diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,16329 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.0,
+  "eval_steps": 500,
+  "global_step": 2718,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6260162601626018e-07,
+      "loss": 1.0542,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.2520325203252037e-07,
+      "loss": 1.0386,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.878048780487805e-07,
+      "loss": 1.0605,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.504065040650407e-07,
+      "loss": 1.0728,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.130081300813009e-07,
+      "loss": 1.043,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.75609756097561e-07,
+      "loss": 0.9863,
+      "step": 6
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.1382113821138213e-06,
+      "loss": 0.9531,
+      "step": 7
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.3008130081300815e-06,
+      "loss": 0.9302,
+      "step": 8
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.4634146341463414e-06,
+      "loss": 0.7334,
+      "step": 9
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6260162601626018e-06,
+      "loss": 0.7007,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.788617886178862e-06,
+      "loss": 0.6694,
+      "step": 11
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.951219512195122e-06,
+      "loss": 0.3904,
+      "step": 12
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1138211382113824e-06,
+      "loss": 0.3696,
+      "step": 13
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2764227642276426e-06,
+      "loss": 0.3311,
+      "step": 14
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4390243902439027e-06,
+      "loss": 0.3379,
+      "step": 15
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.601626016260163e-06,
+      "loss": 0.1733,
+      "step": 16
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.764227642276423e-06,
+      "loss": 0.1578,
+      "step": 17
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.926829268292683e-06,
+      "loss": 0.138,
+      "step": 18
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0894308943089435e-06,
+      "loss": 0.1503,
+      "step": 19
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2520325203252037e-06,
+      "loss": 0.1407,
+      "step": 20
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.414634146341464e-06,
+      "loss": 0.1399,
+      "step": 21
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.577235772357724e-06,
+      "loss": 0.1733,
+      "step": 22
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.7398373983739838e-06,
+      "loss": 0.1832,
+      "step": 23
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 3.902439024390244e-06,
+      "loss": 0.176,
+      "step": 24
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.0650406504065046e-06,
+      "loss": 0.131,
+      "step": 25
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.227642276422765e-06,
+      "loss": 0.1376,
+      "step": 26
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.390243902439025e-06,
+      "loss": 0.1592,
+      "step": 27
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.552845528455285e-06,
+      "loss": 0.141,
+      "step": 28
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.715447154471545e-06,
+      "loss": 0.1349,
+      "step": 29
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 4.8780487804878055e-06,
+      "loss": 0.1295,
+      "step": 30
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.040650406504065e-06,
+      "loss": 0.1256,
+      "step": 31
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.203252032520326e-06,
+      "loss": 0.1315,
+      "step": 32
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.365853658536586e-06,
+      "loss": 0.1288,
+      "step": 33
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.528455284552846e-06,
+      "loss": 0.1216,
+      "step": 34
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.691056910569106e-06,
+      "loss": 0.1336,
+      "step": 35
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 5.853658536585366e-06,
+      "loss": 0.116,
+      "step": 36
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.016260162601627e-06,
+      "loss": 0.1231,
+      "step": 37
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.178861788617887e-06,
+      "loss": 0.1296,
+      "step": 38
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.341463414634147e-06,
+      "loss": 0.1226,
+      "step": 39
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.504065040650407e-06,
+      "loss": 0.1202,
+      "step": 40
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 0.1241,
+      "step": 41
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.829268292682928e-06,
+      "loss": 0.1221,
+      "step": 42
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 6.991869918699188e-06,
+      "loss": 0.1196,
+      "step": 43
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.154471544715448e-06,
+      "loss": 0.1255,
+      "step": 44
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.317073170731707e-06,
+      "loss": 0.1156,
+      "step": 45
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.4796747967479676e-06,
+      "loss": 0.1179,
+      "step": 46
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 7.64227642276423e-06,
+      "loss": 0.1192,
+      "step": 47
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.804878048780489e-06,
+      "loss": 0.1231,
+      "step": 48
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 7.967479674796748e-06,
+      "loss": 0.1182,
+      "step": 49
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.130081300813009e-06,
+      "loss": 0.1152,
+      "step": 50
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.292682926829268e-06,
+      "loss": 0.1221,
+      "step": 51
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.45528455284553e-06,
+      "loss": 0.1194,
+      "step": 52
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.617886178861789e-06,
+      "loss": 0.1189,
+      "step": 53
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.78048780487805e-06,
+      "loss": 0.1107,
+      "step": 54
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 8.94308943089431e-06,
+      "loss": 0.1132,
+      "step": 55
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.10569105691057e-06,
+      "loss": 0.1172,
+      "step": 56
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.268292682926831e-06,
+      "loss": 0.1127,
+      "step": 57
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.43089430894309e-06,
+      "loss": 0.1146,
+      "step": 58
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.59349593495935e-06,
+      "loss": 0.1107,
+      "step": 59
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.756097560975611e-06,
+      "loss": 0.1115,
+      "step": 60
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.91869918699187e-06,
+      "loss": 0.1099,
+      "step": 61
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.008130081300813e-05,
+      "loss": 0.1116,
+      "step": 62
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.024390243902439e-05,
+      "loss": 0.1148,
+      "step": 63
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0406504065040652e-05,
+      "loss": 0.111,
+      "step": 64
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0569105691056911e-05,
+      "loss": 0.1146,
+      "step": 65
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0731707317073172e-05,
+      "loss": 0.1071,
+      "step": 66
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.0894308943089431e-05,
+      "loss": 0.1031,
+      "step": 67
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1056910569105692e-05,
+      "loss": 0.1119,
+      "step": 68
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1219512195121953e-05,
+      "loss": 0.111,
+      "step": 69
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1382113821138213e-05,
+      "loss": 0.1061,
+      "step": 70
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1544715447154474e-05,
+      "loss": 0.1135,
+      "step": 71
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1707317073170731e-05,
+      "loss": 0.1066,
+      "step": 72
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.1869918699186992e-05,
+      "loss": 0.105,
+      "step": 73
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.2032520325203254e-05,
+      "loss": 0.1049,
+      "step": 74
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2195121951219513e-05,
+      "loss": 0.1103,
+      "step": 75
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2357723577235774e-05,
+      "loss": 0.1064,
+      "step": 76
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2520325203252033e-05,
+      "loss": 0.1176,
+      "step": 77
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2682926829268294e-05,
+      "loss": 0.1157,
+      "step": 78
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.2845528455284555e-05,
+      "loss": 0.1082,
+      "step": 79
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3008130081300815e-05,
+      "loss": 0.0941,
+      "step": 80
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3170731707317076e-05,
+      "loss": 0.0921,
+      "step": 81
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 0.0936,
+      "step": 82
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3495934959349594e-05,
+      "loss": 0.1082,
+      "step": 83
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3658536585365855e-05,
+      "loss": 0.0958,
+      "step": 84
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3821138211382115e-05,
+      "loss": 0.0927,
+      "step": 85
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.3983739837398376e-05,
+      "loss": 0.0836,
+      "step": 86
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4146341463414635e-05,
+      "loss": 0.0852,
+      "step": 87
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.4308943089430896e-05,
+      "loss": 0.0822,
+      "step": 88
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4471544715447157e-05,
+      "loss": 0.0806,
+      "step": 89
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4634146341463415e-05,
+      "loss": 0.0892,
+      "step": 90
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4796747967479676e-05,
+      "loss": 0.082,
+      "step": 91
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.4959349593495935e-05,
+      "loss": 0.0712,
+      "step": 92
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5121951219512196e-05,
+      "loss": 0.0825,
+      "step": 93
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.528455284552846e-05,
+      "loss": 0.0763,
+      "step": 94
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5447154471544717e-05,
+      "loss": 0.0786,
+      "step": 95
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5609756097560978e-05,
+      "loss": 0.0875,
+      "step": 96
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5772357723577235e-05,
+      "loss": 0.0778,
+      "step": 97
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.5934959349593496e-05,
+      "loss": 0.0811,
+      "step": 98
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6097560975609757e-05,
+      "loss": 0.0784,
+      "step": 99
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.6260162601626018e-05,
+      "loss": 0.0795,
+      "step": 100
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.642276422764228e-05,
+      "loss": 0.067,
+      "step": 101
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6585365853658537e-05,
+      "loss": 0.0809,
+      "step": 102
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.6747967479674798e-05,
+      "loss": 0.0681,
+      "step": 103
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.691056910569106e-05,
+      "loss": 0.0787,
+      "step": 104
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7073170731707317e-05,
+      "loss": 0.074,
+      "step": 105
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.7235772357723578e-05,
+      "loss": 0.0724,
+      "step": 106
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.739837398373984e-05,
+      "loss": 0.0685,
+      "step": 107
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.75609756097561e-05,
+      "loss": 0.0746,
+      "step": 108
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.772357723577236e-05,
+      "loss": 0.0797,
+      "step": 109
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.788617886178862e-05,
+      "loss": 0.0811,
+      "step": 110
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.804878048780488e-05,
+      "loss": 0.0747,
+      "step": 111
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.821138211382114e-05,
+      "loss": 0.0758,
+      "step": 112
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.83739837398374e-05,
+      "loss": 0.0754,
+      "step": 113
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.8536585365853663e-05,
+      "loss": 0.0721,
+      "step": 114
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.869918699186992e-05,
+      "loss": 0.0696,
+      "step": 115
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.886178861788618e-05,
+      "loss": 0.0737,
+      "step": 116
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.902439024390244e-05,
+      "loss": 0.0796,
+      "step": 117
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.91869918699187e-05,
+      "loss": 0.0642,
+      "step": 118
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.934959349593496e-05,
+      "loss": 0.0761,
+      "step": 119
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9512195121951222e-05,
+      "loss": 0.0623,
+      "step": 120
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9674796747967483e-05,
+      "loss": 0.0603,
+      "step": 121
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983739837398374e-05,
+      "loss": 0.0643,
+      "step": 122
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 2e-05,
+      "loss": 0.08,
+      "step": 123
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.99999968435683e-05,
+      "loss": 0.066,
+      "step": 124
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999987374275175e-05,
+      "loss": 0.0912,
+      "step": 125
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.999997159212662e-05,
+      "loss": 0.0665,
+      "step": 126
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999949497132587e-05,
+      "loss": 0.0747,
+      "step": 127
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999921089307026e-05,
+      "loss": 0.0729,
+      "step": 128
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9999886368667875e-05,
+      "loss": 0.0722,
+      "step": 129
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999984533523705e-05,
+      "loss": 0.0735,
+      "step": 130
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999979798904045e-05,
+      "loss": 0.0655,
+      "step": 131
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999744330107972e-05,
+      "loss": 0.0559,
+      "step": 132
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999684358473485e-05,
+      "loss": 0.0688,
+      "step": 133
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999618074174848e-05,
+      "loss": 0.0634,
+      "step": 134
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.999954547725391e-05,
+      "loss": 0.062,
+      "step": 135
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.99994665677565e-05,
+      "loss": 0.0751,
+      "step": 136
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999381345732423e-05,
+      "loss": 0.0737,
+      "step": 137
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999289811235492e-05,
+      "loss": 0.0576,
+      "step": 138
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999191964323482e-05,
+      "loss": 0.0653,
+      "step": 139
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9999087805058166e-05,
+      "loss": 0.0614,
+      "step": 140
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9998977333505298e-05,
+      "loss": 0.064,
+      "step": 141
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9998860549734618e-05,
+      "loss": 0.0668,
+      "step": 142
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998737453819846e-05,
+      "loss": 0.074,
+      "step": 143
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998608045838696e-05,
+      "loss": 0.0772,
+      "step": 144
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999847232587286e-05,
+      "loss": 0.0667,
+      "step": 145
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998330294008016e-05,
+      "loss": 0.0588,
+      "step": 146
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998181950333825e-05,
+      "loss": 0.0603,
+      "step": 147
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9998027294943934e-05,
+      "loss": 0.0629,
+      "step": 148
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999786632793598e-05,
+      "loss": 0.0616,
+      "step": 149
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.999769904941157e-05,
+      "loss": 0.0559,
+      "step": 150
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997525459476313e-05,
+      "loss": 0.0684,
+      "step": 151
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997345558239794e-05,
+      "loss": 0.0644,
+      "step": 152
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9997159345815577e-05,
+      "loss": 0.0556,
+      "step": 153
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9996966822321217e-05,
+      "loss": 0.0656,
+      "step": 154
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9996767987878254e-05,
+      "loss": 0.0615,
+      "step": 155
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9996562842612208e-05,
+      "loss": 0.0651,
+      "step": 156
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9996351386652583e-05,
+      "loss": 0.0669,
+      "step": 157
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999613362013287e-05,
+      "loss": 0.0646,
+      "step": 158
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999590954319054e-05,
+      "loss": 0.0575,
+      "step": 159
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999567915596705e-05,
+      "loss": 0.062,
+      "step": 160
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995442458607846e-05,
+      "loss": 0.0604,
+      "step": 161
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9995199451262348e-05,
+      "loss": 0.0613,
+      "step": 162
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994950134083954e-05,
+      "loss": 0.0574,
+      "step": 163
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994694507230072e-05,
+      "loss": 0.069,
+      "step": 164
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.999443257086206e-05,
+      "loss": 0.0598,
+      "step": 165
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9994164325145283e-05,
+      "loss": 0.0623,
+      "step": 166
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993889770249084e-05,
+      "loss": 0.059,
+      "step": 167
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993608906346778e-05,
+      "loss": 0.0593,
+      "step": 168
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9993321733615674e-05,
+      "loss": 0.0522,
+      "step": 169
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.999302825223706e-05,
+      "loss": 0.0632,
+      "step": 170
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9992728462396207e-05,
+      "loss": 0.0519,
+      "step": 171
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9992422364282367e-05,
+      "loss": 0.0671,
+      "step": 172
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9992109958088778e-05,
+      "loss": 0.0628,
+      "step": 173
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9991791244012654e-05,
+      "loss": 0.0549,
+      "step": 174
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.99914662222552e-05,
+      "loss": 0.067,
+      "step": 175
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9991134893021594e-05,
+      "loss": 0.0615,
+      "step": 176
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9990797256521e-05,
+      "loss": 0.0705,
+      "step": 177
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9990453312966564e-05,
+      "loss": 0.0566,
+      "step": 178
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9990103062575415e-05,
+      "loss": 0.0602,
+      "step": 179
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9989746505568655e-05,
+      "loss": 0.0573,
+      "step": 180
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9989383642171377e-05,
+      "loss": 0.0612,
+      "step": 181
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9989014472612654e-05,
+      "loss": 0.0518,
+      "step": 182
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.998863899712554e-05,
+      "loss": 0.057,
+      "step": 183
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998825721594706e-05,
+      "loss": 0.0597,
+      "step": 184
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9987869129318232e-05,
+      "loss": 0.0525,
+      "step": 185
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998747473748405e-05,
+      "loss": 0.0627,
+      "step": 186
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998707404069348e-05,
+      "loss": 0.0562,
+      "step": 187
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9986667039199488e-05,
+      "loss": 0.0595,
+      "step": 188
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9986253733259004e-05,
+      "loss": 0.0629,
+      "step": 189
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9985834123132943e-05,
+      "loss": 0.0584,
+      "step": 190
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9985408209086195e-05,
+      "loss": 0.0634,
+      "step": 191
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.998497599138764e-05,
+      "loss": 0.0553,
+      "step": 192
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9984537470310127e-05,
+      "loss": 0.0544,
+      "step": 193
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9984092646130487e-05,
+      "loss": 0.0596,
+      "step": 194
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9983641519129534e-05,
+      "loss": 0.0562,
+      "step": 195
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9983184089592062e-05,
+      "loss": 0.0667,
+      "step": 196
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9982720357806835e-05,
+      "loss": 0.0544,
+      "step": 197
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.99822503240666e-05,
+      "loss": 0.0569,
+      "step": 198
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9981773988668086e-05,
+      "loss": 0.0515,
+      "step": 199
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9981291351911995e-05,
+      "loss": 0.059,
+      "step": 200
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.998080241410301e-05,
+      "loss": 0.0647,
+      "step": 201
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.998030717554979e-05,
+      "loss": 0.0588,
+      "step": 202
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9979805636564974e-05,
+      "loss": 0.0581,
+      "step": 203
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997929779746517e-05,
+      "loss": 0.0641,
+      "step": 204
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9978783658570982e-05,
+      "loss": 0.0562,
+      "step": 205
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9978263220206967e-05,
+      "loss": 0.051,
+      "step": 206
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997773648270168e-05,
+      "loss": 0.0658,
+      "step": 207
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9977203446387638e-05,
+      "loss": 0.0578,
+      "step": 208
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9976664111601337e-05,
+      "loss": 0.0641,
+      "step": 209
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.997611847868326e-05,
+      "loss": 0.068,
+      "step": 210
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997556654797785e-05,
+      "loss": 0.0666,
+      "step": 211
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997500831983354e-05,
+      "loss": 0.0734,
+      "step": 212
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9974443794602723e-05,
+      "loss": 0.0562,
+      "step": 213
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9973872972641783e-05,
+      "loss": 0.0605,
+      "step": 214
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997329585431107e-05,
+      "loss": 0.0662,
+      "step": 215
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9972712439974912e-05,
+      "loss": 0.0486,
+      "step": 216
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997212273000161e-05,
+      "loss": 0.0601,
+      "step": 217
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9971526724763438e-05,
+      "loss": 0.0551,
+      "step": 218
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9970924424636645e-05,
+      "loss": 0.0522,
+      "step": 219
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.997031583000146e-05,
+      "loss": 0.0489,
+      "step": 220
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.996970094124208e-05,
+      "loss": 0.0602,
+      "step": 221
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.996907975874667e-05,
+      "loss": 0.0551,
+      "step": 222
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9968452282907377e-05,
+      "loss": 0.0482,
+      "step": 223
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.996781851412032e-05,
+      "loss": 0.0577,
+      "step": 224
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9967178452785586e-05,
+      "loss": 0.0581,
+      "step": 225
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996653209930724e-05,
+      "loss": 0.0595,
+      "step": 226
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9965879454093313e-05,
+      "loss": 0.0644,
+      "step": 227
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9965220517555814e-05,
+      "loss": 0.0495,
+      "step": 228
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9964555290110718e-05,
+      "loss": 0.0495,
+      "step": 229
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9963883772177973e-05,
+      "loss": 0.0582,
+      "step": 230
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9963205964181503e-05,
+      "loss": 0.0596,
+      "step": 231
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9962521866549194e-05,
+      "loss": 0.0495,
+      "step": 232
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9961831479712916e-05,
+      "loss": 0.0488,
+      "step": 233
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.996113480410849e-05,
+      "loss": 0.0544,
+      "step": 234
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9960431840175722e-05,
+      "loss": 0.0572,
+      "step": 235
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9959722588358387e-05,
+      "loss": 0.046,
+      "step": 236
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9959007049104223e-05,
+      "loss": 0.0505,
+      "step": 237
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995828522286494e-05,
+      "loss": 0.0648,
+      "step": 238
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995755711009622e-05,
+      "loss": 0.0517,
+      "step": 239
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9956822711257708e-05,
+      "loss": 0.0503,
+      "step": 240
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995608202681302e-05,
+      "loss": 0.0508,
+      "step": 241
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9955335057229738e-05,
+      "loss": 0.0589,
+      "step": 242
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995458180297942e-05,
+      "loss": 0.0493,
+      "step": 243
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9953822264537573e-05,
+      "loss": 0.0595,
+      "step": 244
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.99530564423837e-05,
+      "loss": 0.0471,
+      "step": 245
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9952284337001238e-05,
+      "loss": 0.0573,
+      "step": 246
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.995150594887762e-05,
+      "loss": 0.0694,
+      "step": 247
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9950721278504222e-05,
+      "loss": 0.0619,
+      "step": 248
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9949930326376403e-05,
+      "loss": 0.0527,
+      "step": 249
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.9949133092993473e-05,
+      "loss": 0.0566,
+      "step": 250
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.994832957885872e-05,
+      "loss": 0.0569,
+      "step": 251
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994751978447939e-05,
+      "loss": 0.0602,
+      "step": 252
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9946703710366693e-05,
+      "loss": 0.0534,
+      "step": 253
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9945881357035808e-05,
+      "loss": 0.0647,
+      "step": 254
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994505272500588e-05,
+      "loss": 0.0619,
+      "step": 255
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.99442178148e-05,
+      "loss": 0.0567,
+      "step": 256
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9943376626945246e-05,
+      "loss": 0.0482,
+      "step": 257
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9942529161972646e-05,
+      "loss": 0.0493,
+      "step": 258
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994167542041719e-05,
+      "loss": 0.0573,
+      "step": 259
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.994081540281784e-05,
+      "loss": 0.059,
+      "step": 260
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.993994910971751e-05,
+      "loss": 0.0631,
+      "step": 261
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9939076541663077e-05,
+      "loss": 0.0562,
+      "step": 262
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.993819769920538e-05,
+      "loss": 0.0599,
+      "step": 263
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9937312582899224e-05,
+      "loss": 0.0508,
+      "step": 264
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.9936421193303373e-05,
+      "loss": 0.0526,
+      "step": 265
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9935523530980542e-05,
+      "loss": 0.0532,
+      "step": 266
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993461959649742e-05,
+      "loss": 0.0574,
+      "step": 267
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.993370939042464e-05,
+      "loss": 0.0519,
+      "step": 268
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9932792913336814e-05,
+      "loss": 0.0542,
+      "step": 269
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9931870165812492e-05,
+      "loss": 0.0579,
+      "step": 270
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9930941148434197e-05,
+      "loss": 0.0534,
+      "step": 271
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.99300058617884e-05,
+      "loss": 0.0624,
+      "step": 272
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9929064306465543e-05,
+      "loss": 0.0515,
+      "step": 273
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.992811648306001e-05,
+      "loss": 0.0472,
+      "step": 274
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.992716239217015e-05,
+      "loss": 0.0634,
+      "step": 275
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.992620203439827e-05,
+      "loss": 0.0559,
+      "step": 276
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.992523541035063e-05,
+      "loss": 0.0486,
+      "step": 277
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.9924262520637442e-05,
+      "loss": 0.0591,
+      "step": 278
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9923283365872886e-05,
+      "loss": 0.0663,
+      "step": 279
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9922297946675083e-05,
+      "loss": 0.0481,
+      "step": 280
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9921306263666115e-05,
+      "loss": 0.0595,
+      "step": 281
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9920308317472023e-05,
+      "loss": 0.0571,
+      "step": 282
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991930410872279e-05,
+      "loss": 0.0535,
+      "step": 283
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9918293638052365e-05,
+      "loss": 0.0476,
+      "step": 284
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9917276906098643e-05,
+      "loss": 0.0551,
+      "step": 285
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991625391350347e-05,
+      "loss": 0.0536,
+      "step": 286
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9915224660912647e-05,
+      "loss": 0.0519,
+      "step": 287
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.991418914897593e-05,
+      "loss": 0.049,
+      "step": 288
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9913147378347025e-05,
+      "loss": 0.0563,
+      "step": 289
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9912099349683586e-05,
+      "loss": 0.0527,
+      "step": 290
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9911045063647214e-05,
+      "loss": 0.0463,
+      "step": 291
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.9909984520903474e-05,
+      "loss": 0.0576,
+      "step": 292
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9908917722121867e-05,
+      "loss": 0.0522,
+      "step": 293
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9907844667975847e-05,
+      "loss": 0.056,
+      "step": 294
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9906765359142825e-05,
+      "loss": 0.0504,
+      "step": 295
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9905679796304148e-05,
+      "loss": 0.0473,
+      "step": 296
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9904587980145117e-05,
+      "loss": 0.0427,
+      "step": 297
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9903489911354986e-05,
+      "loss": 0.0466,
+      "step": 298
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9902385590626947e-05,
+      "loss": 0.0499,
+      "step": 299
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990127501865814e-05,
+      "loss": 0.042,
+      "step": 300
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.990015819614966e-05,
+      "loss": 0.0458,
+      "step": 301
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9899035123806536e-05,
+      "loss": 0.0556,
+      "step": 302
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.989790580233775e-05,
+      "loss": 0.064,
+      "step": 303
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9896770232456233e-05,
+      "loss": 0.0513,
+      "step": 304
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.9895628414878848e-05,
+      "loss": 0.0491,
+      "step": 305
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.989448035032641e-05,
+      "loss": 0.0575,
+      "step": 306
