diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,9766 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.998769609350969,
+  "global_step": 1625,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.439024390243903e-07,
+      "loss": 1.5052,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.878048780487805e-07,
+      "loss": 1.572,
+      "step": 2
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.317073170731707e-07,
+      "loss": 1.4865,
+      "step": 3
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.75609756097561e-07,
+      "loss": 1.4908,
+      "step": 4
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.2195121951219514e-06,
+      "loss": 1.5907,
+      "step": 5
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4634146341463414e-06,
+      "loss": 1.5447,
+      "step": 6
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.707317073170732e-06,
+      "loss": 1.5681,
+      "step": 7
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.951219512195122e-06,
+      "loss": 1.5166,
+      "step": 8
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.951219512195122e-06,
+      "loss": 1.4772,
+      "step": 9
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 2.1951219512195125e-06,
+      "loss": 1.4818,
+      "step": 10
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.4390243902439027e-06,
+      "loss": 1.5198,
+      "step": 11
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.682926829268293e-06,
+      "loss": 1.5627,
+      "step": 12
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2.926829268292683e-06,
+      "loss": 1.4861,
+      "step": 13
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 3.1707317073170736e-06,
+      "loss": 1.3893,
+      "step": 14
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.414634146341464e-06,
+      "loss": 1.3892,
+      "step": 15
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.6585365853658537e-06,
+      "loss": 1.3815,
+      "step": 16
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 3.902439024390244e-06,
+      "loss": 1.4312,
+      "step": 17
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 4.146341463414634e-06,
+      "loss": 1.2972,
+      "step": 18
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.390243902439025e-06,
+      "loss": 1.3501,
+      "step": 19
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.634146341463416e-06,
+      "loss": 1.2666,
+      "step": 20
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.634146341463416e-06,
+      "loss": 1.293,
+      "step": 21
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 4.634146341463416e-06,
+      "loss": 1.2803,
+      "step": 22
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.634146341463416e-06,
+      "loss": 1.2764,
+      "step": 23
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.634146341463416e-06,
+      "loss": 1.3231,
+      "step": 24
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.8780487804878055e-06,
+      "loss": 1.3313,
+      "step": 25
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 4.8780487804878055e-06,
+      "loss": 1.3092,
+      "step": 26
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.121951219512195e-06,
+      "loss": 1.3163,
+      "step": 27
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.365853658536586e-06,
+      "loss": 1.2831,
+      "step": 28
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.365853658536586e-06,
+      "loss": 1.2422,
+      "step": 29
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 5.365853658536586e-06,
+      "loss": 1.2351,
+      "step": 30
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.609756097560977e-06,
+      "loss": 1.2632,
+      "step": 31
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.853658536585366e-06,
+      "loss": 1.3242,
+      "step": 32
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.853658536585366e-06,
+      "loss": 1.1876,
+      "step": 33
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 5.853658536585366e-06,
+      "loss": 1.2341,
+      "step": 34
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.0975609756097564e-06,
+      "loss": 1.1691,
+      "step": 35
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.341463414634147e-06,
+      "loss": 1.2451,
+      "step": 36
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.585365853658538e-06,
+      "loss": 1.1865,
+      "step": 37
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 6.829268292682928e-06,
+      "loss": 1.2603,
+      "step": 38
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 7.0731707317073175e-06,
+      "loss": 1.2372,
+      "step": 39
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 7.317073170731707e-06,
+      "loss": 1.1166,
+      "step": 40
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 7.560975609756098e-06,
+      "loss": 1.1896,
+      "step": 41
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 7.804878048780489e-06,
+      "loss": 1.2146,
+      "step": 42
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.048780487804879e-06,
+      "loss": 1.1695,
+      "step": 43
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.292682926829268e-06,
+      "loss": 1.1818,
+      "step": 44
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.536585365853658e-06,
+      "loss": 1.1531,
+      "step": 45
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 8.78048780487805e-06,
+      "loss": 1.1287,
+      "step": 46
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.02439024390244e-06,
+      "loss": 1.2141,
+      "step": 47
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.268292682926831e-06,
+      "loss": 1.215,
+      "step": 48
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.51219512195122e-06,
+      "loss": 1.1729,
+      "step": 49
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.756097560975611e-06,
+      "loss": 1.1383,
+      "step": 50
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1e-05,
+      "loss": 1.1179,
+      "step": 51
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.994972347913525e-06,
+      "loss": 1.1498,
+      "step": 52
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.989944695827049e-06,
+      "loss": 1.205,
+      "step": 53
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.984917043740574e-06,
+      "loss": 1.1718,
+      "step": 54
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.979889391654098e-06,
+      "loss": 1.12,
+      "step": 55
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.974861739567622e-06,
+      "loss": 1.1107,
+      "step": 56
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.969834087481146e-06,
+      "loss": 1.1077,
+      "step": 57
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.969834087481146e-06,
+      "loss": 1.1332,
+      "step": 58
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.964806435394672e-06,
+      "loss": 1.1686,
+      "step": 59
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.959778783308196e-06,
+      "loss": 1.2051,
+      "step": 60
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.95475113122172e-06,
+      "loss": 1.0797,
+      "step": 61
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.949723479135244e-06,
+      "loss": 1.1575,
+      "step": 62
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.94469582704877e-06,
+      "loss": 1.1133,
+      "step": 63
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.939668174962294e-06,
+      "loss": 1.1526,
+      "step": 64
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.934640522875818e-06,
+      "loss": 1.2043,
+      "step": 65
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.929612870789342e-06,
+      "loss": 1.188,
+      "step": 66
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.924585218702867e-06,
+      "loss": 1.1436,
+      "step": 67
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.919557566616391e-06,
+      "loss": 1.1758,
+      "step": 68
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.919557566616391e-06,
+      "loss": 1.1718,
+      "step": 69
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.914529914529915e-06,
+      "loss": 1.0834,
+      "step": 70
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.90950226244344e-06,
+      "loss": 1.2388,
+      "step": 71
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.904474610356965e-06,
+      "loss": 1.1532,
+      "step": 72
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.899446958270489e-06,
+      "loss": 1.0919,
+      "step": 73
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.894419306184013e-06,
+      "loss": 1.1178,
+      "step": 74
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.889391654097537e-06,
+      "loss": 1.1158,
+      "step": 75
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.884364002011063e-06,
+      "loss": 1.1752,
+      "step": 76
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.879336349924586e-06,
+      "loss": 1.0917,
+      "step": 77
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.87430869783811e-06,
+      "loss": 1.1442,
+      "step": 78
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.869281045751634e-06,
+      "loss": 1.0964,
+      "step": 79
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.86425339366516e-06,
+      "loss": 1.1464,
+      "step": 80
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.859225741578684e-06,
+      "loss": 1.1819,
+      "step": 81
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.854198089492208e-06,
+      "loss": 1.1531,
+      "step": 82
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.849170437405732e-06,
+      "loss": 1.1154,
+      "step": 83
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.844142785319258e-06,
+      "loss": 1.2208,
+      "step": 84
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.83911513323278e-06,
+      "loss": 1.1102,
+      "step": 85
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.834087481146306e-06,
+      "loss": 1.1169,
+      "step": 86
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.82905982905983e-06,
+      "loss": 1.1643,
+      "step": 87
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.824032176973354e-06,
+      "loss": 1.1638,
+      "step": 88
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.819004524886878e-06,
+      "loss": 1.186,
+      "step": 89
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.813976872800403e-06,
+      "loss": 1.1721,
+      "step": 90
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.808949220713927e-06,
+      "loss": 1.171,
+      "step": 91
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.803921568627451e-06,
+      "loss": 1.0813,
+      "step": 92
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.798893916540975e-06,
+      "loss": 1.1436,
+      "step": 93
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.793866264454501e-06,
+      "loss": 1.2045,
+      "step": 94
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.788838612368025e-06,
+      "loss": 1.2101,
+      "step": 95
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.783810960281549e-06,
+      "loss": 1.124,
+      "step": 96
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.778783308195073e-06,
+      "loss": 1.1404,
+      "step": 97
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.773755656108599e-06,
+      "loss": 1.1417,
+      "step": 98
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.768728004022122e-06,
+      "loss": 1.1674,
+      "step": 99
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.763700351935646e-06,
+      "loss": 1.1144,
+      "step": 100
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.75867269984917e-06,
+      "loss": 1.1797,
+      "step": 101
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.753645047762696e-06,
+      "loss": 1.0946,
+      "step": 102
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.74861739567622e-06,
+      "loss": 1.1472,
+      "step": 103
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.743589743589744e-06,
+      "loss": 1.1713,
+      "step": 104
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.743589743589744e-06,
+      "loss": 1.0588,
+      "step": 105
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.738562091503268e-06,
+      "loss": 1.1245,
+      "step": 106
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.738562091503268e-06,
+      "loss": 1.142,
+      "step": 107
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.738562091503268e-06,
+      "loss": 1.078,
+      "step": 108
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.733534439416794e-06,
+      "loss": 1.0659,
+      "step": 109
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.733534439416794e-06,
+      "loss": 1.1169,
+      "step": 110
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.728506787330318e-06,
+      "loss": 1.1761,
+      "step": 111
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.723479135243842e-06,
+      "loss": 1.1771,
+      "step": 112
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.718451483157366e-06,
+      "loss": 1.0682,
+      "step": 113
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.713423831070891e-06,
+      "loss": 1.1263,
+      "step": 114
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.708396178984415e-06,
+      "loss": 1.0933,
+      "step": 115
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.70336852689794e-06,
+      "loss": 1.1002,
+      "step": 116
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.698340874811463e-06,
+      "loss": 1.0613,
+      "step": 117
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.693313222724989e-06,
+      "loss": 1.026,
+      "step": 118
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.693313222724989e-06,
+      "loss": 1.1002,
+      "step": 119
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.688285570638513e-06,
+      "loss": 1.1436,
+      "step": 120
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.683257918552037e-06,
+      "loss": 1.2072,
+      "step": 121
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.678230266465561e-06,
+      "loss": 1.0698,
+      "step": 122
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.673202614379087e-06,
+      "loss": 1.139,
+      "step": 123
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.66817496229261e-06,
+      "loss": 1.1058,
+      "step": 124
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.663147310206135e-06,
+      "loss": 1.1265,
+      "step": 125
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.658119658119659e-06,
+      "loss": 1.1118,
+      "step": 126
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.653092006033184e-06,
+      "loss": 1.1819,
+      "step": 127
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 9.653092006033184e-06,
+      "loss": 1.1302,
+      "step": 128
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.648064353946708e-06,
+      "loss": 1.1511,
+      "step": 129
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.643036701860232e-06,
+      "loss": 1.0446,
+      "step": 130
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.638009049773756e-06,
+      "loss": 1.1293,
+      "step": 131
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 9.632981397687282e-06,
+      "loss": 1.127,
+      "step": 132
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.627953745600806e-06,
+      "loss": 1.036,
+      "step": 133
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.627953745600806e-06,
+      "loss": 1.1472,
+      "step": 134
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.62292609351433e-06,
+      "loss": 1.088,
+      "step": 135
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 9.62292609351433e-06,
+      "loss": 1.064,
+      "step": 136
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.617898441427854e-06,
+      "loss": 1.1562,
+      "step": 137
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.61287078934138e-06,
+      "loss": 1.1215,
+      "step": 138
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.607843137254903e-06,
+      "loss": 1.1473,
+      "step": 139
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 9.602815485168427e-06,
+      "loss": 1.0519,
+      "step": 140
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.597787833081951e-06,
+      "loss": 1.1394,
+      "step": 141
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.592760180995477e-06,
+      "loss": 1.1033,
+      "step": 142
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.587732528909001e-06,
+      "loss": 1.1238,
+      "step": 143
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 9.582704876822525e-06,
+      "loss": 1.152,
+      "step": 144
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.577677224736049e-06,
+      "loss": 1.0709,
+      "step": 145
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.572649572649575e-06,
+      "loss": 1.1166,
+      "step": 146
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.567621920563099e-06,
+      "loss": 1.0743,
+      "step": 147
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 9.562594268476623e-06,
+      "loss": 1.1672,
+      "step": 148
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.557566616390147e-06,
+      "loss": 1.1285,
+      "step": 149
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.552538964303672e-06,
+      "loss": 1.1067,
+      "step": 150
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.547511312217196e-06,
+      "loss": 1.0565,
+      "step": 151
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 9.54248366013072e-06,
+      "loss": 1.0919,
+      "step": 152
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.537456008044244e-06,
+      "loss": 1.157,
+      "step": 153
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.53242835595777e-06,
+      "loss": 1.0901,
+      "step": 154
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.527400703871292e-06,
+      "loss": 1.1037,
+      "step": 155
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 9.522373051784816e-06,
+      "loss": 1.1274,
+      "step": 156
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.517345399698342e-06,
+      "loss": 1.1002,
+      "step": 157
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.512317747611866e-06,
+      "loss": 1.0648,
+      "step": 158
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.50729009552539e-06,
+      "loss": 1.1337,
+      "step": 159
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 9.502262443438914e-06,
+      "loss": 1.1183,
+      "step": 160
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.49723479135244e-06,
+      "loss": 0.9919,
+      "step": 161
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.492207139265963e-06,
+      "loss": 1.1515,
+      "step": 162
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.487179487179487e-06,
+      "loss": 1.1318,
+      "step": 163
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 9.482151835093011e-06,
+      "loss": 1.1617,
+      "step": 164
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.477124183006537e-06,
+      "loss": 1.0852,
+      "step": 165
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.472096530920061e-06,
+      "loss": 1.176,
+      "step": 166
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.472096530920061e-06,
+      "loss": 1.1772,
+      "step": 167
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 9.467068878833585e-06,
+      "loss": 1.1552,
+      "step": 168
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.462041226747109e-06,
+      "loss": 1.0631,
+      "step": 169
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.457013574660635e-06,
+      "loss": 1.0573,
+      "step": 170
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.451985922574159e-06,
+      "loss": 1.1135,
+      "step": 171
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 9.446958270487683e-06,
+      "loss": 1.0775,
+      "step": 172
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.441930618401207e-06,
+      "loss": 1.1118,
+      "step": 173
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.436902966314732e-06,
+      "loss": 1.1368,
+      "step": 174
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.431875314228256e-06,
+      "loss": 1.0793,
+      "step": 175
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 9.42684766214178e-06,
+      "loss": 1.153,
+      "step": 176
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.421820010055304e-06,
+      "loss": 1.1361,
+      "step": 177
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.41679235796883e-06,
+      "loss": 1.1096,
+      "step": 178
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.411764705882354e-06,
+      "loss": 1.174,
+      "step": 179
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 9.406737053795878e-06,
+      "loss": 1.1367,
+      "step": 180
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.401709401709402e-06,
+      "loss": 1.1951,
+      "step": 181
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.396681749622927e-06,
+      "loss": 1.1035,
+      "step": 182
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.391654097536451e-06,
+      "loss": 1.1809,
+      "step": 183
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 9.386626445449975e-06,
+      "loss": 1.0839,
+      "step": 184
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.3815987933635e-06,
+      "loss": 1.115,
+      "step": 185
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.376571141277025e-06,
+      "loss": 1.1459,
+      "step": 186
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.371543489190549e-06,
+      "loss": 1.1937,
+      "step": 187
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 9.366515837104073e-06,
+      "loss": 1.1211,
+      "step": 188
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.361488185017597e-06,
+      "loss": 1.1536,
+      "step": 189
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.356460532931123e-06,
+      "loss": 1.0728,
+      "step": 190
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.351432880844647e-06,
+      "loss": 1.0959,
+      "step": 191
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.34640522875817e-06,
+      "loss": 1.0895,
+      "step": 192
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 9.341377576671695e-06,
+      "loss": 1.0854,
+      "step": 193
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.33634992458522e-06,
+      "loss": 1.1329,
+      "step": 194
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.331322272498744e-06,
+      "loss": 1.0359,
+      "step": 195
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.326294620412268e-06,
+      "loss": 1.0601,
+      "step": 196
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 9.321266968325792e-06,
+      "loss": 1.0813,
+      "step": 197
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.321266968325792e-06,
+      "loss": 1.0535,
+      "step": 198
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.316239316239318e-06,
+      "loss": 1.1211,
+      "step": 199
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.311211664152842e-06,
+      "loss": 1.11,
+      "step": 200
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 9.306184012066366e-06,
+      "loss": 1.173,
+      "step": 201
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.30115635997989e-06,
+      "loss": 1.1129,
+      "step": 202
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.296128707893415e-06,
+      "loss": 1.1332,
+      "step": 203
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.29110105580694e-06,
+      "loss": 1.1572,
+      "step": 204
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 9.286073403720463e-06,
+      "loss": 1.1409,
+      "step": 205
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.286073403720463e-06,
+      "loss": 1.0966,
