4899328859060405e-06,
+      "loss": 0.814,
+      "step": 5516
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.4865771812080537e-06,
+      "loss": 0.8198,
+      "step": 5517
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.4832214765100672e-06,
+      "loss": 0.8389,
+      "step": 5518
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.4798657718120806e-06,
+      "loss": 0.8003,
+      "step": 5519
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.4765100671140942e-06,
+      "loss": 0.8232,
+      "step": 5520
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.4731543624161075e-06,
+      "loss": 0.812,
+      "step": 5521
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.4697986577181209e-06,
+      "loss": 0.8271,
+      "step": 5522
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.4664429530201343e-06,
+      "loss": 0.8271,
+      "step": 5523
+    },
+    {
+      "epoch": 4.63,
+      "learning_rate": 1.4630872483221477e-06,
+      "loss": 0.8276,
+      "step": 5524
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.4597315436241614e-06,
+      "loss": 0.8408,
+      "step": 5525
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.4563758389261746e-06,
+      "loss": 0.8457,
+      "step": 5526
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.453020134228188e-06,
+      "loss": 0.8564,
+      "step": 5527
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.4496644295302014e-06,
+      "loss": 0.8247,
+      "step": 5528
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.4463087248322149e-06,
+      "loss": 0.835,
+      "step": 5529
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.4429530201342285e-06,
+      "loss": 0.814,
+      "step": 5530
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.4395973154362417e-06,
+      "loss": 0.8301,
+      "step": 5531
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.4362416107382551e-06,
+      "loss": 0.8447,
+      "step": 5532
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.4328859060402686e-06,
+      "loss": 0.791,
+      "step": 5533
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.429530201342282e-06,
+      "loss": 0.8008,
+      "step": 5534
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.4261744966442952e-06,
+      "loss": 0.8252,
+      "step": 5535
+    },
+    {
+      "epoch": 4.64,
+      "learning_rate": 1.4228187919463088e-06,
+      "loss": 0.8066,
+      "step": 5536
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.4194630872483223e-06,
+      "loss": 0.8154,
+      "step": 5537
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.4161073825503357e-06,
+      "loss": 0.8804,
+      "step": 5538
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.4127516778523493e-06,
+      "loss": 0.8403,
+      "step": 5539
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.4093959731543623e-06,
+      "loss": 0.8276,
+      "step": 5540
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.406040268456376e-06,
+      "loss": 0.8643,
+      "step": 5541
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.4026845637583894e-06,
+      "loss": 0.8047,
+      "step": 5542
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.3993288590604028e-06,
+      "loss": 0.8374,
+      "step": 5543
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.395973154362416e-06,
+      "loss": 0.8335,
+      "step": 5544
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.3926174496644295e-06,
+      "loss": 0.8735,
+      "step": 5545
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.3892617449664431e-06,
+      "loss": 0.8325,
+      "step": 5546
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.3859060402684565e-06,
+      "loss": 0.7944,
+      "step": 5547
+    },
+    {
+      "epoch": 4.65,
+      "learning_rate": 1.38255033557047e-06,
+      "loss": 0.856,
+      "step": 5548
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.3791946308724832e-06,
+      "loss": 0.8457,
+      "step": 5549
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.3758389261744968e-06,
+      "loss": 0.8506,
+      "step": 5550
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.3724832214765102e-06,
+      "loss": 0.8159,
+      "step": 5551
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.3691275167785237e-06,
+      "loss": 0.8262,
+      "step": 5552
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.365771812080537e-06,
+      "loss": 0.8315,
+      "step": 5553
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.3624161073825503e-06,
+      "loss": 0.7935,
+      "step": 5554
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.359060402684564e-06,
+      "loss": 0.8179,
+      "step": 5555
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.3557046979865774e-06,
+      "loss": 0.8203,
+      "step": 5556
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.3523489932885908e-06,
+      "loss": 0.834,
+      "step": 5557
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.348993288590604e-06,
+      "loss": 0.835,
+      "step": 5558
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.3456375838926174e-06,
+      "loss": 0.8369,
+      "step": 5559
+    },
+    {
+      "epoch": 4.66,
+      "learning_rate": 1.342281879194631e-06,
+      "loss": 0.8232,
+      "step": 5560
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.3389261744966445e-06,
+      "loss": 0.8306,
+      "step": 5561
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.335570469798658e-06,
+      "loss": 0.8613,
+      "step": 5562
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.3322147651006712e-06,
+      "loss": 0.8262,
+      "step": 5563
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.3288590604026846e-06,
+      "loss": 0.814,
+      "step": 5564
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.3255033557046982e-06,
+      "loss": 0.8525,
+      "step": 5565
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.3221476510067116e-06,
+      "loss": 0.7915,
+      "step": 5566
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.3187919463087249e-06,
+      "loss": 0.8291,
+      "step": 5567
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.3154362416107383e-06,
+      "loss": 0.8169,
+      "step": 5568
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.3120805369127517e-06,
+      "loss": 0.8242,
+      "step": 5569
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.3087248322147653e-06,
+      "loss": 0.7964,
+      "step": 5570
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.3053691275167788e-06,
+      "loss": 0.8242,
+      "step": 5571
+    },
+    {
+      "epoch": 4.67,
+      "learning_rate": 1.302013422818792e-06,