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9893326039523675e-05,
+      "loss": 0.0537,
+      "step": 307
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9892165483199347e-05,
+      "loss": 0.058,
+      "step": 308
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.989099868208607e-05,
+      "loss": 0.0503,
+      "step": 309
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9889825636920427e-05,
+      "loss": 0.0495,
+      "step": 310
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9888646348442944e-05,
+      "loss": 0.0565,
+      "step": 311
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9887460817398093e-05,
+      "loss": 0.0556,
+      "step": 312
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9886269044534283e-05,
+      "loss": 0.0535,
+      "step": 313
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9885071030603864e-05,
+      "loss": 0.0526,
+      "step": 314
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9883866776363123e-05,
+      "loss": 0.0642,
+      "step": 315
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9882656282572293e-05,
+      "loss": 0.05,
+      "step": 316
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.988143954999554e-05,
+      "loss": 0.0482,
+      "step": 317
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9880216579400972e-05,
+      "loss": 0.05,
+      "step": 318
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.9878987371560633e-05,
+      "loss": 0.0563,
+      "step": 319
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9877751927250505e-05,
+      "loss": 0.0515,
+      "step": 320
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9876510247250506e-05,
+      "loss": 0.0564,
+      "step": 321
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9875262332344495e-05,
+      "loss": 0.0465,
+      "step": 322
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.987400818332026e-05,
+      "loss": 0.0415,
+      "step": 323
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9872747800969526e-05,
+      "loss": 0.053,
+      "step": 324
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9871481186087964e-05,
+      "loss": 0.0538,
+      "step": 325
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9870208339475162e-05,
+      "loss": 0.0463,
+      "step": 326
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9868929261934657e-05,
+      "loss": 0.0528,
+      "step": 327
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.986764395427391e-05,
+      "loss": 0.0481,
+      "step": 328
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9866352417304316e-05,
+      "loss": 0.0454,
+      "step": 329
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.986505465184121e-05,
+      "loss": 0.0666,
+      "step": 330
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9863750658703846e-05,
+      "loss": 0.0436,
+      "step": 331
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.9862440438715427e-05,
+      "loss": 0.052,
+      "step": 332
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.986112399270307e-05,
+      "loss": 0.0513,
+      "step": 333
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985980132149783e-05,
+      "loss": 0.0546,
+      "step": 334
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985847242593469e-05,
+      "loss": 0.0443,
+      "step": 335
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985713730685257e-05,
+      "loss": 0.0455,
+      "step": 336
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9855795965094307e-05,
+      "loss": 0.0474,
+      "step": 337
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9854448401506677e-05,
+      "loss": 0.0444,
+      "step": 338
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985309461694037e-05,
+      "loss": 0.0667,
+      "step": 339
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9851734612250015e-05,
+      "loss": 0.0847,
+      "step": 340
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.985036838829417e-05,
+      "loss": 0.068,
+      "step": 341
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9848995945935305e-05,
+      "loss": 0.0545,
+      "step": 342
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9847617286039834e-05,
+      "loss": 0.0635,
+      "step": 343
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9846232409478076e-05,
+      "loss": 0.0602,
+      "step": 344
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.984484131712429e-05,
+      "loss": 0.0695,
+      "step": 345
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.9843444009856652e-05,
+      "loss": 0.0536,
+      "step": 346
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9842040488557267e-05,
+      "loss": 0.0498,
+      "step": 347
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9840630754112152e-05,
+      "loss": 0.0589,
+      "step": 348
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.983921480741126e-05,
+      "loss": 0.0601,
+      "step": 349
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9837792649348454e-05,
+      "loss": 0.0548,
+      "step": 350
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9836364280821522e-05,
+      "loss": 0.0511,
+      "step": 351
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.983492970273218e-05,
+      "loss": 0.0552,
+      "step": 352
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9833488915986054e-05,
+      "loss": 0.0538,
+      "step": 353
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9832041921492688e-05,
+      "loss": 0.0562,
+      "step": 354
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.983058872016556e-05,
+      "loss": 0.0568,
+      "step": 355
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9829129312922044e-05,
+      "loss": 0.0473,
+      "step": 356
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9827663700683454e-05,
+      "loss": 0.0672,
+      "step": 357
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9826191884375007e-05,
+      "loss": 0.0594,
+      "step": 358
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.9824713864925843e-05,
+      "loss": 0.0509,
+      "step": 359
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.982322964326901e-05,
+      "loss": 0.0651,
+      "step": 360
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9821739220341483e-05,
+      "loss": 0.0536,
+      "step": 361
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.982024259708414e-05,
+      "loss": 0.0556,
+      "step": 362
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9818739774441784e-05,
+      "loss": 0.0487,
+      "step": 363
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9817230753363127e-05,
+      "loss": 0.0438,
+      "step": 364
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9815715534800785e-05,
+      "loss": 0.0502,
+      "step": 365
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.98141941197113e-05,
+      "loss": 0.0579,
+      "step": 366
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9812666509055124e-05,
+      "loss": 0.054,
+      "step": 367
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9811132703796615e-05,
+      "loss": 0.061,
+      "step": 368
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.980959270490404e-05,
+      "loss": 0.0472,
+      "step": 369
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9808046513349583e-05,
+      "loss": 0.0553,
+      "step": 370
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.9806494130109327e-05,
+      "loss": 0.0489,
+      "step": 371
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.980493555616328e-05,
+      "loss": 0.0616,
+      "step": 372
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.980337079249534e-05,
+      "loss": 0.0538,
+      "step": 373
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9801799840093326e-05,
+      "loss": 0.0474,
+      "step": 374
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.980022269994896e-05,
+      "loss": 0.0421,
+      "step": 375
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9798639373057864e-05,
+      "loss": 0.0558,
+      "step": 376
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9797049860419572e-05,
+      "loss": 0.0562,
+      "step": 377
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9795454163037523e-05,
+      "loss": 0.0578,
+      "step": 378
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9793852281919062e-05,
+      "loss": 0.0451,
+      "step": 379
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9792244218075426e-05,
+      "loss": 0.0494,
+      "step": 380
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9790629972521772e-05,
+      "loss": 0.0562,
+      "step": 381
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9789009546277146e-05,
+      "loss": 0.0551,
+      "step": 382
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9787382940364506e-05,
+      "loss": 0.0432,
+      "step": 383
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.97857501558107e-05,
+      "loss": 0.0476,
+      "step": 384
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.978411119364649e-05,
+      "loss": 0.0511,
+      "step": 385
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9782466054906525e-05,
+      "loss": 0.0503,
+      "step": 386
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.9780814740629357e-05,
+      "loss": 0.0533,
+      "step": 387
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9779157251857445e-05,
+      "loss": 0.0486,
+      "step": 388
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.977749358963713e-05,
+      "loss": 0.043,
+      "step": 389
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9775823755018665e-05,
+      "loss": 0.0474,
+      "step": 390
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.97741477490562e-05,
+      "loss": 0.0544,
+      "step": 391
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.977246557280776e-05,
+      "loss": 0.0447,
+      "step": 392
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9770777227335292e-05,
+      "loss": 0.0549,
+      "step": 393
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9769082713704616e-05,
+      "loss": 0.0495,
+      "step": 394
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9767382032985466e-05,
+      "loss": 0.0549,
+      "step": 395
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.976567518625145e-05,
+      "loss": 0.0514,
+      "step": 396
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.976396217458008e-05,
+      "loss": 0.0484,
+      "step": 397
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9762242999052756e-05,
+      "loss": 0.0478,
+      "step": 398
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.976051766075477e-05,
+      "loss": 0.0514,
+      "step": 399
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.9758786160775306e-05,
+      "loss": 0.0511,
+      "step": 400
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.975704850020744e-05,
+      "loss": 0.0468,
+      "step": 401
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9755304680148125e-05,
+      "loss": 0.0603,
+      "step": 402
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9753554701698215e-05,
+      "loss": 0.0547,
+      "step": 403
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9751798565962446e-05,
+      "loss": 0.0575,
+      "step": 404
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9750036274049447e-05,
+      "loss": 0.0465,
+      "step": 405
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9748267827071723e-05,
+      "loss": 0.0512,
+      "step": 406
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9746493226145673e-05,
+      "loss": 0.0483,
+      "step": 407
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.974471247239158e-05,
+      "loss": 0.0466,
+      "step": 408
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9742925566933607e-05,
+      "loss": 0.0473,
+      "step": 409
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.97411325108998e-05,
+      "loss": 0.054,
+      "step": 410
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.97393333054221e-05,
+      "loss": 0.0456,
+      "step": 411
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.973752795163631e-05,
+      "loss": 0.05,
+      "step": 412
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9735716450682137e-05,
+      "loss": 0.0508,
+      "step": 413
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.9733898803703145e-05,
+      "loss": 0.0446,
+      "step": 414
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.97320750118468e-05,
+      "loss": 0.0509,
+      "step": 415
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.973024507626443e-05,
+      "loss": 0.0491,
+      "step": 416
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.972840899811125e-05,
+      "loss": 0.049,
+      "step": 417
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.972656677854635e-05,
+      "loss": 0.0488,
+      "step": 418
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.97247184187327e-05,
+      "loss": 0.054,
+      "step": 419
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9722863919837146e-05,
+      "loss": 0.0454,
+      "step": 420
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9721003283030402e-05,
+      "loss": 0.0429,
+      "step": 421
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.971913650948707e-05,
+      "loss": 0.0501,
+      "step": 422
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9717263600385614e-05,
+      "loss": 0.0564,
+      "step": 423
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9715384556908375e-05,
+      "loss": 0.0506,
+      "step": 424
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9713499380241568e-05,
+      "loss": 0.0535,
+      "step": 425
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9711608071575285e-05,
+      "loss": 0.0563,
+      "step": 426
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.9709710632103473e-05,
+      "loss": 0.0496,
+      "step": 427
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.970780706302397e-05,
+      "loss": 0.0452,
+      "step": 428
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.970589736553847e-05,
+      "loss": 0.0478,
+      "step": 429
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9703981540852534e-05,
+      "loss": 0.0427,
+      "step": 430
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.97020595901756e-05,
+      "loss": 0.0418,
+      "step": 431
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.970013151472097e-05,
+      "loss": 0.0518,
+      "step": 432
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.969819731570581e-05,
+      "loss": 0.0402,
+      "step": 433
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.969625699435115e-05,
+      "loss": 0.0478,
+      "step": 434
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.96943105518819e-05,
+      "loss": 0.0465,
+      "step": 435
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.969235798952681e-05,
+      "loss": 0.043,
+      "step": 436
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9690399308518508e-05,
+      "loss": 0.0477,
+      "step": 437
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.968843451009349e-05,
+      "loss": 0.054,
+      "step": 438
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.96864635954921e-05,
+      "loss": 0.0545,
+      "step": 439
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.968448656595855e-05,
+      "loss": 0.0519,
+      "step": 440
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.9682503422740915e-05,
+      "loss": 0.0511,
+      "step": 441
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9680514167091124e-05,
+      "loss": 0.0569,
+      "step": 442
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9678518800264966e-05,
+      "loss": 0.0398,
+      "step": 443
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.967651732352209e-05,
+      "loss": 0.0478,
+      "step": 444
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9674509738126e-05,
+      "loss": 0.0413,
+      "step": 445
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.967249604534406e-05,
+      "loss": 0.0492,
+      "step": 446
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9670476246447484e-05,
+      "loss": 0.0412,
+      "step": 447
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9668450342711343e-05,
+      "loss": 0.0497,
+      "step": 448
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9666418335414565e-05,
+      "loss": 0.0459,
+      "step": 449
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.966438022583993e-05,
+      "loss": 0.0484,
+      "step": 450
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.966233601527406e-05,
+      "loss": 0.041,
+      "step": 451
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.966028570500744e-05,
+      "loss": 0.0507,
+      "step": 452
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9658229296334416e-05,
+      "loss": 0.0442,
+      "step": 453
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9656166790553157e-05,
+      "loss": 0.0461,
+      "step": 454
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.9654098188965695e-05,
+      "loss": 0.0439,
+      "step": 455
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9652023492877915e-05,
+      "loss": 0.0488,
+      "step": 456
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.964994270359954e-05,
+      "loss": 0.0478,
+      "step": 457
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9647855822444152e-05,
+      "loss": 0.0524,
+      "step": 458
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.964576285072916e-05,
+      "loss": 0.0423,
+      "step": 459
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9643663789775836e-05,
+      "loss": 0.0416,
+      "step": 460
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9641558640909286e-05,
+      "loss": 0.0591,
+      "step": 461
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.963944740545846e-05,
+      "loss": 0.042,
+      "step": 462
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9637330084756154e-05,
+      "loss": 0.0593,
+      "step": 463
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9635206680139004e-05,
+      "loss": 0.0489,
+      "step": 464
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9633077192947486e-05,
+      "loss": 0.0409,
+      "step": 465
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9630941624525914e-05,
+      "loss": 0.043,
+      "step": 466
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9628799976222444e-05,
+      "loss": 0.0453,
+      "step": 467
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.9626652249389076e-05,
+      "loss": 0.0508,
+      "step": 468
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.962449844538163e-05,
+      "loss": 0.0442,
+      "step": 469
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9622338565559783e-05,
+      "loss": 0.0576,
+      "step": 470
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9620172611287028e-05,
+      "loss": 0.042,
+      "step": 471
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9618000583930708e-05,
+      "loss": 0.0545,
+      "step": 472
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9615822484861994e-05,
+      "loss": 0.0471,
+      "step": 473
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9613638315455888e-05,
+      "loss": 0.0521,
+      "step": 474
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9611448077091228e-05,
+      "loss": 0.0438,
+      "step": 475
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9609251771150682e-05,
+      "loss": 0.0537,
+      "step": 476
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9607049399020746e-05,
+      "loss": 0.059,
+      "step": 477
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.960484096209175e-05,
+      "loss": 0.0517,
+      "step": 478
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.960262646175785e-05,
+      "loss": 0.0463,
+      "step": 479
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9600405899417026e-05,
+      "loss": 0.0476,
+      "step": 480
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.959817927647109e-05,
+      "loss": 0.0512,
+      "step": 481
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.9595946594325684e-05,
+      "loss": 0.0501,
+      "step": 482
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9593707854390263e-05,
+      "loss": 0.0378,
+      "step": 483
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9591463058078118e-05,
+      "loss": 0.0441,
+      "step": 484
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.958921220680635e-05,
+      "loss": 0.0546,
+      "step": 485
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9586955301995903e-05,
+      "loss": 0.0522,
+      "step": 486
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.958469234507152e-05,
+      "loss": 0.0461,
+      "step": 487
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.958242333746178e-05,
+      "loss": 0.0438,
+      "step": 488
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9580148280599075e-05,
+      "loss": 0.0458,
+      "step": 489
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9577867175919616e-05,
+      "loss": 0.0479,
+      "step": 490
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9575580024863434e-05,
+      "loss": 0.0515,
+      "step": 491
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.957328682887438e-05,
+      "loss": 0.0456,
+      "step": 492
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9570987589400108e-05,
+      "loss": 0.0498,
+      "step": 493
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9568682307892108e-05,
+      "loss": 0.0442,
+      "step": 494
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9566370985805666e-05,
+      "loss": 0.0611,
+      "step": 495
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.9564053624599886e-05,
+      "loss": 0.0479,
+      "step": 496
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9561730225737694e-05,
+      "loss": 0.0443,
+      "step": 497
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9559400790685813e-05,
+      "loss": 0.0408,
+      "step": 498
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9557065320914787e-05,
+      "loss": 0.0423,
+      "step": 499
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9554723817898965e-05,
+      "loss": 0.0404,
+      "step": 500
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9552376283116508e-05,
+      "loss": 0.0542,
+      "step": 501
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9550022718049383e-05,
+      "loss": 0.041,
+      "step": 502
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9547663124183356e-05,
+      "loss": 0.0443,
+      "step": 503
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9545297503008014e-05,
+      "loss": 0.0546,
+      "step": 504
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9542925856016742e-05,
+      "loss": 0.0461,
+      "step": 505
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.954054818470672e-05,
+      "loss": 0.0487,
+      "step": 506
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.953816449057895e-05,
+      "loss": 0.0627,
+      "step": 507
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9535774775138216e-05,
+      "loss": 0.0412,
+      "step": 508
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.9533379039893118e-05,
+      "loss": 0.0523,
+      "step": 509
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9530977286356053e-05,
+      "loss": 0.0422,
+      "step": 510
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.952856951604321e-05,
+      "loss": 0.0435,
+      "step": 511
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9526155730474586e-05,
+      "loss": 0.047,
+      "step": 512
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9523735931173964e-05,
+      "loss": 0.0421,
+      "step": 513
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.952131011966894e-05,
+      "loss": 0.0487,
+      "step": 514
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.951887829749089e-05,
+      "loss": 0.0389,
+      "step": 515
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.951644046617499e-05,
+      "loss": 0.0436,
+      "step": 516
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.951399662726021e-05,
+      "loss": 0.051,
+      "step": 517
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9511546782289307e-05,
+      "loss": 0.0483,
+      "step": 518
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.950909093280885e-05,
+      "loss": 0.049,
+      "step": 519
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9506629080369163e-05,
+      "loss": 0.0372,
+      "step": 520
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9504161226524395e-05,
+      "loss": 0.0416,
+      "step": 521
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9501687372832466e-05,
+      "loss": 0.0487,
+      "step": 522
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.9499207520855085e-05,
+      "loss": 0.0597,
+      "step": 523
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.949672167215774e-05,
+      "loss": 0.0514,
+      "step": 524
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9494229828309724e-05,
+      "loss": 0.0538,
+      "step": 525
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9491731990884105e-05,
+      "loss": 0.0486,
+      "step": 526
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9489228161457724e-05,
+      "loss": 0.0497,
+      "step": 527
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.948671834161122e-05,
+      "loss": 0.0426,
+      "step": 528
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.948420253292901e-05,
+      "loss": 0.0498,
+      "step": 529
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9481680736999283e-05,
+      "loss": 0.0505,
+      "step": 530
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.947915295541402e-05,
+      "loss": 0.0468,
+      "step": 531
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9476619189768978e-05,
+      "loss": 0.0552,
+      "step": 532
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.947407944166368e-05,
+      "loss": 0.0496,
+      "step": 533
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.947153371270144e-05,
+      "loss": 0.0387,
+      "step": 534
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.946898200448934e-05,
+      "loss": 0.0408,
+      "step": 535
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.9466424318638243e-05,
+      "loss": 0.039,
+      "step": 536
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.946386065676277e-05,
+      "loss": 0.0435,
+      "step": 537
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9461291020481336e-05,
+      "loss": 0.0474,
+      "step": 538
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9458715411416115e-05,
+      "loss": 0.0495,
+      "step": 539
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.945613383119305e-05,
+      "loss": 0.0424,
+      "step": 540
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.945354628144186e-05,
+      "loss": 0.0423,
+      "step": 541
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9450952763796035e-05,
+      "loss": 0.0459,
+      "step": 542
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.944835327989282e-05,
+      "loss": 0.0349,
+      "step": 543
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9445747831373234e-05,
+      "loss": 0.0454,
+      "step": 544
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9443136419882064e-05,
+      "loss": 0.0415,
+      "step": 545
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.944051904706786e-05,
+      "loss": 0.0444,
+      "step": 546
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9437895714582926e-05,
+      "loss": 0.0445,
+      "step": 547
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9435266424083346e-05,
+      "loss": 0.0445,
+      "step": 548
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.9432631177228948e-05,
+      "loss": 0.0564,
+      "step": 549
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.942998997568333e-05,
+      "loss": 0.0436,
+      "step": 550
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9427342821113845e-05,
+      "loss": 0.0439,
+      "step": 551
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.942468971519161e-05,
+      "loss": 0.052,
+      "step": 552
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9422030659591485e-05,
+      "loss": 0.0473,
+      "step": 553
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.941936565599211e-05,
+      "loss": 0.038,
+      "step": 554
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.941669470607585e-05,
+      "loss": 0.0493,
+      "step": 555
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.941401781152885e-05,
+      "loss": 0.0448,
+      "step": 556
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9411334974040992e-05,
+      "loss": 0.0398,
+      "step": 557
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9408646195305914e-05,
+      "loss": 0.0366,
+      "step": 558
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.940595147702101e-05,
+      "loss": 0.0485,
+      "step": 559
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.9403250820887416e-05,
+      "loss": 0.0401,
+      "step": 560
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.940054422861002e-05,
+      "loss": 0.0412,
+      "step": 561
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.939783170189745e-05,
+      "loss": 0.0423,
+      "step": 562
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.93951132424621e-05,
+      "loss": 0.0468,
+      "step": 563
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.939238885202009e-05,
+      "loss": 0.0394,
+      "step": 564
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9389658532291286e-05,
+      "loss": 0.0581,
+      "step": 565
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9386922284999305e-05,
+      "loss": 0.0517,
+      "step": 566
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9384180111871502e-05,
+      "loss": 0.0432,
+      "step": 567
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9381432014638977e-05,
+      "loss": 0.0435,
+      "step": 568
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9378677995036564e-05,
+      "loss": 0.0411,
+      "step": 569
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9375918054802836e-05,
+      "loss": 0.0411,
+      "step": 570
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9373152195680106e-05,
+      "loss": 0.0458,
+      "step": 571
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9370380419414427e-05,
+      "loss": 0.0542,
+      "step": 572
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.936760272775558e-05,
+      "loss": 0.0629,
+      "step": 573
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9364819122457085e-05,
+      "loss": 0.0562,
+      "step": 574
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9362029605276193e-05,
+      "loss": 0.0537,
+      "step": 575
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.935923417797389e-05,
+      "loss": 0.0556,
+      "step": 576
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.9356432842314886e-05,
+      "loss": 0.0469,
+      "step": 577
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9353625600067632e-05,
+      "loss": 0.0444,
+      "step": 578
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9350812453004303e-05,
+      "loss": 0.0507,
+      "step": 579
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9347993402900792e-05,
+      "loss": 0.0489,
+      "step": 580
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.934516845153673e-05,
+      "loss": 0.0428,
+      "step": 581
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.934233760069548e-05,
+      "loss": 0.0458,
+      "step": 582
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9339500852164104e-05,
+      "loss": 0.0446,
+      "step": 583
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.933665820773341e-05,
+      "loss": 0.0478,
+      "step": 584
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.933380966919792e-05,
+      "loss": 0.0452,
+      "step": 585
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9330955238355883e-05,
+      "loss": 0.0396,
+      "step": 586
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9328094917009252e-05,
+      "loss": 0.0345,
+      "step": 587
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9325228706963716e-05,
+      "loss": 0.0463,
+      "step": 588
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9322356610028673e-05,
+      "loss": 0.0534,
+      "step": 589
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9319478628017237e-05,
+      "loss": 0.0376,
+      "step": 590
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.9316594762746238e-05,
+      "loss": 0.0491,
+      "step": 591
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9313705016036222e-05,
+      "loss": 0.0375,
+      "step": 592
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9310809389711446e-05,
+      "loss": 0.0479,
+      "step": 593
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9307907885599883e-05,
+      "loss": 0.0409,
+      "step": 594
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.930500050553321e-05,
+      "loss": 0.0511,
+      "step": 595
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9302087251346812e-05,
+      "loss": 0.0406,
+      "step": 596
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9299168124879798e-05,
+      "loss": 0.0436,
+      "step": 597
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.929624312797496e-05,
+      "loss": 0.0453,
+      "step": 598
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9293312262478817e-05,
+      "loss": 0.0412,
+      "step": 599
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9290375530241577e-05,
+      "loss": 0.0374,
+      "step": 600
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.928743293311717e-05,
+      "loss": 0.0438,
+      "step": 601
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.928448447296321e-05,
+      "loss": 0.0497,
+      "step": 602
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9281530151641016e-05,
+      "loss": 0.0484,
+      "step": 603
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.9278569971015617e-05,
+      "loss": 0.0424,
+      "step": 604
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9275603932955733e-05,
+      "loss": 0.0428,
+      "step": 605
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9272632039333784e-05,
+      "loss": 0.0488,
+      "step": 606
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9269654292025885e-05,
+      "loss": 0.0537,
+      "step": 607
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9266670692911848e-05,
+      "loss": 0.0443,
+      "step": 608
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9263681243875173e-05,
+      "loss": 0.0416,
+      "step": 609
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9260685946803067e-05,
+      "loss": 0.0539,
+      "step": 610
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9257684803586415e-05,
+      "loss": 0.0373,
+      "step": 611