+      "step": 206
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.281045751633987e-06,
+      "loss": 1.1974,
+      "step": 207
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.276018099547513e-06,
+      "loss": 1.0828,
+      "step": 208
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 9.276018099547513e-06,
+      "loss": 1.0972,
+      "step": 209
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.270990447461037e-06,
+      "loss": 1.0881,
+      "step": 210
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.265962795374561e-06,
+      "loss": 1.1136,
+      "step": 211
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.260935143288085e-06,
+      "loss": 1.0677,
+      "step": 212
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.25590749120161e-06,
+      "loss": 1.1689,
+      "step": 213
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.250879839115135e-06,
+      "loss": 1.1725,
+      "step": 214
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.250879839115135e-06,
+      "loss": 1.0658,
+      "step": 215
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.245852187028659e-06,
+      "loss": 1.1356,
+      "step": 216
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.240824534942183e-06,
+      "loss": 1.1637,
+      "step": 217
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.235796882855708e-06,
+      "loss": 1.0527,
+      "step": 218
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.230769230769232e-06,
+      "loss": 1.075,
+      "step": 219
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.225741578682756e-06,
+      "loss": 1.1548,
+      "step": 220
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.22071392659628e-06,
+      "loss": 1.16,
+      "step": 221
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.215686274509804e-06,
+      "loss": 1.1689,
+      "step": 222
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.210658622423328e-06,
+      "loss": 1.0433,
+      "step": 223
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.205630970336852e-06,
+      "loss": 1.2013,
+      "step": 224
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.200603318250378e-06,
+      "loss": 1.0817,
+      "step": 225
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.195575666163902e-06,
+      "loss": 1.1093,
+      "step": 226
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.190548014077426e-06,
+      "loss": 1.2129,
+      "step": 227
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.18552036199095e-06,
+      "loss": 1.1028,
+      "step": 228
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 9.180492709904475e-06,
+      "loss": 1.0722,
+      "step": 229
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.175465057818e-06,
+      "loss": 1.1031,
+      "step": 230
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.170437405731523e-06,
+      "loss": 1.093,
+      "step": 231
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.165409753645047e-06,
+      "loss": 1.1296,
+      "step": 232
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 9.160382101558573e-06,
+      "loss": 1.1014,
+      "step": 233
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.155354449472097e-06,
+      "loss": 1.1296,
+      "step": 234
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.150326797385621e-06,
+      "loss": 1.1558,
+      "step": 235
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.145299145299145e-06,
+      "loss": 1.1526,
+      "step": 236
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 9.145299145299145e-06,
+      "loss": 1.101,
+      "step": 237
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.14027149321267e-06,
+      "loss": 1.1084,
+      "step": 238
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.135243841126195e-06,
+      "loss": 1.0861,
+      "step": 239
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.130216189039719e-06,
+      "loss": 1.0865,
+      "step": 240
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 9.125188536953243e-06,
+      "loss": 1.1206,
+      "step": 241
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.120160884866768e-06,
+      "loss": 1.1227,
+      "step": 242
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.115133232780292e-06,
+      "loss": 1.0994,
+      "step": 243
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.110105580693816e-06,
+      "loss": 1.097,
+      "step": 244
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 9.110105580693816e-06,
+      "loss": 0.994,
+      "step": 245
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.10507792860734e-06,
+      "loss": 1.0765,
+      "step": 246
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.100050276520866e-06,
+      "loss": 1.1066,
+      "step": 247
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.09502262443439e-06,
+      "loss": 1.2018,
+      "step": 248
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 9.089994972347914e-06,
+      "loss": 1.191,
+      "step": 249
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.084967320261438e-06,
+      "loss": 1.2052,
+      "step": 250
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.084967320261438e-06,
+      "loss": 1.0844,
+      "step": 251
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.084967320261438e-06,
+      "loss": 1.142,
+      "step": 252
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 9.084967320261438e-06,
+      "loss": 1.1097,
+      "step": 253
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.079939668174964e-06,
+      "loss": 1.1226,
+      "step": 254
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.074912016088488e-06,
+      "loss": 1.1503,
+      "step": 255
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.069884364002011e-06,
+      "loss": 1.098,
+      "step": 256
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.064856711915535e-06,
+      "loss": 1.125,
+      "step": 257
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 9.064856711915535e-06,
+      "loss": 1.1146,
+      "step": 258
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.059829059829061e-06,
+      "loss": 1.1405,
+      "step": 259
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.054801407742585e-06,
+      "loss": 1.0397,
+      "step": 260
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.049773755656109e-06,
+      "loss": 1.1234,
+      "step": 261
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 9.044746103569633e-06,
+      "loss": 1.0797,
+      "step": 262
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.039718451483159e-06,
+      "loss": 1.0865,
+      "step": 263
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.039718451483159e-06,
+      "loss": 1.0652,
+      "step": 264
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.034690799396683e-06,
+      "loss": 1.179,
+      "step": 265
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 9.029663147310207e-06,
+      "loss": 1.1382,
+      "step": 266
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.029663147310207e-06,
+      "loss": 1.1613,
+      "step": 267
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.02463549522373e-06,
+      "loss": 1.1781,
+      "step": 268
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.019607843137256e-06,
+      "loss": 1.1162,
+      "step": 269
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 9.01458019105078e-06,
+      "loss": 1.0447,
+      "step": 270
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.009552538964304e-06,
+      "loss": 1.1487,
+      "step": 271
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 9.004524886877828e-06,
+      "loss": 1.1075,
+      "step": 272
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 8.999497234791354e-06,
+      "loss": 1.0829,
+      "step": 273
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 8.994469582704878e-06,
+      "loss": 1.163,
+      "step": 274
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 8.989441930618402e-06,
+      "loss": 1.0706,
+      "step": 275
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 8.984414278531926e-06,
+      "loss": 1.1335,
+      "step": 276
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 8.979386626445452e-06,
+      "loss": 1.0945,
+      "step": 277
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 8.974358974358976e-06,
+      "loss": 1.098,
+      "step": 278
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 8.9693313222725e-06,
+      "loss": 1.177,
+      "step": 279
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 8.964303670186024e-06,
+      "loss": 1.1027,
+      "step": 280
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 8.95927601809955e-06,
+      "loss": 1.1632,
+      "step": 281
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 8.954248366013073e-06,
+      "loss": 1.1552,
+      "step": 282
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.949220713926597e-06,
+      "loss": 1.1311,
+      "step": 283
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.944193061840121e-06,
+      "loss": 1.1317,
+      "step": 284
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.939165409753647e-06,
+      "loss": 1.1027,
+      "step": 285
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 8.939165409753647e-06,
+      "loss": 1.0862,
+      "step": 286
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.93413775766717e-06,
+      "loss": 1.0892,
+      "step": 287
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.929110105580695e-06,
+      "loss": 1.0663,
+      "step": 288
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.924082453494219e-06,
+      "loss": 1.0297,
+      "step": 289
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 8.924082453494219e-06,
+      "loss": 1.1376,
+      "step": 290
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.919054801407744e-06,
+      "loss": 1.0574,
+      "step": 291
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.914027149321268e-06,
+      "loss": 1.0735,
+      "step": 292
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.908999497234792e-06,
+      "loss": 1.1881,
+      "step": 293
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 8.903971845148316e-06,
+      "loss": 1.0774,
+      "step": 294
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.903971845148316e-06,
+      "loss": 1.1309,
+      "step": 295
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.903971845148316e-06,
+      "loss": 1.084,
+      "step": 296
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.903971845148316e-06,
+      "loss": 1.1427,
+      "step": 297
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 8.89894419306184e-06,
+      "loss": 0.9691,
+      "step": 298
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.89894419306184e-06,
+      "loss": 1.083,
+      "step": 299
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.893916540975364e-06,
+      "loss": 1.1194,
+      "step": 300
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.888888888888888e-06,
+      "loss": 1.0974,
+      "step": 301
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 8.883861236802414e-06,
+      "loss": 1.0333,
+      "step": 302
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.878833584715938e-06,
+      "loss": 1.0977,
+      "step": 303
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.873805932629462e-06,
+      "loss": 1.156,
+      "step": 304
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.868778280542986e-06,
+      "loss": 1.0701,
+      "step": 305
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 8.863750628456512e-06,
+      "loss": 1.1071,
+      "step": 306
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.858722976370036e-06,
+      "loss": 1.0937,
+      "step": 307
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.85369532428356e-06,
+      "loss": 1.1313,
+      "step": 308
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.848667672197084e-06,
+      "loss": 1.0961,
+      "step": 309
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 8.84364002011061e-06,
+      "loss": 1.1019,
+      "step": 310
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 8.838612368024133e-06,
+      "loss": 1.069,
+      "step": 311
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 8.833584715937657e-06,
+      "loss": 1.1506,
+      "step": 312
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 8.828557063851181e-06,
+      "loss": 1.1361,
+      "step": 313
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 8.823529411764707e-06,
+      "loss": 1.1748,
+      "step": 314
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 8.81850175967823e-06,
+      "loss": 1.1688,
+      "step": 315
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 8.813474107591755e-06,
+      "loss": 1.1016,
+      "step": 316
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 8.808446455505279e-06,
+      "loss": 1.1151,
+      "step": 317
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 8.803418803418804e-06,
+      "loss": 1.1172,
+      "step": 318
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 8.798391151332328e-06,
+      "loss": 1.1293,
+      "step": 319
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.793363499245852e-06,
+      "loss": 1.0881,
+      "step": 320
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.788335847159376e-06,
+      "loss": 1.1093,
+      "step": 321
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.783308195072902e-06,
+      "loss": 1.0803,
+      "step": 322
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 8.778280542986426e-06,
+      "loss": 1.1453,
+      "step": 323
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.77325289089995e-06,
+      "loss": 1.1885,
+      "step": 324
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.768225238813474e-06,
+      "loss": 1.0118,
+      "step": 325
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.763197586727e-06,
+      "loss": 1.1207,
+      "step": 326
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 8.758169934640524e-06,
+      "loss": 1.1307,
+      "step": 327
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.753142282554048e-06,
+      "loss": 1.1339,
+      "step": 328
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.748114630467572e-06,
+      "loss": 1.0767,
+      "step": 329
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.743086978381097e-06,
+      "loss": 1.1941,
+      "step": 330
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.738059326294621e-06,
+      "loss": 1.0192,
+      "step": 331
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.733031674208145e-06,
+      "loss": 1.0897,
+      "step": 332
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.72800402212167e-06,
+      "loss": 1.0203,
+      "step": 333
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.722976370035195e-06,
+      "loss": 1.1804,
+      "step": 334
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.717948717948719e-06,
+      "loss": 1.0459,
+      "step": 335
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.712921065862243e-06,
+      "loss": 1.1233,
+      "step": 336
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.712921065862243e-06,
+      "loss": 1.0798,
+      "step": 337
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.712921065862243e-06,
+      "loss": 1.1715,
+      "step": 338
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 8.707893413775767e-06,
+      "loss": 1.0668,
+      "step": 339
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.702865761689292e-06,
+      "loss": 1.0916,
+      "step": 340
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.697838109602816e-06,
+      "loss": 1.1096,
+      "step": 341
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.69281045751634e-06,
+      "loss": 1.082,
+      "step": 342
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 8.687782805429864e-06,
+      "loss": 1.0863,
+      "step": 343
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.68275515334339e-06,
+      "loss": 1.0951,
+      "step": 344
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.677727501256914e-06,
+      "loss": 1.012,
+      "step": 345
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.672699849170438e-06,
+      "loss": 1.0676,
+      "step": 346
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 8.667672197083962e-06,
+      "loss": 1.1877,
+      "step": 347
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.667672197083962e-06,
+      "loss": 1.1178,
+      "step": 348
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.667672197083962e-06,
+      "loss": 1.0876,
+      "step": 349
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.662644544997488e-06,
+      "loss": 1.1227,
+      "step": 350
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 8.657616892911012e-06,
+      "loss": 1.0934,
+      "step": 351
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.652589240824536e-06,
+      "loss": 1.0953,
+      "step": 352
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.652589240824536e-06,
+      "loss": 1.1005,
+      "step": 353
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.64756158873806e-06,
+      "loss": 1.1514,
+      "step": 354
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.642533936651585e-06,
+      "loss": 1.0914,
+      "step": 355
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.63750628456511e-06,
+      "loss": 1.1562,
+      "step": 356
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.632478632478633e-06,
+      "loss": 1.1438,
+      "step": 357
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.627450980392157e-06,
+      "loss": 1.0422,
+      "step": 358
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.622423328305683e-06,
+      "loss": 1.0649,
+      "step": 359
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.617395676219207e-06,
+      "loss": 1.1786,
+      "step": 360
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.612368024132731e-06,
+      "loss": 1.0587,
+      "step": 361
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.607340372046255e-06,
+      "loss": 1.1483,
+      "step": 362
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 8.60231271995978e-06,
+      "loss": 1.014,
+      "step": 363
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.597285067873304e-06,
+      "loss": 1.0784,
+      "step": 364
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.592257415786828e-06,
+      "loss": 1.151,
+      "step": 365
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.587229763700352e-06,
+      "loss": 1.079,
+      "step": 366
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 8.582202111613876e-06,
+      "loss": 1.1434,
+      "step": 367
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.5771744595274e-06,
+      "loss": 1.0912,
+      "step": 368
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.572146807440926e-06,
+      "loss": 1.182,
+      "step": 369
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.56711915535445e-06,
+      "loss": 1.0988,
+      "step": 370
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 8.562091503267974e-06,
+      "loss": 1.1194,
+      "step": 371
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.557063851181498e-06,
+      "loss": 1.0911,
+      "step": 372
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.552036199095024e-06,
+      "loss": 1.0748,
+      "step": 373
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.547008547008548e-06,
+      "loss": 1.1099,
+      "step": 374
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 8.541980894922072e-06,
+      "loss": 1.0873,
+      "step": 375
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.536953242835596e-06,
+      "loss": 1.0328,
+      "step": 376
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.531925590749121e-06,
+      "loss": 1.0603,
+      "step": 377
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.526897938662645e-06,
+      "loss": 1.0138,
+      "step": 378
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 8.52187028657617e-06,
+      "loss": 1.0504,
+      "step": 379
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.516842634489693e-06,
+      "loss": 1.1752,
+      "step": 380
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.511814982403219e-06,
+      "loss": 1.1257,
+      "step": 381
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.511814982403219e-06,
+      "loss": 1.0836,
+      "step": 382
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.506787330316743e-06,
+      "loss": 1.1053,
+      "step": 383
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.501759678230267e-06,
+      "loss": 1.1018,
+      "step": 384
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.501759678230267e-06,
+      "loss": 1.1403,
+      "step": 385
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.496732026143791e-06,
+      "loss": 1.0841,
+      "step": 386
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.491704374057317e-06,
+      "loss": 1.1128,
+      "step": 387
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 8.48667672197084e-06,
+      "loss": 1.1547,
+      "step": 388
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.48667672197084e-06,
+      "loss": 1.1121,
+      "step": 389
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.481649069884364e-06,
+      "loss": 1.0642,
+      "step": 390
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.476621417797888e-06,
+      "loss": 1.0664,
+      "step": 391
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.471593765711414e-06,
+      "loss": 1.0894,
+      "step": 392
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.466566113624938e-06,
+      "loss": 1.175,
+      "step": 393
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.461538461538462e-06,
+      "loss": 1.1177,
+      "step": 394
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.456510809451986e-06,
+      "loss": 1.1189,
+      "step": 395
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 8.451483157365512e-06,
+      "loss": 1.12,
+      "step": 396
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.446455505279036e-06,
+      "loss": 1.1549,
+      "step": 397
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.44142785319256e-06,
+      "loss": 1.1433,
+      "step": 398
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.436400201106084e-06,
+      "loss": 1.0934,
+      "step": 399
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.436400201106084e-06,
+      "loss": 1.0704,
+      "step": 400
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.43137254901961e-06,
+      "loss": 1.1656,
+      "step": 401
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.426344896933133e-06,
+      "loss": 1.0428,
+      "step": 402
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.421317244846657e-06,
+      "loss": 1.0996,
+      "step": 403
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.416289592760181e-06,
+      "loss": 1.1254,
+      "step": 404
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.411261940673707e-06,
+      "loss": 1.1274,
+      "step": 405
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.406234288587231e-06,
+      "loss": 1.2144,
+      "step": 406
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.401206636500755e-06,
+      "loss": 1.0209,
+      "step": 407
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.396178984414279e-06,
+      "loss": 1.0306,
+      "step": 408
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.391151332327805e-06,
+      "loss": 1.1321,
+      "step": 409
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.391151332327805e-06,
+      "loss": 1.0731,
+      "step": 410
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.386123680241329e-06,
+      "loss": 1.0712,
+      "step": 411
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 8.381096028154853e-06,