+      "loss": 0.8354,
+      "step": 5572
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.2986577181208054e-06,
+      "loss": 0.7803,
+      "step": 5573
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.2953020134228188e-06,
+      "loss": 0.8203,
+      "step": 5574
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.2919463087248325e-06,
+      "loss": 0.8101,
+      "step": 5575
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.288590604026846e-06,
+      "loss": 0.8101,
+      "step": 5576
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.2852348993288591e-06,
+      "loss": 0.8042,
+      "step": 5577
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.2818791946308726e-06,
+      "loss": 0.8213,
+      "step": 5578
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.278523489932886e-06,
+      "loss": 0.8081,
+      "step": 5579
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.2751677852348996e-06,
+      "loss": 0.8359,
+      "step": 5580
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.2718120805369128e-06,
+      "loss": 0.8413,
+      "step": 5581
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.2684563758389263e-06,
+      "loss": 0.8325,
+      "step": 5582
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.2651006711409397e-06,
+      "loss": 0.8145,
+      "step": 5583
+    },
+    {
+      "epoch": 4.68,
+      "learning_rate": 1.2617449664429531e-06,
+      "loss": 0.8384,
+      "step": 5584
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.2583892617449667e-06,
+      "loss": 0.8213,
+      "step": 5585
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.25503355704698e-06,
+      "loss": 0.8472,
+      "step": 5586
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.2516778523489934e-06,
+      "loss": 0.8174,
+      "step": 5587
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.2483221476510068e-06,
+      "loss": 0.7974,
+      "step": 5588
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.2449664429530202e-06,
+      "loss": 0.8208,
+      "step": 5589
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.2416107382550337e-06,
+      "loss": 0.8125,
+      "step": 5590
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.238255033557047e-06,
+      "loss": 0.8096,
+      "step": 5591
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.2348993288590605e-06,
+      "loss": 0.8423,
+      "step": 5592
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.231543624161074e-06,
+      "loss": 0.7969,
+      "step": 5593
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.2281879194630874e-06,
+      "loss": 0.8403,
+      "step": 5594
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.2248322147651008e-06,
+      "loss": 0.8057,
+      "step": 5595
+    },
+    {
+      "epoch": 4.69,
+      "learning_rate": 1.2214765100671142e-06,
+      "loss": 0.8345,
+      "step": 5596
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.2181208053691277e-06,
+      "loss": 0.8403,
+      "step": 5597
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.214765100671141e-06,
+      "loss": 0.8281,
+      "step": 5598
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.2114093959731545e-06,
+      "loss": 0.8262,
+      "step": 5599
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.208053691275168e-06,
+      "loss": 0.8213,
+      "step": 5600
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.2046979865771814e-06,
+      "loss": 0.8218,
+      "step": 5601
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.2013422818791948e-06,
+      "loss": 0.7993,
+      "step": 5602
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.1979865771812082e-06,
+      "loss": 0.8452,
+      "step": 5603
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.1946308724832216e-06,
+      "loss": 0.8403,
+      "step": 5604
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.1912751677852349e-06,
+      "loss": 0.7769,
+      "step": 5605
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.1879194630872485e-06,
+      "loss": 0.8169,
+      "step": 5606
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.184563758389262e-06,
+      "loss": 0.8262,
+      "step": 5607
+    },
+    {
+      "epoch": 4.7,
+      "learning_rate": 1.1812080536912753e-06,
+      "loss": 0.8384,
+      "step": 5608
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.1778523489932888e-06,
+      "loss": 0.814,
+      "step": 5609
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.174496644295302e-06,
+      "loss": 0.835,
+      "step": 5610
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.1711409395973156e-06,
+      "loss": 0.8589,
+      "step": 5611
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.1677852348993288e-06,
+      "loss": 0.8032,
+      "step": 5612
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.1644295302013425e-06,
+      "loss": 0.8481,
+      "step": 5613
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.1610738255033557e-06,
+      "loss": 0.8521,
+      "step": 5614
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.1577181208053691e-06,
+      "loss": 0.8647,
+      "step": 5615
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.1543624161073828e-06,
+      "loss": 0.8208,
+      "step": 5616
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.151006711409396e-06,
+      "loss": 0.8223,
+      "step": 5617
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.1476510067114096e-06,
+      "loss": 0.8369,
+      "step": 5618
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.1442953020134228e-06,
+      "loss": 0.835,
+      "step": 5619
+    },
+    {
+      "epoch": 4.71,
+      "learning_rate": 1.1409395973154363e-06,
+      "loss": 0.834,
+      "step": 5620
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.1375838926174497e-06,
+      "loss": 0.832,
+      "step": 5621
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.134228187919463e-06,
+      "loss": 0.8433,
+      "step": 5622
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.1308724832214767e-06,
+      "loss": 0.8027,
+      "step": 5623
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.12751677852349e-06,
+      "loss": 0.8115,
+      "step": 5624
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.1241610738255034e-06,
+      "loss": 0.8301,
+      "step": 5625
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.1208053691275168e-06,
+      "loss": 0.8091,
+      "step": 5626
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.1174496644295302e-06,
+      "loss": 0.8125,
+      "step": 5627
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.1140939597315437e-06,
+      "loss": 0.811,