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.92546778161198e-05,
+      "loss": 0.0434,
+      "step": 612
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9251664986301487e-05,
+      "loss": 0.0437,
+      "step": 613
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9248646316033444e-05,
+      "loss": 0.0425,
+      "step": 614
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9245621807221306e-05,
+      "loss": 0.0461,
+      "step": 615
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.924259146177441e-05,
+      "loss": 0.0446,
+      "step": 616
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.923955528160577e-05,
+      "loss": 0.0413,
+      "step": 617
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.9236513268632085e-05,
+      "loss": 0.0392,
+      "step": 618
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9233465424773737e-05,
+      "loss": 0.0446,
+      "step": 619
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9230411751954788e-05,
+      "loss": 0.0442,
+      "step": 620
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.922735225210298e-05,
+      "loss": 0.0467,
+      "step": 621
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.922428692714973e-05,
+      "loss": 0.0466,
+      "step": 622
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9221215779030143e-05,
+      "loss": 0.0466,
+      "step": 623
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9218138809682988e-05,
+      "loss": 0.0436,
+      "step": 624
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.921505602105071e-05,
+      "loss": 0.0435,
+      "step": 625
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.921196741507944e-05,
+      "loss": 0.0429,
+      "step": 626
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9208872993718967e-05,
+      "loss": 0.0471,
+      "step": 627
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.920577275892276e-05,
+      "loss": 0.042,
+      "step": 628
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9202666712647953e-05,
+      "loss": 0.0396,
+      "step": 629
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.919955485685535e-05,
+      "loss": 0.0443,
+      "step": 630
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.9196437193509426e-05,
+      "loss": 0.0366,
+      "step": 631
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9193313724578316e-05,
+      "loss": 0.047,
+      "step": 632
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9190184452033828e-05,
+      "loss": 0.0488,
+      "step": 633
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.918704937785142e-05,
+      "loss": 0.0422,
+      "step": 634
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.918390850401023e-05,
+      "loss": 0.0482,
+      "step": 635
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9180761832493045e-05,
+      "loss": 0.0469,
+      "step": 636
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9177609365286318e-05,
+      "loss": 0.0372,
+      "step": 637
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9174451104380155e-05,
+      "loss": 0.0456,
+      "step": 638
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.917128705176833e-05,
+      "loss": 0.0451,
+      "step": 639
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.916811720944826e-05,
+      "loss": 0.0514,
+      "step": 640
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9164941579421018e-05,
+      "loss": 0.0491,
+      "step": 641
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9161760163691347e-05,
+      "loss": 0.0355,
+      "step": 642
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9158572964267626e-05,
+      "loss": 0.0533,
+      "step": 643
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.9155379983161895e-05,
+      "loss": 0.0447,
+      "step": 644
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.915218122238983e-05,
+      "loss": 0.0373,
+      "step": 645
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9148976683970774e-05,
+      "loss": 0.0496,
+      "step": 646
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.91457663699277e-05,
+      "loss": 0.0406,
+      "step": 647
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9142550282287247e-05,
+      "loss": 0.0404,
+      "step": 648
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9139328423079674e-05,
+      "loss": 0.0439,
+      "step": 649
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9136100794338907e-05,
+      "loss": 0.0417,
+      "step": 650
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9132867398102498e-05,
+      "loss": 0.0431,
+      "step": 651
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.912962823641165e-05,
+      "loss": 0.046,
+      "step": 652
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.91263833113112e-05,
+      "loss": 0.0455,
+      "step": 653
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.912313262484962e-05,
+      "loss": 0.0436,
+      "step": 654
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9119876179079032e-05,
+      "loss": 0.0429,
+      "step": 655
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.911661397605518e-05,
+      "loss": 0.0501,
+      "step": 656
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.911334601783745e-05,
+      "loss": 0.0485,
+      "step": 657
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.9110072306488862e-05,
+      "loss": 0.05,
+      "step": 658
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.910679284407606e-05,
+      "loss": 0.0434,
+      "step": 659
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.910350763266933e-05,
+      "loss": 0.0383,
+      "step": 660
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9100216674342575e-05,
+      "loss": 0.0439,
+      "step": 661
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9096919971173336e-05,
+      "loss": 0.0406,
+      "step": 662
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9093617525242772e-05,
+      "loss": 0.0475,
+      "step": 663
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.909030933863568e-05,
+      "loss": 0.0522,
+      "step": 664
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.908699541344047e-05,
+      "loss": 0.0458,
+      "step": 665
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9083675751749174e-05,
+      "loss": 0.0502,
+      "step": 666
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.908035035565745e-05,
+      "loss": 0.0383,
+      "step": 667
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9077019227264573e-05,
+      "loss": 0.055,
+      "step": 668
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.907368236867345e-05,
+      "loss": 0.0533,
+      "step": 669
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.907033978199058e-05,
+      "loss": 0.0481,
+      "step": 670
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9066991469326103e-05,
+      "loss": 0.0393,
+      "step": 671
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.9063637432793757e-05,
+      "loss": 0.039,
+      "step": 672
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9060277674510904e-05,
+      "loss": 0.045,
+      "step": 673
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.905691219659851e-05,
+      "loss": 0.0452,
+      "step": 674
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9053541001181156e-05,
+      "loss": 0.0433,
+      "step": 675
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.905016409038703e-05,
+      "loss": 0.0536,
+      "step": 676
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9046781466347935e-05,
+      "loss": 0.0441,
+      "step": 677
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9043393131199266e-05,
+      "loss": 0.0369,
+      "step": 678
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9039999087080043e-05,
+      "loss": 0.0482,
+      "step": 679
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9036599336132876e-05,
+      "loss": 0.0477,
+      "step": 680
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9033193880503976e-05,
+      "loss": 0.0446,
+      "step": 681
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9029782722343168e-05,
+      "loss": 0.0486,
+      "step": 682
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9026365863803863e-05,
+      "loss": 0.0366,
+      "step": 683
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9022943307043085e-05,
+      "loss": 0.0509,
+      "step": 684
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9019515054221443e-05,
+      "loss": 0.0559,
+      "step": 685
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.9016081107503145e-05,
+      "loss": 0.0485,
+      "step": 686
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9012641469055998e-05,
+      "loss": 0.0412,
+      "step": 687
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9009196141051392e-05,
+      "loss": 0.0425,
+      "step": 688
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9005745125664324e-05,
+      "loss": 0.0511,
+      "step": 689
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.9002288425073367e-05,
+      "loss": 0.0421,
+      "step": 690
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.899882604146069e-05,
+      "loss": 0.0399,
+      "step": 691
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8995357977012054e-05,
+      "loss": 0.0397,
+      "step": 692
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8991884233916795e-05,
+      "loss": 0.0456,
+      "step": 693
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8988404814367837e-05,
+      "loss": 0.0463,
+      "step": 694
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.89849197205617e-05,
+      "loss": 0.0464,
+      "step": 695
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8981428954698466e-05,
+      "loss": 0.0395,
+      "step": 696
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.897793251898181e-05,
+      "loss": 0.0499,
+      "step": 697
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8974430415618987e-05,
+      "loss": 0.0463,
+      "step": 698
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.8970922646820825e-05,
+      "loss": 0.0406,
+      "step": 699
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8967409214801737e-05,
+      "loss": 0.0434,
+      "step": 700
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.896389012177969e-05,
+      "loss": 0.0505,
+      "step": 701
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8960365369976256e-05,
+      "loss": 0.038,
+      "step": 702
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.895683496161655e-05,
+      "loss": 0.0447,
+      "step": 703
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.895329889892928e-05,
+      "loss": 0.0424,
+      "step": 704
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8949757184146706e-05,
+      "loss": 0.0446,
+      "step": 705
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.894620981950467e-05,
+      "loss": 0.0529,
+      "step": 706
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.894265680724257e-05,
+      "loss": 0.0413,
+      "step": 707
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.893909814960338e-05,
+      "loss": 0.0373,
+      "step": 708
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8935533848833626e-05,
+      "loss": 0.0414,
+      "step": 709
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8931963907183406e-05,
+      "loss": 0.0363,
+      "step": 710
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8928388326906376e-05,
+      "loss": 0.0409,
+      "step": 711
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.892480711025975e-05,
+      "loss": 0.0427,
+      "step": 712
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.8921220259504297e-05,
+      "loss": 0.0512,
+      "step": 713
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8917627776904352e-05,
+      "loss": 0.0514,
+      "step": 714
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8914029664727803e-05,
+      "loss": 0.0392,
+      "step": 715
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.891042592524608e-05,
+      "loss": 0.0398,
+      "step": 716
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8906816560734182e-05,
+      "loss": 0.0346,
+      "step": 717
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8903201573470645e-05,
+      "loss": 0.0398,
+      "step": 718
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.889958096573757e-05,
+      "loss": 0.0488,
+      "step": 719
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.889595473982059e-05,
+      "loss": 0.0395,
+      "step": 720
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8892322898008893e-05,
+      "loss": 0.0429,
+      "step": 721
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8888685442595213e-05,
+      "loss": 0.0376,
+      "step": 722
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8885042375875825e-05,
+      "loss": 0.0418,
+      "step": 723
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.888139370015055e-05,
+      "loss": 0.0358,
+      "step": 724
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.887773941772274e-05,
+      "loss": 0.0377,
+      "step": 725
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8874079530899298e-05,
+      "loss": 0.0398,
+      "step": 726
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 1.8870414041990662e-05,
+      "loss": 0.0386,
+      "step": 727
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8866742953310806e-05,
+      "loss": 0.0333,
+      "step": 728
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8863066267177234e-05,
+      "loss": 0.0423,
+      "step": 729
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.885938398591099e-05,
+      "loss": 0.0452,
+      "step": 730
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8855696111836645e-05,
+      "loss": 0.0401,
+      "step": 731
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.885200264728231e-05,
+      "loss": 0.0414,
+      "step": 732
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8848303594579614e-05,
+      "loss": 0.0446,
+      "step": 733
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8844598956063716e-05,
+      "loss": 0.0429,
+      "step": 734
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.884088873407331e-05,
+      "loss": 0.041,
+      "step": 735
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8837172930950602e-05,
+      "loss": 0.0413,
+      "step": 736
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8833451549041337e-05,
+      "loss": 0.0484,
+      "step": 737
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.882972459069476e-05,
+      "loss": 0.0394,
+      "step": 738
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8825992058263656e-05,
+      "loss": 0.0558,
+      "step": 739
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 1.8822253954104325e-05,
+      "loss": 0.0497,
+      "step": 740
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8818510280576577e-05,
+      "loss": 0.0454,
+      "step": 741
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.881476104004374e-05,
+      "loss": 0.0468,
+      "step": 742
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8811006234872665e-05,
+      "loss": 0.0472,
+      "step": 743
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.88072458674337e-05,
+      "loss": 0.0433,
+      "step": 744
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8803479940100723e-05,
+      "loss": 0.0426,
+      "step": 745
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8799708455251103e-05,
+      "loss": 0.0347,
+      "step": 746
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8795931415265735e-05,
+      "loss": 0.0549,
+      "step": 747
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.879214882252901e-05,
+      "loss": 0.0406,
+      "step": 748
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8788360679428822e-05,
+      "loss": 0.0491,
+      "step": 749
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8784566988356586e-05,
+      "loss": 0.0449,
+      "step": 750
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8780767751707193e-05,
+      "loss": 0.0447,
+      "step": 751
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8776962971879064e-05,
+      "loss": 0.0417,
+      "step": 752
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.877315265127409e-05,
+      "loss": 0.0414,
+      "step": 753
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 1.8769336792297683e-05,
+      "loss": 0.0401,
+      "step": 754
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8765515397358744e-05,
+      "loss": 0.0374,
+      "step": 755
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8761688468869658e-05,
+      "loss": 0.0396,
+      "step": 756
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8757856009246325e-05,
+      "loss": 0.0446,
+      "step": 757
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8754018020908113e-05,
+      "loss": 0.0393,
+      "step": 758
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8750174506277902e-05,
+      "loss": 0.0421,
+      "step": 759
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.874632546778204e-05,
+      "loss": 0.0462,
+      "step": 760
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8742470907850384e-05,
+      "loss": 0.0454,
+      "step": 761
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8738610828916255e-05,
+      "loss": 0.0415,
+      "step": 762
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8734745233416475e-05,
+      "loss": 0.0358,
+      "step": 763
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8730874123791334e-05,
+      "loss": 0.0422,
+      "step": 764
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8726997502484617e-05,
+      "loss": 0.0442,
+      "step": 765
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.872311537194358e-05,
+      "loss": 0.0403,
+      "step": 766
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 1.8719227734618955e-05,
+      "loss": 0.0313,
+      "step": 767
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8715334592964964e-05,
+      "loss": 0.0459,
+      "step": 768
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.871143594943929e-05,
+      "loss": 0.0517,
+      "step": 769
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8707531806503086e-05,
+      "loss": 0.0402,
+      "step": 770
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8703622166620995e-05,
+      "loss": 0.0376,
+      "step": 771
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8699707032261112e-05,
+      "loss": 0.0424,
+      "step": 772
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8695786405895013e-05,
+      "loss": 0.0363,
+      "step": 773
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.869186028999773e-05,
+      "loss": 0.0351,
+      "step": 774
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.868792868704777e-05,
+      "loss": 0.0483,
+      "step": 775
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.86839915995271e-05,
+      "loss": 0.0399,
+      "step": 776
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.868004902992115e-05,
+      "loss": 0.0343,
+      "step": 777
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.867610098071881e-05,
+      "loss": 0.0374,
+      "step": 778
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.867214745441243e-05,
+      "loss": 0.0431,
+      "step": 779
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8668188453497814e-05,
+      "loss": 0.0384,
+      "step": 780
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 1.8664223980474226e-05,
+      "loss": 0.0396,
+      "step": 781
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.866025403784439e-05,
+      "loss": 0.036,
+      "step": 782
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.865627862811447e-05,
+      "loss": 0.0354,
+      "step": 783
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8652297753794087e-05,
+      "loss": 0.0411,
+      "step": 784
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8648311417396314e-05,
+      "loss": 0.0458,
+      "step": 785
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8644319621437677e-05,
+      "loss": 0.0494,
+      "step": 786
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8640322368438134e-05,
+      "loss": 0.0424,
+      "step": 787
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8636319660921098e-05,
+      "loss": 0.035,
+      "step": 788
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.863231150141343e-05,
+      "loss": 0.0376,
+      "step": 789
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8628297892445417e-05,
+      "loss": 0.0357,
+      "step": 790
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8624278836550803e-05,
+      "loss": 0.0634,
+      "step": 791
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8620254336266757e-05,
+      "loss": 0.0391,
+      "step": 792
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8616224394133896e-05,
+      "loss": 0.0437,
+      "step": 793
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8612189012696263e-05,
+      "loss": 0.0411,
+      "step": 794
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 1.8608148194501343e-05,
+      "loss": 0.0348,
+      "step": 795
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.860410194210005e-05,
+      "loss": 0.0365,
+      "step": 796
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.860005025804672e-05,
+      "loss": 0.0407,
+      "step": 797
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8595993144899135e-05,
+      "loss": 0.0394,
+      "step": 798
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8591930605218494e-05,
+      "loss": 0.0422,
+      "step": 799
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8587862641569417e-05,
+      "loss": 0.0391,
+      "step": 800
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.858378925651996e-05,
+      "loss": 0.0363,
+      "step": 801
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.857971045264159e-05,
+      "loss": 0.0469,
+      "step": 802
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8575626232509206e-05,
+      "loss": 0.0408,
+      "step": 803
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8571536598701114e-05,
+      "loss": 0.0345,
+      "step": 804
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.856744155379905e-05,
+      "loss": 0.0395,
+      "step": 805
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8563341100388157e-05,
+      "loss": 0.0433,
+      "step": 806
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8559235241056994e-05,
+      "loss": 0.0396,
+      "step": 807
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 1.8555123978397536e-05,
+      "loss": 0.0611,
+      "step": 808
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8551007315005165e-05,
+      "loss": 0.045,
+      "step": 809
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8546885253478678e-05,
+      "loss": 0.0439,
+      "step": 810
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8542757796420273e-05,
+      "loss": 0.0365,
+      "step": 811
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8538624946435555e-05,
+      "loss": 0.0396,
+      "step": 812
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.853448670613354e-05,
+      "loss": 0.0412,
+      "step": 813
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8530343078126645e-05,
+      "loss": 0.0316,
+      "step": 814
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.852619406503068e-05,
+      "loss": 0.0479,
+      "step": 815
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8522039669464863e-05,
+      "loss": 0.0398,
+      "step": 816
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8517879894051804e-05,
+      "loss": 0.0486,
+      "step": 817
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8513714741417517e-05,
+      "loss": 0.046,
+      "step": 818
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8509544214191403e-05,
+      "loss": 0.0513,
+      "step": 819
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8505368315006264e-05,
+      "loss": 0.0365,
+      "step": 820
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.850118704649828e-05,
+      "loss": 0.041,
+      "step": 821
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 1.8497000411307035e-05,
+      "loss": 0.0433,
+      "step": 822
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8492808412075487e-05,
+      "loss": 0.0383,
+      "step": 823
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.848861105145e-05,
+      "loss": 0.0349,
+      "step": 824
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8484408332080298e-05,
+      "loss": 0.0355,
+      "step": 825
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8480200256619507e-05,
+      "loss": 0.0365,
+      "step": 826
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8475986827724125e-05,
+      "loss": 0.0391,
+      "step": 827
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.847176804805404e-05,
+      "loss": 0.0412,
+      "step": 828
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8467543920272498e-05,
+      "loss": 0.0409,
+      "step": 829
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.846331444704614e-05,
+      "loss": 0.0433,
+      "step": 830
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.845907963104497e-05,
+      "loss": 0.0326,
+      "step": 831
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.845483947494238e-05,
+      "loss": 0.0435,
+      "step": 832
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.845059398141511e-05,
+      "loss": 0.039,
+      "step": 833
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.844634315314329e-05,
+      "loss": 0.0396,
+      "step": 834
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 1.8442086992810407e-05,
+      "loss": 0.0388,
+      "step": 835
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8437825503103317e-05,
+      "loss": 0.0367,
+      "step": 836
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.843355868671224e-05,
+      "loss": 0.0362,
+      "step": 837
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.842928654633076e-05,
+      "loss": 0.0379,
+      "step": 838
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.842500908465582e-05,
+      "loss": 0.0322,
+      "step": 839
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8420726304387723e-05,
+      "loss": 0.0453,
+      "step": 840
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.841643820823013e-05,
+      "loss": 0.04,
+      "step": 841
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.841214479889006e-05,
+      "loss": 0.0291,
+      "step": 842
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.840784607907788e-05,
+      "loss": 0.0354,
+      "step": 843
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8403542051507314e-05,
+      "loss": 0.0394,
+      "step": 844
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8399232718895435e-05,
+      "loss": 0.0382,
+      "step": 845
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.839491808396267e-05,
+      "loss": 0.0504,
+      "step": 846
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8390598149432784e-05,
+      "loss": 0.037,
+      "step": 847
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8386272918032896e-05,
+      "loss": 0.0407,
+      "step": 848
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 1.8381942392493464e-05,
+      "loss": 0.0413,
+      "step": 849
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8377606575548288e-05,
+      "loss": 0.0397,
+      "step": 850
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8373265469934513e-05,
+      "loss": 0.0446,
+      "step": 851
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.836891907839262e-05,
+      "loss": 0.038,
+      "step": 852
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8364567403666426e-05,
+      "loss": 0.0435,
+      "step": 853
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.836021044850308e-05,
+      "loss": 0.0439,
+      "step": 854
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8355848215653073e-05,
+      "loss": 0.0464,
+      "step": 855
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.835148070787022e-05,
+      "loss": 0.0417,
+      "step": 856
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8347107927911673e-05,
+      "loss": 0.0458,
+      "step": 857
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8342729878537903e-05,
+      "loss": 0.0386,
+      "step": 858
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8338346562512716e-05,
+      "loss": 0.037,
+      "step": 859
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.833395798260324e-05,
+      "loss": 0.038,
+      "step": 860
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.8329564141579924e-05,
+      "loss": 0.0421,
+      "step": 861
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 1.832516504221654e-05,
+      "loss": 0.0388,
+      "step": 862
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8320760687290177e-05,
+      "loss": 0.0447,
+      "step": 863
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.831635107958125e-05,
+      "loss": 0.0487,
+      "step": 864
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8311936221873482e-05,
+      "loss": 0.0354,
+      "step": 865
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8307516116953906e-05,
+      "loss": 0.0356,
+      "step": 866
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8303090767612882e-05,
+      "loss": 0.0346,
+      "step": 867
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8298660176644075e-05,
+      "loss": 0.0338,
+      "step": 868
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8294224346844445e-05,
+      "loss": 0.039,
+      "step": 869
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.828978328101428e-05,
+      "loss": 0.0365,
+      "step": 870
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.828533698195716e-05,
+      "loss": 0.0407,
+      "step": 871
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.828088545247997e-05,
+      "loss": 0.0414,
+      "step": 872
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8276428695392908e-05,
+      "loss": 0.0423,
+      "step": 873
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.827196671350946e-05,
+      "loss": 0.0293,
+      "step": 874
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8267499509646414e-05,
+      "loss": 0.0398,
+      "step": 875
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 1.8263027086623852e-05,
+      "loss": 0.0371,
+      "step": 876
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.825854944726516e-05,
+      "loss": 0.0386,
+      "step": 877
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8254066594397004e-05,
+      "loss": 0.0341,
+      "step": 878
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.824957853084935e-05,
+      "loss": 0.046,
+      "step": 879
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8245085259455453e-05,
+      "loss": 0.042,
+      "step": 880
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8240586783051854e-05,
+      "loss": 0.0438,
+      "step": 881
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8236083104478373e-05,
+      "loss": 0.0457,
+      "step": 882
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8231574226578134e-05,
+      "loss": 0.0449,
+      "step": 883
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.822706015219752e-05,
+      "loss": 0.0478,
+      "step": 884
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.82225408841862e-05,
+      "loss": 0.0331,
+      "step": 885
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8218016425397142e-05,
+      "loss": 0.0404,
+      "step": 886
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.821348677868656e-05,
+      "loss": 0.048,
+      "step": 887
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8208951946913965e-05,
+      "loss": 0.0448,
+      "step": 888
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8204411932942137e-05,
+      "loss": 0.0408,
+      "step": 889
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 1.8199866739637117e-05,
+      "loss": 0.0387,
+      "step": 890
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.819531636986823e-05,
+      "loss": 0.0432,
+      "step": 891
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8190760826508058e-05,
+      "loss": 0.0406,
+      "step": 892
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.818620011243246e-05,
+      "loss": 0.0398,
+      "step": 893
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8181634230520537e-05,
+      "loss": 0.0532,
+      "step": 894
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8177063183654688e-05,
+      "loss": 0.0376,
+      "step": 895
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.817248697472054e-05,
+      "loss": 0.0439,
+      "step": 896
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8167905606606995e-05,
+      "loss": 0.0371,
+      "step": 897
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8163319082206206e-05,
+      "loss": 0.0398,
+      "step": 898
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8158727404413584e-05,
+      "loss": 0.0472,
+      "step": 899
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8154130576127794e-05,
+      "loss": 0.0426,
+      "step": 900
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.814952860025075e-05,
+      "loss": 0.0462,
+      "step": 901
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8144921479687613e-05,
+      "loss": 0.0351,
+      "step": 902
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 1.8140309217346805e-05,
+      "loss": 0.0569,
+      "step": 903
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8135691816139974e-05,
+      "loss": 0.0433,
+      "step": 904
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8131069278982026e-05,
+      "loss": 0.0476,
+      "step": 905
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.812644160879111e-05,
+      "loss": 0.0409,
+      "step": 906
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.81218088084886e-05,
+      "loss": 0.0406,
+      "step": 907
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8117170880999132e-05,
+      "loss": 0.04,
+      "step": 908
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8112527829250558e-05,
+      "loss": 0.0371,
+      "step": 909
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8107879656173977e-05,
+      "loss": 0.0343,
+      "step": 910
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8103226364703714e-05,
+      "loss": 0.0419,
+      "step": 911
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.809856795777733e-05,
+      "loss": 0.0399,
+      "step": 912
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8093904438335617e-05,
+      "loss": 0.0397,
+      "step": 913
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8089235809322585e-05,
+      "loss": 0.0437,
+      "step": 914
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8084562073685482e-05,
+      "loss": 0.0457,
+      "step": 915
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8079883234374765e-05,