+      "loss": 1.1002,
+      "step": 412
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.376068376068377e-06,
+      "loss": 0.988,
+      "step": 413
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.371040723981902e-06,
+      "loss": 1.1023,
+      "step": 414
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.366013071895426e-06,
+      "loss": 1.0493,
+      "step": 415
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 8.36098541980895e-06,
+      "loss": 1.0704,
+      "step": 416
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 8.355957767722474e-06,
+      "loss": 1.0471,
+      "step": 417
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 8.350930115636e-06,
+      "loss": 1.0442,
+      "step": 418
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 8.345902463549524e-06,
+      "loss": 1.1829,
+      "step": 419
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 8.345902463549524e-06,
+      "loss": 1.0622,
+      "step": 420
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 8.340874811463048e-06,
+      "loss": 0.9978,
+      "step": 421
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 8.340874811463048e-06,
+      "loss": 1.043,
+      "step": 422
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 8.335847159376572e-06,
+      "loss": 1.0677,
+      "step": 423
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 8.330819507290097e-06,
+      "loss": 1.0815,
+      "step": 424
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 8.325791855203621e-06,
+      "loss": 1.0336,
+      "step": 425
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 8.320764203117145e-06,
+      "loss": 1.0765,
+      "step": 426
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 8.31573655103067e-06,
+      "loss": 1.0204,
+      "step": 427
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 8.310708898944195e-06,
+      "loss": 1.0996,
+      "step": 428
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 8.305681246857719e-06,
+      "loss": 1.1169,
+      "step": 429
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 8.300653594771243e-06,
+      "loss": 1.0546,
+      "step": 430
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 8.300653594771243e-06,
+      "loss": 1.0941,
+      "step": 431
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 8.295625942684767e-06,
+      "loss": 1.0643,
+      "step": 432
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.290598290598293e-06,
+      "loss": 1.0492,
+      "step": 433
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.285570638511817e-06,
+      "loss": 1.1198,
+      "step": 434
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.280542986425339e-06,
+      "loss": 1.083,
+      "step": 435
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 8.275515334338865e-06,
+      "loss": 1.0736,
+      "step": 436
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.270487682252389e-06,
+      "loss": 1.1097,
+      "step": 437
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.265460030165913e-06,
+      "loss": 1.0579,
+      "step": 438
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.260432378079437e-06,
+      "loss": 1.0468,
+      "step": 439
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 8.255404725992962e-06,
+      "loss": 0.9857,
+      "step": 440
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.250377073906486e-06,
+      "loss": 1.0321,
+      "step": 441
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.24534942182001e-06,
+      "loss": 1.0587,
+      "step": 442
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.240321769733534e-06,
+      "loss": 1.1685,
+      "step": 443
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 8.23529411764706e-06,
+      "loss": 1.132,
+      "step": 444
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.230266465560584e-06,
+      "loss": 1.1348,
+      "step": 445
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.225238813474108e-06,
+      "loss": 1.1426,
+      "step": 446
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.220211161387632e-06,
+      "loss": 1.0645,
+      "step": 447
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.215183509301157e-06,
+      "loss": 1.0247,
+      "step": 448
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 8.210155857214681e-06,
+      "loss": 1.0654,
+      "step": 449
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.210155857214681e-06,
+      "loss": 1.1022,
+      "step": 450
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.205128205128205e-06,
+      "loss": 1.0197,
+      "step": 451
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.20010055304173e-06,
+      "loss": 1.1475,
+      "step": 452
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 8.195072900955255e-06,
+      "loss": 1.0955,
+      "step": 453
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.190045248868779e-06,
+      "loss": 1.0451,
+      "step": 454
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.185017596782303e-06,
+      "loss": 1.1487,
+      "step": 455
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.179989944695827e-06,
+      "loss": 1.0176,
+      "step": 456
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 8.174962292609353e-06,
+      "loss": 1.1408,
+      "step": 457
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.169934640522877e-06,
+      "loss": 1.0934,
+      "step": 458
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.1649069884364e-06,
+      "loss": 1.1102,
+      "step": 459
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.159879336349925e-06,
+      "loss": 1.1284,
+      "step": 460
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 8.15485168426345e-06,
+      "loss": 1.058,
+      "step": 461
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.149824032176974e-06,
+      "loss": 1.1205,
+      "step": 462
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.149824032176974e-06,
+      "loss": 1.0611,
+      "step": 463
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.144796380090498e-06,
+      "loss": 1.0791,
+      "step": 464
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 8.139768728004022e-06,
+      "loss": 1.1124,
+      "step": 465
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.134741075917548e-06,
+      "loss": 1.0876,
+      "step": 466
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.129713423831072e-06,
+      "loss": 1.0551,
+      "step": 467
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.124685771744596e-06,
+      "loss": 1.1099,
+      "step": 468
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 8.11965811965812e-06,
+      "loss": 1.1108,
+      "step": 469
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.114630467571645e-06,
+      "loss": 1.0007,
+      "step": 470
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.10960281548517e-06,
+      "loss": 1.0353,
+      "step": 471
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.104575163398693e-06,
+      "loss": 1.0686,
+      "step": 472
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 8.099547511312217e-06,
+      "loss": 1.089,
+      "step": 473
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.094519859225743e-06,
+      "loss": 1.0957,
+      "step": 474
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.089492207139267e-06,
+      "loss": 1.1387,
+      "step": 475
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.084464555052791e-06,
+      "loss": 1.1342,
+      "step": 476
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 8.079436902966315e-06,
+      "loss": 1.0477,
+      "step": 477
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.07440925087984e-06,
+      "loss": 1.0841,
+      "step": 478
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.069381598793365e-06,
+      "loss": 1.0779,
+      "step": 479
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.064353946706889e-06,
+      "loss": 1.1367,
+      "step": 480
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 8.059326294620413e-06,
+      "loss": 1.0736,
+      "step": 481
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.054298642533938e-06,
+      "loss": 1.0947,
+      "step": 482
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.049270990447462e-06,
+      "loss": 1.0749,
+      "step": 483
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.049270990447462e-06,
+      "loss": 1.1614,
+      "step": 484
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 8.049270990447462e-06,
+      "loss": 1.0818,
+      "step": 485
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.044243338360986e-06,
+      "loss": 1.0519,
+      "step": 486
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.03921568627451e-06,
+      "loss": 1.0625,
+      "step": 487
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.034188034188036e-06,
+      "loss": 1.1364,
+      "step": 488
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 8.02916038210156e-06,
+      "loss": 1.1316,
+      "step": 489
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.024132730015084e-06,
+      "loss": 1.0356,
+      "step": 490
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.019105077928608e-06,
+      "loss": 1.1344,
+      "step": 491
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.014077425842134e-06,
+      "loss": 1.1553,
+      "step": 492
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 8.014077425842134e-06,
+      "loss": 1.1786,
+      "step": 493
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.009049773755657e-06,
+      "loss": 1.1207,
+      "step": 494
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 8.004022121669181e-06,
+      "loss": 1.1176,
+      "step": 495
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.998994469582705e-06,
+      "loss": 1.0796,
+      "step": 496
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 7.993966817496231e-06,
+      "loss": 1.0215,
+      "step": 497
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.988939165409755e-06,
+      "loss": 1.095,
+      "step": 498
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.988939165409755e-06,
+      "loss": 1.0681,
+      "step": 499
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.988939165409755e-06,
+      "loss": 1.0773,
+      "step": 500
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 7.983911513323279e-06,
+      "loss": 1.0659,
+      "step": 501
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.978883861236803e-06,
+      "loss": 1.0473,
+      "step": 502
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.973856209150329e-06,
+      "loss": 1.0809,
+      "step": 503
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.968828557063853e-06,
+      "loss": 1.1124,
+      "step": 504
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 7.963800904977375e-06,
+      "loss": 1.0956,
+      "step": 505
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.963800904977375e-06,
+      "loss": 1.0533,
+      "step": 506
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.9587732528909e-06,
+      "loss": 1.0287,
+      "step": 507
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.953745600804425e-06,
+      "loss": 1.1111,
+      "step": 508
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.948717948717949e-06,
+      "loss": 1.025,
+      "step": 509
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 7.943690296631473e-06,
+      "loss": 1.0852,
+      "step": 510
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.938662644544998e-06,
+      "loss": 1.0355,
+      "step": 511
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.933634992458522e-06,
+      "loss": 1.09,
+      "step": 512
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.928607340372046e-06,
+      "loss": 1.0739,
+      "step": 513
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 7.92357968828557e-06,
+      "loss": 1.088,
+      "step": 514
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.92357968828557e-06,
+      "loss": 1.1022,
+      "step": 515
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.918552036199096e-06,
+      "loss": 1.0812,
+      "step": 516
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.91352438411262e-06,
+      "loss": 1.0299,
+      "step": 517
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 7.908496732026144e-06,
+      "loss": 1.0425,
+      "step": 518
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.903469079939668e-06,
+      "loss": 1.0363,
+      "step": 519
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.898441427853194e-06,
+      "loss": 1.0756,
+      "step": 520
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.893413775766717e-06,
+      "loss": 1.0869,
+      "step": 521
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 7.888386123680241e-06,
+      "loss": 0.9648,
+      "step": 522
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.883358471593765e-06,
+      "loss": 1.075,
+      "step": 523
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.878330819507291e-06,
+      "loss": 1.0431,
+      "step": 524
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.873303167420815e-06,
+      "loss": 1.0386,
+      "step": 525
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 7.868275515334339e-06,
+      "loss": 1.0607,
+      "step": 526
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.863247863247863e-06,
+      "loss": 1.0703,
+      "step": 527
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.858220211161389e-06,
+      "loss": 1.0798,
+      "step": 528
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.858220211161389e-06,
+      "loss": 0.9873,
+      "step": 529
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 7.853192559074913e-06,
+      "loss": 1.0835,
+      "step": 530
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.853192559074913e-06,
+      "loss": 1.0945,
+      "step": 531
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.848164906988437e-06,
+      "loss": 1.105,
+      "step": 532
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.848164906988437e-06,
+      "loss": 1.1248,
+      "step": 533
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 7.848164906988437e-06,
+      "loss": 1.1423,
+      "step": 534
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.84313725490196e-06,
+      "loss": 1.014,
+      "step": 535
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.84313725490196e-06,
+      "loss": 1.0922,
+      "step": 536
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.838109602815486e-06,
+      "loss": 1.0255,
+      "step": 537
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 7.83308195072901e-06,
+      "loss": 1.051,
+      "step": 538
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.828054298642534e-06,
+      "loss": 1.1718,
+      "step": 539
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.823026646556058e-06,
+      "loss": 1.0556,
+      "step": 540
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.817998994469584e-06,
+      "loss": 1.1516,
+      "step": 541
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 7.812971342383108e-06,
+      "loss": 0.9913,
+      "step": 542
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.807943690296632e-06,
+      "loss": 1.0677,
+      "step": 543
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.802916038210156e-06,
+      "loss": 1.0448,
+      "step": 544
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.797888386123682e-06,
+      "loss": 0.9994,
+      "step": 545
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 7.792860734037206e-06,
+      "loss": 1.0258,
+      "step": 546
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.78783308195073e-06,
+      "loss": 1.1271,
+      "step": 547
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.782805429864253e-06,
+      "loss": 1.0228,
+      "step": 548
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.77777777777778e-06,
+      "loss": 1.1024,
+      "step": 549
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 7.772750125691303e-06,
+      "loss": 1.1003,
+      "step": 550
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.767722473604827e-06,
+      "loss": 1.0325,
+      "step": 551
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.762694821518351e-06,
+      "loss": 1.0178,
+      "step": 552
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.757667169431877e-06,
+      "loss": 1.0815,
+      "step": 553
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 7.7526395173454e-06,
+      "loss": 1.0939,
+      "step": 554
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.747611865258925e-06,
+      "loss": 1.0069,
+      "step": 555
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.742584213172449e-06,
+      "loss": 1.002,
+      "step": 556
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.737556561085974e-06,
+      "loss": 1.1107,
+      "step": 557
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 7.732528908999498e-06,
+      "loss": 1.0203,
+      "step": 558
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.727501256913022e-06,
+      "loss": 1.0807,
+      "step": 559
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.722473604826546e-06,
+      "loss": 1.0411,
+      "step": 560
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.717445952740072e-06,
+      "loss": 1.007,
+      "step": 561
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 7.717445952740072e-06,
+      "loss": 1.1029,
+      "step": 562
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 7.712418300653596e-06,
+      "loss": 1.1437,
+      "step": 563
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 7.70739064856712e-06,
+      "loss": 1.0375,
+      "step": 564
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 7.702362996480644e-06,
+      "loss": 1.0687,
+      "step": 565
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 7.69733534439417e-06,
+      "loss": 0.9667,
+      "step": 566
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 7.692307692307694e-06,
+      "loss": 1.0521,
+      "step": 567
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 7.687280040221218e-06,
+      "loss": 1.0782,
+      "step": 568
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 7.682252388134742e-06,
+      "loss": 1.088,
+      "step": 569
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 7.677224736048267e-06,
+      "loss": 1.0748,
+      "step": 570
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.672197083961791e-06,
+      "loss": 1.0666,
+      "step": 571
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.667169431875315e-06,
+      "loss": 1.0376,
+      "step": 572
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.66214177978884e-06,
+      "loss": 1.1128,
+      "step": 573
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.657114127702365e-06,
+      "loss": 1.1569,
+      "step": 574
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 7.657114127702365e-06,
+      "loss": 1.0523,
+      "step": 575
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.652086475615887e-06,
+      "loss": 1.0963,
+      "step": 576
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.647058823529411e-06,
+      "loss": 1.0911,
+      "step": 577
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.642031171442937e-06,
+      "loss": 1.0583,
+      "step": 578
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 7.642031171442937e-06,
+      "loss": 1.0038,
+      "step": 579
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.63700351935646e-06,
+      "loss": 1.0646,
+      "step": 580
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.631975867269985e-06,
+      "loss": 1.0892,
+      "step": 581
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.6269482151835096e-06,
+      "loss": 1.1478,
+      "step": 582
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 7.6219205630970335e-06,
+      "loss": 1.1033,
+      "step": 583
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.616892911010558e-06,
+      "loss": 1.0645,
+      "step": 584
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.616892911010558e-06,
+      "loss": 1.0616,
+      "step": 585
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.611865258924082e-06,
+      "loss": 1.1763,
+      "step": 586
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 7.606837606837607e-06,
+      "loss": 1.0802,
+      "step": 587
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.606837606837607e-06,
+      "loss": 1.1511,
+      "step": 588
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.601809954751131e-06,
+      "loss": 1.1436,
+      "step": 589
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.596782302664656e-06,
+      "loss": 1.0367,
+      "step": 590
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 7.59175465057818e-06,
+      "loss": 1.1174,
+      "step": 591
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.586726998491705e-06,
+      "loss": 1.0055,
+      "step": 592
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.581699346405229e-06,
+      "loss": 1.0846,
+      "step": 593
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.576671694318754e-06,
+      "loss": 0.9659,
+      "step": 594
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 7.5716440422322776e-06,
+      "loss": 1.0747,
+      "step": 595
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.566616390145802e-06,
+      "loss": 1.064,
+      "step": 596
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.566616390145802e-06,
+      "loss": 1.0863,
+      "step": 597
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.561588738059326e-06,
+      "loss": 1.047,
+      "step": 598
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 7.556561085972851e-06,
+      "loss": 1.1231,
+      "step": 599
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.551533433886375e-06,
+      "loss": 1.0532,
+      "step": 600
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.5465057817999e-06,
+      "loss": 0.9944,
+      "step": 601
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.541478129713424e-06,
+      "loss": 1.0676,
+      "step": 602
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 7.541478129713424e-06,
+      "loss": 0.9994,
+      "step": 603
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.536450477626949e-06,
+      "loss": 1.0962,
+      "step": 604
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.531422825540473e-06,
+      "loss": 1.0969,
+      "step": 605
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.526395173453998e-06,
+      "loss": 1.1008,
+      "step": 606
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 7.521367521367522e-06,
+      "loss": 1.0513,
+      "step": 607
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.516339869281046e-06,
+      "loss": 1.0573,
+      "step": 608
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.51131221719457e-06,
+      "loss": 1.1256,
+      "step": 609
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.506284565108095e-06,
+      "loss": 1.0668,
+      "step": 610
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 7.501256913021619e-06,
+      "loss": 1.1032,
+      "step": 611
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.496229260935144e-06,
+      "loss": 1.0746,
+      "step": 612
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.496229260935144e-06,
+      "loss": 1.1156,
+      "step": 613
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.491201608848668e-06,
+      "loss": 1.1969,
+      "step": 614
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 7.486173956762193e-06,
+      "loss": 1.1018,
+      "step": 615
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 7.481146304675717e-06,
+      "loss": 1.1019,
+      "step": 616
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 7.476118652589242e-06,
+      "loss": 1.1147,