+      "step": 5628
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.110738255033557e-06,
+      "loss": 0.8286,
+      "step": 5629
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.1073825503355707e-06,
+      "loss": 0.8613,
+      "step": 5630
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.104026845637584e-06,
+      "loss": 0.8066,
+      "step": 5631
+    },
+    {
+      "epoch": 4.72,
+      "learning_rate": 1.1006711409395974e-06,
+      "loss": 0.8242,
+      "step": 5632
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.0973154362416108e-06,
+      "loss": 0.8267,
+      "step": 5633
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.0939597315436242e-06,
+      "loss": 0.8276,
+      "step": 5634
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.0906040268456377e-06,
+      "loss": 0.8154,
+      "step": 5635
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.087248322147651e-06,
+      "loss": 0.8359,
+      "step": 5636
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.0838926174496645e-06,
+      "loss": 0.8145,
+      "step": 5637
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.080536912751678e-06,
+      "loss": 0.8311,
+      "step": 5638
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.0771812080536914e-06,
+      "loss": 0.8203,
+      "step": 5639
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.0738255033557048e-06,
+      "loss": 0.8208,
+      "step": 5640
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.0704697986577182e-06,
+      "loss": 0.8408,
+      "step": 5641
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.0671140939597316e-06,
+      "loss": 0.834,
+      "step": 5642
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.063758389261745e-06,
+      "loss": 0.8467,
+      "step": 5643
+    },
+    {
+      "epoch": 4.73,
+      "learning_rate": 1.0604026845637585e-06,
+      "loss": 0.8159,
+      "step": 5644
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.057046979865772e-06,
+      "loss": 0.8335,
+      "step": 5645
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.0536912751677853e-06,
+      "loss": 0.8208,
+      "step": 5646
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.0503355704697988e-06,
+      "loss": 0.8252,
+      "step": 5647
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.0469798657718122e-06,
+      "loss": 0.814,
+      "step": 5648
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.0436241610738256e-06,
+      "loss": 0.7905,
+      "step": 5649
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.040268456375839e-06,
+      "loss": 0.8008,
+      "step": 5650
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.0369127516778525e-06,
+      "loss": 0.8237,
+      "step": 5651
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.033557046979866e-06,
+      "loss": 0.8164,
+      "step": 5652
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.0302013422818793e-06,
+      "loss": 0.835,
+      "step": 5653
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.0268456375838928e-06,
+      "loss": 0.8418,
+      "step": 5654
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.0234899328859062e-06,
+      "loss": 0.8345,
+      "step": 5655
+    },
+    {
+      "epoch": 4.74,
+      "learning_rate": 1.0201342281879196e-06,
+      "loss": 0.8052,
+      "step": 5656
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.016778523489933e-06,
+      "loss": 0.8384,
+      "step": 5657
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.0134228187919462e-06,
+      "loss": 0.8062,
+      "step": 5658
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.0100671140939599e-06,
+      "loss": 0.8257,
+      "step": 5659
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.006711409395973e-06,
+      "loss": 0.8291,
+      "step": 5660
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.0033557046979867e-06,
+      "loss": 0.813,
+      "step": 5661
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 1.0000000000000002e-06,
+      "loss": 0.854,
+      "step": 5662
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 9.966442953020134e-07,
+      "loss": 0.7998,
+      "step": 5663
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 9.93288590604027e-07,
+      "loss": 0.8477,
+      "step": 5664
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 9.899328859060402e-07,
+      "loss": 0.8281,
+      "step": 5665
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 9.865771812080539e-07,
+      "loss": 0.8008,
+      "step": 5666
+    },
+    {
+      "epoch": 4.75,
+      "learning_rate": 9.83221476510067e-07,
+      "loss": 0.8076,
+      "step": 5667
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 9.798657718120805e-07,
+      "loss": 0.8169,
+      "step": 5668
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 9.765100671140942e-07,
+      "loss": 0.8198,
+      "step": 5669
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 9.731543624161074e-07,
+      "loss": 0.8569,
+      "step": 5670
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 9.69798657718121e-07,
+      "loss": 0.8022,
+      "step": 5671
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 9.664429530201342e-07,
+      "loss": 0.7988,
+      "step": 5672
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 9.630872483221479e-07,
+      "loss": 0.8242,
+      "step": 5673
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 9.59731543624161e-07,
+      "loss": 0.8335,
+      "step": 5674
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 9.563758389261745e-07,
+      "loss": 0.8193,
+      "step": 5675
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 9.53020134228188e-07,
+      "loss": 0.8027,
+      "step": 5676
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 9.496644295302015e-07,
+      "loss": 0.833,
+      "step": 5677
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 9.463087248322149e-07,
+      "loss": 0.8193,
+      "step": 5678
+    },
+    {
+      "epoch": 4.76,
+      "learning_rate": 9.429530201342282e-07,
+      "loss": 0.8003,
+      "step": 5679
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 9.395973154362417e-07,
+      "loss": 0.8086,
+      "step": 5680
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 9.362416107382551e-07,
+      "loss": 0.7671,
+      "step": 5681
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 9.328859060402686e-07,
+      "loss": 0.8296,
+      "step": 5682
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 9.295302013422819e-07,
+      "loss": 0.8403,
+      "step": 5683
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 9.261744966442953e-07,