+      "loss": 0.0435,
+      "step": 916
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 1.8075199294344132e-05,
+      "loss": 0.0399,
+      "step": 917
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.807051025655048e-05,
+      "loss": 0.0396,
+      "step": 918
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8065816123953944e-05,
+      "loss": 0.0322,
+      "step": 919
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8061116899517863e-05,
+      "loss": 0.0407,
+      "step": 920
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8056412586208784e-05,
+      "loss": 0.0379,
+      "step": 921
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.805170318699649e-05,
+      "loss": 0.0355,
+      "step": 922
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8046988704853946e-05,
+      "loss": 0.0439,
+      "step": 923
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8042269142757354e-05,
+      "loss": 0.037,
+      "step": 924
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.80375445036861e-05,
+      "loss": 0.0473,
+      "step": 925
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.803281479062279e-05,
+      "loss": 0.0431,
+      "step": 926
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8028080006553223e-05,
+      "loss": 0.04,
+      "step": 927
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8023340154466402e-05,
+      "loss": 0.0453,
+      "step": 928
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8018595237354536e-05,
+      "loss": 0.0376,
+      "step": 929
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 1.8013845258213024e-05,
+      "loss": 0.036,
+      "step": 930
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.800909022004046e-05,
+      "loss": 0.0423,
+      "step": 931
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.8004330125838642e-05,
+      "loss": 0.0453,
+      "step": 932
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7999564978612544e-05,
+      "loss": 0.048,
+      "step": 933
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.799479478137035e-05,
+      "loss": 0.0452,
+      "step": 934
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7990019537123403e-05,
+      "loss": 0.0378,
+      "step": 935
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7985239248886264e-05,
+      "loss": 0.0448,
+      "step": 936
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7980453919676658e-05,
+      "loss": 0.0387,
+      "step": 937
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.79756635525155e-05,
+      "loss": 0.039,
+      "step": 938
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.797086815042688e-05,
+      "loss": 0.0418,
+      "step": 939
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7966067716438075e-05,
+      "loss": 0.04,
+      "step": 940
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7961262253579526e-05,
+      "loss": 0.0396,
+      "step": 941
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7956451764884862e-05,
+      "loss": 0.0416,
+      "step": 942
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.795163625339088e-05,
+      "loss": 0.0521,
+      "step": 943
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 1.7946815722137543e-05,
+      "loss": 0.0389,
+      "step": 944
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7941990174167987e-05,
+      "loss": 0.0389,
+      "step": 945
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7937159612528514e-05,
+      "loss": 0.0466,
+      "step": 946
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7932324040268594e-05,
+      "loss": 0.0472,
+      "step": 947
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7927483460440857e-05,
+      "loss": 0.0375,
+      "step": 948
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.792263787610109e-05,
+      "loss": 0.0445,
+      "step": 949
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.791778729030825e-05,
+      "loss": 0.0413,
+      "step": 950
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7912931706124447e-05,
+      "loss": 0.0419,
+      "step": 951
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.790807112661494e-05,
+      "loss": 0.0343,
+      "step": 952
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7903205554848148e-05,
+      "loss": 0.0419,
+      "step": 953
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.789833499389564e-05,
+      "loss": 0.0384,
+      "step": 954
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7893459446832135e-05,
+      "loss": 0.0466,
+      "step": 955
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7888578916735502e-05,
+      "loss": 0.0435,
+      "step": 956
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.7883693406686746e-05,
+      "loss": 0.0476,
+      "step": 957
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 1.787880291977003e-05,
+      "loss": 0.0532,
+      "step": 958
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7873907459072648e-05,
+      "loss": 0.04,
+      "step": 959
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.786900702768504e-05,
+      "loss": 0.0411,
+      "step": 960
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7864101628700776e-05,
+      "loss": 0.0458,
+      "step": 961
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7859191265216572e-05,
+      "loss": 0.042,
+      "step": 962
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7854275940332272e-05,
+      "loss": 0.0328,
+      "step": 963
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7849355657150857e-05,
+      "loss": 0.0362,
+      "step": 964
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7844430418778426e-05,
+      "loss": 0.0367,
+      "step": 965
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7839500228324223e-05,
+      "loss": 0.0482,
+      "step": 966
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7834565088900605e-05,
+      "loss": 0.0367,
+      "step": 967
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7829625003623063e-05,
+      "loss": 0.0308,
+      "step": 968
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.78246799756102e-05,
+      "loss": 0.0498,
+      "step": 969
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7819730007983747e-05,
+      "loss": 0.0353,
+      "step": 970
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.7814775103868553e-05,
+      "loss": 0.0424,
+      "step": 971
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7809815266392575e-05,
+      "loss": 0.036,
+      "step": 972
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.78048504986869e-05,
+      "loss": 0.0454,
+      "step": 973
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.779988080388571e-05,
+      "loss": 0.0392,
+      "step": 974
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.779490618512631e-05,
+      "loss": 0.0308,
+      "step": 975
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7789926645549103e-05,
+      "loss": 0.0313,
+      "step": 976
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7784942188297614e-05,
+      "loss": 0.0355,
+      "step": 977
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7779952816518454e-05,
+      "loss": 0.0427,
+      "step": 978
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.777495853336135e-05,
+      "loss": 0.0412,
+      "step": 979
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7769959341979123e-05,
+      "loss": 0.0323,
+      "step": 980
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7764955245527693e-05,
+      "loss": 0.0442,
+      "step": 981
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7759946247166083e-05,
+      "loss": 0.0454,
+      "step": 982
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7754932350056395e-05,
+      "loss": 0.0456,
+      "step": 983
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7749913557363844e-05,
+      "loss": 0.0359,
+      "step": 984
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7744889872256724e-05,
+      "loss": 0.0387,
+      "step": 985
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7739861297906413e-05,
+      "loss": 0.0378,
+      "step": 986
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7734827837487386e-05,
+      "loss": 0.0333,
+      "step": 987
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.77297894941772e-05,
+      "loss": 0.039,
+      "step": 988
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7724746271156482e-05,
+      "loss": 0.037,
+      "step": 989
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.771969817160896e-05,
+      "loss": 0.0349,
+      "step": 990
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.771464519872143e-05,
+      "loss": 0.0386,
+      "step": 991
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7709587355683756e-05,
+      "loss": 0.0357,
+      "step": 992
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.770452464568889e-05,
+      "loss": 0.0427,
+      "step": 993
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7699457071932858e-05,
+      "loss": 0.0399,
+      "step": 994
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7694384637614745e-05,
+      "loss": 0.0448,
+      "step": 995
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7689307345936705e-05,
+      "loss": 0.0416,
+      "step": 996
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.768422520010397e-05,
+      "loss": 0.0374,
+      "step": 997
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7679138203324823e-05,
+      "loss": 0.0433,
+      "step": 998
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.767404635881062e-05,
+      "loss": 0.0413,
+      "step": 999
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.766894966977577e-05,
+      "loss": 0.046,
+      "step": 1000
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7663848139437747e-05,
+      "loss": 0.0409,
+      "step": 1001
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7658741771017076e-05,
+      "loss": 0.0328,
+      "step": 1002
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7653630567737337e-05,
+      "loss": 0.0538,
+      "step": 1003
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.764851453282516e-05,
+      "loss": 0.0354,
+      "step": 1004
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7643393669510236e-05,
+      "loss": 0.0429,
+      "step": 1005
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7638267981025286e-05,
+      "loss": 0.0356,
+      "step": 1006
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7633137470606095e-05,
+      "loss": 0.0366,
+      "step": 1007
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7628002141491477e-05,
+      "loss": 0.038,
+      "step": 1008
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7622861996923305e-05,
+      "loss": 0.0412,
+      "step": 1009
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.761771704014647e-05,
+      "loss": 0.0357,
+      "step": 1010
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7612567274408925e-05,
+      "loss": 0.0407,
+      "step": 1011
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7607412702961638e-05,
+      "loss": 0.0347,
+      "step": 1012
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.760225332905862e-05,
+      "loss": 0.0435,
+      "step": 1013
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.759708915595692e-05,
+      "loss": 0.0454,
+      "step": 1014
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7591920186916605e-05,
+      "loss": 0.0469,
+      "step": 1015
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7586746425200773e-05,
+      "loss": 0.0314,
+      "step": 1016
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7581567874075552e-05,
+      "loss": 0.0325,
+      "step": 1017
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.757638453681009e-05,
+      "loss": 0.0365,
+      "step": 1018
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.757119641667656e-05,
+      "loss": 0.0389,
+      "step": 1019
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7566003516950146e-05,
+      "loss": 0.038,
+      "step": 1020
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7560805840909058e-05,
+      "loss": 0.0382,
+      "step": 1021
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7555603391834514e-05,
+      "loss": 0.0412,
+      "step": 1022
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.755039617301075e-05,
+      "loss": 0.0498,
+      "step": 1023
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.754518418772502e-05,
+      "loss": 0.044,
+      "step": 1024
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.7539967439267574e-05,
+      "loss": 0.0435,
+      "step": 1025
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.753474593093167e-05,
+      "loss": 0.0367,
+      "step": 1026
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7529519666013578e-05,
+      "loss": 0.0375,
+      "step": 1027
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.752428864781257e-05,
+      "loss": 0.0378,
+      "step": 1028
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.751905287963091e-05,
+      "loss": 0.041,
+      "step": 1029
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7513812364773876e-05,
+      "loss": 0.0434,
+      "step": 1030
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7508567106549725e-05,
+      "loss": 0.0382,
+      "step": 1031
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7503317108269722e-05,
+      "loss": 0.031,
+      "step": 1032
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7498062373248118e-05,
+      "loss": 0.0354,
+      "step": 1033
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7492802904802155e-05,
+      "loss": 0.0321,
+      "step": 1034
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7487538706252062e-05,
+      "loss": 0.0377,
+      "step": 1035
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7482269780921056e-05,
+      "loss": 0.0353,
+      "step": 1036
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7476996132135342e-05,
+      "loss": 0.0375,
+      "step": 1037
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7471717763224096e-05,
+      "loss": 0.0556,
+      "step": 1038
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.7466434677519485e-05,
+      "loss": 0.0516,
+      "step": 1039
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.746114687835665e-05,
+      "loss": 0.0371,
+      "step": 1040
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7455854369073703e-05,
+      "loss": 0.0306,
+      "step": 1041
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.745055715301173e-05,
+      "loss": 0.035,
+      "step": 1042
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7445255233514794e-05,
+      "loss": 0.032,
+      "step": 1043
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7439948613929928e-05,
+      "loss": 0.0431,
+      "step": 1044
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7434637297607125e-05,
+      "loss": 0.0462,
+      "step": 1045
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.742932128789935e-05,
+      "loss": 0.0449,
+      "step": 1046
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.742400058816252e-05,
+      "loss": 0.0354,
+      "step": 1047
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7418675201755525e-05,
+      "loss": 0.0364,
+      "step": 1048
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.741334513204021e-05,
+      "loss": 0.0408,
+      "step": 1049
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.740801038238137e-05,
+      "loss": 0.0399,
+      "step": 1050
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7402670956146766e-05,
+      "loss": 0.0312,
+      "step": 1051
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7397326856707102e-05,
+      "loss": 0.0458,
+      "step": 1052
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.7391978087436032e-05,
+      "loss": 0.0391,
+      "step": 1053
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7386624651710165e-05,
+      "loss": 0.0312,
+      "step": 1054
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.738126655290905e-05,
+      "loss": 0.0475,
+      "step": 1055
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.737590379441518e-05,
+      "loss": 0.0362,
+      "step": 1056
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7370536379613998e-05,
+      "loss": 0.0457,
+      "step": 1057
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7365164311893868e-05,
+      "loss": 0.038,
+      "step": 1058
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.735978759464612e-05,
+      "loss": 0.0414,
+      "step": 1059
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7354406231264983e-05,
+      "loss": 0.035,
+      "step": 1060
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7349020225147653e-05,
+      "loss": 0.0452,
+      "step": 1061
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.734362957969423e-05,
+      "loss": 0.0424,
+      "step": 1062
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7338234298307764e-05,
+      "loss": 0.0411,
+      "step": 1063
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7332834384394225e-05,
+      "loss": 0.0332,
+      "step": 1064
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.7327429841362494e-05,
+      "loss": 0.0393,
+      "step": 1065
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.732202067262439e-05,
+      "loss": 0.0301,
+      "step": 1066
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7316606881594652e-05,
+      "loss": 0.0418,
+      "step": 1067
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7311188471690925e-05,
+      "loss": 0.04,
+      "step": 1068
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7305765446333783e-05,
+      "loss": 0.0367,
+      "step": 1069
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7300337808946703e-05,
+      "loss": 0.0383,
+      "step": 1070
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.729490556295608e-05,
+      "loss": 0.0355,
+      "step": 1071
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.728946871179122e-05,
+      "loss": 0.0427,
+      "step": 1072
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.728402725888433e-05,
+      "loss": 0.0432,
+      "step": 1073
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7278581207670522e-05,
+      "loss": 0.0339,
+      "step": 1074
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7273130561587822e-05,
+      "loss": 0.0335,
+      "step": 1075
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7267675324077142e-05,
+      "loss": 0.0364,
+      "step": 1076
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.72622154985823e-05,
+      "loss": 0.0353,
+      "step": 1077
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7256751088550007e-05,
+      "loss": 0.0366,
+      "step": 1078
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7251282097429876e-05,
+      "loss": 0.0417,
+      "step": 1079
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.7245808528674403e-05,
+      "loss": 0.0426,
+      "step": 1080
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7240330385738974e-05,
+      "loss": 0.0452,
+      "step": 1081
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7234847672081875e-05,
+      "loss": 0.0415,
+      "step": 1082
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7229360391164256e-05,
+      "loss": 0.0339,
+      "step": 1083
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7223868546450172e-05,
+      "loss": 0.0456,
+      "step": 1084
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.721837214140655e-05,
+      "loss": 0.0475,
+      "step": 1085
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7212871179503188e-05,
+      "loss": 0.0411,
+      "step": 1086
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7207365664212773e-05,
+      "loss": 0.0349,
+      "step": 1087
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.720185559901086e-05,
+      "loss": 0.0484,
+      "step": 1088
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.719634098737588e-05,
+      "loss": 0.0361,
+      "step": 1089
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.719082183278913e-05,
+      "loss": 0.0335,
+      "step": 1090
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.718529813873478e-05,
+      "loss": 0.0324,
+      "step": 1091
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.7179769908699856e-05,
+      "loss": 0.0476,
+      "step": 1092
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.717423714617426e-05,
+      "loss": 0.0393,
+      "step": 1093
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.716869985465075e-05,
+      "loss": 0.0393,
+      "step": 1094
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.716315803762494e-05,
+      "loss": 0.0287,
+      "step": 1095
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.71576116985953e-05,
+      "loss": 0.0399,
+      "step": 1096
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7152060841063163e-05,
+      "loss": 0.0385,
+      "step": 1097
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7146505468532707e-05,
+      "loss": 0.0396,
+      "step": 1098
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.714094558451097e-05,
+      "loss": 0.0364,
+      "step": 1099
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7135381192507817e-05,
+      "loss": 0.0379,
+      "step": 1100
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7129812296035985e-05,
+      "loss": 0.0326,
+      "step": 1101
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7124238898611035e-05,
+      "loss": 0.051,
+      "step": 1102
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7118661003751383e-05,
+      "loss": 0.0348,
+      "step": 1103
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.711307861497827e-05,
+      "loss": 0.0358,
+      "step": 1104
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.710749173581579e-05,
+      "loss": 0.0434,
+      "step": 1105
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7101900369790857e-05,
+      "loss": 0.034,
+      "step": 1106
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.709630452043323e-05,
+      "loss": 0.0425,
+      "step": 1107
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7090704191275486e-05,
+      "loss": 0.0322,
+      "step": 1108
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7085099385853036e-05,
+      "loss": 0.0391,
+      "step": 1109
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.707949010770413e-05,
+      "loss": 0.0302,
+      "step": 1110
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.707387636036981e-05,
+      "loss": 0.0285,
+      "step": 1111
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7068258147393967e-05,
+      "loss": 0.0332,
+      "step": 1112
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7062635472323306e-05,
+      "loss": 0.0356,
+      "step": 1113
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7057008338707342e-05,
+      "loss": 0.0323,
+      "step": 1114
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7051376750098404e-05,
+      "loss": 0.0371,
+      "step": 1115
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7045740710051637e-05,
+      "loss": 0.0331,
+      "step": 1116
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7040100222125e-05,
+      "loss": 0.0466,
+      "step": 1117
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.703445528987926e-05,
+      "loss": 0.0375,
+      "step": 1118
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7028805916877975e-05,
+      "loss": 0.0396,
+      "step": 1119
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7023152106687525e-05,
+      "loss": 0.0314,
+      "step": 1120
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.701749386287708e-05,
+      "loss": 0.0365,
+      "step": 1121
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7011831189018607e-05,
+      "loss": 0.0348,
+      "step": 1122
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.700616408868689e-05,
+      "loss": 0.0317,
+      "step": 1123
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.7000492565459476e-05,
+      "loss": 0.0308,
+      "step": 1124
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6994816622916726e-05,
+      "loss": 0.0355,
+      "step": 1125
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6989136264641785e-05,
+      "loss": 0.0401,
+      "step": 1126
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6983451494220587e-05,
+      "loss": 0.0288,
+      "step": 1127
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.697776231524185e-05,
+      "loss": 0.0336,
+      "step": 1128
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6972068731297075e-05,
+      "loss": 0.0332,
+      "step": 1129
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.696637074598054e-05,
+      "loss": 0.0376,
+      "step": 1130
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.696066836288931e-05,
+      "loss": 0.0307,
+      "step": 1131
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6954961585623214e-05,
+      "loss": 0.0443,
+      "step": 1132
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.6949250417784873e-05,
+      "loss": 0.0368,
+      "step": 1133
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.694353486297966e-05,
+      "loss": 0.0316,
+      "step": 1134
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6937814924815733e-05,
+      "loss": 0.0487,
+      "step": 1135
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6932090606904012e-05,
+      "loss": 0.0467,
+      "step": 1136
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6926361912858172e-05,
+      "loss": 0.0419,
+      "step": 1137
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.692062884629467e-05,
+      "loss": 0.0359,
+      "step": 1138
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6914891410832702e-05,
+      "loss": 0.038,
+      "step": 1139
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6909149610094245e-05,
+      "loss": 0.0449,
+      "step": 1140
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6903403447704004e-05,
+      "loss": 0.0387,
+      "step": 1141
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6897652927289466e-05,
+      "loss": 0.0378,
+      "step": 1142
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.689189805248085e-05,
+      "loss": 0.0445,
+      "step": 1143
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6886138826911132e-05,
+      "loss": 0.0435,
+      "step": 1144
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.688037525421603e-05,
+      "loss": 0.0357,
+      "step": 1145
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6874607338034015e-05,
+      "loss": 0.0368,
+      "step": 1146
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6868835082006283e-05,
+      "loss": 0.0382,
+      "step": 1147
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.6863058489776788e-05,
+      "loss": 0.0342,
+      "step": 1148
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6857277564992212e-05,
+      "loss": 0.0382,
+      "step": 1149
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6851492311301973e-05,
+      "loss": 0.0459,
+      "step": 1150
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6845702732358224e-05,
+      "loss": 0.033,
+      "step": 1151
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.683990883181585e-05,
+      "loss": 0.0336,
+      "step": 1152
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6834110613332453e-05,
+      "loss": 0.0447,
+      "step": 1153
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6828308080568376e-05,
+      "loss": 0.0425,
+      "step": 1154
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6822501237186677e-05,
+      "loss": 0.0436,
+      "step": 1155
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.681669008685314e-05,
+      "loss": 0.0403,
+      "step": 1156
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6810874633236255e-05,
+      "loss": 0.0408,
+      "step": 1157
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.680505488000725e-05,
+      "loss": 0.0462,
+      "step": 1158
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6799230830840048e-05,
+      "loss": 0.05,
+      "step": 1159
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.6793402489411293e-05,
+      "loss": 0.0385,
+      "step": 1160
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.678756985940034e-05,
+      "loss": 0.0376,
+      "step": 1161
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6781732944489248e-05,
+      "loss": 0.0415,
+      "step": 1162
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.677589174836278e-05,
+      "loss": 0.0319,
+      "step": 1163
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6770046274708404e-05,
+      "loss": 0.0351,
+      "step": 1164
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.676419652721629e-05,
+      "loss": 0.0404,
+      "step": 1165
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.67583425095793e-05,
+      "loss": 0.0442,
+      "step": 1166
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6752484225493e-05,
+      "loss": 0.0405,
+      "step": 1167
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6746621678655636e-05,
+      "loss": 0.0384,
+      "step": 1168
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6740754872768165e-05,
+      "loss": 0.0277,
+      "step": 1169
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.673488381153421e-05,
+      "loss": 0.0376,
+      "step": 1170
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6729008498660104e-05,
+      "loss": 0.0348,
+      "step": 1171
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.672312893785484e-05,
+      "loss": 0.0257,
+      "step": 1172
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6717245132830114e-05,
+      "loss": 0.0513,
+      "step": 1173
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.6711357087300287e-05,
+      "loss": 0.0375,
+      "step": 1174
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.67054648049824e-05,
+      "loss": 0.0357,
+      "step": 1175
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6699568289596175e-05,
+      "loss": 0.0354,
+      "step": 1176
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6693667544863998e-05,
+      "loss": 0.0382,
+      "step": 1177
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6687762574510927e-05,
+      "loss": 0.0316,
+      "step": 1178
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.66818533822647e-05,
+      "loss": 0.0443,
+      "step": 1179
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6675939971855694e-05,
+      "loss": 0.0312,
+      "step": 1180
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6670022347016972e-05,
+      "loss": 0.0478,
+      "step": 1181
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6664100511484252e-05,
+      "loss": 0.0319,
+      "step": 1182
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6658174468995903e-05,
+      "loss": 0.0346,
+      "step": 1183
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.665224422329296e-05,
+      "loss": 0.0382,
+      "step": 1184
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.66463097781191e-05,
+      "loss": 0.0264,
+      "step": 1185
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.664037113722066e-05,
+      "loss": 0.0328,
+      "step": 1186
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.6634428304346623e-05,
+      "loss": 0.0308,
+      "step": 1187
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.662848128324862e-05,
+      "loss": 0.0454,
+      "step": 1188
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 1.662253007768092e-05,
+      "loss": 0.0431,
+      "step": 1189
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6616574691400443e-05,
+      "loss": 0.0399,
+      "step": 1190
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6610615128166738e-05,
+      "loss": 0.0404,
+      "step": 1191
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6604651391742e-05,
+      "loss": 0.0488,
+      "step": 1192
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6598683485891053e-05,
+      "loss": 0.0339,
+      "step": 1193
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.659271141438135e-05,
+      "loss": 0.0428,
+      "step": 1194
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6586735180982986e-05,
+      "loss": 0.033,
+      "step": 1195
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6580754789468666e-05,
+      "loss": 0.0413,
+      "step": 1196
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.657477024361374e-05,
+      "loss": 0.0382,
+      "step": 1197
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6568781547196163e-05,
+      "loss": 0.0434,
+      "step": 1198
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.656278870399652e-05,
+      "loss": 0.0316,
+      "step": 1199
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6556791717798013e-05,
+      "loss": 0.0394,
+      "step": 1200
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.655079059238645e-05,
+      "loss": 0.0472,
+      "step": 1201
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 1.6544785331550265e-05,
+      "loss": 0.0295,
+      "step": 1202
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.65387759390805e-05,
+      "loss": 0.0406,
+      "step": 1203
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6532762418770797e-05,
+      "loss": 0.0443,
+      "step": 1204
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.652674477441741e-05,
+      "loss": 0.0322,
+      "step": 1205
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.65207230098192e-05,
+      "loss": 0.0365,
+      "step": 1206
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6514697128777615e-05,
+      "loss": 0.0412,
+      "step": 1207
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6508667135096727e-05,
+      "loss": 0.0392,
+      "step": 1208
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6502633032583173e-05,
+      "loss": 0.0353,
+      "step": 1209
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6496594825046206e-05,
+      "loss": 0.0361,
+      "step": 1210
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6490552516297667e-05,
+      "loss": 0.0307,
+      "step": 1211
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6484506110151977e-05,
+      "loss": 0.0383,
+      "step": 1212
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6478455610426157e-05,
+      "loss": 0.0382,
+      "step": 1213
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6472401020939797e-05,
+      "loss": 0.0414,
+      "step": 1214
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.646634234551508e-05,
+      "loss": 0.031,
+      "step": 1215
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 1.6460279587976765e-05,
+      "loss": 0.0423,
+      "step": 1216
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.645421275215219e-05,
+      "loss": 0.0421,
+      "step": 1217
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6448141841871262e-05,
+      "loss": 0.0376,
+      "step": 1218
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6442066860966466e-05,