+      "step": 617
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 7.471091000502766e-06,
+      "loss": 1.0922,
+      "step": 618
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 7.4660633484162904e-06,
+      "loss": 0.9863,
+      "step": 619
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.461035696329814e-06,
+      "loss": 0.9868,
+      "step": 620
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.456008044243339e-06,
+      "loss": 1.1194,
+      "step": 621
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.456008044243339e-06,
+      "loss": 1.0356,
+      "step": 622
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 7.450980392156863e-06,
+      "loss": 1.0305,
+      "step": 623
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 7.445952740070388e-06,
+      "loss": 1.0053,
+      "step": 624
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 7.440925087983912e-06,
+      "loss": 1.0355,
+      "step": 625
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 7.435897435897437e-06,
+      "loss": 1.0706,
+      "step": 626
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 7.430869783810961e-06,
+      "loss": 1.0918,
+      "step": 627
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 7.425842131724486e-06,
+      "loss": 1.0645,
+      "step": 628
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 7.42081447963801e-06,
+      "loss": 1.037,
+      "step": 629
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 7.4157868275515345e-06,
+      "loss": 1.0671,
+      "step": 630
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 7.4107591754650584e-06,
+      "loss": 1.1272,
+      "step": 631
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 7.405731523378583e-06,
+      "loss": 1.1066,
+      "step": 632
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 7.400703871292107e-06,
+      "loss": 1.0895,
+      "step": 633
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 7.395676219205632e-06,
+      "loss": 1.113,
+      "step": 634
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 7.390648567119156e-06,
+      "loss": 1.0244,
+      "step": 635
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 7.385620915032681e-06,
+      "loss": 1.0728,
+      "step": 636
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 7.380593262946205e-06,
+      "loss": 1.0979,
+      "step": 637
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 7.37556561085973e-06,
+      "loss": 1.0448,
+      "step": 638
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 7.37556561085973e-06,
+      "loss": 1.0902,
+      "step": 639
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 7.370537958773254e-06,
+      "loss": 1.0812,
+      "step": 640
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 7.370537958773254e-06,
+      "loss": 1.0151,
+      "step": 641
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 7.3655103066867785e-06,
+      "loss": 1.0353,
+      "step": 642
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 7.3604826546003025e-06,
+      "loss": 1.0134,
+      "step": 643
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 7.355455002513827e-06,
+      "loss": 1.1379,
+      "step": 644
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 7.350427350427351e-06,
+      "loss": 1.0094,
+      "step": 645
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 7.350427350427351e-06,
+      "loss": 0.9897,
+      "step": 646
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 7.345399698340876e-06,
+      "loss": 1.1305,
+      "step": 647
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 7.340372046254399e-06,
+      "loss": 1.166,
+      "step": 648
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 7.340372046254399e-06,
+      "loss": 1.0832,
+      "step": 649
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 7.335344394167924e-06,
+      "loss": 1.1058,
+      "step": 650
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 7.330316742081448e-06,
+      "loss": 0.9921,
+      "step": 651
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 7.325289089994973e-06,
+      "loss": 1.1078,
+      "step": 652
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 7.320261437908497e-06,
+      "loss": 1.0937,
+      "step": 653
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 7.320261437908497e-06,
+      "loss": 0.9514,
+      "step": 654
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 7.315233785822022e-06,
+      "loss": 1.0948,
+      "step": 655
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 7.310206133735546e-06,
+      "loss": 1.0516,
+      "step": 656
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 7.3051784816490705e-06,
+      "loss": 1.0662,
+      "step": 657
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 7.3001508295625945e-06,
+      "loss": 1.0352,
+      "step": 658
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 7.295123177476119e-06,
+      "loss": 1.0633,
+      "step": 659
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 7.290095525389643e-06,
+      "loss": 1.0676,
+      "step": 660
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 7.285067873303168e-06,
+      "loss": 1.0967,
+      "step": 661
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 7.280040221216692e-06,
+      "loss": 1.0534,
+      "step": 662
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 7.275012569130217e-06,
+      "loss": 1.0681,
+      "step": 663
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 7.269984917043741e-06,
+      "loss": 1.0842,
+      "step": 664
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 7.264957264957266e-06,
+      "loss": 1.0358,
+      "step": 665
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 7.25992961287079e-06,
+      "loss": 1.0246,
+      "step": 666
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 7.2549019607843145e-06,
+      "loss": 0.9478,
+      "step": 667
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 7.2498743086978385e-06,
+      "loss": 1.0258,
+      "step": 668
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.2448466566113625e-06,
+      "loss": 1.0536,
+      "step": 669
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.239819004524887e-06,
+      "loss": 1.1248,
+      "step": 670
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.239819004524887e-06,
+      "loss": 1.1466,
+      "step": 671
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 7.234791352438411e-06,
+      "loss": 1.0674,
+      "step": 672
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.229763700351936e-06,
+      "loss": 1.0085,
+      "step": 673
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.22473604826546e-06,
+      "loss": 1.1295,
+      "step": 674
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.219708396178985e-06,
+      "loss": 1.0961,
+      "step": 675
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 7.214680744092509e-06,
+      "loss": 1.0154,
+      "step": 676
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.209653092006034e-06,
+      "loss": 0.9776,
+      "step": 677
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.204625439919558e-06,
+      "loss": 1.0436,
+      "step": 678
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.204625439919558e-06,
+      "loss": 1.0962,
+      "step": 679
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 7.1995977878330825e-06,
+      "loss": 1.1237,
+      "step": 680
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.1945701357466065e-06,
+      "loss": 1.0508,
+      "step": 681
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.189542483660131e-06,
+      "loss": 0.9941,
+      "step": 682
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.184514831573655e-06,
+      "loss": 0.9881,
+      "step": 683
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 7.17948717948718e-06,
+      "loss": 1.0638,
+      "step": 684
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.174459527400704e-06,
+      "loss": 1.0582,
+      "step": 685
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.169431875314229e-06,
+      "loss": 1.0927,
+      "step": 686
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.164404223227753e-06,
+      "loss": 1.1293,
+      "step": 687
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 7.159376571141278e-06,
+      "loss": 1.0398,
+      "step": 688
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.154348919054802e-06,
+      "loss": 1.0667,
+      "step": 689
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.1493212669683265e-06,
+      "loss": 1.0911,
+      "step": 690
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.1493212669683265e-06,
+      "loss": 1.0892,
+      "step": 691
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 7.1442936148818505e-06,
+      "loss": 1.0458,
+      "step": 692
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.139265962795375e-06,
+      "loss": 1.0938,
+      "step": 693
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.139265962795375e-06,
+      "loss": 1.0463,
+      "step": 694
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.134238310708899e-06,
+      "loss": 1.0481,
+      "step": 695
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 7.129210658622424e-06,
+      "loss": 1.0699,
+      "step": 696
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.124183006535948e-06,
+      "loss": 1.1143,
+      "step": 697
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.124183006535948e-06,
+      "loss": 1.0712,
+      "step": 698
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.124183006535948e-06,
+      "loss": 1.0452,
+      "step": 699
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 7.119155354449473e-06,
+      "loss": 1.0402,
+      "step": 700
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.114127702362997e-06,
+      "loss": 1.1025,
+      "step": 701
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.109100050276522e-06,
+      "loss": 1.0691,
+      "step": 702
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.104072398190046e-06,
+      "loss": 1.1263,
+      "step": 703
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.104072398190046e-06,
+      "loss": 1.0457,
+      "step": 704
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.0990447461035706e-06,
+      "loss": 1.1024,
+      "step": 705
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.0940170940170945e-06,
+      "loss": 1.1003,
+      "step": 706
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.088989441930619e-06,
+      "loss": 1.0306,
+      "step": 707
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.083961789844143e-06,
+      "loss": 1.0459,
+      "step": 708
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.078934137757668e-06,
+      "loss": 1.0746,
+      "step": 709
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.073906485671192e-06,
+      "loss": 1.094,
+      "step": 710
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.068878833584717e-06,
+      "loss": 1.0489,
+      "step": 711
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.063851181498241e-06,
+      "loss": 1.052,
+      "step": 712
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.058823529411766e-06,
+      "loss": 1.1377,
+      "step": 713
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.05379587732529e-06,
+      "loss": 1.0826,
+      "step": 714
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.048768225238815e-06,
+      "loss": 1.0959,
+      "step": 715
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.0437405731523386e-06,
+      "loss": 1.1202,
+      "step": 716
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.038712921065863e-06,
+      "loss": 1.0867,
+      "step": 717
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.033685268979387e-06,
+      "loss": 1.0941,
+      "step": 718
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.028657616892911e-06,
+      "loss": 1.0648,
+      "step": 719
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.023629964806435e-06,
+      "loss": 1.1042,
+      "step": 720
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.01860231271996e-06,
+      "loss": 1.0492,
+      "step": 721
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.013574660633484e-06,
+      "loss": 1.0654,
+      "step": 722
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.008547008547009e-06,
+      "loss": 1.0619,
+      "step": 723
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.003519356460533e-06,
+      "loss": 1.0326,
+      "step": 724
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 6.998491704374058e-06,
+      "loss": 1.1339,
+      "step": 725
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.993464052287582e-06,
+      "loss": 1.0102,
+      "step": 726
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.9884364002011066e-06,
+      "loss": 1.0297,
+      "step": 727
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.9884364002011066e-06,
+      "loss": 1.0505,
+      "step": 728
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 6.9834087481146306e-06,
+      "loss": 1.0654,
+      "step": 729
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.978381096028155e-06,
+      "loss": 1.0425,
+      "step": 730
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.973353443941679e-06,
+      "loss": 1.1211,
+      "step": 731
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.968325791855204e-06,
+      "loss": 1.0192,
+      "step": 732
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 6.963298139768728e-06,
+      "loss": 1.0706,
+      "step": 733
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.963298139768728e-06,
+      "loss": 1.1045,
+      "step": 734
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.958270487682253e-06,
+      "loss": 1.0723,
+      "step": 735
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.953242835595777e-06,
+      "loss": 1.0961,
+      "step": 736
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 6.948215183509302e-06,
+      "loss": 1.1306,
+      "step": 737
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.943187531422826e-06,
+      "loss": 1.0774,
+      "step": 738
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.938159879336351e-06,
+      "loss": 1.1198,
+      "step": 739
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.933132227249875e-06,
+      "loss": 1.0814,
+      "step": 740
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 6.933132227249875e-06,
+      "loss": 1.0791,
+      "step": 741
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.928104575163399e-06,
+      "loss": 1.078,
+      "step": 742
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.923076923076923e-06,
+      "loss": 1.0336,
+      "step": 743
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.918049270990448e-06,
+      "loss": 1.0624,
+      "step": 744
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.913021618903972e-06,
+      "loss": 1.0673,
+      "step": 745
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.907993966817497e-06,
+      "loss": 1.0266,
+      "step": 746
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.902966314731021e-06,
+      "loss": 0.9991,
+      "step": 747
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.897938662644546e-06,
+      "loss": 1.025,
+      "step": 748
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.89291101055807e-06,
+      "loss": 1.0756,
+      "step": 749
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.89291101055807e-06,
+      "loss": 1.0773,
+      "step": 750
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.887883358471595e-06,
+      "loss": 1.064,
+      "step": 751
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.882855706385119e-06,
+      "loss": 1.113,
+      "step": 752
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.8778280542986434e-06,
+      "loss": 1.0341,
+      "step": 753
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.872800402212167e-06,
+      "loss": 1.0941,
+      "step": 754
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.867772750125692e-06,
+      "loss": 1.0146,
+      "step": 755
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.862745098039216e-06,
+      "loss": 1.1323,
+      "step": 756
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.857717445952741e-06,
+      "loss": 1.0707,
+      "step": 757
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.852689793866265e-06,
+      "loss": 1.0431,
+      "step": 758
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.84766214177979e-06,
+      "loss": 1.0507,
+      "step": 759
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.842634489693314e-06,
+      "loss": 1.0594,
+      "step": 760
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.837606837606839e-06,
+      "loss": 1.0815,
+      "step": 761
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.832579185520363e-06,
+      "loss": 1.0544,
+      "step": 762
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.8275515334338875e-06,
+      "loss": 0.9192,
+      "step": 763
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.8225238813474114e-06,
+      "loss": 1.0455,
+      "step": 764
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.817496229260936e-06,
+      "loss": 1.1034,
+      "step": 765
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.81246857717446e-06,
+      "loss": 1.0657,
+      "step": 766
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.807440925087985e-06,
+      "loss": 1.0487,
+      "step": 767
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.802413273001509e-06,
+      "loss": 1.0012,
+      "step": 768
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.797385620915034e-06,
+      "loss": 1.0791,
+      "step": 769
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.792357968828558e-06,
+      "loss": 1.0894,
+      "step": 770
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.787330316742083e-06,
+      "loss": 1.1,
+      "step": 771
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.782302664655607e-06,
+      "loss": 1.0281,
+      "step": 772
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.7772750125691315e-06,
+      "loss": 0.98,
+      "step": 773
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.7722473604826555e-06,
+      "loss": 1.0684,
+      "step": 774
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.76721970839618e-06,
+      "loss": 1.0419,
+      "step": 775
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.762192056309704e-06,
+      "loss": 1.0903,
+      "step": 776
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.757164404223229e-06,
+      "loss": 1.0861,
+      "step": 777
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.752136752136753e-06,
+      "loss": 1.0541,
+      "step": 778
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.747109100050278e-06,
+      "loss": 1.1064,
+      "step": 779
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.742081447963802e-06,
+      "loss": 1.0392,
+      "step": 780
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.737053795877327e-06,
+      "loss": 1.1147,
+      "step": 781
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.732026143790851e-06,
+      "loss": 1.0884,
+      "step": 782
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.7269984917043755e-06,
+      "loss": 1.085,
+      "step": 783
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.7219708396178995e-06,
+      "loss": 1.0922,
+      "step": 784
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.716943187531423e-06,
+      "loss": 1.0577,
+      "step": 785
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.7119155354449474e-06,
+      "loss": 1.0647,
+      "step": 786
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.7068878833584714e-06,
+      "loss": 1.0957,
+      "step": 787
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.701860231271996e-06,
+      "loss": 1.0017,
+      "step": 788
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.69683257918552e-06,
+      "loss": 1.0574,
+      "step": 789
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 6.691804927099045e-06,
+      "loss": 1.0835,
+      "step": 790
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.686777275012569e-06,
+      "loss": 1.0829,
+      "step": 791
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.681749622926094e-06,
+      "loss": 1.0903,
+      "step": 792
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.676721970839618e-06,
+      "loss": 1.1085,
+      "step": 793
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 6.671694318753143e-06,
+      "loss": 1.0651,
+      "step": 794
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.671694318753143e-06,
+      "loss": 1.0381,
+      "step": 795
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 1.0496,
+      "step": 796
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.6616390145801915e-06,
+      "loss": 1.093,
+      "step": 797
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 6.6566113624937155e-06,
+      "loss": 1.1029,
+      "step": 798
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.65158371040724e-06,
+      "loss": 1.05,
+      "step": 799
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.65158371040724e-06,
+      "loss": 1.054,
+      "step": 800
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.646556058320764e-06,
+      "loss": 1.0604,
+      "step": 801
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 6.641528406234289e-06,
+      "loss": 1.0473,
+      "step": 802
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.636500754147813e-06,
+      "loss": 1.1103,
+      "step": 803
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.631473102061338e-06,
+      "loss": 1.1183,
+      "step": 804
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.626445449974862e-06,
+      "loss": 1.074,
+      "step": 805
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 6.621417797888387e-06,
+      "loss": 1.0648,
+      "step": 806
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.616390145801911e-06,
+      "loss": 0.9885,
+      "step": 807
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.6113624937154355e-06,
+      "loss": 1.0173,
+      "step": 808
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.6063348416289595e-06,
+      "loss": 1.017,
+      "step": 809
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 6.601307189542484e-06,
+      "loss": 1.1294,
+      "step": 810
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.596279537456008e-06,
+      "loss": 1.0962,
+      "step": 811
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.591251885369533e-06,
+      "loss": 1.1165,
+      "step": 812
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.586224233283057e-06,
+      "loss": 1.0452,
+      "step": 813
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 6.581196581196582e-06,
+      "loss": 0.9868,
+      "step": 814
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.576168929110106e-06,
+      "loss": 1.0186,
+      "step": 815
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.571141277023631e-06,
+      "loss": 1.0504,
+      "step": 816
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.566113624937155e-06,
+      "loss": 1.0339,
+      "step": 817
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 6.5610859728506795e-06,
+      "loss": 1.0605,
+      "step": 818
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.5560583207642035e-06,
+      "loss": 0.9807,
+      "step": 819
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.5560583207642035e-06,
+      "loss": 1.0631,
+      "step": 820
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.551030668677728e-06,
+      "loss": 1.0861,
+      "step": 821
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 6.546003016591252e-06,
+      "loss": 1.0976,
+      "step": 822