+      "loss": 0.8076,
+      "step": 5684
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 9.228187919463089e-07,
+      "loss": 0.8291,
+      "step": 5685
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 9.194630872483222e-07,
+      "loss": 0.8442,
+      "step": 5686
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 9.161073825503357e-07,
+      "loss": 0.8296,
+      "step": 5687
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 9.12751677852349e-07,
+      "loss": 0.8145,
+      "step": 5688
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 9.093959731543625e-07,
+      "loss": 0.8184,
+      "step": 5689
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 9.060402684563759e-07,
+      "loss": 0.8076,
+      "step": 5690
+    },
+    {
+      "epoch": 4.77,
+      "learning_rate": 9.026845637583893e-07,
+      "loss": 0.8184,
+      "step": 5691
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 8.993288590604029e-07,
+      "loss": 0.7964,
+      "step": 5692
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 8.959731543624162e-07,
+      "loss": 0.8481,
+      "step": 5693
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 8.926174496644296e-07,
+      "loss": 0.8379,
+      "step": 5694
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 8.89261744966443e-07,
+      "loss": 0.792,
+      "step": 5695
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 8.859060402684565e-07,
+      "loss": 0.8481,
+      "step": 5696
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 8.825503355704698e-07,
+      "loss": 0.8271,
+      "step": 5697
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 8.791946308724833e-07,
+      "loss": 0.8203,
+      "step": 5698
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 8.758389261744967e-07,
+      "loss": 0.8276,
+      "step": 5699
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 8.724832214765102e-07,
+      "loss": 0.8325,
+      "step": 5700
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 8.691275167785236e-07,
+      "loss": 0.8096,
+      "step": 5701
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 8.657718120805369e-07,
+      "loss": 0.8564,
+      "step": 5702
+    },
+    {
+      "epoch": 4.78,
+      "learning_rate": 8.624161073825504e-07,
+      "loss": 0.8311,
+      "step": 5703
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.590604026845638e-07,
+      "loss": 0.7983,
+      "step": 5704
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.557046979865773e-07,
+      "loss": 0.8477,
+      "step": 5705
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.523489932885906e-07,
+      "loss": 0.8101,
+      "step": 5706
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.48993288590604e-07,
+      "loss": 0.79,
+      "step": 5707
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.456375838926176e-07,
+      "loss": 0.8389,
+      "step": 5708
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.422818791946309e-07,
+      "loss": 0.8125,
+      "step": 5709
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.389261744966444e-07,
+      "loss": 0.7935,
+      "step": 5710
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.355704697986577e-07,
+      "loss": 0.8101,
+      "step": 5711
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.322147651006713e-07,
+      "loss": 0.8433,
+      "step": 5712
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.288590604026846e-07,
+      "loss": 0.8462,
+      "step": 5713
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.25503355704698e-07,
+      "loss": 0.8311,
+      "step": 5714
+    },
+    {
+      "epoch": 4.79,
+      "learning_rate": 8.221476510067116e-07,
+      "loss": 0.8374,
+      "step": 5715
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.187919463087249e-07,
+      "loss": 0.8193,
+      "step": 5716
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.154362416107384e-07,
+      "loss": 0.7886,
+      "step": 5717
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.120805369127517e-07,
+      "loss": 0.8271,
+      "step": 5718
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.087248322147652e-07,
+      "loss": 0.8257,
+      "step": 5719
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.053691275167786e-07,
+      "loss": 0.8208,
+      "step": 5720
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 8.02013422818792e-07,
+      "loss": 0.8213,
+      "step": 5721
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 7.986577181208055e-07,
+      "loss": 0.8374,
+      "step": 5722
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 7.953020134228189e-07,
+      "loss": 0.7837,
+      "step": 5723
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 7.919463087248323e-07,
+      "loss": 0.8442,
+      "step": 5724
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 7.885906040268457e-07,
+      "loss": 0.8306,
+      "step": 5725
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 7.852348993288591e-07,
+      "loss": 0.8472,
+      "step": 5726
+    },
+    {
+      "epoch": 4.8,
+      "learning_rate": 7.818791946308725e-07,
+      "loss": 0.8081,
+      "step": 5727
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.78523489932886e-07,
+      "loss": 0.8276,
+      "step": 5728
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.751677852348993e-07,
+      "loss": 0.8398,
+      "step": 5729
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.718120805369129e-07,
+      "loss": 0.855,
+      "step": 5730
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.684563758389263e-07,
+      "loss": 0.834,
+      "step": 5731
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.651006711409396e-07,
+      "loss": 0.8403,
+      "step": 5732
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.617449664429531e-07,
+      "loss": 0.8467,
+      "step": 5733
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.583892617449665e-07,
+      "loss": 0.8184,
+      "step": 5734
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.5503355704698e-07,
+      "loss": 0.8198,
+      "step": 5735
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.516778523489933e-07,
+      "loss": 0.8496,
+      "step": 5736
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.483221476510067e-07,
+      "loss": 0.8516,
+      "step": 5737
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.449664429530203e-07,
+      "loss": 0.8032,
+      "step": 5738
+    },
+    {
+      "epoch": 4.81,
+      "learning_rate": 7.416107382550336e-07,
+      "loss": 0.8433,
+      "step": 5739
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 7.382550335570471e-07,
+      "loss": 0.8242,