+      "loss": 0.0241,
+      "step": 1219
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6435987813272854e-05,
+      "loss": 0.0428,
+      "step": 1220
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6429904702628044e-05,
+      "loss": 0.0304,
+      "step": 1221
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.642381753287223e-05,
+      "loss": 0.0388,
+      "step": 1222
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6417726307848143e-05,
+      "loss": 0.0394,
+      "step": 1223
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.64116310314011e-05,
+      "loss": 0.0392,
+      "step": 1224
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6405531707378965e-05,
+      "loss": 0.0307,
+      "step": 1225
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.639942833963216e-05,
+      "loss": 0.0384,
+      "step": 1226
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.639332093201365e-05,
+      "loss": 0.0386,
+      "step": 1227
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6387209488378966e-05,
+      "loss": 0.0338,
+      "step": 1228
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 1.6381094012586176e-05,
+      "loss": 0.0343,
+      "step": 1229
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6374974508495895e-05,
+      "loss": 0.035,
+      "step": 1230
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6368850979971287e-05,
+      "loss": 0.0403,
+      "step": 1231
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6362723430878045e-05,
+      "loss": 0.0258,
+      "step": 1232
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6356591865084413e-05,
+      "loss": 0.0383,
+      "step": 1233
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.635045628646116e-05,
+      "loss": 0.0347,
+      "step": 1234
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6344316698881595e-05,
+      "loss": 0.0435,
+      "step": 1235
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.633817310622156e-05,
+      "loss": 0.0545,
+      "step": 1236
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6332025512359412e-05,
+      "loss": 0.0296,
+      "step": 1237
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.632587392117605e-05,
+      "loss": 0.0412,
+      "step": 1238
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.631971833655489e-05,
+      "loss": 0.0381,
+      "step": 1239
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6313558762381865e-05,
+      "loss": 0.0425,
+      "step": 1240
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6307395202545427e-05,
+      "loss": 0.0402,
+      "step": 1241
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.630122766093656e-05,
+      "loss": 0.0444,
+      "step": 1242
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.6295056141448734e-05,
+      "loss": 0.035,
+      "step": 1243
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6288880647977953e-05,
+      "loss": 0.0419,
+      "step": 1244
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6282701184422717e-05,
+      "loss": 0.0307,
+      "step": 1245
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6276517754684044e-05,
+      "loss": 0.0319,
+      "step": 1246
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.627033036266544e-05,
+      "loss": 0.029,
+      "step": 1247
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6264139012272927e-05,
+      "loss": 0.0274,
+      "step": 1248
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.625794370741502e-05,
+      "loss": 0.0323,
+      "step": 1249
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.625174445200273e-05,
+      "loss": 0.0311,
+      "step": 1250
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6245541249949558e-05,
+      "loss": 0.0352,
+      "step": 1251
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6239334105171506e-05,
+      "loss": 0.0348,
+      "step": 1252
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6233123021587057e-05,
+      "loss": 0.0368,
+      "step": 1253
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.622690800311718e-05,
+      "loss": 0.0395,
+      "step": 1254
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.622068905368534e-05,
+      "loss": 0.0375,
+      "step": 1255
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.621446617721747e-05,
+      "loss": 0.0424,
+      "step": 1256
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.620823937764198e-05,
+      "loss": 0.0373,
+      "step": 1257
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6202008658889776e-05,
+      "loss": 0.0413,
+      "step": 1258
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.619577402489422e-05,
+      "loss": 0.038,
+      "step": 1259
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.618953547959115e-05,
+      "loss": 0.0373,
+      "step": 1260
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6183293026918875e-05,
+      "loss": 0.029,
+      "step": 1261
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.617704667081817e-05,
+      "loss": 0.028,
+      "step": 1262
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6170796415232278e-05,
+      "loss": 0.0457,
+      "step": 1263
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6164542264106895e-05,
+      "loss": 0.0299,
+      "step": 1264
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.615828422139018e-05,
+      "loss": 0.0385,
+      "step": 1265
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.615202229103276e-05,
+      "loss": 0.0329,
+      "step": 1266
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6145756476987692e-05,
+      "loss": 0.0373,
+      "step": 1267
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.613948678321051e-05,
+      "loss": 0.0305,
+      "step": 1268
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.613321321365918e-05,
+      "loss": 0.0408,
+      "step": 1269
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.6126935772294127e-05,
+      "loss": 0.0385,
+      "step": 1270
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.612065446307821e-05,
+      "loss": 0.0375,
+      "step": 1271
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6114369289976727e-05,
+      "loss": 0.0301,
+      "step": 1272
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6108080256957434e-05,
+      "loss": 0.0351,
+      "step": 1273
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6101787367990503e-05,
+      "loss": 0.0455,
+      "step": 1274
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.609549062704855e-05,
+      "loss": 0.034,
+      "step": 1275
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6089190038106628e-05,
+      "loss": 0.03,
+      "step": 1276
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.608288560514221e-05,
+      "loss": 0.0394,
+      "step": 1277
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.607657733213519e-05,
+      "loss": 0.0392,
+      "step": 1278
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6070265223067904e-05,
+      "loss": 0.0336,
+      "step": 1279
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6063949281925095e-05,
+      "loss": 0.0299,
+      "step": 1280
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6057629512693938e-05,
+      "loss": 0.0315,
+      "step": 1281
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6051305919364008e-05,
+      "loss": 0.0344,
+      "step": 1282
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.604497850592731e-05,
+      "loss": 0.0329,
+      "step": 1283
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6038647276378246e-05,
+      "loss": 0.0334,
+      "step": 1284
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.603231223471364e-05,
+      "loss": 0.0316,
+      "step": 1285
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6025973384932722e-05,
+      "loss": 0.0299,
+      "step": 1286
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.601963073103711e-05,
+      "loss": 0.0358,
+      "step": 1287
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6013284277030844e-05,
+      "loss": 0.0361,
+      "step": 1288
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6006934026920347e-05,
+      "loss": 0.0412,
+      "step": 1289
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6000579984714453e-05,
+      "loss": 0.042,
+      "step": 1290
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5994222154424376e-05,
+      "loss": 0.0353,
+      "step": 1291
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5987860540063725e-05,
+      "loss": 0.0374,
+      "step": 1292
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5981495145648507e-05,
+      "loss": 0.0361,
+      "step": 1293
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5975125975197107e-05,
+      "loss": 0.0389,
+      "step": 1294
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.596875303273029e-05,
+      "loss": 0.0389,
+      "step": 1295
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5962376322271218e-05,
+      "loss": 0.0321,
+      "step": 1296
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5955995847845415e-05,
+      "loss": 0.0312,
+      "step": 1297
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5949611613480784e-05,
+      "loss": 0.0367,
+      "step": 1298
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5943223623207608e-05,
+      "loss": 0.0329,
+      "step": 1299
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5936831881058533e-05,
+      "loss": 0.0317,
+      "step": 1300
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.593043639106859e-05,
+      "loss": 0.0383,
+      "step": 1301
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5924037157275156e-05,
+      "loss": 0.0346,
+      "step": 1302
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5917634183717985e-05,
+      "loss": 0.0418,
+      "step": 1303
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.591122747443918e-05,
+      "loss": 0.044,
+      "step": 1304
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5904817033483216e-05,
+      "loss": 0.0225,
+      "step": 1305
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5898402864896908e-05,
+      "loss": 0.028,
+      "step": 1306
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5891984972729448e-05,
+      "loss": 0.0321,
+      "step": 1307
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.588556336103235e-05,
+      "loss": 0.0437,
+      "step": 1308
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.587913803385949e-05,
+      "loss": 0.0377,
+      "step": 1309
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.5872708995267094e-05,
+      "loss": 0.0415,
+      "step": 1310
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.586627624931373e-05,
+      "loss": 0.0266,
+      "step": 1311
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5859839800060295e-05,
+      "loss": 0.039,
+      "step": 1312
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5853399651570036e-05,
+      "loss": 0.0397,
+      "step": 1313
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.584695580790853e-05,
+      "loss": 0.0328,
+      "step": 1314
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5840508273143687e-05,
+      "loss": 0.0469,
+      "step": 1315
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5834057051345745e-05,
+      "loss": 0.0305,
+      "step": 1316
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5827602146587277e-05,
+      "loss": 0.0373,
+      "step": 1317
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.582114356294317e-05,
+      "loss": 0.0325,
+      "step": 1318
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5814681304490653e-05,
+      "loss": 0.0362,
+      "step": 1319
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5808215375309243e-05,
+      "loss": 0.0365,
+      "step": 1320
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5801745779480807e-05,
+      "loss": 0.0294,
+      "step": 1321
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5795272521089503e-05,
+      "loss": 0.028,
+      "step": 1322
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.578879560422182e-05,
+      "loss": 0.0258,
+      "step": 1323
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.578231503296654e-05,
+      "loss": 0.0438,
+      "step": 1324
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.5775830811414762e-05,
+      "loss": 0.0462,
+      "step": 1325
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.576934294365988e-05,
+      "loss": 0.0315,
+      "step": 1326
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.576285143379761e-05,
+      "loss": 0.0307,
+      "step": 1327
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5756356285925943e-05,
+      "loss": 0.0359,
+      "step": 1328
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.574985750414518e-05,
+      "loss": 0.0409,
+      "step": 1329
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5743355092557912e-05,
+      "loss": 0.0303,
+      "step": 1330
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5736849055269024e-05,
+      "loss": 0.0332,
+      "step": 1331
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5730339396385684e-05,
+      "loss": 0.0319,
+      "step": 1332
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5723826120017363e-05,
+      "loss": 0.0261,
+      "step": 1333
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.571730923027579e-05,
+      "loss": 0.0365,
+      "step": 1334
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.571078873127499e-05,
+      "loss": 0.0294,
+      "step": 1335
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5704264627131274e-05,
+      "loss": 0.0352,
+      "step": 1336
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5697736921963212e-05,
+      "loss": 0.0329,
+      "step": 1337
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.569120561989166e-05,
+      "loss": 0.03,
+      "step": 1338
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5684670725039736e-05,
+      "loss": 0.0418,
+      "step": 1339
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.567813224153283e-05,
+      "loss": 0.0355,
+      "step": 1340
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5671590173498602e-05,
+      "loss": 0.0322,
+      "step": 1341
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5665044525066963e-05,
+      "loss": 0.0309,
+      "step": 1342
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5658495300370093e-05,
+      "loss": 0.0261,
+      "step": 1343
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5651942503542435e-05,
+      "loss": 0.0356,
+      "step": 1344
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5645386138720672e-05,
+      "loss": 0.0316,
+      "step": 1345
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5638826210043752e-05,
+      "loss": 0.0325,
+      "step": 1346
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.563226272165287e-05,
+      "loss": 0.038,
+      "step": 1347
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.562569567769146e-05,
+      "loss": 0.0332,
+      "step": 1348
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.561912508230521e-05,
+      "loss": 0.0458,
+      "step": 1349
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.561255093964205e-05,
+      "loss": 0.0298,
+      "step": 1350
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5605973253852148e-05,
+      "loss": 0.036,
+      "step": 1351
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5599392029087897e-05,
+      "loss": 0.0278,
+      "step": 1352
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.559280726950395e-05,
+      "loss": 0.0313,
+      "step": 1353
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5586218979257158e-05,
+      "loss": 0.0332,
+      "step": 1354
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.557962716250663e-05,
+      "loss": 0.0379,
+      "step": 1355
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.557303182341369e-05,
+      "loss": 0.0363,
+      "step": 1356
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.556643296614188e-05,
+      "loss": 0.0339,
+      "step": 1357
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5559830594856973e-05,
+      "loss": 0.0348,
+      "step": 1358
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5553224713726954e-05,
+      "loss": 0.0299,
+      "step": 1359
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5546615326922024e-05,
+      "loss": 0.0268,
+      "step": 1360
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5540002438614605e-05,
+      "loss": 0.032,
+      "step": 1361
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.553338605297931e-05,
+      "loss": 0.031,
+      "step": 1362
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.552676617419299e-05,
+      "loss": 0.0345,
+      "step": 1363
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5520142806434667e-05,
+      "loss": 0.0453,
+      "step": 1364
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.55135159538856e-05,
+      "loss": 0.0233,
+      "step": 1365
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5506885620729215e-05,
+      "loss": 0.0288,
+      "step": 1366
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.550025181115116e-05,
+      "loss": 0.0371,
+      "step": 1367
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.549361452933926e-05,
+      "loss": 0.0281,
+      "step": 1368
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.548697377948355e-05,
+      "loss": 0.0253,
+      "step": 1369
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.548032956577624e-05,
+      "loss": 0.0328,
+      "step": 1370
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5473681892411733e-05,
+      "loss": 0.0286,
+      "step": 1371
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.546703076358661e-05,
+      "loss": 0.0373,
+      "step": 1372
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5460376183499645e-05,
+      "loss": 0.0288,
+      "step": 1373
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5453718156351775e-05,
+      "loss": 0.0288,
+      "step": 1374
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.544705668634613e-05,
+      "loss": 0.0327,
+      "step": 1375
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5440391777688e-05,
+      "loss": 0.0306,
+      "step": 1376
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.543372343458485e-05,
+      "loss": 0.0255,
+      "step": 1377
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.542705166124632e-05,
+      "loss": 0.0319,
+      "step": 1378
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 1.5420376461884203e-05,
+      "loss": 0.0292,
+      "step": 1379
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.541369784071246e-05,
+      "loss": 0.0317,
+      "step": 1380
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5407015801947224e-05,
+      "loss": 0.0249,
+      "step": 1381
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.540033034980676e-05,
+      "loss": 0.0337,
+      "step": 1382
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5393641488511514e-05,
+      "loss": 0.0266,
+      "step": 1383
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5386949222284072e-05,
+      "loss": 0.0289,
+      "step": 1384
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5380253555349163e-05,
+      "loss": 0.0321,
+      "step": 1385
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.537355449193367e-05,
+      "loss": 0.0315,
+      "step": 1386
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5366852036266636e-05,
+      "loss": 0.039,
+      "step": 1387
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.536014619257921e-05,
+      "loss": 0.0398,
+      "step": 1388
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5353436965104708e-05,
+      "loss": 0.028,
+      "step": 1389
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5346724358078576e-05,
+      "loss": 0.0296,
+      "step": 1390
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.5340008375738385e-05,
+      "loss": 0.0323,
+      "step": 1391
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 1.533328902232385e-05,
+      "loss": 0.0268,
+      "step": 1392
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.53265663020768e-05,
+      "loss": 0.0304,
+      "step": 1393
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5319840219241205e-05,
+      "loss": 0.0226,
+      "step": 1394
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5313110778063142e-05,
+      "loss": 0.0321,
+      "step": 1395
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5306377982790818e-05,
+      "loss": 0.0212,
+      "step": 1396
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.529964183767455e-05,
+      "loss": 0.0321,
+      "step": 1397
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5292902346966782e-05,
+      "loss": 0.0306,
+      "step": 1398
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.528615951492206e-05,
+      "loss": 0.0278,
+      "step": 1399
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5279413345797037e-05,
+      "loss": 0.0251,
+      "step": 1400
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5272663843850484e-05,
+      "loss": 0.0263,
+      "step": 1401
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5265911013343265e-05,
+      "loss": 0.0217,
+      "step": 1402
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5259154858538354e-05,
+      "loss": 0.0278,
+      "step": 1403
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5252395383700815e-05,
+      "loss": 0.0257,
+      "step": 1404
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5245632593097814e-05,
+      "loss": 0.03,
+      "step": 1405
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 1.5238866490998607e-05,
+      "loss": 0.0251,
+      "step": 1406
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5232097081674542e-05,
+      "loss": 0.0265,
+      "step": 1407
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5225324369399058e-05,
+      "loss": 0.0234,
+      "step": 1408
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5218548358447668e-05,
+      "loss": 0.0332,
+      "step": 1409
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5211769053097984e-05,
+      "loss": 0.0354,
+      "step": 1410
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5204986457629685e-05,
+      "loss": 0.0262,
+      "step": 1411
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5198200576324534e-05,
+      "loss": 0.029,
+      "step": 1412
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5191411413466357e-05,
+      "loss": 0.0229,
+      "step": 1413
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5184618973341068e-05,
+      "loss": 0.0289,
+      "step": 1414
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5177823260236636e-05,
+      "loss": 0.0275,
+      "step": 1415
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5171024278443108e-05,
+      "loss": 0.0331,
+      "step": 1416
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5164222032252585e-05,
+      "loss": 0.0311,
+      "step": 1417
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5157416525959227e-05,
+      "loss": 0.0231,
+      "step": 1418
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5150607763859266e-05,
+      "loss": 0.0358,
+      "step": 1419
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 1.5143795750250976e-05,
+      "loss": 0.0307,
+      "step": 1420
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5136980489434691e-05,
+      "loss": 0.0223,
+      "step": 1421
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5130161985712786e-05,
+      "loss": 0.0232,
+      "step": 1422
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5123340243389697e-05,
+      "loss": 0.0243,
+      "step": 1423
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5116515266771888e-05,
+      "loss": 0.0303,
+      "step": 1424
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.510968706016788e-05,
+      "loss": 0.0266,
+      "step": 1425
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5102855627888228e-05,
+      "loss": 0.0388,
+      "step": 1426
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5096020974245518e-05,
+      "loss": 0.0399,
+      "step": 1427
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5089183103554372e-05,
+      "loss": 0.0317,
+      "step": 1428
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5082342020131448e-05,
+      "loss": 0.0335,
+      "step": 1429
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5075497728295427e-05,
+      "loss": 0.0323,
+      "step": 1430
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.506865023236702e-05,
+      "loss": 0.0368,
+      "step": 1431
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5061799536668951e-05,
+      "loss": 0.0452,
+      "step": 1432
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 1.5054945645525973e-05,
+      "loss": 0.0322,
+      "step": 1433
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.504808856326486e-05,
+      "loss": 0.0287,
+      "step": 1434
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5041228294214388e-05,
+      "loss": 0.0283,
+      "step": 1435
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5034364842705352e-05,
+      "loss": 0.0342,
+      "step": 1436
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5027498213070558e-05,
+      "loss": 0.0283,
+      "step": 1437
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5020628409644813e-05,
+      "loss": 0.0274,
+      "step": 1438
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5013755436764928e-05,
+      "loss": 0.0316,
+      "step": 1439
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5006879298769721e-05,
+      "loss": 0.026,
+      "step": 1440
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 0.0228,
+      "step": 1441
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4993117544798578e-05,
+      "loss": 0.0239,
+      "step": 1442
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.498623193751025e-05,
+      "loss": 0.0366,
+      "step": 1443
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4979343182481812e-05,
+      "loss": 0.0324,
+      "step": 1444
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.497245128406203e-05,
+      "loss": 0.0316,
+      "step": 1445
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.4965556246601677e-05,
+      "loss": 0.0382,
+      "step": 1446
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 1.495865807445349e-05,
+      "loss": 0.0349,
+      "step": 1447
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.495175677197219e-05,
+      "loss": 0.0355,
+      "step": 1448
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4944852343514478e-05,
+      "loss": 0.0315,
+      "step": 1449
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4937944793439023e-05,
+      "loss": 0.0284,
+      "step": 1450
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4931034126106468e-05,
+      "loss": 0.0347,
+      "step": 1451
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4924120345879422e-05,
+      "loss": 0.0421,
+      "step": 1452
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4917203457122465e-05,
+      "loss": 0.0331,
+      "step": 1453
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4910283464202126e-05,
+      "loss": 0.0333,
+      "step": 1454
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.490336037148691e-05,
+      "loss": 0.0312,
+      "step": 1455
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4896434183347266e-05,
+      "loss": 0.0309,
+      "step": 1456
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4889504904155602e-05,
+      "loss": 0.0292,
+      "step": 1457
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4882572538286279e-05,
+      "loss": 0.0353,
+      "step": 1458
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4875637090115605e-05,
+      "loss": 0.03,
+      "step": 1459
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 1.4868698564021833e-05,
+      "loss": 0.0325,
+      "step": 1460
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.486175696438516e-05,
+      "loss": 0.0305,
+      "step": 1461
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4854812295587722e-05,
+      "loss": 0.0276,
+      "step": 1462
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4847864562013596e-05,
+      "loss": 0.033,
+      "step": 1463
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4840913768048788e-05,
+      "loss": 0.0394,
+      "step": 1464
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4833959918081244e-05,
+      "loss": 0.027,
+      "step": 1465
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4827003016500827e-05,
+      "loss": 0.0337,
+      "step": 1466
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4820043067699342e-05,
+      "loss": 0.0419,
+      "step": 1467
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4813080076070505e-05,
+      "loss": 0.027,
+      "step": 1468
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4806114046009958e-05,
+      "loss": 0.0276,
+      "step": 1469
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.479914498191526e-05,
+      "loss": 0.0268,
+      "step": 1470
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4792172888185891e-05,
+      "loss": 0.0248,
+      "step": 1471
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4785197769223234e-05,
+      "loss": 0.0249,
+      "step": 1472
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4778219629430585e-05,
+      "loss": 0.0317,
+      "step": 1473
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 1.4771238473213153e-05,
+      "loss": 0.028,
+      "step": 1474
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4764254304978042e-05,
+      "loss": 0.0389,
+      "step": 1475
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4757267129134266e-05,
+      "loss": 0.0267,
+      "step": 1476
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4750276950092728e-05,
+      "loss": 0.0259,
+      "step": 1477
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4743283772266238e-05,
+      "loss": 0.0348,
+      "step": 1478
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4736287600069493e-05,
+      "loss": 0.0364,
+      "step": 1479
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4729288437919078e-05,
+      "loss": 0.0319,
+      "step": 1480
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4722286290233474e-05,
+      "loss": 0.0262,
+      "step": 1481
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4715281161433032e-05,
+      "loss": 0.0288,
+      "step": 1482
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4708273055940003e-05,
+      "loss": 0.0297,
+      "step": 1483
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4701261978178506e-05,
+      "loss": 0.0261,
+      "step": 1484
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4694247932574533e-05,
+      "loss": 0.0371,
+      "step": 1485
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4687230923555961e-05,
+      "loss": 0.0274,
+      "step": 1486
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4680210955552535e-05,
+      "loss": 0.0288,
+      "step": 1487
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 1.4673188032995858e-05,
+      "loss": 0.0407,
+      "step": 1488
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4666162160319406e-05,
+      "loss": 0.0335,
+      "step": 1489
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.465913334195852e-05,
+      "loss": 0.0389,
+      "step": 1490
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4652101582350394e-05,
+      "loss": 0.0378,
+      "step": 1491
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.464506688593408e-05,
+      "loss": 0.0283,
+      "step": 1492
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4638029257150494e-05,
+      "loss": 0.0314,
+      "step": 1493
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4630988700442386e-05,
+      "loss": 0.0275,
+      "step": 1494
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4623945220254367e-05,
+      "loss": 0.0378,
+      "step": 1495
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4616898821032892e-05,
+      "loss": 0.0291,
+      "step": 1496
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4609849507226254e-05,
+      "loss": 0.0314,
+      "step": 1497
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4602797283284586e-05,
+      "loss": 0.0307,
+      "step": 1498
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4595742153659867e-05,
+      "loss": 0.0329,
+      "step": 1499
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4588684122805895e-05,
+      "loss": 0.037,
+      "step": 1500
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 1.4581623195178318e-05,
+      "loss": 0.0245,
+      "step": 1501
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.45745593752346e-05,
+      "loss": 0.0254,
+      "step": 1502
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4567492667434031e-05,
+      "loss": 0.0326,
+      "step": 1503
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4560423076237729e-05,
+      "loss": 0.0283,
+      "step": 1504
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4553350606108632e-05,
+      "loss": 0.0334,
+      "step": 1505
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4546275261511493e-05,
+      "loss": 0.0324,
+      "step": 1506
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4539197046912876e-05,
+      "loss": 0.0294,
+      "step": 1507
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4532115966781168e-05,
+      "loss": 0.0352,
+      "step": 1508
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4525032025586555e-05,
+      "loss": 0.0328,
+      "step": 1509
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4517945227801029e-05,
+      "loss": 0.0292,
+      "step": 1510
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4510855577898397e-05,
+      "loss": 0.0321,
+      "step": 1511
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.450376308035425e-05,
+      "loss": 0.0414,
+      "step": 1512
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4496667739645986e-05,
+      "loss": 0.0331,
+      "step": 1513
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4489569560252804e-05,
+      "loss": 0.0322,
+      "step": 1514
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 1.4482468546655679e-05,
+      "loss": 0.0267,
+      "step": 1515
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.447536470333739e-05,
+      "loss": 0.0307,
+      "step": 1516
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.446825803478249e-05,
+      "loss": 0.0284,
+      "step": 1517
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4461148545477328e-05,
+      "loss": 0.0334,
+      "step": 1518
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4454036239910022e-05,
+      "loss": 0.0239,
+      "step": 1519
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4446921122570478e-05,
+      "loss": 0.0272,
+      "step": 1520
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.443980319795037e-05,