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.540975364504777e-06,
+      "loss": 1.033,
+      "step": 823
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.535947712418301e-06,
+      "loss": 1.0124,
+      "step": 824
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.530920060331826e-06,
+      "loss": 1.1107,
+      "step": 825
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 6.52589240824535e-06,
+      "loss": 1.0775,
+      "step": 826
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.520864756158875e-06,
+      "loss": 1.0519,
+      "step": 827
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.515837104072399e-06,
+      "loss": 1.0991,
+      "step": 828
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.515837104072399e-06,
+      "loss": 1.0215,
+      "step": 829
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.5108094519859236e-06,
+      "loss": 1.0356,
+      "step": 830
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 6.5108094519859236e-06,
+      "loss": 1.0874,
+      "step": 831
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.5057817998994475e-06,
+      "loss": 1.0532,
+      "step": 832
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.500754147812972e-06,
+      "loss": 1.0721,
+      "step": 833
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.495726495726496e-06,
+      "loss": 0.9975,
+      "step": 834
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 6.490698843640021e-06,
+      "loss": 0.9376,
+      "step": 835
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.485671191553545e-06,
+      "loss": 1.0181,
+      "step": 836
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.485671191553545e-06,
+      "loss": 1.0458,
+      "step": 837
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.48064353946707e-06,
+      "loss": 1.0159,
+      "step": 838
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 6.475615887380594e-06,
+      "loss": 1.0464,
+      "step": 839
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.470588235294119e-06,
+      "loss": 1.0654,
+      "step": 840
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.465560583207643e-06,
+      "loss": 1.0724,
+      "step": 841
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.460532931121168e-06,
+      "loss": 1.0245,
+      "step": 842
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 6.4555052790346916e-06,
+      "loss": 1.0573,
+      "step": 843
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.450477626948216e-06,
+      "loss": 1.0502,
+      "step": 844
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.44544997486174e-06,
+      "loss": 1.095,
+      "step": 845
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.44544997486174e-06,
+      "loss": 0.9805,
+      "step": 846
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 6.440422322775265e-06,
+      "loss": 0.9622,
+      "step": 847
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.435394670688789e-06,
+      "loss": 0.9741,
+      "step": 848
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.430367018602314e-06,
+      "loss": 1.0959,
+      "step": 849
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.425339366515838e-06,
+      "loss": 1.0941,
+      "step": 850
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 6.420311714429363e-06,
+      "loss": 0.9695,
+      "step": 851
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.415284062342887e-06,
+      "loss": 1.0156,
+      "step": 852
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.410256410256412e-06,
+      "loss": 1.0682,
+      "step": 853
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.405228758169935e-06,
+      "loss": 1.0104,
+      "step": 854
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 6.400201106083459e-06,
+      "loss": 1.1437,
+      "step": 855
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.3951734539969835e-06,
+      "loss": 1.0494,
+      "step": 856
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.3901458019105075e-06,
+      "loss": 1.0314,
+      "step": 857
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.385118149824032e-06,
+      "loss": 0.9581,
+      "step": 858
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 6.380090497737556e-06,
+      "loss": 0.9487,
+      "step": 859
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.375062845651081e-06,
+      "loss": 0.9805,
+      "step": 860
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.375062845651081e-06,
+      "loss": 0.9496,
+      "step": 861
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.370035193564605e-06,
+      "loss": 1.0637,
+      "step": 862
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 6.370035193564605e-06,
+      "loss": 1.0301,
+      "step": 863
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 6.36500754147813e-06,
+      "loss": 1.0847,
+      "step": 864
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 6.359979889391654e-06,
+      "loss": 1.0935,
+      "step": 865
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 6.354952237305179e-06,
+      "loss": 1.0006,
+      "step": 866
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 6.349924585218703e-06,
+      "loss": 1.052,
+      "step": 867
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 6.3448969331322276e-06,
+      "loss": 1.0266,
+      "step": 868
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 6.3398692810457515e-06,
+      "loss": 1.0989,
+      "step": 869
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 6.3398692810457515e-06,
+      "loss": 0.9818,
+      "step": 870
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 6.334841628959276e-06,
+      "loss": 1.0447,
+      "step": 871
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 6.3298139768728e-06,
+      "loss": 1.0886,
+      "step": 872
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 6.324786324786325e-06,
+      "loss": 1.0238,
+      "step": 873
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 6.319758672699849e-06,
+      "loss": 1.1001,
+      "step": 874
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 6.314731020613374e-06,
+      "loss": 1.0465,
+      "step": 875
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 6.309703368526898e-06,
+      "loss": 1.0441,
+      "step": 876
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 6.304675716440423e-06,
+      "loss": 1.0877,
+      "step": 877
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 6.299648064353947e-06,
+      "loss": 1.0355,
+      "step": 878
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 6.299648064353947e-06,
+      "loss": 1.0714,
+      "step": 879
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 6.294620412267472e-06,
+      "loss": 1.1055,
+      "step": 880
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 6.2895927601809956e-06,
+      "loss": 1.0566,
+      "step": 881
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 6.28456510809452e-06,
+      "loss": 1.0179,
+      "step": 882
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 6.279537456008044e-06,
+      "loss": 1.0208,
+      "step": 883
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 6.274509803921569e-06,
+      "loss": 1.0834,
+      "step": 884
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 6.269482151835093e-06,
+      "loss": 1.1047,
+      "step": 885
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 6.264454499748618e-06,
+      "loss": 1.0728,
+      "step": 886
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 6.259426847662142e-06,
+      "loss": 0.9218,
+      "step": 887
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 6.254399195575667e-06,
+      "loss": 0.9885,
+      "step": 888
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 6.249371543489191e-06,
+      "loss": 0.9574,
+      "step": 889
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 6.244343891402716e-06,
+      "loss": 1.0079,
+      "step": 890
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 6.23931623931624e-06,
+      "loss": 1.0436,
+      "step": 891
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 6.2342885872297644e-06,
+      "loss": 0.9941,
+      "step": 892
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 6.229260935143288e-06,
+      "loss": 1.0424,
+      "step": 893
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 6.229260935143288e-06,
+      "loss": 0.9977,
+      "step": 894
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 6.224233283056813e-06,
+      "loss": 1.0682,
+      "step": 895
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 6.224233283056813e-06,
+      "loss": 1.004,
+      "step": 896
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 6.224233283056813e-06,
+      "loss": 1.0278,
+      "step": 897
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 6.224233283056813e-06,
+      "loss": 1.0286,
+      "step": 898
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 6.219205630970337e-06,
+      "loss": 1.0496,
+      "step": 899
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 6.214177978883862e-06,
+      "loss": 1.0062,
+      "step": 900
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 6.209150326797386e-06,
+      "loss": 1.0302,
+      "step": 901
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 6.209150326797386e-06,
+      "loss": 1.1785,
+      "step": 902
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 6.204122674710911e-06,
+      "loss": 0.9739,
+      "step": 903
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 6.199095022624435e-06,
+      "loss": 1.0064,
+      "step": 904
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 6.19406737053796e-06,
+      "loss": 1.0574,
+      "step": 905
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 6.189039718451484e-06,
+      "loss": 1.0834,
+      "step": 906
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 6.1840120663650085e-06,
+      "loss": 0.9951,
+      "step": 907
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 6.1789844142785324e-06,
+      "loss": 1.0331,
+      "step": 908
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 6.173956762192057e-06,
+      "loss": 1.0255,
+      "step": 909
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 6.168929110105581e-06,
+      "loss": 0.9549,
+      "step": 910
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 6.163901458019106e-06,
+      "loss": 1.1123,
+      "step": 911
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 6.15887380593263e-06,
+      "loss": 1.0134,
+      "step": 912
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 6.153846153846155e-06,
+      "loss": 1.0298,
+      "step": 913
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 6.148818501759679e-06,
+      "loss": 0.976,
+      "step": 914
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 6.143790849673204e-06,
+      "loss": 0.9862,
+      "step": 915
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 6.138763197586728e-06,
+      "loss": 0.9812,
+      "step": 916
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 6.138763197586728e-06,
+      "loss": 1.0077,
+      "step": 917
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 6.138763197586728e-06,
+      "loss": 0.9752,
+      "step": 918
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 6.1337355455002525e-06,
+      "loss": 1.0181,
+      "step": 919
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 6.1287078934137765e-06,
+      "loss": 0.9729,
+      "step": 920
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 6.123680241327301e-06,
+      "loss": 1.033,
+      "step": 921
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 6.118652589240825e-06,
+      "loss": 1.0361,
+      "step": 922
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 6.11362493715435e-06,
+      "loss": 1.0142,
+      "step": 923
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 6.108597285067874e-06,
+      "loss": 0.9928,
+      "step": 924
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 6.103569632981399e-06,
+      "loss": 1.0162,
+      "step": 925
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 6.098541980894923e-06,
+      "loss": 1.0857,
+      "step": 926
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 6.093514328808446e-06,
+      "loss": 1.0046,
+      "step": 927
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 6.088486676721971e-06,
+      "loss": 1.0239,
+      "step": 928
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 6.083459024635495e-06,
+      "loss": 1.0192,
+      "step": 929
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 6.07843137254902e-06,
+      "loss": 0.9696,
+      "step": 930
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 6.073403720462544e-06,
+      "loss": 0.9674,
+      "step": 931
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 6.0683760683760684e-06,
+      "loss": 1.105,
+      "step": 932
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 6.0633484162895924e-06,
+      "loss": 1.0001,
+      "step": 933
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 6.058320764203117e-06,
+      "loss": 1.0505,
+      "step": 934
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 6.058320764203117e-06,
+      "loss": 1.0048,
+      "step": 935
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 6.053293112116641e-06,
+      "loss": 1.1254,
+      "step": 936
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 6.048265460030166e-06,
+      "loss": 1.0764,
+      "step": 937
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 6.04323780794369e-06,
+      "loss": 1.0105,
+      "step": 938
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 6.038210155857215e-06,
+      "loss": 0.9719,
+      "step": 939
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 6.033182503770739e-06,
+      "loss": 1.034,
+      "step": 940
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 6.028154851684264e-06,
+      "loss": 1.056,
+      "step": 941
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 6.023127199597788e-06,
+      "loss": 1.0515,
+      "step": 942
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 6.0180995475113125e-06,
+      "loss": 0.9946,
+      "step": 943
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 6.0180995475113125e-06,
+      "loss": 0.9728,
+      "step": 944
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 6.0180995475113125e-06,
+      "loss": 1.0051,
+      "step": 945
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 6.0130718954248365e-06,
+      "loss": 1.0201,
+      "step": 946
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 6.008044243338361e-06,
+      "loss": 1.0569,
+      "step": 947
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 6.003016591251885e-06,
+      "loss": 1.0433,
+      "step": 948
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 5.99798893916541e-06,
+      "loss": 0.937,
+      "step": 949
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 5.992961287078934e-06,
+      "loss": 1.0012,
+      "step": 950
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 5.992961287078934e-06,
+      "loss": 1.0898,
+      "step": 951
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 5.987933634992459e-06,
+      "loss": 1.0867,
+      "step": 952
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 5.982905982905983e-06,
+      "loss": 1.0012,
+      "step": 953
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 5.977878330819508e-06,
+      "loss": 0.9677,
+      "step": 954
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 5.972850678733032e-06,
+      "loss": 1.0197,
+      "step": 955
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 5.9678230266465565e-06,
+      "loss": 1.1641,
+      "step": 956
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 5.9627953745600805e-06,
+      "loss": 1.0432,
+      "step": 957
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5.957767722473605e-06,
+      "loss": 1.0131,
+      "step": 958
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5.952740070387129e-06,
+      "loss": 1.0599,
+      "step": 959
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5.952740070387129e-06,
+      "loss": 1.05,
+      "step": 960
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 5.947712418300654e-06,
+      "loss": 1.0319,
+      "step": 961
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 5.942684766214178e-06,
+      "loss": 1.1218,
+      "step": 962
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 5.937657114127703e-06,
+      "loss": 1.0615,
+      "step": 963
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 5.932629462041227e-06,
+      "loss": 0.9901,
+      "step": 964
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 5.932629462041227e-06,
+      "loss": 1.0104,
+      "step": 965
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 5.927601809954752e-06,
+      "loss": 0.9924,
+      "step": 966
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 5.927601809954752e-06,
+      "loss": 1.0775,
+      "step": 967
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 5.922574157868276e-06,
+      "loss": 1.0579,
+      "step": 968
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 5.9175465057818005e-06,
+      "loss": 1.1329,
+      "step": 969
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 5.9125188536953245e-06,
+      "loss": 1.0764,
+      "step": 970
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 5.907491201608849e-06,
+      "loss": 1.0405,
+      "step": 971
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 5.902463549522373e-06,
+      "loss": 1.095,
+      "step": 972
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 5.897435897435898e-06,
+      "loss": 0.9991,
+      "step": 973
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 5.892408245349422e-06,
+      "loss": 1.0284,
+      "step": 974
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 5.887380593262947e-06,
+      "loss": 1.0044,
+      "step": 975
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 5.887380593262947e-06,
+      "loss": 1.0432,
+      "step": 976
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 5.882352941176471e-06,
+      "loss": 1.0978,
+      "step": 977
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 5.877325289089996e-06,
+      "loss": 1.0023,
+      "step": 978
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 5.877325289089996e-06,
+      "loss": 1.0684,
+      "step": 979
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 5.877325289089996e-06,
+      "loss": 1.0986,
+      "step": 980
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 5.87229763700352e-06,
+      "loss": 1.0325,
+      "step": 981
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 5.8672699849170446e-06,
+      "loss": 0.9828,
+      "step": 982
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 5.8622423328305685e-06,
+      "loss": 1.0103,
+      "step": 983
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 5.8622423328305685e-06,
+      "loss": 1.0698,
+      "step": 984
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 5.8622423328305685e-06,
+      "loss": 1.0632,
+      "step": 985
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.857214680744093e-06,
+      "loss": 1.0886,
+      "step": 986
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.852187028657617e-06,
+      "loss": 1.1171,
+      "step": 987
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.847159376571142e-06,
+      "loss": 1.0223,
+      "step": 988
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 5.842131724484666e-06,
+      "loss": 1.0329,
+      "step": 989
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.837104072398191e-06,
+      "loss": 0.97,
+      "step": 990
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.832076420311715e-06,
+      "loss": 1.0031,
+      "step": 991
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.82704876822524e-06,
+      "loss": 0.9675,
+      "step": 992
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 5.822021116138764e-06,
+      "loss": 1.0986,
+      "step": 993
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.816993464052289e-06,
+      "loss": 1.0442,
+      "step": 994
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.8119658119658126e-06,
+      "loss": 1.0245,
+      "step": 995
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.806938159879337e-06,
+      "loss": 1.0504,
+      "step": 996
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 5.801910507792861e-06,
+      "loss": 1.0769,
+      "step": 997
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.796882855706386e-06,
+      "loss": 1.0466,
+      "step": 998
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.79185520361991e-06,
+      "loss": 1.0361,
+      "step": 999
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.786827551533435e-06,
+      "loss": 1.0723,
+      "step": 1000
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 5.781799899446959e-06,
+      "loss": 1.0724,
+      "step": 1001
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.776772247360483e-06,
+      "loss": 1.03,
+      "step": 1002
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.776772247360483e-06,
+      "loss": 0.9916,
+      "step": 1003
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.771744595274007e-06,
+      "loss": 1.1066,
+      "step": 1004
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 5.766716943187532e-06,
+      "loss": 1.0405,
+      "step": 1005
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.761689291101056e-06,
+      "loss": 1.1182,
+      "step": 1006
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.7566616390145806e-06,
+      "loss": 1.1123,
+      "step": 1007
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.7516339869281045e-06,
+      "loss": 1.0347,
+      "step": 1008
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 5.746606334841629e-06,
+      "loss": 0.9921,
+      "step": 1009
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.741578682755153e-06,
+      "loss": 0.9741,
+      "step": 1010
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.736551030668678e-06,
+      "loss": 1.0395,
+      "step": 1011
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.731523378582202e-06,
+      "loss": 1.04,
+      "step": 1012
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 5.731523378582202e-06,
+      "loss": 0.9943,
+      "step": 1013
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.726495726495727e-06,
+      "loss": 1.0116,
+      "step": 1014
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.721468074409251e-06,
+      "loss": 1.0055,
+      "step": 1015
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.716440422322776e-06,
+      "loss": 1.0289,
+      "step": 1016
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 5.716440422322776e-06,
+      "loss": 1.0193,
+      "step": 1017
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.7114127702363e-06,
+      "loss": 1.0439,
+      "step": 1018
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.706385118149825e-06,
+      "loss": 1.0928,
+      "step": 1019
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.706385118149825e-06,
+      "loss": 1.0449,
+      "step": 1020
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.706385118149825e-06,
+      "loss": 1.035,
+      "step": 1021
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 5.7013574660633486e-06,
+      "loss": 1.0033,
+      "step": 1022
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.696329813976873e-06,
+      "loss": 1.0115,
+      "step": 1023
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.691302161890397e-06,
+      "loss": 0.958,
+      "step": 1024
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.686274509803922e-06,
+      "loss": 1.0032,
+      "step": 1025
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 5.681246857717446e-06,