+      "step": 5740
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 7.348993288590604e-07,
+      "loss": 0.8218,
+      "step": 5741
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 7.315436241610739e-07,
+      "loss": 0.793,
+      "step": 5742
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 7.281879194630873e-07,
+      "loss": 0.8418,
+      "step": 5743
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 7.248322147651007e-07,
+      "loss": 0.8149,
+      "step": 5744
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 7.214765100671143e-07,
+      "loss": 0.7866,
+      "step": 5745
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 7.181208053691276e-07,
+      "loss": 0.8159,
+      "step": 5746
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 7.14765100671141e-07,
+      "loss": 0.8042,
+      "step": 5747
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 7.114093959731544e-07,
+      "loss": 0.8306,
+      "step": 5748
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 7.080536912751679e-07,
+      "loss": 0.8252,
+      "step": 5749
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 7.046979865771812e-07,
+      "loss": 0.8022,
+      "step": 5750
+    },
+    {
+      "epoch": 4.82,
+      "learning_rate": 7.013422818791947e-07,
+      "loss": 0.8477,
+      "step": 5751
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.97986577181208e-07,
+      "loss": 0.8032,
+      "step": 5752
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.946308724832216e-07,
+      "loss": 0.8203,
+      "step": 5753
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.91275167785235e-07,
+      "loss": 0.8066,
+      "step": 5754
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.879194630872484e-07,
+      "loss": 0.8354,
+      "step": 5755
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.845637583892618e-07,
+      "loss": 0.832,
+      "step": 5756
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.812080536912752e-07,
+      "loss": 0.8442,
+      "step": 5757
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.778523489932887e-07,
+      "loss": 0.8286,
+      "step": 5758
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.74496644295302e-07,
+      "loss": 0.8306,
+      "step": 5759
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.711409395973155e-07,
+      "loss": 0.8359,
+      "step": 5760
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.67785234899329e-07,
+      "loss": 0.835,
+      "step": 5761
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.644295302013423e-07,
+      "loss": 0.8198,
+      "step": 5762
+    },
+    {
+      "epoch": 4.83,
+      "learning_rate": 6.610738255033558e-07,
+      "loss": 0.833,
+      "step": 5763
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 6.577181208053691e-07,
+      "loss": 0.8604,
+      "step": 5764
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 6.543624161073827e-07,
+      "loss": 0.8091,
+      "step": 5765
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 6.51006711409396e-07,
+      "loss": 0.8579,
+      "step": 5766
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 6.476510067114094e-07,
+      "loss": 0.7949,
+      "step": 5767
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 6.44295302013423e-07,
+      "loss": 0.8242,
+      "step": 5768
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 6.409395973154363e-07,
+      "loss": 0.8301,
+      "step": 5769
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 6.375838926174498e-07,
+      "loss": 0.8184,
+      "step": 5770
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 6.342281879194631e-07,
+      "loss": 0.8447,
+      "step": 5771
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 6.308724832214766e-07,
+      "loss": 0.8193,
+      "step": 5772
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 6.2751677852349e-07,
+      "loss": 0.8228,
+      "step": 5773
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 6.241610738255034e-07,
+      "loss": 0.8301,
+      "step": 5774
+    },
+    {
+      "epoch": 4.84,
+      "learning_rate": 6.208053691275168e-07,
+      "loss": 0.8657,
+      "step": 5775
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 6.174496644295303e-07,
+      "loss": 0.8159,
+      "step": 5776
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 6.140939597315437e-07,
+      "loss": 0.8179,
+      "step": 5777
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 6.107382550335571e-07,
+      "loss": 0.8364,
+      "step": 5778
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 6.073825503355705e-07,
+      "loss": 0.8057,
+      "step": 5779
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 6.04026845637584e-07,
+      "loss": 0.8008,
+      "step": 5780
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 6.006711409395974e-07,
+      "loss": 0.7876,
+      "step": 5781
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 5.973154362416108e-07,
+      "loss": 0.8472,
+      "step": 5782
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 5.939597315436242e-07,
+      "loss": 0.8232,
+      "step": 5783
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 5.906040268456377e-07,
+      "loss": 0.8589,
+      "step": 5784
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 5.87248322147651e-07,
+      "loss": 0.8257,
+      "step": 5785
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 5.838926174496644e-07,
+      "loss": 0.8345,
+      "step": 5786
+    },
+    {
+      "epoch": 4.85,
+      "learning_rate": 5.805369127516778e-07,
+      "loss": 0.8506,
+      "step": 5787
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 5.771812080536914e-07,
+      "loss": 0.8252,
+      "step": 5788
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 5.738255033557048e-07,
+      "loss": 0.8262,
+      "step": 5789
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 5.704697986577181e-07,
+      "loss": 0.8315,
+      "step": 5790
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 5.671140939597316e-07,
+      "loss": 0.8496,
+      "step": 5791
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 5.63758389261745e-07,
+      "loss": 0.8301,
+      "step": 5792
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 5.604026845637584e-07,
+      "loss": 0.8086,
+      "step": 5793
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 5.570469798657718e-07,
+      "loss": 0.8169,
+      "step": 5794
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 5.536912751677854e-07,
+      "loss": 0.8262,
+      "step": 5795
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 5.503355704697987e-07,
+      "loss": 0.8384,