+      "loss": 0.0305,
+      "step": 1521
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4432682470543149e-05,
+      "loss": 0.0388,
+      "step": 1522
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4425558944844027e-05,
+      "loss": 0.0344,
+      "step": 1523
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4418432625349997e-05,
+      "loss": 0.0291,
+      "step": 1524
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.44113035165598e-05,
+      "loss": 0.0313,
+      "step": 1525
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.440417162297395e-05,
+      "loss": 0.0293,
+      "step": 1526
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.439703694909471e-05,
+      "loss": 0.0305,
+      "step": 1527
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 1.4389899499426103e-05,
+      "loss": 0.0217,
+      "step": 1528
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4382759278473906e-05,
+      "loss": 0.024,
+      "step": 1529
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.437561629074564e-05,
+      "loss": 0.0336,
+      "step": 1530
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4368470540750578e-05,
+      "loss": 0.0379,
+      "step": 1531
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4361322032999736e-05,
+      "loss": 0.0272,
+      "step": 1532
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4354170772005862e-05,
+      "loss": 0.0255,
+      "step": 1533
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4347016762283454e-05,
+      "loss": 0.0273,
+      "step": 1534
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4339860008348743e-05,
+      "loss": 0.0293,
+      "step": 1535
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4332700514719687e-05,
+      "loss": 0.025,
+      "step": 1536
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4325538285915974e-05,
+      "loss": 0.0346,
+      "step": 1537
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4318373326459027e-05,
+      "loss": 0.0266,
+      "step": 1538
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4311205640871985e-05,
+      "loss": 0.0222,
+      "step": 1539
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.430403523367971e-05,
+      "loss": 0.0211,
+      "step": 1540
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.429686210940878e-05,
+      "loss": 0.0335,
+      "step": 1541
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 1.4289686272587493e-05,
+      "loss": 0.0249,
+      "step": 1542
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4282507727745853e-05,
+      "loss": 0.0254,
+      "step": 1543
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4275326479415581e-05,
+      "loss": 0.0486,
+      "step": 1544
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4268142532130102e-05,
+      "loss": 0.0398,
+      "step": 1545
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4260955890424539e-05,
+      "loss": 0.0314,
+      "step": 1546
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4253766558835724e-05,
+      "loss": 0.0371,
+      "step": 1547
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4246574541902187e-05,
+      "loss": 0.0344,
+      "step": 1548
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4239379844164144e-05,
+      "loss": 0.0333,
+      "step": 1549
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.423218247016351e-05,
+      "loss": 0.0378,
+      "step": 1550
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4224982424443894e-05,
+      "loss": 0.0278,
+      "step": 1551
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4217779711550578e-05,
+      "loss": 0.0361,
+      "step": 1552
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4210574336030545e-05,
+      "loss": 0.0382,
+      "step": 1553
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4203366302432447e-05,
+      "loss": 0.0389,
+      "step": 1554
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.4196155615306615e-05,
+      "loss": 0.0278,
+      "step": 1555
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 1.418894227920506e-05,
+      "loss": 0.0319,
+      "step": 1556
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4181726298681462e-05,
+      "loss": 0.0335,
+      "step": 1557
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4174507678291169e-05,
+      "loss": 0.0349,
+      "step": 1558
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.41672864225912e-05,
+      "loss": 0.0322,
+      "step": 1559
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4160062536140235e-05,
+      "loss": 0.0338,
+      "step": 1560
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4152836023498613e-05,
+      "loss": 0.0317,
+      "step": 1561
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4145606889228331e-05,
+      "loss": 0.0265,
+      "step": 1562
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.413837513789305e-05,
+      "loss": 0.0302,
+      "step": 1563
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4131140774058067e-05,
+      "loss": 0.0311,
+      "step": 1564
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4123903802290346e-05,
+      "loss": 0.0359,
+      "step": 1565
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4116664227158481e-05,
+      "loss": 0.0231,
+      "step": 1566
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.410942205323272e-05,
+      "loss": 0.0286,
+      "step": 1567
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4102177285084946e-05,
+      "loss": 0.0272,
+      "step": 1568
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 1.4094929927288688e-05,
+      "loss": 0.0338,
+      "step": 1569
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4087679984419096e-05,
+      "loss": 0.0314,
+      "step": 1570
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4080427461052966e-05,
+      "loss": 0.0374,
+      "step": 1571
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4073172361768715e-05,
+      "loss": 0.0332,
+      "step": 1572
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4065914691146388e-05,
+      "loss": 0.033,
+      "step": 1573
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4058654453767655e-05,
+      "loss": 0.0396,
+      "step": 1574
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4051391654215803e-05,
+      "loss": 0.0357,
+      "step": 1575
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4044126297075735e-05,
+      "loss": 0.0383,
+      "step": 1576
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4036858386933977e-05,
+      "loss": 0.0218,
+      "step": 1577
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.402958792837866e-05,
+      "loss": 0.033,
+      "step": 1578
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4022314925999525e-05,
+      "loss": 0.0351,
+      "step": 1579
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4015039384387916e-05,
+      "loss": 0.0278,
+      "step": 1580
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4007761308136791e-05,
+      "loss": 0.0426,
+      "step": 1581
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.4000480701840688e-05,
+      "loss": 0.0292,
+      "step": 1582
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 1.3993197570095766e-05,
+      "loss": 0.0274,
+      "step": 1583
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3985911917499764e-05,
+      "loss": 0.0386,
+      "step": 1584
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3978623748652007e-05,
+      "loss": 0.0277,
+      "step": 1585
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3971333068153423e-05,
+      "loss": 0.0316,
+      "step": 1586
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3964039880606522e-05,
+      "loss": 0.034,
+      "step": 1587
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3956744190615385e-05,
+      "loss": 0.0215,
+      "step": 1588
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.394944600278569e-05,
+      "loss": 0.0392,
+      "step": 1589
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3942145321724678e-05,
+      "loss": 0.0289,
+      "step": 1590
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3934842152041173e-05,
+      "loss": 0.0326,
+      "step": 1591
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.392753649834556e-05,
+      "loss": 0.0275,
+      "step": 1592
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3920228365249807e-05,
+      "loss": 0.0266,
+      "step": 1593
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3912917757367429e-05,
+      "loss": 0.0349,
+      "step": 1594
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.390560467931352e-05,
+      "loss": 0.0382,
+      "step": 1595
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 1.3898289135704726e-05,
+      "loss": 0.0347,
+      "step": 1596
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3890971131159246e-05,
+      "loss": 0.0309,
+      "step": 1597
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3883650670296838e-05,
+      "loss": 0.0355,
+      "step": 1598
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.387632775773881e-05,
+      "loss": 0.0335,
+      "step": 1599
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3869002398108017e-05,
+      "loss": 0.0276,
+      "step": 1600
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3861674596028855e-05,
+      "loss": 0.0277,
+      "step": 1601
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3854344356127272e-05,
+      "loss": 0.0394,
+      "step": 1602
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.384701168303074e-05,
+      "loss": 0.0256,
+      "step": 1603
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.383967658136828e-05,
+      "loss": 0.0428,
+      "step": 1604
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3832339055770443e-05,
+      "loss": 0.0214,
+      "step": 1605
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3824999110869306e-05,
+      "loss": 0.0354,
+      "step": 1606
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3817656751298474e-05,
+      "loss": 0.0316,
+      "step": 1607
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3810311981693084e-05,
+      "loss": 0.0364,
+      "step": 1608
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3802964806689784e-05,
+      "loss": 0.028,
+      "step": 1609
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 1.3795615230926748e-05,
+      "loss": 0.0283,
+      "step": 1610
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.378826325904366e-05,
+      "loss": 0.0238,
+      "step": 1611
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3780908895681725e-05,
+      "loss": 0.0249,
+      "step": 1612
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3773552145483642e-05,
+      "loss": 0.0217,
+      "step": 1613
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3766193013093637e-05,
+      "loss": 0.0314,
+      "step": 1614
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3758831503157426e-05,
+      "loss": 0.034,
+      "step": 1615
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3751467620322228e-05,
+      "loss": 0.0365,
+      "step": 1616
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3744101369236765e-05,
+      "loss": 0.0309,
+      "step": 1617
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.373673275455125e-05,
+      "loss": 0.0279,
+      "step": 1618
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3729361780917383e-05,
+      "loss": 0.0393,
+      "step": 1619
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3721988452988366e-05,
+      "loss": 0.0354,
+      "step": 1620
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3714612775418878e-05,
+      "loss": 0.0275,
+      "step": 1621
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3707234752865084e-05,
+      "loss": 0.0268,
+      "step": 1622
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3699854389984626e-05,
+      "loss": 0.0231,
+      "step": 1623
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 1.3692471691436632e-05,
+      "loss": 0.0308,
+      "step": 1624
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3685086661881695e-05,
+      "loss": 0.0334,
+      "step": 1625
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.367769930598188e-05,
+      "loss": 0.0313,
+      "step": 1626
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3670309628400733e-05,
+      "loss": 0.0246,
+      "step": 1627
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3662917633803246e-05,
+      "loss": 0.0323,
+      "step": 1628
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3655523326855889e-05,
+      "loss": 0.0311,
+      "step": 1629
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.364812671222659e-05,
+      "loss": 0.0298,
+      "step": 1630
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3640727794584725e-05,
+      "loss": 0.0178,
+      "step": 1631
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3633326578601133e-05,
+      "loss": 0.0288,
+      "step": 1632
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3625923068948103e-05,
+      "loss": 0.0259,
+      "step": 1633
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3618517270299367e-05,
+      "loss": 0.0311,
+      "step": 1634
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.36111091873301e-05,
+      "loss": 0.0276,
+      "step": 1635
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.360369882471693e-05,
+      "loss": 0.027,
+      "step": 1636
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 1.3596286187137917e-05,
+      "loss": 0.0352,
+      "step": 1637
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3588871279272553e-05,
+      "loss": 0.0289,
+      "step": 1638
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3581454105801775e-05,
+      "loss": 0.0247,
+      "step": 1639
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3574034671407937e-05,
+      "loss": 0.0401,
+      "step": 1640
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.356661298077483e-05,
+      "loss": 0.0323,
+      "step": 1641
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3559189038587665e-05,
+      "loss": 0.0328,
+      "step": 1642
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3551762849533077e-05,
+      "loss": 0.0249,
+      "step": 1643
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3544334418299115e-05,
+      "loss": 0.0313,
+      "step": 1644
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3536903749575248e-05,
+      "loss": 0.0391,
+      "step": 1645
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3529470848052353e-05,
+      "loss": 0.0369,
+      "step": 1646
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3522035718422722e-05,
+      "loss": 0.0239,
+      "step": 1647
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3514598365380051e-05,
+      "loss": 0.0262,
+      "step": 1648
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3507158793619438e-05,
+      "loss": 0.0301,
+      "step": 1649
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3499717007837381e-05,
+      "loss": 0.0236,
+      "step": 1650
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 1.3492273012731785e-05,
+      "loss": 0.0312,
+      "step": 1651
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3484826813001934e-05,
+      "loss": 0.0421,
+      "step": 1652
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3477378413348516e-05,
+      "loss": 0.0262,
+      "step": 1653
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3469927818473607e-05,
+      "loss": 0.0279,
+      "step": 1654
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.346247503308066e-05,
+      "loss": 0.0311,
+      "step": 1655
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3455020061874517e-05,
+      "loss": 0.0378,
+      "step": 1656
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3447562909561405e-05,
+      "loss": 0.0256,
+      "step": 1657
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3440103580848919e-05,
+      "loss": 0.0277,
+      "step": 1658
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.343264208044603e-05,
+      "loss": 0.0212,
+      "step": 1659
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3425178413063082e-05,
+      "loss": 0.0299,
+      "step": 1660
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3417712583411787e-05,
+      "loss": 0.0337,
+      "step": 1661
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3410244596205222e-05,
+      "loss": 0.0365,
+      "step": 1662
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3402774456157825e-05,
+      "loss": 0.0244,
+      "step": 1663
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 1.3395302167985393e-05,
+      "loss": 0.0395,
+      "step": 1664
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3387827736405079e-05,
+      "loss": 0.0261,
+      "step": 1665
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.338035116613539e-05,
+      "loss": 0.0241,
+      "step": 1666
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.337287246189618e-05,
+      "loss": 0.0285,
+      "step": 1667
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.336539162840866e-05,
+      "loss": 0.0225,
+      "step": 1668
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3357908670395371e-05,
+      "loss": 0.0282,
+      "step": 1669
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3350423592580204e-05,
+      "loss": 0.026,
+      "step": 1670
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3342936399688387e-05,
+      "loss": 0.0339,
+      "step": 1671
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3335447096446481e-05,
+      "loss": 0.0323,
+      "step": 1672
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3327955687582389e-05,
+      "loss": 0.0329,
+      "step": 1673
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3320462177825321e-05,
+      "loss": 0.0244,
+      "step": 1674
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3312966571905836e-05,
+      "loss": 0.0252,
+      "step": 1675
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3305468874555811e-05,
+      "loss": 0.0279,
+      "step": 1676
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3297969090508434e-05,
+      "loss": 0.0278,
+      "step": 1677
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 1.3290467224498214e-05,
+      "loss": 0.025,
+      "step": 1678
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3282963281260985e-05,
+      "loss": 0.0307,
+      "step": 1679
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3275457265533876e-05,
+      "loss": 0.0391,
+      "step": 1680
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3267949182055335e-05,
+      "loss": 0.0369,
+      "step": 1681
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3260439035565114e-05,
+      "loss": 0.0315,
+      "step": 1682
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3252926830804264e-05,
+      "loss": 0.0214,
+      "step": 1683
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3245412572515137e-05,
+      "loss": 0.026,
+      "step": 1684
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3237896265441386e-05,
+      "loss": 0.0366,
+      "step": 1685
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.323037791432795e-05,
+      "loss": 0.0347,
+      "step": 1686
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3222857523921058e-05,
+      "loss": 0.021,
+      "step": 1687
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3215335098968234e-05,
+      "loss": 0.0257,
+      "step": 1688
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3207810644218277e-05,
+      "loss": 0.022,
+      "step": 1689
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.320028416442128e-05,
+      "loss": 0.0416,
+      "step": 1690
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 1.3192755664328605e-05,
+      "loss": 0.0297,
+      "step": 1691
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3185225148692884e-05,
+      "loss": 0.038,
+      "step": 1692
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3177692622268039e-05,
+      "loss": 0.0351,
+      "step": 1693
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3170158089809246e-05,
+      "loss": 0.0272,
+      "step": 1694
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.316262155607295e-05,
+      "loss": 0.0298,
+      "step": 1695
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3155083025816864e-05,
+      "loss": 0.0344,
+      "step": 1696
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3147542503799963e-05,
+      "loss": 0.0335,
+      "step": 1697
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3139999994782468e-05,
+      "loss": 0.0289,
+      "step": 1698
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.313245550352587e-05,
+      "loss": 0.0329,
+      "step": 1699
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3124909034792896e-05,
+      "loss": 0.0255,
+      "step": 1700
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3117360593347535e-05,
+      "loss": 0.0244,
+      "step": 1701
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3109810183955012e-05,
+      "loss": 0.026,
+      "step": 1702
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3102257811381795e-05,
+      "loss": 0.0198,
+      "step": 1703
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3094703480395597e-05,
+      "loss": 0.0272,
+      "step": 1704
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 1.3087147195765361e-05,
+      "loss": 0.0324,
+      "step": 1705
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3079588962261271e-05,
+      "loss": 0.0274,
+      "step": 1706
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3072028784654732e-05,
+      "loss": 0.0401,
+      "step": 1707
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.306446666771838e-05,
+      "loss": 0.0237,
+      "step": 1708
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3056902616226084e-05,
+      "loss": 0.0283,
+      "step": 1709
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3049336634952918e-05,
+      "loss": 0.0257,
+      "step": 1710
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3041768728675189e-05,
+      "loss": 0.0239,
+      "step": 1711
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3034198902170407e-05,
+      "loss": 0.032,
+      "step": 1712
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3026627160217302e-05,
+      "loss": 0.0349,
+      "step": 1713
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3019053507595814e-05,
+      "loss": 0.0289,
+      "step": 1714
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3011477949087083e-05,
+      "loss": 0.0317,
+      "step": 1715
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.3003900489473455e-05,
+      "loss": 0.0237,
+      "step": 1716
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2996321133538482e-05,
+      "loss": 0.0348,
+      "step": 1717
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2988739886066904e-05,
+      "loss": 0.0276,
+      "step": 1718
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 1.2981156751844659e-05,
+      "loss": 0.026,
+      "step": 1719
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2973571735658878e-05,
+      "loss": 0.0271,
+      "step": 1720
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2965984842297876e-05,
+      "loss": 0.032,
+      "step": 1721
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2958396076551157e-05,
+      "loss": 0.0259,
+      "step": 1722
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.29508054432094e-05,
+      "loss": 0.0331,
+      "step": 1723
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2943212947064477e-05,
+      "loss": 0.0288,
+      "step": 1724
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2935618592909419e-05,
+      "loss": 0.0337,
+      "step": 1725
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2928022385538441e-05,
+      "loss": 0.0342,
+      "step": 1726
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2920424329746928e-05,
+      "loss": 0.0324,
+      "step": 1727
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2912824430331425e-05,
+      "loss": 0.0249,
+      "step": 1728
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2905222692089647e-05,
+      "loss": 0.033,
+      "step": 1729
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2897619119820466e-05,
+      "loss": 0.0266,
+      "step": 1730
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2890013718323913e-05,
+      "loss": 0.0234,
+      "step": 1731
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 1.2882406492401174e-05,
+      "loss": 0.0314,
+      "step": 1732
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.287479744685459e-05,
+      "loss": 0.0272,
+      "step": 1733
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2867186586487642e-05,
+      "loss": 0.0265,
+      "step": 1734
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2859573916104966e-05,
+      "loss": 0.0235,
+      "step": 1735
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2851959440512338e-05,
+      "loss": 0.0312,
+      "step": 1736
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2844343164516671e-05,
+      "loss": 0.0312,
+      "step": 1737
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2836725092926013e-05,
+      "loss": 0.0255,
+      "step": 1738
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2829105230549552e-05,
+      "loss": 0.0237,
+      "step": 1739
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2821483582197604e-05,
+      "loss": 0.0271,
+      "step": 1740
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2813860152681607e-05,
+      "loss": 0.0309,
+      "step": 1741
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.280623494681413e-05,
+      "loss": 0.0296,
+      "step": 1742
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2798607969408865e-05,
+      "loss": 0.0272,
+      "step": 1743
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2790979225280614e-05,
+      "loss": 0.0348,
+      "step": 1744
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2783348719245301e-05,
+      "loss": 0.0267,
+      "step": 1745
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 1.2775716456119962e-05,
+      "loss": 0.0297,
+      "step": 1746
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2768082440722735e-05,
+      "loss": 0.0334,
+      "step": 1747
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2760446677872873e-05,
+      "loss": 0.0256,
+      "step": 1748
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.275280917239073e-05,
+      "loss": 0.027,
+      "step": 1749
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2745169929097756e-05,
+      "loss": 0.0315,
+      "step": 1750
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2737528952816503e-05,
+      "loss": 0.0291,
+      "step": 1751
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2729886248370616e-05,
+      "loss": 0.0352,
+      "step": 1752
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2722241820584827e-05,
+      "loss": 0.0262,
+      "step": 1753
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.271459567428496e-05,
+      "loss": 0.0316,
+      "step": 1754
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2706947814297923e-05,
+      "loss": 0.0251,
+      "step": 1755
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2699298245451705e-05,
+      "loss": 0.0367,
+      "step": 1756
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2691646972575376e-05,
+      "loss": 0.0272,
+      "step": 1757
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2683994000499078e-05,
+      "loss": 0.0293,
+      "step": 1758
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 1.2676339334054029e-05,
+      "loss": 0.0269,
+      "step": 1759
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2668682978072512e-05,
+      "loss": 0.0292,
+      "step": 1760
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2661024937387888e-05,
+      "loss": 0.033,
+      "step": 1761
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2653365216834567e-05,
+      "loss": 0.0227,
+      "step": 1762
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2645703821248026e-05,
+      "loss": 0.0216,
+      "step": 1763
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2638040755464802e-05,
+      "loss": 0.0217,
+      "step": 1764
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2630376024322481e-05,
+      "loss": 0.0323,
+      "step": 1765
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2622709632659706e-05,
+      "loss": 0.0222,
+      "step": 1766
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2615041585316163e-05,
+      "loss": 0.0277,
+      "step": 1767
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2607371887132586e-05,
+      "loss": 0.0315,
+      "step": 1768
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2599700542950752e-05,
+      "loss": 0.0244,
+      "step": 1769
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2592027557613476e-05,
+      "loss": 0.026,
+      "step": 1770
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2584352935964608e-05,
+      "loss": 0.0284,
+      "step": 1771
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.257667668284903e-05,
+      "loss": 0.0354,
+      "step": 1772
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 1.2568998803112658e-05,
+      "loss": 0.0295,
+      "step": 1773
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.256131930160243e-05,
+      "loss": 0.0257,
+      "step": 1774
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2553638183166315e-05,
+      "loss": 0.0251,
+      "step": 1775
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2545955452653294e-05,
+      "loss": 0.0369,
+      "step": 1776
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2538271114913372e-05,
+      "loss": 0.0264,
+      "step": 1777
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2530585174797563e-05,
+      "loss": 0.0234,
+      "step": 1778
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2522897637157905e-05,
+      "loss": 0.0321,
+      "step": 1779
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2515208506847424e-05,
+      "loss": 0.029,
+      "step": 1780
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2507517788720172e-05,
+      "loss": 0.0301,
+      "step": 1781
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.249982548763119e-05,
+      "loss": 0.033,
+      "step": 1782
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2492131608436519e-05,
+      "loss": 0.0315,
+      "step": 1783
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2484436155993207e-05,
+      "loss": 0.031,
+      "step": 1784
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2476739135159286e-05,
+      "loss": 0.0344,
+      "step": 1785
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2469040550793777e-05,
+      "loss": 0.022,
+      "step": 1786
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 1.2461340407756692e-05,
+      "loss": 0.03,
+      "step": 1787
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2453638710909033e-05,
+      "loss": 0.0296,
+      "step": 1788
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2445935465112769e-05,
+      "loss": 0.0375,
+      "step": 1789
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2438230675230851e-05,
+      "loss": 0.0402,
+      "step": 1790
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2430524346127215e-05,
+      "loss": 0.0328,
+      "step": 1791
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2422816482666757e-05,
+      "loss": 0.0344,
+      "step": 1792
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2415107089715348e-05,
+      "loss": 0.0307,
+      "step": 1793
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2407396172139822e-05,
+      "loss": 0.0319,
+      "step": 1794
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2399683734807974e-05,
+      "loss": 0.0284,
+      "step": 1795
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2391969782588564e-05,
+      "loss": 0.0311,
+      "step": 1796
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2384254320351301e-05,
+      "loss": 0.0396,
+      "step": 1797
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2376537352966856e-05,
+      "loss": 0.0365,
+      "step": 1798
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2368818885306838e-05,
+      "loss": 0.0443,
+      "step": 1799
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 1.2361098922243812e-05,
+      "loss": 0.0298,
+      "step": 1800
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2353377468651288e-05,
+      "loss": 0.0287,
+      "step": 1801
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2345654529403714e-05,
+      "loss": 0.0291,
+      "step": 1802
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.233793010937648e-05,
+      "loss": 0.0287,
+      "step": 1803
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2330204213445896e-05,
+      "loss": 0.0333,
+      "step": 1804
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2322476846489222e-05,
+      "loss": 0.0391,
+      "step": 1805
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2314748013384639e-05,
+      "loss": 0.0269,
+      "step": 1806
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2307017719011253e-05,
+      "loss": 0.0351,
+      "step": 1807
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2299285968249091e-05,
+      "loss": 0.0264,
+      "step": 1808
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2291552765979104e-05,
+      "loss": 0.0356,
+      "step": 1809
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2283818117083157e-05,
+      "loss": 0.0267,
+      "step": 1810
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2276082026444026e-05,
+      "loss": 0.041,
+      "step": 1811
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2268344498945404e-05,
+      "loss": 0.0273,
+      "step": 1812
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2260605539471884e-05,
+      "loss": 0.0314,
+      "step": 1813
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 1.2252865152908964e-05,
+      "loss": 0.0318,
+      "step": 1814
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2245123344143044e-05,
+      "loss": 0.0271,
+      "step": 1815
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2237380118061423e-05,
+      "loss": 0.0305,
+      "step": 1816
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2229635479552295e-05,
+      "loss": 0.0314,
+      "step": 1817
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2221889433504743e-05,
+      "loss": 0.029,
+      "step": 1818
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2214141984808742e-05,
+      "loss": 0.0216,
+      "step": 1819
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.220639313835515e-05,
+      "loss": 0.0297,
+      "step": 1820
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2198642899035704e-05,
+      "loss": 0.026,
+      "step": 1821
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2190891271743032e-05,
+      "loss": 0.022,
+      "step": 1822
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2183138261370624e-05,