+      "loss": 1.007,
+      "step": 1026
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.676219205630971e-06,
+      "loss": 0.9705,
+      "step": 1027
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.671191553544495e-06,
+      "loss": 1.0398,
+      "step": 1028
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.66616390145802e-06,
+      "loss": 1.0709,
+      "step": 1029
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 5.661136249371544e-06,
+      "loss": 1.0634,
+      "step": 1030
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.661136249371544e-06,
+      "loss": 1.0704,
+      "step": 1031
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.656108597285069e-06,
+      "loss": 0.9435,
+      "step": 1032
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.651080945198593e-06,
+      "loss": 1.0982,
+      "step": 1033
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 5.651080945198593e-06,
+      "loss": 1.0319,
+      "step": 1034
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.646053293112117e-06,
+      "loss": 0.9944,
+      "step": 1035
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.641025641025641e-06,
+      "loss": 1.0668,
+      "step": 1036
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.635997988939166e-06,
+      "loss": 1.0323,
+      "step": 1037
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 5.63097033685269e-06,
+      "loss": 0.9835,
+      "step": 1038
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.625942684766215e-06,
+      "loss": 1.0623,
+      "step": 1039
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.620915032679739e-06,
+      "loss": 0.9791,
+      "step": 1040
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.615887380593264e-06,
+      "loss": 1.0042,
+      "step": 1041
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 5.610859728506788e-06,
+      "loss": 1.083,
+      "step": 1042
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.610859728506788e-06,
+      "loss": 1.0041,
+      "step": 1043
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.605832076420313e-06,
+      "loss": 1.0084,
+      "step": 1044
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.600804424333837e-06,
+      "loss": 1.0568,
+      "step": 1045
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 5.5957767722473614e-06,
+      "loss": 1.0757,
+      "step": 1046
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.5907491201608854e-06,
+      "loss": 1.0405,
+      "step": 1047
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.58572146807441e-06,
+      "loss": 1.0092,
+      "step": 1048
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.58572146807441e-06,
+      "loss": 1.0247,
+      "step": 1049
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 5.580693815987934e-06,
+      "loss": 1.0721,
+      "step": 1050
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.575666163901459e-06,
+      "loss": 1.0107,
+      "step": 1051
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.570638511814983e-06,
+      "loss": 1.0363,
+      "step": 1052
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.565610859728508e-06,
+      "loss": 1.0665,
+      "step": 1053
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 5.560583207642032e-06,
+      "loss": 1.0897,
+      "step": 1054
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.555555555555557e-06,
+      "loss": 1.0718,
+      "step": 1055
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.550527903469081e-06,
+      "loss": 1.0349,
+      "step": 1056
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.5455002513826055e-06,
+      "loss": 0.9821,
+      "step": 1057
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 5.5404725992961295e-06,
+      "loss": 1.1006,
+      "step": 1058
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.535444947209654e-06,
+      "loss": 1.0389,
+      "step": 1059
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.530417295123178e-06,
+      "loss": 1.0804,
+      "step": 1060
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.525389643036703e-06,
+      "loss": 1.0651,
+      "step": 1061
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 5.520361990950227e-06,
+      "loss": 1.0833,
+      "step": 1062
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.520361990950227e-06,
+      "loss": 1.025,
+      "step": 1063
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.515334338863752e-06,
+      "loss": 1.0626,
+      "step": 1064
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.510306686777276e-06,
+      "loss": 1.0336,
+      "step": 1065
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 5.505279034690801e-06,
+      "loss": 1.0273,
+      "step": 1066
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.500251382604325e-06,
+      "loss": 1.0735,
+      "step": 1067
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.495223730517849e-06,
+      "loss": 0.9867,
+      "step": 1068
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.4901960784313735e-06,
+      "loss": 1.0463,
+      "step": 1069
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 5.4851684263448975e-06,
+      "loss": 1.0773,
+      "step": 1070
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.480140774258422e-06,
+      "loss": 1.0906,
+      "step": 1071
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.475113122171946e-06,
+      "loss": 1.1234,
+      "step": 1072
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.470085470085471e-06,
+      "loss": 1.0728,
+      "step": 1073
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 5.465057817998994e-06,
+      "loss": 1.0229,
+      "step": 1074
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.460030165912519e-06,
+      "loss": 0.9673,
+      "step": 1075
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.455002513826043e-06,
+      "loss": 1.0762,
+      "step": 1076
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.449974861739568e-06,
+      "loss": 1.0164,
+      "step": 1077
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 5.444947209653092e-06,
+      "loss": 1.0243,
+      "step": 1078
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.439919557566617e-06,
+      "loss": 1.0952,
+      "step": 1079
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.434891905480141e-06,
+      "loss": 1.0795,
+      "step": 1080
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.434891905480141e-06,
+      "loss": 0.98,
+      "step": 1081
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 5.4298642533936655e-06,
+      "loss": 1.0566,
+      "step": 1082
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.4248366013071894e-06,
+      "loss": 1.1371,
+      "step": 1083
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.419808949220714e-06,
+      "loss": 1.0516,
+      "step": 1084
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.414781297134238e-06,
+      "loss": 1.0509,
+      "step": 1085
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.409753645047763e-06,
+      "loss": 0.9852,
+      "step": 1086
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 5.404725992961287e-06,
+      "loss": 1.0632,
+      "step": 1087
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.399698340874812e-06,
+      "loss": 1.0364,
+      "step": 1088
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.399698340874812e-06,
+      "loss": 1.0129,
+      "step": 1089
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.394670688788336e-06,
+      "loss": 1.0219,
+      "step": 1090
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 5.389643036701861e-06,
+      "loss": 1.0841,
+      "step": 1091
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.384615384615385e-06,
+      "loss": 1.0865,
+      "step": 1092
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.3795877325289095e-06,
+      "loss": 0.9805,
+      "step": 1093
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.3795877325289095e-06,
+      "loss": 1.0121,
+      "step": 1094
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 5.3745600804424335e-06,
+      "loss": 1.0266,
+      "step": 1095
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.369532428355958e-06,
+      "loss": 1.016,
+      "step": 1096
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.364504776269482e-06,
+      "loss": 1.1315,
+      "step": 1097
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.359477124183007e-06,
+      "loss": 1.0706,
+      "step": 1098
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 5.354449472096531e-06,
+      "loss": 1.0275,
+      "step": 1099
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.349421820010056e-06,
+      "loss": 1.123,
+      "step": 1100
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.34439416792358e-06,
+      "loss": 1.0006,
+      "step": 1101
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.339366515837105e-06,
+      "loss": 1.0391,
+      "step": 1102
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 5.334338863750629e-06,
+      "loss": 1.0747,
+      "step": 1103
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 5.3293112116641535e-06,
+      "loss": 1.0698,
+      "step": 1104
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 5.3242835595776775e-06,
+      "loss": 1.029,
+      "step": 1105
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 5.319255907491202e-06,
+      "loss": 1.178,
+      "step": 1106
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 5.314228255404726e-06,
+      "loss": 1.0129,
+      "step": 1107
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 5.309200603318251e-06,
+      "loss": 1.1259,
+      "step": 1108
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 5.304172951231775e-06,
+      "loss": 1.0435,
+      "step": 1109
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 5.2991452991453e-06,
+      "loss": 1.0612,
+      "step": 1110
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 5.294117647058824e-06,
+      "loss": 0.9994,
+      "step": 1111
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 5.289089994972349e-06,
+      "loss": 0.9298,
+      "step": 1112
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 5.284062342885873e-06,
+      "loss": 1.1001,
+      "step": 1113
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 5.2790346907993975e-06,
+      "loss": 1.0891,
+      "step": 1114
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 5.2740070387129215e-06,
+      "loss": 1.0367,
+      "step": 1115
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 5.268979386626446e-06,
+      "loss": 1.0603,
+      "step": 1116
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 5.26395173453997e-06,
+      "loss": 1.0434,
+      "step": 1117
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 5.258924082453495e-06,
+      "loss": 1.0022,
+      "step": 1118
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 5.253896430367019e-06,
+      "loss": 1.0613,
+      "step": 1119
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 5.248868778280544e-06,
+      "loss": 1.0064,
+      "step": 1120
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 5.243841126194068e-06,
+      "loss": 1.0244,
+      "step": 1121
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 5.243841126194068e-06,
+      "loss": 1.0074,
+      "step": 1122
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 5.238813474107593e-06,
+      "loss": 0.9493,
+      "step": 1123
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 5.233785822021117e-06,
+      "loss": 1.0469,
+      "step": 1124
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 5.2287581699346416e-06,
+      "loss": 1.0635,
+      "step": 1125
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 5.2237305178481655e-06,
+      "loss": 0.9956,
+      "step": 1126
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 5.21870286576169e-06,
+      "loss": 1.0343,
+      "step": 1127
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 5.213675213675214e-06,
+      "loss": 0.9998,
+      "step": 1128
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 5.208647561588739e-06,
+      "loss": 1.0009,
+      "step": 1129
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 5.203619909502263e-06,
+      "loss": 1.0081,
+      "step": 1130
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 5.198592257415788e-06,
+      "loss": 1.0316,
+      "step": 1131
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 5.193564605329312e-06,
+      "loss": 1.005,
+      "step": 1132
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 5.188536953242837e-06,
+      "loss": 1.0332,
+      "step": 1133
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 5.183509301156361e-06,
+      "loss": 1.052,
+      "step": 1134
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 5.183509301156361e-06,
+      "loss": 1.0422,
+      "step": 1135
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 5.178481649069886e-06,
+      "loss": 1.0842,
+      "step": 1136
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 5.1734539969834096e-06,
+      "loss": 0.9758,
+      "step": 1137
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 5.168426344896934e-06,
+      "loss": 1.0163,
+      "step": 1138
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 5.163398692810458e-06,
+      "loss": 1.0847,
+      "step": 1139
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 5.158371040723983e-06,
+      "loss": 1.0306,
+      "step": 1140
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 5.153343388637506e-06,
+      "loss": 1.0382,
+      "step": 1141
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 5.14831573655103e-06,
+      "loss": 1.0919,
+      "step": 1142
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 5.143288084464555e-06,
+      "loss": 0.9483,
+      "step": 1143
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 5.138260432378079e-06,
+      "loss": 1.0193,
+      "step": 1144
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 5.133232780291604e-06,
+      "loss": 0.999,
+      "step": 1145
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 5.128205128205128e-06,
+      "loss": 0.9937,
+      "step": 1146
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 5.128205128205128e-06,
+      "loss": 1.0297,
+      "step": 1147
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 5.123177476118653e-06,
+      "loss": 1.0577,
+      "step": 1148
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 5.118149824032177e-06,
+      "loss": 0.9882,
+      "step": 1149
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 5.1131221719457016e-06,
+      "loss": 1.1034,
+      "step": 1150
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 5.1080945198592255e-06,
+      "loss": 1.0044,
+      "step": 1151
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 5.10306686777275e-06,
+      "loss": 0.9795,
+      "step": 1152
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 5.098039215686274e-06,
+      "loss": 1.0828,
+      "step": 1153
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 5.093011563599799e-06,
+      "loss": 1.0471,
+      "step": 1154
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 5.087983911513323e-06,
+      "loss": 0.9488,
+      "step": 1155
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 5.082956259426848e-06,
+      "loss": 1.014,
+      "step": 1156
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5.077928607340372e-06,
+      "loss": 0.9932,
+      "step": 1157
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5.072900955253897e-06,
+      "loss": 1.0341,
+      "step": 1158
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5.067873303167421e-06,
+      "loss": 1.115,
+      "step": 1159
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5.062845651080946e-06,
+      "loss": 0.9652,
+      "step": 1160
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.0578179989944696e-06,
+      "loss": 0.9869,
+      "step": 1161
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.052790346907994e-06,
+      "loss": 1.0743,
+      "step": 1162
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.047762694821518e-06,
+      "loss": 1.0135,
+      "step": 1163
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 5.042735042735043e-06,
+      "loss": 0.9766,
+      "step": 1164
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 5.037707390648567e-06,
+      "loss": 1.0286,
+      "step": 1165
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 5.032679738562092e-06,
+      "loss": 1.0582,
+      "step": 1166
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 5.027652086475616e-06,
+      "loss": 1.1067,
+      "step": 1167
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 5.022624434389141e-06,
+      "loss": 1.1434,
+      "step": 1168
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 5.017596782302665e-06,
+      "loss": 1.0093,
+      "step": 1169
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 5.01256913021619e-06,
+      "loss": 0.9982,
+      "step": 1170
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 5.01256913021619e-06,
+      "loss": 1.0051,
+      "step": 1171
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 5.007541478129714e-06,
+      "loss": 1.0841,
+      "step": 1172
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 5.002513826043238e-06,
+      "loss": 0.9974,
+      "step": 1173
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.997486173956762e-06,
+      "loss": 1.0757,
+      "step": 1174
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.992458521870287e-06,
+      "loss": 1.0454,
+      "step": 1175
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 4.987430869783811e-06,
+      "loss": 1.0418,
+      "step": 1176
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.982403217697336e-06,
+      "loss": 1.0468,
+      "step": 1177
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.97737556561086e-06,
+      "loss": 1.0474,
+      "step": 1178
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.972347913524385e-06,
+      "loss": 1.0331,
+      "step": 1179
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 4.967320261437909e-06,
+      "loss": 1.0256,
+      "step": 1180
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.962292609351434e-06,
+      "loss": 1.0612,
+      "step": 1181
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.957264957264958e-06,
+      "loss": 1.0647,
+      "step": 1182
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.9522373051784824e-06,
+      "loss": 0.9313,
+      "step": 1183
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 4.9472096530920064e-06,
+      "loss": 1.0606,
+      "step": 1184
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.942182001005531e-06,
+      "loss": 1.0488,
+      "step": 1185
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.937154348919055e-06,
+      "loss": 0.97,
+      "step": 1186
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.937154348919055e-06,
+      "loss": 0.9802,
+      "step": 1187
+    },
+    {
+      "epoch": 2.92,
+      "learning_rate": 4.93212669683258e-06,
+      "loss": 1.0625,
+      "step": 1188
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.927099044746104e-06,
+      "loss": 1.0308,
+      "step": 1189
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.922071392659629e-06,
+      "loss": 1.0813,
+      "step": 1190
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.917043740573153e-06,
+      "loss": 0.9568,
+      "step": 1191
+    },
+    {
+      "epoch": 2.93,
+      "learning_rate": 4.917043740573153e-06,
+      "loss": 1.0972,
+      "step": 1192
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.912016088486677e-06,
+      "loss": 1.0796,
+      "step": 1193
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.906988436400202e-06,
+      "loss": 1.0552,
+      "step": 1194
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.901960784313726e-06,
+      "loss": 1.0424,
+      "step": 1195
+    },
+    {
+      "epoch": 2.94,
+      "learning_rate": 4.8969331322272505e-06,
+      "loss": 1.0428,
+      "step": 1196
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.8919054801407744e-06,
+      "loss": 1.1007,
+      "step": 1197
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.886877828054299e-06,
+      "loss": 1.093,
+      "step": 1198
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.881850175967823e-06,
+      "loss": 1.0549,
+      "step": 1199
+    },
+    {
+      "epoch": 2.95,
+      "learning_rate": 4.876822523881348e-06,
+      "loss": 1.0211,
+      "step": 1200
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.871794871794872e-06,
+      "loss": 1.0034,
+      "step": 1201
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.866767219708397e-06,
+      "loss": 0.9501,
+      "step": 1202
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.861739567621921e-06,
+      "loss": 1.055,
+      "step": 1203
+    },
+    {
+      "epoch": 2.96,
+      "learning_rate": 4.856711915535446e-06,
+      "loss": 1.0736,
+      "step": 1204
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.85168426344897e-06,
+      "loss": 1.0099,
+      "step": 1205
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.8466566113624945e-06,
+      "loss": 1.0934,
+      "step": 1206
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.8466566113624945e-06,
+      "loss": 1.1116,
+      "step": 1207
+    },
+    {
+      "epoch": 2.97,
+      "learning_rate": 4.8466566113624945e-06,
+      "loss": 1.121,
+      "step": 1208
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.8416289592760185e-06,
+      "loss": 0.9148,
+      "step": 1209
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.836601307189543e-06,
+      "loss": 1.0393,
+      "step": 1210
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.831573655103067e-06,
+      "loss": 1.0936,
+      "step": 1211
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.826546003016592e-06,
+      "loss": 1.115,
+      "step": 1212
+    },
+    {
+      "epoch": 2.98,
+      "learning_rate": 4.821518350930116e-06,
+      "loss": 1.0529,
+      "step": 1213
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.816490698843641e-06,
+      "loss": 0.9846,
+      "step": 1214
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.811463046757165e-06,
+      "loss": 1.1254,
+      "step": 1215
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.80643539467069e-06,
+      "loss": 0.9551,
+      "step": 1216
+    },
+    {
+      "epoch": 2.99,
+      "learning_rate": 4.801407742584214e-06,
+      "loss": 1.0193,
+      "step": 1217
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.7963800904977385e-06,
+      "loss": 1.0356,
+      "step": 1218
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.7913524384112625e-06,
+      "loss": 1.0467,
+      "step": 1219
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.786324786324787e-06,
+      "loss": 0.9777,
+      "step": 1220
+    },
+    {
+      "epoch": 3.0,
+      "learning_rate": 4.781297134238311e-06,
+      "loss": 1.0151,
+      "step": 1221
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 4.776269482151836e-06,
+      "loss": 1.0221,
+      "step": 1222
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 4.77124183006536e-06,
+      "loss": 1.0883,
+      "step": 1223
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 4.766214177978885e-06,
+      "loss": 1.1123,
+      "step": 1224
+    },
+    {
+      "epoch": 3.01,
+      "learning_rate": 4.761186525892408e-06,
+      "loss": 0.8692,
+      "step": 1225
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 4.756158873805933e-06,
+      "loss": 1.0304,
+      "step": 1226
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 4.751131221719457e-06,
+      "loss": 0.9949,
+      "step": 1227
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 4.746103569632982e-06,