+      "step": 5796
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 5.469798657718121e-07,
+      "loss": 0.8276,
+      "step": 5797
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 5.436241610738255e-07,
+      "loss": 0.8242,
+      "step": 5798
+    },
+    {
+      "epoch": 4.86,
+      "learning_rate": 5.40268456375839e-07,
+      "loss": 0.7983,
+      "step": 5799
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 5.369127516778524e-07,
+      "loss": 0.8359,
+      "step": 5800
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 5.335570469798658e-07,
+      "loss": 0.813,
+      "step": 5801
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 5.302013422818792e-07,
+      "loss": 0.8066,
+      "step": 5802
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 5.268456375838927e-07,
+      "loss": 0.8071,
+      "step": 5803
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 5.234899328859061e-07,
+      "loss": 0.8125,
+      "step": 5804
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 5.201342281879195e-07,
+      "loss": 0.8364,
+      "step": 5805
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 5.16778523489933e-07,
+      "loss": 0.8247,
+      "step": 5806
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 5.134228187919464e-07,
+      "loss": 0.8447,
+      "step": 5807
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 5.100671140939598e-07,
+      "loss": 0.7998,
+      "step": 5808
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 5.067114093959731e-07,
+      "loss": 0.8066,
+      "step": 5809
+    },
+    {
+      "epoch": 4.87,
+      "learning_rate": 5.033557046979866e-07,
+      "loss": 0.8232,
+      "step": 5810
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 5.000000000000001e-07,
+      "loss": 0.8369,
+      "step": 5811
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.966442953020135e-07,
+      "loss": 0.8154,
+      "step": 5812
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.932885906040269e-07,
+      "loss": 0.8403,
+      "step": 5813
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.899328859060403e-07,
+      "loss": 0.7993,
+      "step": 5814
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.865771812080537e-07,
+      "loss": 0.8174,
+      "step": 5815
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.832214765100671e-07,
+      "loss": 0.8071,
+      "step": 5816
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.798657718120805e-07,
+      "loss": 0.8184,
+      "step": 5817
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.76510067114094e-07,
+      "loss": 0.835,
+      "step": 5818
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.7315436241610744e-07,
+      "loss": 0.8188,
+      "step": 5819
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.6979865771812087e-07,
+      "loss": 0.8301,
+      "step": 5820
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.664429530201343e-07,
+      "loss": 0.8149,
+      "step": 5821
+    },
+    {
+      "epoch": 4.88,
+      "learning_rate": 4.6308724832214767e-07,
+      "loss": 0.8423,
+      "step": 5822
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 4.597315436241611e-07,
+      "loss": 0.8169,
+      "step": 5823
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 4.563758389261745e-07,
+      "loss": 0.8135,
+      "step": 5824
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 4.5302013422818795e-07,
+      "loss": 0.8262,
+      "step": 5825
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 4.4966442953020143e-07,
+      "loss": 0.8218,
+      "step": 5826
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 4.463087248322148e-07,
+      "loss": 0.8345,
+      "step": 5827
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 4.4295302013422823e-07,
+      "loss": 0.8687,
+      "step": 5828
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 4.3959731543624165e-07,
+      "loss": 0.812,
+      "step": 5829
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 4.362416107382551e-07,
+      "loss": 0.8623,
+      "step": 5830
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 4.3288590604026845e-07,
+      "loss": 0.8804,
+      "step": 5831
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 4.295302013422819e-07,
+      "loss": 0.8003,
+      "step": 5832
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 4.261744966442953e-07,
+      "loss": 0.8389,
+      "step": 5833
+    },
+    {
+      "epoch": 4.89,
+      "learning_rate": 4.228187919463088e-07,
+      "loss": 0.8262,
+      "step": 5834
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 4.194630872483222e-07,
+      "loss": 0.812,
+      "step": 5835
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 4.1610738255033564e-07,
+      "loss": 0.8301,
+      "step": 5836
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 4.12751677852349e-07,
+      "loss": 0.8564,
+      "step": 5837
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 4.0939597315436244e-07,
+      "loss": 0.8296,
+      "step": 5838
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 4.0604026845637587e-07,
+      "loss": 0.814,
+      "step": 5839
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 4.026845637583893e-07,
+      "loss": 0.8452,
+      "step": 5840
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 3.9932885906040277e-07,
+      "loss": 0.7964,
+      "step": 5841
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 3.9597315436241615e-07,
+      "loss": 0.8071,
+      "step": 5842
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 3.9261744966442957e-07,
+      "loss": 0.8369,
+      "step": 5843
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 3.89261744966443e-07,
+      "loss": 0.8159,
+      "step": 5844
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 3.859060402684564e-07,
+      "loss": 0.8579,
+      "step": 5845
+    },
+    {
+      "epoch": 4.9,
+      "learning_rate": 3.825503355704698e-07,
+      "loss": 0.8228,
+      "step": 5846
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 3.791946308724832e-07,
+      "loss": 0.8213,
+      "step": 5847
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 3.7583892617449665e-07,
+      "loss": 0.8198,
+      "step": 5848
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 3.7248322147651013e-07,
+      "loss": 0.8066,
+      "step": 5849
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 3.6912751677852356e-07,
+      "loss": 0.8364,
+      "step": 5850
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 3.6577181208053693e-07,
+      "loss": 0.8662,
+      "step": 5851
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 3.6241610738255036e-07,