+      "loss": 0.0327,
+      "step": 1823
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2175383872812851e-05,
+      "loss": 0.025,
+      "step": 1824
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2167628110964953e-05,
+      "loss": 0.0276,
+      "step": 1825
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2159870980723037e-05,
+      "loss": 0.0318,
+      "step": 1826
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 1.2152112486984071e-05,
+      "loss": 0.0274,
+      "step": 1827
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.214435263464589e-05,
+      "loss": 0.0295,
+      "step": 1828
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2136591428607181e-05,
+      "loss": 0.0295,
+      "step": 1829
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2128828873767487e-05,
+      "loss": 0.0201,
+      "step": 1830
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2121064975027203e-05,
+      "loss": 0.0272,
+      "step": 1831
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2113299737287571e-05,
+      "loss": 0.0332,
+      "step": 1832
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.210553316545068e-05,
+      "loss": 0.0296,
+      "step": 1833
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2097765264419463e-05,
+      "loss": 0.0359,
+      "step": 1834
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.208999603909769e-05,
+      "loss": 0.035,
+      "step": 1835
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2082225494389961e-05,
+      "loss": 0.0278,
+      "step": 1836
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2074453635201722e-05,
+      "loss": 0.0273,
+      "step": 1837
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2066680466439236e-05,
+      "loss": 0.03,
+      "step": 1838
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2058905993009604e-05,
+      "loss": 0.026,
+      "step": 1839
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2051130219820741e-05,
+      "loss": 0.0311,
+      "step": 1840
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 1.2043353151781384e-05,
+      "loss": 0.0341,
+      "step": 1841
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2035574793801095e-05,
+      "loss": 0.029,
+      "step": 1842
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2027795150790242e-05,
+      "loss": 0.0342,
+      "step": 1843
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2020014227660008e-05,
+      "loss": 0.0256,
+      "step": 1844
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2012232029322384e-05,
+      "loss": 0.033,
+      "step": 1845
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.2004448560690167e-05,
+      "loss": 0.0347,
+      "step": 1846
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.199666382667695e-05,
+      "loss": 0.0246,
+      "step": 1847
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1988877832197135e-05,
+      "loss": 0.0292,
+      "step": 1848
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1981090582165911e-05,
+      "loss": 0.0219,
+      "step": 1849
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1973302081499259e-05,
+      "loss": 0.0316,
+      "step": 1850
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1965512335113958e-05,
+      "loss": 0.023,
+      "step": 1851
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1957721347927568e-05,
+      "loss": 0.0238,
+      "step": 1852
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.194992912485843e-05,
+      "loss": 0.0368,
+      "step": 1853
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1942135670825672e-05,
+      "loss": 0.0295,
+      "step": 1854
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 1.1934340990749193e-05,
+      "loss": 0.024,
+      "step": 1855
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1926545089549666e-05,
+      "loss": 0.0267,
+      "step": 1856
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1918747972148541e-05,
+      "loss": 0.0235,
+      "step": 1857
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.191094964346803e-05,
+      "loss": 0.0401,
+      "step": 1858
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.190315010843111e-05,
+      "loss": 0.0314,
+      "step": 1859
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.189534937196152e-05,
+      "loss": 0.0221,
+      "step": 1860
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1887547438983764e-05,
+      "loss": 0.0314,
+      "step": 1861
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.187974431442309e-05,
+      "loss": 0.0298,
+      "step": 1862
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1871940003205505e-05,
+      "loss": 0.0307,
+      "step": 1863
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1864134510257767e-05,
+      "loss": 0.0223,
+      "step": 1864
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1856327840507377e-05,
+      "loss": 0.0231,
+      "step": 1865
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1848519998882572e-05,
+      "loss": 0.0295,
+      "step": 1866
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1840710990312347e-05,
+      "loss": 0.0345,
+      "step": 1867
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 1.1832900819726412e-05,
+      "loss": 0.0282,
+      "step": 1868
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1825089492055227e-05,
+      "loss": 0.0295,
+      "step": 1869
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1817277012229973e-05,
+      "loss": 0.0331,
+      "step": 1870
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1809463385182568e-05,
+      "loss": 0.0191,
+      "step": 1871
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1801648615845638e-05,
+      "loss": 0.0288,
+      "step": 1872
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1793832709152548e-05,
+      "loss": 0.0215,
+      "step": 1873
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1786015670037374e-05,
+      "loss": 0.0374,
+      "step": 1874
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.17781975034349e-05,
+      "loss": 0.0358,
+      "step": 1875
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1770378214280631e-05,
+      "loss": 0.0315,
+      "step": 1876
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.176255780751078e-05,
+      "loss": 0.027,
+      "step": 1877
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1754736288062256e-05,
+      "loss": 0.0307,
+      "step": 1878
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1746913660872681e-05,
+      "loss": 0.0231,
+      "step": 1879
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1739089930880377e-05,
+      "loss": 0.0202,
+      "step": 1880
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1731265103024351e-05,
+      "loss": 0.0208,
+      "step": 1881
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 1.1723439182244316e-05,
+      "loss": 0.0316,
+      "step": 1882
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1715612173480668e-05,
+      "loss": 0.0322,
+      "step": 1883
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1707784081674482e-05,
+      "loss": 0.0227,
+      "step": 1884
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1699954911767537e-05,
+      "loss": 0.0274,
+      "step": 1885
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1692124668702276e-05,
+      "loss": 0.0332,
+      "step": 1886
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1684293357421824e-05,
+      "loss": 0.0255,
+      "step": 1887
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1676460982869977e-05,
+      "loss": 0.0305,
+      "step": 1888
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1668627549991216e-05,
+      "loss": 0.0248,
+      "step": 1889
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1660793063730674e-05,
+      "loss": 0.0332,
+      "step": 1890
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1652957529034155e-05,
+      "loss": 0.0297,
+      "step": 1891
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1645120950848127e-05,
+      "loss": 0.0274,
+      "step": 1892
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1637283334119713e-05,
+      "loss": 0.0247,
+      "step": 1893
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1629444683796694e-05,
+      "loss": 0.0208,
+      "step": 1894
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 1.1621605004827503e-05,
+      "loss": 0.0247,
+      "step": 1895
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1613764302161222e-05,
+      "loss": 0.0377,
+      "step": 1896
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1605922580747578e-05,
+      "loss": 0.0255,
+      "step": 1897
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1598079845536947e-05,
+      "loss": 0.0238,
+      "step": 1898
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1590236101480339e-05,
+      "loss": 0.0265,
+      "step": 1899
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1582391353529399e-05,
+      "loss": 0.0275,
+      "step": 1900
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1574545606636413e-05,
+      "loss": 0.0264,
+      "step": 1901
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1566698865754291e-05,
+      "loss": 0.0233,
+      "step": 1902
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1558851135836573e-05,
+      "loss": 0.027,
+      "step": 1903
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.155100242183743e-05,
+      "loss": 0.0281,
+      "step": 1904
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.154315272871164e-05,
+      "loss": 0.0264,
+      "step": 1905
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1535302061414613e-05,
+      "loss": 0.027,
+      "step": 1906
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1527450424902364e-05,
+      "loss": 0.0351,
+      "step": 1907
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1519597824131527e-05,
+      "loss": 0.0282,
+      "step": 1908
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 1.1511744264059339e-05,
+      "loss": 0.029,
+      "step": 1909
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1503889749643649e-05,
+      "loss": 0.0283,
+      "step": 1910
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1496034285842897e-05,
+      "loss": 0.032,
+      "step": 1911
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1488177877616136e-05,
+      "loss": 0.0362,
+      "step": 1912
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.148032052992301e-05,
+      "loss": 0.0284,
+      "step": 1913
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1472462247723752e-05,
+      "loss": 0.0298,
+      "step": 1914
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1464603035979192e-05,
+      "loss": 0.0256,
+      "step": 1915
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1456742899650739e-05,
+      "loss": 0.0304,
+      "step": 1916
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1448881843700392e-05,
+      "loss": 0.0316,
+      "step": 1917
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1441019873090726e-05,
+      "loss": 0.0382,
+      "step": 1918
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1433156992784899e-05,
+      "loss": 0.024,
+      "step": 1919
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1425293207746638e-05,
+      "loss": 0.0278,
+      "step": 1920
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.1417428522940242e-05,
+      "loss": 0.0232,
+      "step": 1921
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 1.140956294333058e-05,
+      "loss": 0.0243,
+      "step": 1922
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1401696473883086e-05,
+      "loss": 0.024,
+      "step": 1923
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1393829119563752e-05,
+      "loss": 0.0244,
+      "step": 1924
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1385960885339138e-05,
+      "loss": 0.0316,
+      "step": 1925
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1378091776176348e-05,
+      "loss": 0.0302,
+      "step": 1926
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1370221797043039e-05,
+      "loss": 0.0276,
+      "step": 1927
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.136235095290743e-05,
+      "loss": 0.0244,
+      "step": 1928
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1354479248738271e-05,
+      "loss": 0.028,
+      "step": 1929
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1346606689504862e-05,
+      "loss": 0.0286,
+      "step": 1930
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1338733280177046e-05,
+      "loss": 0.0264,
+      "step": 1931
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1330859025725193e-05,
+      "loss": 0.0205,
+      "step": 1932
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1322983931120217e-05,
+      "loss": 0.0216,
+      "step": 1933
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1315108001333556e-05,
+      "loss": 0.0279,
+      "step": 1934
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.130723124133718e-05,
+      "loss": 0.0413,
+      "step": 1935
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 1.1299353656103573e-05,
+      "loss": 0.0324,
+      "step": 1936
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.129147525060575e-05,
+      "loss": 0.0351,
+      "step": 1937
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1283596029817248e-05,
+      "loss": 0.0274,
+      "step": 1938
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1275715998712098e-05,
+      "loss": 0.0283,
+      "step": 1939
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1267835162264867e-05,
+      "loss": 0.0319,
+      "step": 1940
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1259953525450616e-05,
+      "loss": 0.0299,
+      "step": 1941
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1252071093244915e-05,
+      "loss": 0.0222,
+      "step": 1942
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1244187870623832e-05,
+      "loss": 0.028,
+      "step": 1943
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1236303862563945e-05,
+      "loss": 0.0369,
+      "step": 1944
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1228419074042312e-05,
+      "loss": 0.0306,
+      "step": 1945
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1220533510036497e-05,
+      "loss": 0.0273,
+      "step": 1946
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1212647175524551e-05,
+      "loss": 0.0294,
+      "step": 1947
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1204760075485007e-05,
+      "loss": 0.0305,
+      "step": 1948
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1196872214896881e-05,
+      "loss": 0.0316,
+      "step": 1949
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 1.1188983598739675e-05,
+      "loss": 0.0331,
+      "step": 1950
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1181094231993365e-05,
+      "loss": 0.0337,
+      "step": 1951
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1173204119638399e-05,
+      "loss": 0.0288,
+      "step": 1952
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1165313266655698e-05,
+      "loss": 0.0256,
+      "step": 1953
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1157421678026645e-05,
+      "loss": 0.0268,
+      "step": 1954
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1149529358733098e-05,
+      "loss": 0.0221,
+      "step": 1955
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1141636313757369e-05,
+      "loss": 0.03,
+      "step": 1956
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.113374254808223e-05,
+      "loss": 0.0254,
+      "step": 1957
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1125848066690903e-05,
+      "loss": 0.0252,
+      "step": 1958
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1117952874567073e-05,
+      "loss": 0.0254,
+      "step": 1959
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1110056976694862e-05,
+      "loss": 0.0266,
+      "step": 1960
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1102160378058844e-05,
+      "loss": 0.0223,
+      "step": 1961
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1094263083644036e-05,
+      "loss": 0.0228,
+      "step": 1962
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 1.1086365098435888e-05,
+      "loss": 0.0267,
+      "step": 1963
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1078466427420292e-05,
+      "loss": 0.0351,
+      "step": 1964
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1070567075583572e-05,
+      "loss": 0.028,
+      "step": 1965
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.106266704791248e-05,
+      "loss": 0.0358,
+      "step": 1966
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1054766349394198e-05,
+      "loss": 0.0209,
+      "step": 1967
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1046864985016326e-05,
+      "loss": 0.0277,
+      "step": 1968
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1038962959766883e-05,
+      "loss": 0.0257,
+      "step": 1969
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.103106027863432e-05,
+      "loss": 0.0218,
+      "step": 1970
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1023156946607485e-05,
+      "loss": 0.0276,
+      "step": 1971
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1015252968675645e-05,
+      "loss": 0.032,
+      "step": 1972
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.1007348349828473e-05,
+      "loss": 0.0224,
+      "step": 1973
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0999443095056051e-05,
+      "loss": 0.0272,
+      "step": 1974
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.099153720934885e-05,
+      "loss": 0.0293,
+      "step": 1975
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0983630697697755e-05,
+      "loss": 0.024,
+      "step": 1976
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 1.0975723565094036e-05,
+      "loss": 0.0261,
+      "step": 1977
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0967815816529358e-05,
+      "loss": 0.0316,
+      "step": 1978
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.095990745699577e-05,
+      "loss": 0.0281,
+      "step": 1979
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0951998491485722e-05,
+      "loss": 0.0255,
+      "step": 1980
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0944088924992026e-05,
+      "loss": 0.0254,
+      "step": 1981
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0936178762507887e-05,
+      "loss": 0.028,
+      "step": 1982
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0928268009026885e-05,
+      "loss": 0.0191,
+      "step": 1983
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.092035666954297e-05,
+      "loss": 0.0296,
+      "step": 1984
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0912444749050457e-05,
+      "loss": 0.0227,
+      "step": 1985
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.090453225254404e-05,
+      "loss": 0.0265,
+      "step": 1986
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0896619185018764e-05,
+      "loss": 0.0238,
+      "step": 1987
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0888705551470045e-05,
+      "loss": 0.024,
+      "step": 1988
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0880791356893652e-05,
+      "loss": 0.0269,
+      "step": 1989
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 1.0872876606285702e-05,
+      "loss": 0.03,
+      "step": 1990
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0864961304642676e-05,
+      "loss": 0.0306,
+      "step": 1991
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0857045456961394e-05,
+      "loss": 0.0399,
+      "step": 1992
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.084912906823902e-05,
+      "loss": 0.0266,
+      "step": 1993
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0841212143473065e-05,
+      "loss": 0.0225,
+      "step": 1994
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0833294687661376e-05,
+      "loss": 0.0286,
+      "step": 1995
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0825376705802129e-05,
+      "loss": 0.0345,
+      "step": 1996
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.081745820289384e-05,
+      "loss": 0.027,
+      "step": 1997
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0809539183935358e-05,
+      "loss": 0.029,
+      "step": 1998
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0801619653925843e-05,
+      "loss": 0.0248,
+      "step": 1999
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.079369961786479e-05,
+      "loss": 0.03,
+      "step": 2000
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0785779080752012e-05,
+      "loss": 0.0315,
+      "step": 2001
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0777858047587635e-05,
+      "loss": 0.0314,
+      "step": 2002
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0769936523372094e-05,
+      "loss": 0.0229,
+      "step": 2003
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 1.0762014513106143e-05,
+      "loss": 0.022,
+      "step": 2004
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0754092021790838e-05,
+      "loss": 0.0319,
+      "step": 2005
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0746169054427543e-05,
+      "loss": 0.0328,
+      "step": 2006
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.073824561601791e-05,
+      "loss": 0.0238,
+      "step": 2007
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0730321711563909e-05,
+      "loss": 0.0296,
+      "step": 2008
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0722397346067785e-05,
+      "loss": 0.0229,
+      "step": 2009
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0714472524532085e-05,
+      "loss": 0.0232,
+      "step": 2010
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.070654725195964e-05,
+      "loss": 0.0243,
+      "step": 2011
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0698621533353562e-05,
+      "loss": 0.023,
+      "step": 2012
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0690695373717254e-05,
+      "loss": 0.0293,
+      "step": 2013
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0682768778054394e-05,
+      "loss": 0.0331,
+      "step": 2014
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0674841751368925e-05,
+      "loss": 0.0242,
+      "step": 2015
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0666914298665079e-05,
+      "loss": 0.0343,
+      "step": 2016
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0658986424947347e-05,
+      "loss": 0.027,
+      "step": 2017
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 1.0651058135220484e-05,
+      "loss": 0.0265,
+      "step": 2018
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0643129434489514e-05,
+      "loss": 0.0234,
+      "step": 2019
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0635200327759719e-05,
+      "loss": 0.0239,
+      "step": 2020
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.062727082003663e-05,
+      "loss": 0.0231,
+      "step": 2021
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0619340916326039e-05,
+      "loss": 0.0334,
+      "step": 2022
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0611410621633992e-05,
+      "loss": 0.0222,
+      "step": 2023
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0603479940966769e-05,
+      "loss": 0.0283,
+      "step": 2024
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.05955488793309e-05,
+      "loss": 0.0241,
+      "step": 2025
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.058761744173316e-05,
+      "loss": 0.0175,
+      "step": 2026
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0579685633180556e-05,
+      "loss": 0.0247,
+      "step": 2027
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0571753458680329e-05,
+      "loss": 0.0319,
+      "step": 2028
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0563820923239956e-05,
+      "loss": 0.0404,
+      "step": 2029
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0555888031867132e-05,
+      "loss": 0.0298,
+      "step": 2030
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 1.0547954789569785e-05,
+      "loss": 0.0232,
+      "step": 2031
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0540021201356067e-05,
+      "loss": 0.023,
+      "step": 2032
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0532087272234341e-05,
+      "loss": 0.0295,
+      "step": 2033
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0524153007213185e-05,
+      "loss": 0.0324,
+      "step": 2034
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.05162184113014e-05,
+      "loss": 0.0209,
+      "step": 2035
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.050828348950798e-05,
+      "loss": 0.0339,
+      "step": 2036
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0500348246842136e-05,
+      "loss": 0.0389,
+      "step": 2037
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0492412688313279e-05,
+      "loss": 0.0308,
+      "step": 2038
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0484476818931015e-05,
+      "loss": 0.0277,
+      "step": 2039
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0476540643705153e-05,
+      "loss": 0.026,
+      "step": 2040
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0468604167645695e-05,
+      "loss": 0.0187,
+      "step": 2041
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0460667395762825e-05,
+      "loss": 0.0301,
+      "step": 2042
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0452730333066919e-05,
+      "loss": 0.0292,
+      "step": 2043
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.044479298456854e-05,
+      "loss": 0.0235,
+      "step": 2044
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 1.0436855355278426e-05,
+      "loss": 0.0172,
+      "step": 2045
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0428917450207489e-05,
+      "loss": 0.0234,
+      "step": 2046
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.042097927436683e-05,
+      "loss": 0.0291,
+      "step": 2047
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0413040832767702e-05,
+      "loss": 0.0278,
+      "step": 2048
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0405102130421536e-05,
+      "loss": 0.0243,
+      "step": 2049
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0397163172339932e-05,
+      "loss": 0.0307,
+      "step": 2050
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.038922396353464e-05,
+      "loss": 0.0345,
+      "step": 2051
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0381284509017578e-05,
+      "loss": 0.0373,
+      "step": 2052
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0373344813800814e-05,
+      "loss": 0.0262,
+      "step": 2053
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0365404882896568e-05,
+      "loss": 0.0247,
+      "step": 2054
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.035746472131721e-05,
+      "loss": 0.0246,
+      "step": 2055
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0349524334075256e-05,
+      "loss": 0.0228,
+      "step": 2056
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.0341583726183365e-05,
+      "loss": 0.0308,
+      "step": 2057
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 1.033364290265433e-05,
+      "loss": 0.029,
+      "step": 2058
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0325701868501092e-05,
+      "loss": 0.0294,
+      "step": 2059
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.031776062873671e-05,
+      "loss": 0.0341,
+      "step": 2060
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0309819188374386e-05,
+      "loss": 0.0257,
+      "step": 2061
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.030187755242744e-05,
+      "loss": 0.0265,
+      "step": 2062
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0293935725909318e-05,
+      "loss": 0.0214,
+      "step": 2063
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0285993713833586e-05,
+      "loss": 0.0286,
+      "step": 2064
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0278051521213931e-05,
+      "loss": 0.0226,
+      "step": 2065
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0270109153064147e-05,
+      "loss": 0.0331,
+      "step": 2066
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0262166614398144e-05,
+      "loss": 0.0236,
+      "step": 2067
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0254223910229939e-05,
+      "loss": 0.0257,
+      "step": 2068
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0246281045573654e-05,
+      "loss": 0.0302,
+      "step": 2069
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.0238338025443507e-05,
+      "loss": 0.0246,
+      "step": 2070
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.023039485485382e-05,
+      "loss": 0.0233,
+      "step": 2071
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 1.022245153881901e-05,
+      "loss": 0.0225,
+      "step": 2072
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0214508082353578e-05,
+      "loss": 0.0221,
+      "step": 2073
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0206564490472124e-05,
+      "loss": 0.0183,
+      "step": 2074
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.019862076818933e-05,
+      "loss": 0.0235,
+      "step": 2075
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.019067692051996e-05,
+      "loss": 0.0261,
+      "step": 2076
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0182732952478851e-05,
+      "loss": 0.0305,
+      "step": 2077
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0174788869080929e-05,
+      "loss": 0.0263,
+      "step": 2078
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.016684467534118e-05,
+      "loss": 0.0367,
+      "step": 2079
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0158900376274665e-05,
+      "loss": 0.0228,
+      "step": 2080
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0150955976896515e-05,
+      "loss": 0.026,
+      "step": 2081
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0143011482221916e-05,
+      "loss": 0.0266,
+      "step": 2082
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0135066897266124e-05,
+      "loss": 0.022,
+      "step": 2083
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0127122227044444e-05,
+      "loss": 0.0212,
+      "step": 2084
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0119177476572237e-05,
+      "loss": 0.0267,
+      "step": 2085
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 1.0111232650864918e-05,
+      "loss": 0.0339,
+      "step": 2086
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0103287754937946e-05,
+      "loss": 0.0279,
+      "step": 2087
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0095342793806828e-05,
+      "loss": 0.0294,
+      "step": 2088
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0087397772487102e-05,
+      "loss": 0.0365,
+      "step": 2089
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0079452695994358e-05,
+      "loss": 0.0212,
+      "step": 2090
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0071507569344213e-05,
+      "loss": 0.0278,
+      "step": 2091
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0063562397552314e-05,
+      "loss": 0.0195,
+      "step": 2092
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0055617185634344e-05,
+      "loss": 0.0351,
+      "step": 2093
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0047671938606002e-05,
+      "loss": 0.0288,
+      "step": 2094
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0039726661483019e-05,
+      "loss": 0.0261,
+      "step": 2095
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0031781359281135e-05,
+      "loss": 0.0241,
+      "step": 2096
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0023836037016115e-05,
+      "loss": 0.0305,
+      "step": 2097
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0015890699703727e-05,
+      "loss": 0.0245,
+      "step": 2098
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 1.0007945352359763e-05,
+      "loss": 0.0254,
+      "step": 2099
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 1e-05,
+      "loss": 0.0208,
+      "step": 2100
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.992054647640242e-06,
+      "loss": 0.0292,
+      "step": 2101
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.984109300296271e-06,
+      "loss": 0.0311,
+      "step": 2102
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.976163962983889e-06,
+      "loss": 0.0308,
+      "step": 2103
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.968218640718867e-06,
+      "loss": 0.0204,
+      "step": 2104
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.960273338516983e-06,
+      "loss": 0.0274,
+      "step": 2105
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.952328061394001e-06,
+      "loss": 0.0255,
+      "step": 2106
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.944382814365661e-06,
+      "loss": 0.0277,
+      "step": 2107
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.936437602447687e-06,
+      "loss": 0.0289,
+      "step": 2108
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.92849243065579e-06,
+      "loss": 0.0371,
+      "step": 2109
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.920547304005645e-06,
+      "loss": 0.0234,
+      "step": 2110
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.912602227512898e-06,
+      "loss": 0.0258,
+      "step": 2111
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.904657206193175e-06,
+      "loss": 0.0261,
+      "step": 2112
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 9.896712245062055e-06,
+      "loss": 0.0215,
+      "step": 2113
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.888767349135084e-06,
+      "loss": 0.0343,
+      "step": 2114
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.880822523427766e-06,
+      "loss": 0.0206,
+      "step": 2115
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.872877772955561e-06,
+      "loss": 0.0279,
+      "step": 2116
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.864933102733878e-06,
+      "loss": 0.0281,
+      "step": 2117
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.856988517778086e-06,
+      "loss": 0.0358,
+      "step": 2118
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.84904402310349e-06,
+      "loss": 0.027,
+      "step": 2119
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.841099623725335e-06,
+      "loss": 0.0251,
+      "step": 2120
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.833155324658823e-06,
+      "loss": 0.0283,
+      "step": 2121
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.825211130919075e-06,
+      "loss": 0.0231,
+      "step": 2122
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.81726704752115e-06,
+      "loss": 0.0232,
+      "step": 2123
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.809323079480043e-06,
+      "loss": 0.0277,
+      "step": 2124
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.801379231810673e-06,