+      "loss": 0.9952,
+      "step": 1228
+    },
+    {
+      "epoch": 3.02,
+      "learning_rate": 4.741075917546506e-06,
+      "loss": 1.0029,
+      "step": 1229
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 4.7360482654600305e-06,
+      "loss": 1.1117,
+      "step": 1230
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 4.7310206133735545e-06,
+      "loss": 1.025,
+      "step": 1231
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 4.725992961287079e-06,
+      "loss": 1.0587,
+      "step": 1232
+    },
+    {
+      "epoch": 3.03,
+      "learning_rate": 4.720965309200603e-06,
+      "loss": 1.0082,
+      "step": 1233
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 4.715937657114128e-06,
+      "loss": 0.9791,
+      "step": 1234
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 4.710910005027652e-06,
+      "loss": 1.051,
+      "step": 1235
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 4.710910005027652e-06,
+      "loss": 1.028,
+      "step": 1236
+    },
+    {
+      "epoch": 3.04,
+      "learning_rate": 4.705882352941177e-06,
+      "loss": 0.9673,
+      "step": 1237
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 4.700854700854701e-06,
+      "loss": 0.9814,
+      "step": 1238
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 4.695827048768226e-06,
+      "loss": 0.9973,
+      "step": 1239
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 4.69079939668175e-06,
+      "loss": 0.9497,
+      "step": 1240
+    },
+    {
+      "epoch": 3.05,
+      "learning_rate": 4.6857717445952745e-06,
+      "loss": 1.0564,
+      "step": 1241
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 4.6807440925087985e-06,
+      "loss": 0.9787,
+      "step": 1242
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 4.675716440422323e-06,
+      "loss": 1.0671,
+      "step": 1243
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 4.670688788335847e-06,
+      "loss": 0.9872,
+      "step": 1244
+    },
+    {
+      "epoch": 3.06,
+      "learning_rate": 4.665661136249372e-06,
+      "loss": 1.0057,
+      "step": 1245
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 4.660633484162896e-06,
+      "loss": 1.0767,
+      "step": 1246
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 4.655605832076421e-06,
+      "loss": 0.9487,
+      "step": 1247
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 4.650578179989945e-06,
+      "loss": 1.0334,
+      "step": 1248
+    },
+    {
+      "epoch": 3.07,
+      "learning_rate": 4.64555052790347e-06,
+      "loss": 1.0316,
+      "step": 1249
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 4.640522875816994e-06,
+      "loss": 0.9885,
+      "step": 1250
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 4.6354952237305185e-06,
+      "loss": 0.9638,
+      "step": 1251
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 4.6304675716440425e-06,
+      "loss": 0.9534,
+      "step": 1252
+    },
+    {
+      "epoch": 3.08,
+      "learning_rate": 4.625439919557567e-06,
+      "loss": 0.9637,
+      "step": 1253
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 4.620412267471091e-06,
+      "loss": 1.0745,
+      "step": 1254
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 4.615384615384616e-06,
+      "loss": 1.0359,
+      "step": 1255
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 4.615384615384616e-06,
+      "loss": 1.035,
+      "step": 1256
+    },
+    {
+      "epoch": 3.09,
+      "learning_rate": 4.61035696329814e-06,
+      "loss": 1.0442,
+      "step": 1257
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 4.605329311211664e-06,
+      "loss": 1.0485,
+      "step": 1258
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 4.600301659125189e-06,
+      "loss": 0.9763,
+      "step": 1259
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 4.595274007038713e-06,
+      "loss": 1.0372,
+      "step": 1260
+    },
+    {
+      "epoch": 3.1,
+      "learning_rate": 4.590246354952238e-06,
+      "loss": 1.0365,
+      "step": 1261
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 4.585218702865762e-06,
+      "loss": 0.9702,
+      "step": 1262
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 4.5801910507792865e-06,
+      "loss": 1.0086,
+      "step": 1263
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 4.5801910507792865e-06,
+      "loss": 1.0427,
+      "step": 1264
+    },
+    {
+      "epoch": 3.11,
+      "learning_rate": 4.5751633986928105e-06,
+      "loss": 1.1077,
+      "step": 1265
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 4.570135746606335e-06,
+      "loss": 1.0967,
+      "step": 1266
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 4.565108094519859e-06,
+      "loss": 0.9045,
+      "step": 1267
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 4.560080442433384e-06,
+      "loss": 1.0549,
+      "step": 1268
+    },
+    {
+      "epoch": 3.12,
+      "learning_rate": 4.555052790346908e-06,
+      "loss": 0.9922,
+      "step": 1269
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 4.555052790346908e-06,
+      "loss": 0.9169,
+      "step": 1270
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 4.550025138260433e-06,
+      "loss": 1.0284,
+      "step": 1271
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 4.544997486173957e-06,
+      "loss": 0.9789,
+      "step": 1272
+    },
+    {
+      "epoch": 3.13,
+      "learning_rate": 4.539969834087482e-06,
+      "loss": 1.0535,
+      "step": 1273
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 4.534942182001006e-06,
+      "loss": 1.0003,
+      "step": 1274
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 4.5299145299145306e-06,
+      "loss": 1.0361,
+      "step": 1275
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 4.5248868778280546e-06,
+      "loss": 0.9832,
+      "step": 1276
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 4.519859225741579e-06,
+      "loss": 1.0499,
+      "step": 1277
+    },
+    {
+      "epoch": 3.14,
+      "learning_rate": 4.514831573655103e-06,
+      "loss": 0.9619,
+      "step": 1278
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 4.509803921568628e-06,
+      "loss": 1.0335,
+      "step": 1279
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 4.504776269482152e-06,
+      "loss": 0.9619,
+      "step": 1280
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 4.499748617395677e-06,
+      "loss": 1.0486,
+      "step": 1281
+    },
+    {
+      "epoch": 3.15,
+      "learning_rate": 4.494720965309201e-06,
+      "loss": 1.033,
+      "step": 1282
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 4.489693313222726e-06,
+      "loss": 1.0345,
+      "step": 1283
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 4.48466566113625e-06,
+      "loss": 1.0469,
+      "step": 1284
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 4.479638009049775e-06,
+      "loss": 1.0073,
+      "step": 1285
+    },
+    {
+      "epoch": 3.16,
+      "learning_rate": 4.479638009049775e-06,
+      "loss": 1.043,
+      "step": 1286
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 4.474610356963299e-06,
+      "loss": 0.9522,
+      "step": 1287
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 4.469582704876823e-06,
+      "loss": 1.0076,
+      "step": 1288
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 4.464555052790347e-06,
+      "loss": 1.026,
+      "step": 1289
+    },
+    {
+      "epoch": 3.17,
+      "learning_rate": 4.459527400703872e-06,
+      "loss": 1.0505,
+      "step": 1290
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 4.454499748617396e-06,
+      "loss": 0.9733,
+      "step": 1291
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 4.44947209653092e-06,
+      "loss": 1.014,
+      "step": 1292
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 4.444444444444444e-06,
+      "loss": 1.0285,
+      "step": 1293
+    },
+    {
+      "epoch": 3.18,
+      "learning_rate": 4.439416792357969e-06,
+      "loss": 0.9498,
+      "step": 1294
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 4.434389140271493e-06,
+      "loss": 1.0458,
+      "step": 1295
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 4.429361488185018e-06,
+      "loss": 1.0018,
+      "step": 1296
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 4.424333836098542e-06,
+      "loss": 0.967,
+      "step": 1297
+    },
+    {
+      "epoch": 3.19,
+      "learning_rate": 4.419306184012067e-06,
+      "loss": 0.9126,
+      "step": 1298
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 4.4142785319255906e-06,
+      "loss": 1.051,
+      "step": 1299
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 4.409250879839115e-06,
+      "loss": 0.9121,
+      "step": 1300
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 4.409250879839115e-06,
+      "loss": 1.069,
+      "step": 1301
+    },
+    {
+      "epoch": 3.2,
+      "learning_rate": 4.404223227752639e-06,
+      "loss": 0.9854,
+      "step": 1302
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 4.404223227752639e-06,
+      "loss": 1.0146,
+      "step": 1303
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 4.399195575666164e-06,
+      "loss": 0.9857,
+      "step": 1304
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 4.399195575666164e-06,
+      "loss": 1.0837,
+      "step": 1305
+    },
+    {
+      "epoch": 3.21,
+      "learning_rate": 4.394167923579688e-06,
+      "loss": 0.9842,
+      "step": 1306
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 4.389140271493213e-06,
+      "loss": 0.9866,
+      "step": 1307
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 4.384112619406737e-06,
+      "loss": 1.0452,
+      "step": 1308
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 4.379084967320262e-06,
+      "loss": 1.0741,
+      "step": 1309
+    },
+    {
+      "epoch": 3.22,
+      "learning_rate": 4.374057315233786e-06,
+      "loss": 0.9783,
+      "step": 1310
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 4.369029663147311e-06,
+      "loss": 1.056,
+      "step": 1311
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 4.364002011060835e-06,
+      "loss": 1.0657,
+      "step": 1312
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 4.358974358974359e-06,
+      "loss": 1.0291,
+      "step": 1313
+    },
+    {
+      "epoch": 3.23,
+      "learning_rate": 4.353946706887883e-06,
+      "loss": 1.0281,
+      "step": 1314
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 4.348919054801408e-06,
+      "loss": 1.0731,
+      "step": 1315
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 4.343891402714932e-06,
+      "loss": 0.9855,
+      "step": 1316
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 4.338863750628457e-06,
+      "loss": 1.024,
+      "step": 1317
+    },
+    {
+      "epoch": 3.24,
+      "learning_rate": 4.333836098541981e-06,
+      "loss": 0.9509,
+      "step": 1318
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 4.328808446455506e-06,
+      "loss": 1.0137,
+      "step": 1319
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 4.32378079436903e-06,
+      "loss": 1.0516,
+      "step": 1320
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 4.318753142282555e-06,
+      "loss": 0.953,
+      "step": 1321
+    },
+    {
+      "epoch": 3.25,
+      "learning_rate": 4.313725490196079e-06,
+      "loss": 1.0229,
+      "step": 1322
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 4.3086978381096034e-06,
+      "loss": 1.0613,
+      "step": 1323
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 4.303670186023127e-06,
+      "loss": 1.0347,
+      "step": 1324
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 4.303670186023127e-06,
+      "loss": 1.0728,
+      "step": 1325
+    },
+    {
+      "epoch": 3.26,
+      "learning_rate": 4.298642533936652e-06,
+      "loss": 0.9842,
+      "step": 1326
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 4.293614881850176e-06,
+      "loss": 1.0716,
+      "step": 1327
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 4.2885872297637e-06,
+      "loss": 1.0897,
+      "step": 1328
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 4.283559577677225e-06,
+      "loss": 0.979,
+      "step": 1329
+    },
+    {
+      "epoch": 3.27,
+      "learning_rate": 4.278531925590749e-06,
+      "loss": 0.9684,
+      "step": 1330
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 4.273504273504274e-06,
+      "loss": 1.0178,
+      "step": 1331
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 4.268476621417798e-06,
+      "loss": 0.9565,
+      "step": 1332
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 4.263448969331323e-06,
+      "loss": 1.001,
+      "step": 1333
+    },
+    {
+      "epoch": 3.28,
+      "learning_rate": 4.258421317244847e-06,
+      "loss": 1.0311,
+      "step": 1334
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 4.258421317244847e-06,
+      "loss": 1.0222,
+      "step": 1335
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 4.2533936651583714e-06,
+      "loss": 1.0315,
+      "step": 1336
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 4.2483660130718954e-06,
+      "loss": 1.0172,
+      "step": 1337
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 4.24333836098542e-06,
+      "loss": 1.0432,
+      "step": 1338
+    },
+    {
+      "epoch": 3.29,
+      "learning_rate": 4.24333836098542e-06,
+      "loss": 0.9611,
+      "step": 1339
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 4.238310708898944e-06,
+      "loss": 1.0641,
+      "step": 1340
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 4.233283056812469e-06,
+      "loss": 0.9781,
+      "step": 1341
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 4.228255404725993e-06,
+      "loss": 0.9818,
+      "step": 1342
+    },
+    {
+      "epoch": 3.3,
+      "learning_rate": 4.223227752639518e-06,
+      "loss": 1.0084,
+      "step": 1343
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 4.218200100553042e-06,
+      "loss": 1.0247,
+      "step": 1344
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 4.218200100553042e-06,
+      "loss": 1.0054,
+      "step": 1345
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 4.213172448466567e-06,
+      "loss": 1.0112,
+      "step": 1346
+    },
+    {
+      "epoch": 3.31,
+      "learning_rate": 4.208144796380091e-06,
+      "loss": 1.0309,
+      "step": 1347
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 4.2031171442936155e-06,
+      "loss": 0.9512,
+      "step": 1348
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 4.1980894922071395e-06,
+      "loss": 0.9635,
+      "step": 1349
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 4.193061840120664e-06,
+      "loss": 1.0276,
+      "step": 1350
+    },
+    {
+      "epoch": 3.32,
+      "learning_rate": 4.188034188034188e-06,
+      "loss": 0.9983,
+      "step": 1351
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 4.183006535947713e-06,
+      "loss": 0.9722,
+      "step": 1352
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 4.177978883861237e-06,
+      "loss": 0.9775,
+      "step": 1353
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 4.172951231774762e-06,
+      "loss": 0.9747,
+      "step": 1354
+    },
+    {
+      "epoch": 3.33,
+      "learning_rate": 4.167923579688286e-06,
+      "loss": 0.9657,
+      "step": 1355
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 4.162895927601811e-06,
+      "loss": 1.0234,
+      "step": 1356
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 4.162895927601811e-06,
+      "loss": 1.0621,
+      "step": 1357
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 4.157868275515335e-06,
+      "loss": 1.1243,
+      "step": 1358
+    },
+    {
+      "epoch": 3.34,
+      "learning_rate": 4.1528406234288595e-06,
+      "loss": 1.0297,
+      "step": 1359
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 4.1478129713423835e-06,
+      "loss": 0.9841,
+      "step": 1360
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 4.142785319255908e-06,
+      "loss": 0.9758,
+      "step": 1361
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 4.137757667169432e-06,
+      "loss": 1.0468,
+      "step": 1362
+    },
+    {
+      "epoch": 3.35,
+      "learning_rate": 4.132730015082956e-06,
+      "loss": 1.0569,
+      "step": 1363
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 4.127702362996481e-06,
+      "loss": 1.0338,
+      "step": 1364
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 4.122674710910005e-06,
+      "loss": 0.9949,
+      "step": 1365
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 4.11764705882353e-06,
+      "loss": 1.0096,
+      "step": 1366
+    },
+    {
+      "epoch": 3.36,
+      "learning_rate": 4.112619406737054e-06,
+      "loss": 1.008,
+      "step": 1367
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 4.107591754650579e-06,
+      "loss": 1.0685,
+      "step": 1368
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 4.107591754650579e-06,
+      "loss": 1.0154,
+      "step": 1369
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 4.107591754650579e-06,
+      "loss": 0.9684,
+      "step": 1370
+    },
+    {
+      "epoch": 3.37,
+      "learning_rate": 4.102564102564103e-06,
+      "loss": 0.9789,
+      "step": 1371
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 4.0975364504776275e-06,
+      "loss": 1.0791,
+      "step": 1372
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 4.0975364504776275e-06,
+      "loss": 1.0391,
+      "step": 1373
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 4.0925087983911515e-06,
+      "loss": 0.9828,
+      "step": 1374
+    },
+    {
+      "epoch": 3.38,
+      "learning_rate": 4.087481146304676e-06,
+      "loss": 0.9753,
+      "step": 1375
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.0824534942182e-06,
+      "loss": 1.0133,
+      "step": 1376
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.0824534942182e-06,
+      "loss": 0.9793,
+      "step": 1377
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.077425842131725e-06,
+      "loss": 1.0217,
+      "step": 1378
+    },
+    {
+      "epoch": 3.39,
+      "learning_rate": 4.072398190045249e-06,
+      "loss": 1.0496,
+      "step": 1379
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.067370537958774e-06,
+      "loss": 0.9963,
+      "step": 1380
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.062342885872298e-06,
+      "loss": 1.1093,
+      "step": 1381
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.057315233785823e-06,
+      "loss": 0.9639,
+      "step": 1382
+    },
+    {
+      "epoch": 3.4,
+      "learning_rate": 4.052287581699347e-06,
+      "loss": 1.016,
+      "step": 1383
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.052287581699347e-06,
+      "loss": 0.9473,
+      "step": 1384
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.0472599296128715e-06,
+      "loss": 1.0027,
+      "step": 1385
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.0422322775263955e-06,
+      "loss": 0.9884,
+      "step": 1386
+    },
+    {
+      "epoch": 3.41,
+      "learning_rate": 4.03720462543992e-06,
+      "loss": 1.0104,
+      "step": 1387
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.032176973353444e-06,
+      "loss": 0.9805,
+      "step": 1388
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.027149321266969e-06,
+      "loss": 1.0068,
+      "step": 1389
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.022121669180493e-06,
+      "loss": 0.9692,
+      "step": 1390
+    },
+    {
+      "epoch": 3.42,
+      "learning_rate": 4.017094017094018e-06,
+      "loss": 1.0592,
+      "step": 1391
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.012066365007542e-06,
+      "loss": 1.0448,
+      "step": 1392
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.007038712921067e-06,
+      "loss": 1.0164,
+      "step": 1393
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 4.002011060834591e-06,
+      "loss": 1.0275,
+      "step": 1394
+    },
+    {
+      "epoch": 3.43,
+      "learning_rate": 3.9969834087481156e-06,
+      "loss": 1.0112,
+      "step": 1395
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.9919557566616395e-06,
+      "loss": 0.9833,
+      "step": 1396
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.986928104575164e-06,
+      "loss": 1.0035,
+      "step": 1397
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.9819004524886875e-06,
+      "loss": 0.9941,
+      "step": 1398
+    },
+    {
+      "epoch": 3.44,
+      "learning_rate": 3.976872800402212e-06,
+      "loss": 1.0229,
+      "step": 1399
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.971845148315736e-06,
+      "loss": 1.0061,
+      "step": 1400
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.966817496229261e-06,
+      "loss": 0.9886,
+      "step": 1401
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.961789844142785e-06,
+      "loss": 1.0707,
+      "step": 1402
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.95676219205631e-06,
+      "loss": 1.0232,
+      "step": 1403
+    },
+    {
+      "epoch": 3.45,
+      "learning_rate": 3.951734539969834e-06,
+      "loss": 1.0127,
+      "step": 1404
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.946706887883359e-06,
+      "loss": 1.0618,
+      "step": 1405
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.941679235796883e-06,
+      "loss": 0.9695,
+      "step": 1406
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.9366515837104075e-06,
+      "loss": 0.9896,
+      "step": 1407
+    },
+    {
+      "epoch": 3.46,
+      "learning_rate": 3.9366515837104075e-06,
+      "loss": 1.0712,
+      "step": 1408
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.9316239316239315e-06,
+      "loss": 1.0096,
+      "step": 1409
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.926596279537456e-06,
+      "loss": 1.0155,
+      "step": 1410
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.926596279537456e-06,
+      "loss": 1.021,
+      "step": 1411
+    },
+    {
+      "epoch": 3.47,
+      "learning_rate": 3.92156862745098e-06,
+      "loss": 1.0099,
+      "step": 1412
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.916540975364505e-06,
+      "loss": 1.0813,
+      "step": 1413
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.911513323278029e-06,
+      "loss": 0.9805,
+      "step": 1414
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.906485671191554e-06,
+      "loss": 0.9904,
+      "step": 1415
+    },
+    {
+      "epoch": 3.48,
+      "learning_rate": 3.901458019105078e-06,
+      "loss": 1.0444,
+      "step": 1416
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.896430367018603e-06,
+      "loss": 1.0217,
+      "step": 1417
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.891402714932127e-06,
+      "loss": 0.9539,
+      "step": 1418
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.891402714932127e-06,
+      "loss": 1.0318,
+      "step": 1419
+    },
+    {
+      "epoch": 3.49,
+      "learning_rate": 3.8863750628456516e-06,
+      "loss": 1.0457,
+      "step": 1420
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.8813474107591756e-06,
+      "loss": 0.973,
+      "step": 1421
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.8763197586727e-06,
+      "loss": 1.0605,
+      "step": 1422
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.871292106586224e-06,
+      "loss": 0.9982,
+      "step": 1423
+    },
+    {
+      "epoch": 3.5,
+      "learning_rate": 3.866264454499749e-06,
+      "loss": 1.0266,
+      "step": 1424
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.866264454499749e-06,
+      "loss": 0.9447,
+      "step": 1425
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.861236802413273e-06,
+      "loss": 1.0283,
+      "step": 1426