+      "loss": 0.8281,
+      "step": 5852
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 3.590604026845638e-07,
+      "loss": 0.8291,
+      "step": 5853
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 3.557046979865772e-07,
+      "loss": 0.8223,
+      "step": 5854
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 3.523489932885906e-07,
+      "loss": 0.8018,
+      "step": 5855
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 3.48993288590604e-07,
+      "loss": 0.8296,
+      "step": 5856
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 3.456375838926175e-07,
+      "loss": 0.7891,
+      "step": 5857
+    },
+    {
+      "epoch": 4.91,
+      "learning_rate": 3.422818791946309e-07,
+      "loss": 0.8125,
+      "step": 5858
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.3892617449664434e-07,
+      "loss": 0.8398,
+      "step": 5859
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.3557046979865777e-07,
+      "loss": 0.8115,
+      "step": 5860
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.3221476510067114e-07,
+      "loss": 0.8174,
+      "step": 5861
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.2885906040268457e-07,
+      "loss": 0.7905,
+      "step": 5862
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.25503355704698e-07,
+      "loss": 0.8242,
+      "step": 5863
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.221476510067115e-07,
+      "loss": 0.8521,
+      "step": 5864
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.187919463087249e-07,
+      "loss": 0.813,
+      "step": 5865
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.154362416107383e-07,
+      "loss": 0.8232,
+      "step": 5866
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.120805369127517e-07,
+      "loss": 0.8232,
+      "step": 5867
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.0872483221476513e-07,
+      "loss": 0.7842,
+      "step": 5868
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.0536912751677856e-07,
+      "loss": 0.8174,
+      "step": 5869
+    },
+    {
+      "epoch": 4.92,
+      "learning_rate": 3.02013422818792e-07,
+      "loss": 0.8608,
+      "step": 5870
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.986577181208054e-07,
+      "loss": 0.8379,
+      "step": 5871
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.9530201342281884e-07,
+      "loss": 0.8203,
+      "step": 5872
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.919463087248322e-07,
+      "loss": 0.8076,
+      "step": 5873
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.885906040268457e-07,
+      "loss": 0.8428,
+      "step": 5874
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.8523489932885906e-07,
+      "loss": 0.8691,
+      "step": 5875
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.818791946308725e-07,
+      "loss": 0.8638,
+      "step": 5876
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.785234899328859e-07,
+      "loss": 0.8218,
+      "step": 5877
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.7516778523489934e-07,
+      "loss": 0.8384,
+      "step": 5878
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.7181208053691277e-07,
+      "loss": 0.8403,
+      "step": 5879
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.684563758389262e-07,
+      "loss": 0.8135,
+      "step": 5880
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.651006711409396e-07,
+      "loss": 0.8086,
+      "step": 5881
+    },
+    {
+      "epoch": 4.93,
+      "learning_rate": 2.6174496644295305e-07,
+      "loss": 0.8267,
+      "step": 5882
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.583892617449665e-07,
+      "loss": 0.8057,
+      "step": 5883
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.550335570469799e-07,
+      "loss": 0.8428,
+      "step": 5884
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.516778523489933e-07,
+      "loss": 0.8428,
+      "step": 5885
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.4832214765100675e-07,
+      "loss": 0.8154,
+      "step": 5886
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.4496644295302013e-07,
+      "loss": 0.8188,
+      "step": 5887
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.4161073825503355e-07,
+      "loss": 0.8188,
+      "step": 5888
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.38255033557047e-07,
+      "loss": 0.8359,
+      "step": 5889
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.3489932885906043e-07,
+      "loss": 0.7993,
+      "step": 5890
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.3154362416107383e-07,
+      "loss": 0.8262,
+      "step": 5891
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.2818791946308726e-07,
+      "loss": 0.8569,
+      "step": 5892
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.2483221476510071e-07,
+      "loss": 0.855,
+      "step": 5893
+    },
+    {
+      "epoch": 4.94,
+      "learning_rate": 2.2147651006711411e-07,
+      "loss": 0.833,
+      "step": 5894
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.1812080536912754e-07,
+      "loss": 0.8296,
+      "step": 5895
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.1476510067114094e-07,
+      "loss": 0.8364,
+      "step": 5896
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.114093959731544e-07,
+      "loss": 0.8247,
+      "step": 5897
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.0805369127516782e-07,
+      "loss": 0.8276,
+      "step": 5898
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.0469798657718122e-07,
+      "loss": 0.8037,
+      "step": 5899
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 2.0134228187919465e-07,
+      "loss": 0.8164,
+      "step": 5900
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.9798657718120807e-07,
+      "loss": 0.8359,
+      "step": 5901
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.946308724832215e-07,
+      "loss": 0.8462,
+      "step": 5902
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.912751677852349e-07,
+      "loss": 0.8149,
+      "step": 5903
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.8791946308724833e-07,
+      "loss": 0.8408,
+      "step": 5904
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.8456375838926178e-07,
+      "loss": 0.8228,
+      "step": 5905
+    },
+    {
+      "epoch": 4.95,
+      "learning_rate": 1.8120805369127518e-07,
+      "loss": 0.8228,
+      "step": 5906
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.778523489932886e-07,
+      "loss": 0.8433,
+      "step": 5907
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.74496644295302e-07,