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.856209150326798e-06,
+      "loss": 0.9956,
+      "step": 1427
+    },
+    {
+      "epoch": 3.51,
+      "learning_rate": 3.851181498240322e-06,
+      "loss": 1.0065,
+      "step": 1428
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.846153846153847e-06,
+      "loss": 0.976,
+      "step": 1429
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.841126194067371e-06,
+      "loss": 0.9547,
+      "step": 1430
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.841126194067371e-06,
+      "loss": 1.012,
+      "step": 1431
+    },
+    {
+      "epoch": 3.52,
+      "learning_rate": 3.841126194067371e-06,
+      "loss": 1.0308,
+      "step": 1432
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.841126194067371e-06,
+      "loss": 1.0,
+      "step": 1433
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.836098541980896e-06,
+      "loss": 1.0381,
+      "step": 1434
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.83107088989442e-06,
+      "loss": 1.0721,
+      "step": 1435
+    },
+    {
+      "epoch": 3.53,
+      "learning_rate": 3.8260432378079436e-06,
+      "loss": 1.0399,
+      "step": 1436
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.821015585721468e-06,
+      "loss": 0.9896,
+      "step": 1437
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.815987933634992e-06,
+      "loss": 1.0643,
+      "step": 1438
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.8109602815485168e-06,
+      "loss": 1.0875,
+      "step": 1439
+    },
+    {
+      "epoch": 3.54,
+      "learning_rate": 3.805932629462041e-06,
+      "loss": 1.0502,
+      "step": 1440
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.805932629462041e-06,
+      "loss": 0.9774,
+      "step": 1441
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.8009049773755656e-06,
+      "loss": 1.0952,
+      "step": 1442
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.79587732528909e-06,
+      "loss": 0.971,
+      "step": 1443
+    },
+    {
+      "epoch": 3.55,
+      "learning_rate": 3.7908496732026144e-06,
+      "loss": 1.0538,
+      "step": 1444
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.7858220211161388e-06,
+      "loss": 1.0017,
+      "step": 1445
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.780794369029663e-06,
+      "loss": 1.0657,
+      "step": 1446
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.7757667169431876e-06,
+      "loss": 1.0145,
+      "step": 1447
+    },
+    {
+      "epoch": 3.56,
+      "learning_rate": 3.770739064856712e-06,
+      "loss": 1.0435,
+      "step": 1448
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.7657114127702364e-06,
+      "loss": 0.9629,
+      "step": 1449
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.760683760683761e-06,
+      "loss": 1.0187,
+      "step": 1450
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.755656108597285e-06,
+      "loss": 0.9977,
+      "step": 1451
+    },
+    {
+      "epoch": 3.57,
+      "learning_rate": 3.7506284565108096e-06,
+      "loss": 0.948,
+      "step": 1452
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.745600804424334e-06,
+      "loss": 1.0317,
+      "step": 1453
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.7405731523378584e-06,
+      "loss": 0.9651,
+      "step": 1454
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.735545500251383e-06,
+      "loss": 1.065,
+      "step": 1455
+    },
+    {
+      "epoch": 3.58,
+      "learning_rate": 3.730517848164907e-06,
+      "loss": 1.0273,
+      "step": 1456
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.7254901960784316e-06,
+      "loss": 1.014,
+      "step": 1457
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.720462543991956e-06,
+      "loss": 0.9596,
+      "step": 1458
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.7154348919054804e-06,
+      "loss": 1.0174,
+      "step": 1459
+    },
+    {
+      "epoch": 3.59,
+      "learning_rate": 3.710407239819005e-06,
+      "loss": 1.174,
+      "step": 1460
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.7053795877325292e-06,
+      "loss": 1.0287,
+      "step": 1461
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.7003519356460536e-06,
+      "loss": 1.0287,
+      "step": 1462
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.695324283559578e-06,
+      "loss": 1.028,
+      "step": 1463
+    },
+    {
+      "epoch": 3.6,
+      "learning_rate": 3.6902966314731024e-06,
+      "loss": 0.9818,
+      "step": 1464
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.685268979386627e-06,
+      "loss": 1.0331,
+      "step": 1465
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.6802413273001512e-06,
+      "loss": 1.0668,
+      "step": 1466
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.6752136752136756e-06,
+      "loss": 1.0145,
+      "step": 1467
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.6701860231271996e-06,
+      "loss": 1.0527,
+      "step": 1468
+    },
+    {
+      "epoch": 3.61,
+      "learning_rate": 3.665158371040724e-06,
+      "loss": 1.0179,
+      "step": 1469
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.6601307189542484e-06,
+      "loss": 1.0017,
+      "step": 1470
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.655103066867773e-06,
+      "loss": 0.9849,
+      "step": 1471
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.6500754147812972e-06,
+      "loss": 1.0463,
+      "step": 1472
+    },
+    {
+      "epoch": 3.62,
+      "learning_rate": 3.6450477626948216e-06,
+      "loss": 0.9673,
+      "step": 1473
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.640020110608346e-06,
+      "loss": 1.0344,
+      "step": 1474
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6349924585218704e-06,
+      "loss": 1.0065,
+      "step": 1475
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.629964806435395e-06,
+      "loss": 0.9932,
+      "step": 1476
+    },
+    {
+      "epoch": 3.63,
+      "learning_rate": 3.6249371543489192e-06,
+      "loss": 1.0391,
+      "step": 1477
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6199095022624436e-06,
+      "loss": 0.9669,
+      "step": 1478
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.614881850175968e-06,
+      "loss": 1.0028,
+      "step": 1479
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.614881850175968e-06,
+      "loss": 0.9011,
+      "step": 1480
+    },
+    {
+      "epoch": 3.64,
+      "learning_rate": 3.6098541980894924e-06,
+      "loss": 1.019,
+      "step": 1481
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.604826546003017e-06,
+      "loss": 1.0533,
+      "step": 1482
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5997988939165413e-06,
+      "loss": 1.0521,
+      "step": 1483
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.5947712418300657e-06,
+      "loss": 1.0276,
+      "step": 1484
+    },
+    {
+      "epoch": 3.65,
+      "learning_rate": 3.58974358974359e-06,
+      "loss": 0.9965,
+      "step": 1485
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5847159376571145e-06,
+      "loss": 1.0624,
+      "step": 1486
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.579688285570639e-06,
+      "loss": 0.9472,
+      "step": 1487
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5746606334841633e-06,
+      "loss": 1.0351,
+      "step": 1488
+    },
+    {
+      "epoch": 3.66,
+      "learning_rate": 3.5696329813976877e-06,
+      "loss": 1.03,
+      "step": 1489
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.564605329311212e-06,
+      "loss": 1.0428,
+      "step": 1490
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.5595776772247365e-06,
+      "loss": 0.9865,
+      "step": 1491
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.554550025138261e-06,
+      "loss": 1.0483,
+      "step": 1492
+    },
+    {
+      "epoch": 3.67,
+      "learning_rate": 3.5495223730517853e-06,
+      "loss": 0.9657,
+      "step": 1493
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.5444947209653097e-06,
+      "loss": 1.0593,
+      "step": 1494
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.539467068878834e-06,
+      "loss": 0.9946,
+      "step": 1495
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.5344394167923585e-06,
+      "loss": 1.0476,
+      "step": 1496
+    },
+    {
+      "epoch": 3.68,
+      "learning_rate": 3.529411764705883e-06,
+      "loss": 1.084,
+      "step": 1497
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.529411764705883e-06,
+      "loss": 0.9495,
+      "step": 1498
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.5243841126194073e-06,
+      "loss": 1.1079,
+      "step": 1499
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.5193564605329317e-06,
+      "loss": 0.9404,
+      "step": 1500
+    },
+    {
+      "epoch": 3.69,
+      "learning_rate": 3.5193564605329317e-06,
+      "loss": 1.0254,
+      "step": 1501
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.5143288084464557e-06,
+      "loss": 0.9391,
+      "step": 1502
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.50930115635998e-06,
+      "loss": 0.9937,
+      "step": 1503
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.5042735042735045e-06,
+      "loss": 1.0103,
+      "step": 1504
+    },
+    {
+      "epoch": 3.7,
+      "learning_rate": 3.499245852187029e-06,
+      "loss": 1.043,
+      "step": 1505
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.4942182001005533e-06,
+      "loss": 0.9211,
+      "step": 1506
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.4891905480140777e-06,
+      "loss": 1.0321,
+      "step": 1507
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.4891905480140777e-06,
+      "loss": 0.9351,
+      "step": 1508
+    },
+    {
+      "epoch": 3.71,
+      "learning_rate": 3.484162895927602e-06,
+      "loss": 0.9733,
+      "step": 1509
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.4791352438411265e-06,
+      "loss": 0.9879,
+      "step": 1510
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.474107591754651e-06,
+      "loss": 0.9309,
+      "step": 1511
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.4690799396681753e-06,
+      "loss": 1.012,
+      "step": 1512
+    },
+    {
+      "epoch": 3.72,
+      "learning_rate": 3.4640522875816997e-06,
+      "loss": 1.1077,
+      "step": 1513
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.459024635495224e-06,
+      "loss": 1.0022,
+      "step": 1514
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.4539969834087485e-06,
+      "loss": 0.9499,
+      "step": 1515
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.4539969834087485e-06,
+      "loss": 1.0043,
+      "step": 1516
+    },
+    {
+      "epoch": 3.73,
+      "learning_rate": 3.448969331322273e-06,
+      "loss": 1.0856,
+      "step": 1517
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.4439416792357973e-06,
+      "loss": 1.078,
+      "step": 1518
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.4439416792357973e-06,
+      "loss": 1.0502,
+      "step": 1519
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.4389140271493217e-06,
+      "loss": 0.986,
+      "step": 1520
+    },
+    {
+      "epoch": 3.74,
+      "learning_rate": 3.433886375062846e-06,
+      "loss": 0.9989,
+      "step": 1521
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.4288587229763705e-06,
+      "loss": 1.01,
+      "step": 1522
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.423831070889895e-06,
+      "loss": 0.9619,
+      "step": 1523
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.423831070889895e-06,
+      "loss": 0.9854,
+      "step": 1524
+    },
+    {
+      "epoch": 3.75,
+      "learning_rate": 3.4188034188034193e-06,
+      "loss": 1.0089,
+      "step": 1525
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.4137757667169437e-06,
+      "loss": 1.0546,
+      "step": 1526
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.4137757667169437e-06,
+      "loss": 0.978,
+      "step": 1527
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.4137757667169437e-06,
+      "loss": 1.0791,
+      "step": 1528
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.408748114630468e-06,
+      "loss": 1.0681,
+      "step": 1529
+    },
+    {
+      "epoch": 3.76,
+      "learning_rate": 3.4037204625439925e-06,
+      "loss": 1.0609,
+      "step": 1530
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.398692810457517e-06,
+      "loss": 1.022,
+      "step": 1531
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.3936651583710413e-06,
+      "loss": 1.0048,
+      "step": 1532
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.3936651583710413e-06,
+      "loss": 1.0693,
+      "step": 1533
+    },
+    {
+      "epoch": 3.77,
+      "learning_rate": 3.3936651583710413e-06,
+      "loss": 0.9377,
+      "step": 1534
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.3886375062845657e-06,
+      "loss": 1.0574,
+      "step": 1535
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.38360985419809e-06,
+      "loss": 0.9526,
+      "step": 1536
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.3785822021116145e-06,
+      "loss": 1.0015,
+      "step": 1537
+    },
+    {
+      "epoch": 3.78,
+      "learning_rate": 3.373554550025139e-06,
+      "loss": 1.0521,
+      "step": 1538
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.3685268979386634e-06,
+      "loss": 1.0194,
+      "step": 1539
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.3634992458521878e-06,
+      "loss": 1.0588,
+      "step": 1540
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.3584715937657113e-06,
+      "loss": 1.0259,
+      "step": 1541
+    },
+    {
+      "epoch": 3.79,
+      "learning_rate": 3.3584715937657113e-06,
+      "loss": 0.9166,
+      "step": 1542
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.3534439416792357e-06,
+      "loss": 1.0056,
+      "step": 1543
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.34841628959276e-06,
+      "loss": 1.0948,
+      "step": 1544
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.3433886375062845e-06,
+      "loss": 0.9531,
+      "step": 1545
+    },
+    {
+      "epoch": 3.8,
+      "learning_rate": 3.338360985419809e-06,
+      "loss": 0.9349,
+      "step": 1546
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.3333333333333333e-06,
+      "loss": 1.0946,
+      "step": 1547
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.3283056812468577e-06,
+      "loss": 0.9987,
+      "step": 1548
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.323278029160382e-06,
+      "loss": 1.0383,
+      "step": 1549
+    },
+    {
+      "epoch": 3.81,
+      "learning_rate": 3.3182503770739065e-06,
+      "loss": 0.9905,
+      "step": 1550
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.313222724987431e-06,
+      "loss": 1.0309,
+      "step": 1551
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.3081950729009553e-06,
+      "loss": 0.9871,
+      "step": 1552
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.3081950729009553e-06,
+      "loss": 0.9934,
+      "step": 1553
+    },
+    {
+      "epoch": 3.82,
+      "learning_rate": 3.3031674208144797e-06,
+      "loss": 0.9498,
+      "step": 1554
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.298139768728004e-06,
+      "loss": 1.0497,
+      "step": 1555
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.2931121166415285e-06,
+      "loss": 0.9726,
+      "step": 1556
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.288084464555053e-06,
+      "loss": 1.0443,
+      "step": 1557
+    },
+    {
+      "epoch": 3.83,
+      "learning_rate": 3.2830568124685774e-06,
+      "loss": 1.008,
+      "step": 1558
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.2780291603821018e-06,
+      "loss": 1.0084,
+      "step": 1559
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.2780291603821018e-06,
+      "loss": 1.0435,
+      "step": 1560
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.273001508295626e-06,
+      "loss": 0.9755,
+      "step": 1561
+    },
+    {
+      "epoch": 3.84,
+      "learning_rate": 3.2679738562091506e-06,
+      "loss": 1.0446,
+      "step": 1562
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.262946204122675e-06,
+      "loss": 1.0071,
+      "step": 1563
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.2579185520361994e-06,
+      "loss": 1.001,
+      "step": 1564
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.2579185520361994e-06,
+      "loss": 1.0597,
+      "step": 1565
+    },
+    {
+      "epoch": 3.85,
+      "learning_rate": 3.2528908999497238e-06,
+      "loss": 0.984,
+      "step": 1566
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.247863247863248e-06,
+      "loss": 1.0196,
+      "step": 1567
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.2428355957767726e-06,
+      "loss": 0.9414,
+      "step": 1568
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.237807943690297e-06,
+      "loss": 1.0125,
+      "step": 1569
+    },
+    {
+      "epoch": 3.86,
+      "learning_rate": 3.237807943690297e-06,
+      "loss": 0.9719,
+      "step": 1570
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.237807943690297e-06,
+      "loss": 0.9884,
+      "step": 1571
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.2327802916038214e-06,
+      "loss": 1.0679,
+      "step": 1572
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.2277526395173458e-06,
+      "loss": 0.9326,
+      "step": 1573
+    },
+    {
+      "epoch": 3.87,
+      "learning_rate": 3.22272498743087e-06,
+      "loss": 1.0062,
+      "step": 1574
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.2176973353443946e-06,
+      "loss": 1.03,
+      "step": 1575
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.212669683257919e-06,
+      "loss": 1.0509,
+      "step": 1576
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.2076420311714434e-06,
+      "loss": 0.9342,
+      "step": 1577
+    },
+    {
+      "epoch": 3.88,
+      "learning_rate": 3.2026143790849674e-06,
+      "loss": 0.9928,
+      "step": 1578
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.1975867269984918e-06,
+      "loss": 1.0308,
+      "step": 1579
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.192559074912016e-06,
+      "loss": 1.0239,
+      "step": 1580
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.1875314228255406e-06,
+      "loss": 1.0692,
+      "step": 1581
+    },
+    {
+      "epoch": 3.89,
+      "learning_rate": 3.182503770739065e-06,
+      "loss": 0.9062,
+      "step": 1582
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.1774761186525894e-06,
+      "loss": 1.033,
+      "step": 1583
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.1724484665661138e-06,
+      "loss": 1.0757,
+      "step": 1584
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.167420814479638e-06,
+      "loss": 1.0308,
+      "step": 1585
+    },
+    {
+      "epoch": 3.9,
+      "learning_rate": 3.1623931623931626e-06,
+      "loss": 1.0281,
+      "step": 1586
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.157365510306687e-06,
+      "loss": 0.9901,
+      "step": 1587
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.1523378582202114e-06,
+      "loss": 0.9718,
+      "step": 1588
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.147310206133736e-06,
+      "loss": 0.9839,
+      "step": 1589
+    },
+    {
+      "epoch": 3.91,
+      "learning_rate": 3.14228255404726e-06,
+      "loss": 1.0007,
+      "step": 1590
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.1372549019607846e-06,
+      "loss": 0.9995,
+      "step": 1591
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.132227249874309e-06,
+      "loss": 0.9874,
+      "step": 1592
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.1271995977878334e-06,
+      "loss": 0.9717,
+      "step": 1593
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.122171945701358e-06,
+      "loss": 0.9908,
+      "step": 1594
+    },
+    {
+      "epoch": 3.92,
+      "learning_rate": 3.1171442936148822e-06,
+      "loss": 1.0078,
+      "step": 1595
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.1121166415284066e-06,
+      "loss": 1.0638,
+      "step": 1596
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.107088989441931e-06,
+      "loss": 1.0658,
+      "step": 1597
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.1020613373554554e-06,
+      "loss": 0.9933,
+      "step": 1598
+    },
+    {
+      "epoch": 3.93,
+      "learning_rate": 3.09703368526898e-06,
+      "loss": 0.9838,
+      "step": 1599
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.0920060331825042e-06,
+      "loss": 1.017,
+      "step": 1600
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.0869783810960286e-06,
+      "loss": 0.9851,
+      "step": 1601
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.081950729009553e-06,
+      "loss": 0.955,
+      "step": 1602
+    },
+    {
+      "epoch": 3.94,
+      "learning_rate": 3.0769230769230774e-06,
+      "loss": 0.9934,
+      "step": 1603
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.071895424836602e-06,
+      "loss": 0.9789,
+      "step": 1604
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.0668677727501262e-06,
+      "loss": 0.9977,
+      "step": 1605
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.0618401206636506e-06,
+      "loss": 0.9827,
+      "step": 1606
+    },
+    {
+      "epoch": 3.95,
+      "learning_rate": 3.056812468577175e-06,
+      "loss": 1.0683,
+      "step": 1607
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.0517848164906994e-06,
+      "loss": 0.9897,
+      "step": 1608
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.046757164404223e-06,
+      "loss": 1.0035,
+      "step": 1609
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.0417295123177474e-06,
+      "loss": 0.934,
+      "step": 1610
+    },
+    {
+      "epoch": 3.96,
+      "learning_rate": 3.036701860231272e-06,
+      "loss": 1.0045,
+      "step": 1611
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.0316742081447962e-06,
+      "loss": 0.9189,
+      "step": 1612
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.0266465560583206e-06,
+      "loss": 1.1182,
+      "step": 1613
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.021618903971845e-06,
+      "loss": 0.9685,
+      "step": 1614
+    },
+    {
+      "epoch": 3.97,
+      "learning_rate": 3.0165912518853694e-06,
+      "loss": 1.0292,
+      "step": 1615
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.011563599798894e-06,
+      "loss": 1.0653,
+      "step": 1616
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0065359477124182e-06,
+      "loss": 1.1378,
+      "step": 1617
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 3.0015082956259426e-06,
+      "loss": 1.0615,
+      "step": 1618
+    },
+    {
+      "epoch": 3.98,
+      "learning_rate": 2.996480643539467e-06,
+      "loss": 0.9875,
+      "step": 1619
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.9914529914529914e-06,
+      "loss": 1.0432,
+      "step": 1620
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.986425339366516e-06,
+      "loss": 1.0042,
+      "step": 1621
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.9813976872800402e-06,
+      "loss": 0.9847,
+      "step": 1622
+    },
+    {
+      "epoch": 3.99,
+      "learning_rate": 2.9763700351935646e-06,
+      "loss": 1.0173,
+      "step": 1623
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.971342383107089e-06,
+      "loss": 1.0487,
+      "step": 1624
+    },
+    {
+      "epoch": 4.0,
+      "learning_rate": 2.9663147310206134e-06,
+      "loss": 1.0081,
+      "step": 1625
+    }
+  ],
+  "max_steps": 2030,
+  "num_train_epochs": 5,
+  "total_flos": 1.6041977106074173e+18,
+  "trial_name": null,
+  "trial_params": null
+}