+      "loss": 0.8359,
+      "step": 5908
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.7114093959731546e-07,
+      "loss": 0.813,
+      "step": 5909
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.6778523489932889e-07,
+      "loss": 0.8179,
+      "step": 5910
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.6442953020134229e-07,
+      "loss": 0.855,
+      "step": 5911
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.6107382550335574e-07,
+      "loss": 0.8618,
+      "step": 5912
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.5771812080536914e-07,
+      "loss": 0.8135,
+      "step": 5913
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.5436241610738257e-07,
+      "loss": 0.8271,
+      "step": 5914
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.51006711409396e-07,
+      "loss": 0.8398,
+      "step": 5915
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.4765100671140942e-07,
+      "loss": 0.8091,
+      "step": 5916
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.4429530201342284e-07,
+      "loss": 0.8677,
+      "step": 5917
+    },
+    {
+      "epoch": 4.96,
+      "learning_rate": 1.4093959731543624e-07,
+      "loss": 0.8022,
+      "step": 5918
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.3758389261744967e-07,
+      "loss": 0.8306,
+      "step": 5919
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.342281879194631e-07,
+      "loss": 0.8398,
+      "step": 5920
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.3087248322147652e-07,
+      "loss": 0.8408,
+      "step": 5921
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.2751677852348995e-07,
+      "loss": 0.8169,
+      "step": 5922
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.2416107382550338e-07,
+      "loss": 0.8042,
+      "step": 5923
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.2080536912751678e-07,
+      "loss": 0.8232,
+      "step": 5924
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.1744966442953022e-07,
+      "loss": 0.8145,
+      "step": 5925
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.1409395973154363e-07,
+      "loss": 0.8203,
+      "step": 5926
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.1073825503355706e-07,
+      "loss": 0.8052,
+      "step": 5927
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.0738255033557047e-07,
+      "loss": 0.8052,
+      "step": 5928
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.0402684563758391e-07,
+      "loss": 0.8218,
+      "step": 5929
+    },
+    {
+      "epoch": 4.97,
+      "learning_rate": 1.0067114093959732e-07,
+      "loss": 0.8472,
+      "step": 5930
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 9.731543624161075e-08,
+      "loss": 0.8062,
+      "step": 5931
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 9.395973154362416e-08,
+      "loss": 0.8271,
+      "step": 5932
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 9.060402684563759e-08,
+      "loss": 0.833,
+      "step": 5933
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 8.7248322147651e-08,
+      "loss": 0.8247,
+      "step": 5934
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 8.389261744966444e-08,
+      "loss": 0.8174,
+      "step": 5935
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 8.053691275167787e-08,
+      "loss": 0.8301,
+      "step": 5936
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 7.718120805369128e-08,
+      "loss": 0.8296,
+      "step": 5937
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 7.382550335570471e-08,
+      "loss": 0.8569,
+      "step": 5938
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 7.046979865771812e-08,
+      "loss": 0.8149,
+      "step": 5939
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 6.711409395973155e-08,
+      "loss": 0.8081,
+      "step": 5940
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 6.375838926174498e-08,
+      "loss": 0.8545,
+      "step": 5941
+    },
+    {
+      "epoch": 4.98,
+      "learning_rate": 6.040268456375839e-08,
+      "loss": 0.8467,
+      "step": 5942
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 5.7046979865771815e-08,
+      "loss": 0.8169,
+      "step": 5943
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 5.3691275167785235e-08,
+      "loss": 0.8125,
+      "step": 5944
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 5.033557046979866e-08,
+      "loss": 0.8203,
+      "step": 5945
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 4.697986577181208e-08,
+      "loss": 0.8535,
+      "step": 5946
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 4.36241610738255e-08,
+      "loss": 0.8472,
+      "step": 5947
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 4.0268456375838935e-08,
+      "loss": 0.8423,
+      "step": 5948
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 3.6912751677852355e-08,
+      "loss": 0.8091,
+      "step": 5949
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 3.3557046979865774e-08,
+      "loss": 0.8291,
+      "step": 5950
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 3.0201342281879194e-08,
+      "loss": 0.8384,
+      "step": 5951
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.6845637583892618e-08,
+      "loss": 0.7935,
+      "step": 5952
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.348993288590604e-08,
+      "loss": 0.8325,
+      "step": 5953
+    },
+    {
+      "epoch": 4.99,
+      "learning_rate": 2.0134228187919467e-08,
+      "loss": 0.8008,
+      "step": 5954
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.6778523489932887e-08,
+      "loss": 0.7944,
+      "step": 5955
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.3422818791946309e-08,
+      "loss": 0.8223,
+      "step": 5956
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 1.0067114093959734e-08,
+      "loss": 0.8184,
+      "step": 5957
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 6.711409395973154e-09,
+      "loss": 0.8188,
+      "step": 5958
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 3.355704697986577e-09,
+      "loss": 0.8208,
+      "step": 5959
+    },
+    {
+      "epoch": 5.0,
+      "learning_rate": 0.0,
+      "loss": 0.8066,
+      "step": 5960
+    },
+    {
+      "epoch": 5.0,
+      "step": 5960,
+      "total_flos": 4.4219999153291264e+18,
+      "train_loss": 0.9227586579802852,
+      "train_runtime": 27558.1748,
+      "train_samples_per_second": 13.832,
+      "train_steps_per_second": 0.216
+    }
+  ],
+  "max_steps": 5960,
+  "num_train_epochs": 5,
+  "total_flos": 4.4219999153291264e+18,
+  "trial_name": null,
+  "trial_params": null
+}