+      "loss": 0.0249,
+      "step": 2125
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 9.793435509527876e-06,
+      "loss": 0.0253,
+      "step": 2126
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.785491917646425e-06,
+      "loss": 0.0253,
+      "step": 2127
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.777548461180997e-06,
+      "loss": 0.0216,
+      "step": 2128
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.769605145146184e-06,
+      "loss": 0.0301,
+      "step": 2129
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.761661974556495e-06,
+      "loss": 0.0244,
+      "step": 2130
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.75371895442635e-06,
+      "loss": 0.0292,
+      "step": 2131
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.745776089770061e-06,
+      "loss": 0.0233,
+      "step": 2132
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.737833385601858e-06,
+      "loss": 0.0245,
+      "step": 2133
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.729890846935856e-06,
+      "loss": 0.0278,
+      "step": 2134
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.72194847878607e-06,
+      "loss": 0.0336,
+      "step": 2135
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.714006286166416e-06,
+      "loss": 0.0261,
+      "step": 2136
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.706064274090687e-06,
+      "loss": 0.0317,
+      "step": 2137
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.698122447572563e-06,
+      "loss": 0.0254,
+      "step": 2138
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.690180811625618e-06,
+      "loss": 0.0247,
+      "step": 2139
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 9.682239371263293e-06,
+      "loss": 0.0334,
+      "step": 2140
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.67429813149891e-06,
+      "loss": 0.0364,
+      "step": 2141
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.666357097345672e-06,
+      "loss": 0.0282,
+      "step": 2142
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.65841627381664e-06,
+      "loss": 0.0217,
+      "step": 2143
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.650475665924747e-06,
+      "loss": 0.0238,
+      "step": 2144
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.642535278682795e-06,
+      "loss": 0.0229,
+      "step": 2145
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.634595117103437e-06,
+      "loss": 0.0297,
+      "step": 2146
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.62665518619919e-06,
+      "loss": 0.0271,
+      "step": 2147
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.618715490982425e-06,
+      "loss": 0.0276,
+      "step": 2148
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.610776036465364e-06,
+      "loss": 0.0244,
+      "step": 2149
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.60283682766007e-06,
+      "loss": 0.0189,
+      "step": 2150
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.594897869578466e-06,
+      "loss": 0.0236,
+      "step": 2151
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.586959167232304e-06,
+      "loss": 0.0199,
+      "step": 2152
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.579020725633172e-06,
+      "loss": 0.0266,
+      "step": 2153
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 9.571082549792513e-06,
+      "loss": 0.0329,
+      "step": 2154
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.56314464472158e-06,
+      "loss": 0.0223,
+      "step": 2155
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.555207015431462e-06,
+      "loss": 0.024,
+      "step": 2156
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.547269666933085e-06,
+      "loss": 0.0289,
+      "step": 2157
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.53933260423718e-06,
+      "loss": 0.0325,
+      "step": 2158
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.531395832354308e-06,
+      "loss": 0.0289,
+      "step": 2159
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.523459356294849e-06,
+      "loss": 0.0211,
+      "step": 2160
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.51552318106899e-06,
+      "loss": 0.0169,
+      "step": 2161
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.507587311686725e-06,
+      "loss": 0.0298,
+      "step": 2162
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.499651753157869e-06,
+      "loss": 0.0271,
+      "step": 2163
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.491716510492025e-06,
+      "loss": 0.0335,
+      "step": 2164
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.483781588698603e-06,
+      "loss": 0.0313,
+      "step": 2165
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.475846992786817e-06,
+      "loss": 0.0306,
+      "step": 2166
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 9.46791272776566e-06,
+      "loss": 0.0255,
+      "step": 2167
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.459978798643934e-06,
+      "loss": 0.0247,
+      "step": 2168
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.452045210430218e-06,
+      "loss": 0.0259,
+      "step": 2169
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.444111968132868e-06,
+      "loss": 0.0266,
+      "step": 2170
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.436179076760047e-06,
+      "loss": 0.0197,
+      "step": 2171
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.428246541319673e-06,
+      "loss": 0.0244,
+      "step": 2172
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.420314366819446e-06,
+      "loss": 0.0279,
+      "step": 2173
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.412382558266841e-06,
+      "loss": 0.0243,
+      "step": 2174
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.404451120669102e-06,
+      "loss": 0.0239,
+      "step": 2175
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.396520059033233e-06,
+      "loss": 0.0294,
+      "step": 2176
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.388589378366011e-06,
+      "loss": 0.0331,
+      "step": 2177
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.380659083673963e-06,
+      "loss": 0.0253,
+      "step": 2178
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.37272917996337e-06,
+      "loss": 0.0233,
+      "step": 2179
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.364799672240283e-06,
+      "loss": 0.0259,
+      "step": 2180
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 9.356870565510488e-06,
+      "loss": 0.0245,
+      "step": 2181
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.348941864779517e-06,
+      "loss": 0.0367,
+      "step": 2182
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.341013575052656e-06,
+      "loss": 0.0365,
+      "step": 2183
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.333085701334925e-06,
+      "loss": 0.0311,
+      "step": 2184
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.325158248631075e-06,
+      "loss": 0.0276,
+      "step": 2185
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.317231221945611e-06,
+      "loss": 0.0274,
+      "step": 2186
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.30930462628275e-06,
+      "loss": 0.0221,
+      "step": 2187
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.301378466646438e-06,
+      "loss": 0.0205,
+      "step": 2188
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.293452748040363e-06,
+      "loss": 0.0234,
+      "step": 2189
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.285527475467918e-06,
+      "loss": 0.0363,
+      "step": 2190
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.277602653932217e-06,
+      "loss": 0.0242,
+      "step": 2191
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.269678288436094e-06,
+      "loss": 0.0289,
+      "step": 2192
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.261754383982093e-06,
+      "loss": 0.0269,
+      "step": 2193
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 9.253830945572459e-06,
+      "loss": 0.0263,
+      "step": 2194
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.245907978209164e-06,
+      "loss": 0.0271,
+      "step": 2195
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.23798548689386e-06,
+      "loss": 0.0299,
+      "step": 2196
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.230063476627908e-06,
+      "loss": 0.0218,
+      "step": 2197
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.222141952412369e-06,
+      "loss": 0.026,
+      "step": 2198
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.21422091924799e-06,
+      "loss": 0.0353,
+      "step": 2199
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.20630038213521e-06,
+      "loss": 0.0205,
+      "step": 2200
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.198380346074159e-06,
+      "loss": 0.0213,
+      "step": 2201
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.190460816064649e-06,
+      "loss": 0.0252,
+      "step": 2202
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.18254179710616e-06,
+      "loss": 0.0247,
+      "step": 2203
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.174623294197875e-06,
+      "loss": 0.0252,
+      "step": 2204
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.16670531233863e-06,
+      "loss": 0.0275,
+      "step": 2205
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.158787856526937e-06,
+      "loss": 0.0361,
+      "step": 2206
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.150870931760982e-06,
+      "loss": 0.0269,
+      "step": 2207
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 9.14295454303861e-06,
+      "loss": 0.0239,
+      "step": 2208
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.135038695357324e-06,
+      "loss": 0.0247,
+      "step": 2209
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.1271233937143e-06,
+      "loss": 0.0152,
+      "step": 2210
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.119208643106353e-06,
+      "loss": 0.0268,
+      "step": 2211
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.111294448529955e-06,
+      "loss": 0.03,
+      "step": 2212
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.10338081498124e-06,
+      "loss": 0.0256,
+      "step": 2213
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.095467747455965e-06,
+      "loss": 0.0254,
+      "step": 2214
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.087555250949546e-06,
+      "loss": 0.0306,
+      "step": 2215
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.079643330457036e-06,
+      "loss": 0.0265,
+      "step": 2216
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.071731990973118e-06,
+      "loss": 0.0291,
+      "step": 2217
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.063821237492113e-06,
+      "loss": 0.0348,
+      "step": 2218
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.055911075007977e-06,
+      "loss": 0.0273,
+      "step": 2219
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.048001508514283e-06,
+      "loss": 0.0228,
+      "step": 2220
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 9.04009254300423e-06,
+      "loss": 0.0263,
+      "step": 2221
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.032184183470648e-06,
+      "loss": 0.028,
+      "step": 2222
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.024276434905969e-06,
+      "loss": 0.0395,
+      "step": 2223
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.016369302302248e-06,
+      "loss": 0.0198,
+      "step": 2224
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.008462790651154e-06,
+      "loss": 0.0322,
+      "step": 2225
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 9.000556904943956e-06,
+      "loss": 0.0184,
+      "step": 2226
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.992651650171528e-06,
+      "loss": 0.0216,
+      "step": 2227
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.984747031324357e-06,
+      "loss": 0.026,
+      "step": 2228
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.976843053392518e-06,
+      "loss": 0.0269,
+      "step": 2229
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.968939721365682e-06,
+      "loss": 0.0251,
+      "step": 2230
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.961037040233119e-06,
+      "loss": 0.0264,
+      "step": 2231
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.95313501498368e-06,
+      "loss": 0.0312,
+      "step": 2232
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.945233650605808e-06,
+      "loss": 0.0235,
+      "step": 2233
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.937332952087522e-06,
+      "loss": 0.0183,
+      "step": 2234
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 8.929432924416433e-06,
+      "loss": 0.0257,
+      "step": 2235
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.92153357257971e-06,
+      "loss": 0.0291,
+      "step": 2236
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.913634901564115e-06,
+      "loss": 0.0255,
+      "step": 2237
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.905736916355969e-06,
+      "loss": 0.0277,
+      "step": 2238
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.897839621941156e-06,
+      "loss": 0.0231,
+      "step": 2239
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.88994302330514e-06,
+      "loss": 0.034,
+      "step": 2240
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.882047125432929e-06,
+      "loss": 0.0156,
+      "step": 2241
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.874151933309099e-06,
+      "loss": 0.0204,
+      "step": 2242
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.866257451917774e-06,
+      "loss": 0.0249,
+      "step": 2243
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.858363686242635e-06,
+      "loss": 0.0226,
+      "step": 2244
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.850470641266903e-06,
+      "loss": 0.0252,
+      "step": 2245
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.842578321973358e-06,
+      "loss": 0.0218,
+      "step": 2246
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.834686733344309e-06,
+      "loss": 0.0268,
+      "step": 2247
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.826795880361603e-06,
+      "loss": 0.0252,
+      "step": 2248
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 8.818905768006637e-06,
+      "loss": 0.0233,
+      "step": 2249
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.811016401260327e-06,
+      "loss": 0.0327,
+      "step": 2250
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.80312778510312e-06,
+      "loss": 0.022,
+      "step": 2251
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.795239924514997e-06,
+      "loss": 0.0232,
+      "step": 2252
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.787352824475454e-06,
+      "loss": 0.023,
+      "step": 2253
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.779466489963503e-06,
+      "loss": 0.0392,
+      "step": 2254
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.771580925957691e-06,
+      "loss": 0.0253,
+      "step": 2255
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.76369613743606e-06,
+      "loss": 0.0265,
+      "step": 2256
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.75581212937617e-06,
+      "loss": 0.0251,
+      "step": 2257
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.747928906755087e-06,
+      "loss": 0.0172,
+      "step": 2258
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.740046474549387e-06,
+      "loss": 0.0182,
+      "step": 2259
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.732164837735135e-06,
+      "loss": 0.0323,
+      "step": 2260
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.724284001287903e-06,
+      "loss": 0.0263,
+      "step": 2261
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 8.716403970182759e-06,
+      "loss": 0.0229,
+      "step": 2262
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.708524749394249e-06,
+      "loss": 0.0271,
+      "step": 2263
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.70064634389643e-06,
+      "loss": 0.025,
+      "step": 2264
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.692768758662827e-06,
+      "loss": 0.0291,
+      "step": 2265
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.684891998666445e-06,
+      "loss": 0.0269,
+      "step": 2266
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.677016068879785e-06,
+      "loss": 0.0292,
+      "step": 2267
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.66914097427481e-06,
+      "loss": 0.0131,
+      "step": 2268
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.661266719822959e-06,
+      "loss": 0.0294,
+      "step": 2269
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.653393310495141e-06,
+      "loss": 0.0299,
+      "step": 2270
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.645520751261736e-06,
+      "loss": 0.0151,
+      "step": 2271
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.637649047092573e-06,
+      "loss": 0.0251,
+      "step": 2272
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.629778202956963e-06,
+      "loss": 0.0247,
+      "step": 2273
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.621908223823657e-06,
+      "loss": 0.0213,
+      "step": 2274
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.614039114660864e-06,
+      "loss": 0.0339,
+      "step": 2275
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 8.60617088043625e-06,
+      "loss": 0.0288,
+      "step": 2276
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.598303526116916e-06,
+      "loss": 0.0229,
+      "step": 2277
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.590437056669423e-06,
+      "loss": 0.0313,
+      "step": 2278
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.582571477059763e-06,
+      "loss": 0.0279,
+      "step": 2279
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.574706792253367e-06,
+      "loss": 0.0278,
+      "step": 2280
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.566843007215103e-06,
+      "loss": 0.027,
+      "step": 2281
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.558980126909277e-06,
+      "loss": 0.0253,
+      "step": 2282
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.551118156299613e-06,
+      "loss": 0.0276,
+      "step": 2283
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.543257100349265e-06,
+      "loss": 0.0235,
+      "step": 2284
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.535396964020812e-06,
+      "loss": 0.0354,
+      "step": 2285
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.527537752276251e-06,
+      "loss": 0.0255,
+      "step": 2286
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.519679470076992e-06,
+      "loss": 0.0316,
+      "step": 2287
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.511822122383866e-06,
+      "loss": 0.0274,
+      "step": 2288
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 8.503965714157108e-06,
+      "loss": 0.0252,
+      "step": 2289
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.496110250356355e-06,
+      "loss": 0.0366,
+      "step": 2290
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.488255735940663e-06,
+      "loss": 0.0245,
+      "step": 2291
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.480402175868477e-06,
+      "loss": 0.0219,
+      "step": 2292
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.472549575097639e-06,
+      "loss": 0.0268,
+      "step": 2293
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.464697938585392e-06,
+      "loss": 0.0285,
+      "step": 2294
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.456847271288365e-06,
+      "loss": 0.0255,
+      "step": 2295
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.448997578162572e-06,
+      "loss": 0.0252,
+      "step": 2296
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.441148864163429e-06,
+      "loss": 0.0203,
+      "step": 2297
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.43330113424571e-06,
+      "loss": 0.0212,
+      "step": 2298
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.42545439336359e-06,
+      "loss": 0.0324,
+      "step": 2299
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.417608646470605e-06,
+      "loss": 0.0266,
+      "step": 2300
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.409763898519665e-06,
+      "loss": 0.0222,
+      "step": 2301
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.401920154463056e-06,
+      "loss": 0.0223,
+      "step": 2302
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 8.394077419252425e-06,
+      "loss": 0.0304,
+      "step": 2303
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.38623569783878e-06,
+      "loss": 0.0222,
+      "step": 2304
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.3783949951725e-06,
+      "loss": 0.0268,
+      "step": 2305
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.37055531620331e-06,
+      "loss": 0.028,
+      "step": 2306
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.362716665880289e-06,
+      "loss": 0.0303,
+      "step": 2307
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.354879049151875e-06,
+      "loss": 0.0199,
+      "step": 2308
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.347042470965846e-06,
+      "loss": 0.0333,
+      "step": 2309
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.339206936269328e-06,
+      "loss": 0.0228,
+      "step": 2310
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.331372450008786e-06,
+      "loss": 0.0205,
+      "step": 2311
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.323539017130024e-06,
+      "loss": 0.023,
+      "step": 2312
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.315706642578178e-06,
+      "loss": 0.0216,
+      "step": 2313
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.307875331297727e-06,
+      "loss": 0.0246,
+      "step": 2314
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.300045088232468e-06,
+      "loss": 0.0266,
+      "step": 2315
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.292215918325518e-06,
+      "loss": 0.0229,
+      "step": 2316
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 8.284387826519335e-06,
+      "loss": 0.028,
+      "step": 2317
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.276560817755686e-06,
+      "loss": 0.0301,
+      "step": 2318
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.268734896975649e-06,
+      "loss": 0.0219,
+      "step": 2319
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.260910069119625e-06,
+      "loss": 0.0221,
+      "step": 2320
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.25308633912732e-06,
+      "loss": 0.0232,
+      "step": 2321
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.245263711937746e-06,
+      "loss": 0.0245,
+      "step": 2322
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.237442192489225e-06,
+      "loss": 0.0235,
+      "step": 2323
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.229621785719372e-06,
+      "loss": 0.0225,
+      "step": 2324
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.221802496565102e-06,
+      "loss": 0.0255,
+      "step": 2325
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.213984329962629e-06,
+      "loss": 0.0255,
+      "step": 2326
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.206167290847454e-06,
+      "loss": 0.0284,
+      "step": 2327
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.198351384154363e-06,
+      "loss": 0.0257,
+      "step": 2328
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.190536614817437e-06,
+      "loss": 0.0235,
+      "step": 2329
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 8.18272298777003e-06,
+      "loss": 0.0243,
+      "step": 2330
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.174910507944775e-06,
+      "loss": 0.0233,
+      "step": 2331
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.167099180273591e-06,
+      "loss": 0.0244,
+      "step": 2332
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.159289009687658e-06,
+      "loss": 0.0227,
+      "step": 2333
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.15148000111743e-06,
+      "loss": 0.0281,
+      "step": 2334
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.143672159492626e-06,
+      "loss": 0.0364,
+      "step": 2335
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.135865489742235e-06,
+      "loss": 0.0372,
+      "step": 2336
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.128059996794495e-06,
+      "loss": 0.0277,
+      "step": 2337
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.120255685576913e-06,
+      "loss": 0.0254,
+      "step": 2338
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.112452561016241e-06,
+      "loss": 0.0372,
+      "step": 2339
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.10465062803848e-06,
+      "loss": 0.0256,
+      "step": 2340
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.096849891568894e-06,
+      "loss": 0.0333,
+      "step": 2341
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.089050356531975e-06,
+      "loss": 0.0259,
+      "step": 2342
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.08125202785146e-06,
+      "loss": 0.0216,
+      "step": 2343
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 8.073454910450337e-06,
+      "loss": 0.037,
+      "step": 2344
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.065659009250812e-06,
+      "loss": 0.0222,
+      "step": 2345
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.05786432917433e-06,
+      "loss": 0.0247,
+      "step": 2346
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.050070875141572e-06,
+      "loss": 0.0223,
+      "step": 2347
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.042278652072437e-06,
+      "loss": 0.0296,
+      "step": 2348
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.034487664886042e-06,
+      "loss": 0.0227,
+      "step": 2349
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.026697918500743e-06,
+      "loss": 0.0333,
+      "step": 2350
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.018909417834096e-06,
+      "loss": 0.0241,
+      "step": 2351
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.011122167802869e-06,
+      "loss": 0.0241,
+      "step": 2352
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 8.003336173323053e-06,
+      "loss": 0.02,
+      "step": 2353
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.99555143930984e-06,
+      "loss": 0.0195,
+      "step": 2354
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.987767970677618e-06,
+      "loss": 0.0175,
+      "step": 2355
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.979985772339995e-06,
+      "loss": 0.0254,
+      "step": 2356
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 7.972204849209763e-06,
+      "loss": 0.0297,
+      "step": 2357
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.964425206198907e-06,
+      "loss": 0.0287,
+      "step": 2358
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.956646848218619e-06,
+      "loss": 0.0221,
+      "step": 2359
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.948869780179266e-06,
+      "loss": 0.0173,
+      "step": 2360
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.941094006990398e-06,
+      "loss": 0.0181,
+      "step": 2361
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.933319533560766e-06,
+      "loss": 0.0196,
+      "step": 2362
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.925546364798283e-06,
+      "loss": 0.0242,
+      "step": 2363
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.917774505610039e-06,
+      "loss": 0.0219,
+      "step": 2364
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.910003960902314e-06,
+      "loss": 0.0207,
+      "step": 2365
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.90223473558054e-06,
+      "loss": 0.0236,
+      "step": 2366
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.89446683454932e-06,
+      "loss": 0.0282,
+      "step": 2367
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.886700262712432e-06,
+      "loss": 0.0245,
+      "step": 2368
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.878935024972802e-06,
+      "loss": 0.0248,
+      "step": 2369
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.871171126232516e-06,
+      "loss": 0.0264,
+      "step": 2370
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 7.863408571392824e-06,
+      "loss": 0.0226,
+      "step": 2371
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.855647365354114e-06,
+      "loss": 0.0223,
+      "step": 2372
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.84788751301593e-06,
+      "loss": 0.0214,
+      "step": 2373
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.840129019276966e-06,
+      "loss": 0.0145,
+      "step": 2374
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.832371889035052e-06,
+      "loss": 0.0243,
+      "step": 2375
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.82461612718715e-06,
+      "loss": 0.0194,
+      "step": 2376
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.816861738629377e-06,
+      "loss": 0.0221,
+      "step": 2377
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.809108728256971e-06,
+      "loss": 0.0255,
+      "step": 2378
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.801357100964295e-06,
+      "loss": 0.0243,
+      "step": 2379
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.793606861644854e-06,
+      "loss": 0.0248,
+      "step": 2380
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.785858015191261e-06,
+      "loss": 0.0195,
+      "step": 2381
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.778110566495256e-06,
+      "loss": 0.0201,
+      "step": 2382
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.770364520447706e-06,
+      "loss": 0.0267,
+      "step": 2383
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.76261988193858e-06,
+      "loss": 0.0196,
+      "step": 2384
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 7.754876655856957e-06,
+      "loss": 0.022,
+      "step": 2385
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.747134847091039e-06,
+      "loss": 0.024,
+      "step": 2386
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.73939446052812e-06,
+      "loss": 0.0279,
+      "step": 2387
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.731655501054597e-06,
+      "loss": 0.0257,
+      "step": 2388
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.723917973555975e-06,
+      "loss": 0.0256,
+      "step": 2389
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.716181882916846e-06,
+      "loss": 0.0241,
+      "step": 2390
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.708447234020898e-06,
+      "loss": 0.0218,
+      "step": 2391
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.700714031750912e-06,
+      "loss": 0.0249,
+      "step": 2392
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.692982280988752e-06,
+      "loss": 0.0251,
+      "step": 2393
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.685251986615363e-06,
+      "loss": 0.0248,
+      "step": 2394
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.67752315351078e-06,
+      "loss": 0.0225,
+      "step": 2395
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.669795786554107e-06,
+      "loss": 0.0193,
+      "step": 2396
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.662069890623525e-06,
+      "loss": 0.0187,
+      "step": 2397
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 7.654345470596288e-06,
+      "loss": 0.027,
+      "step": 2398
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.646622531348715e-06,
+      "loss": 0.023,
+      "step": 2399
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.63890107775619e-06,
+      "loss": 0.025,
+      "step": 2400
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.631181114693167e-06,
+      "loss": 0.0311,
+      "step": 2401
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.623462647033151e-06,
+      "loss": 0.0334,
+      "step": 2402
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.615745679648702e-06,
+      "loss": 0.0311,
+      "step": 2403
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.608030217411438e-06,
+      "loss": 0.0292,
+      "step": 2404
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.600316265192028e-06,
+      "loss": 0.0217,
+      "step": 2405
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.59260382786018e-06,
+      "loss": 0.0163,
+      "step": 2406
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.584892910284655e-06,
+      "loss": 0.0173,
+      "step": 2407
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.577183517333247e-06,
+      "loss": 0.0177,
+      "step": 2408
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.569475653872787e-06,
+      "loss": 0.0267,
+      "step": 2409
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.561769324769151e-06,
+      "loss": 0.0248,
+      "step": 2410
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.554064534887238e-06,
+      "loss": 0.0203,
+      "step": 2411
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 7.546361289090971e-06,
+      "loss": 0.0174,
+      "step": 2412
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.538659592243309e-06,
+      "loss": 0.0176,
+      "step": 2413
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.5309594492062256e-06,
+      "loss": 0.0217,
+      "step": 2414
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.5232608648407166e-06,
+      "loss": 0.0171,
+      "step": 2415
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.515563844006796e-06,
+      "loss": 0.026,
+      "step": 2416
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.507868391563485e-06,
+      "loss": 0.0254,
+      "step": 2417
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.500174512368814e-06,
+      "loss": 0.0201,
+      "step": 2418
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.492482211279832e-06,
+      "loss": 0.018,
+      "step": 2419
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.484791493152579e-06,
+      "loss": 0.0205,
+      "step": 2420
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.477102362842099e-06,
+      "loss": 0.0219,
+      "step": 2421
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.469414825202439e-06,
+      "loss": 0.0271,
+      "step": 2422
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.461728885086634e-06,
+      "loss": 0.0339,
+      "step": 2423
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.454044547346708e-06,
+      "loss": 0.0262,
+      "step": 2424
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 7.44636181683369e-06,
+      "loss": 0.0187,
+      "step": 2425
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.438680698397575e-06,
+      "loss": 0.0233,
+      "step": 2426
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.431001196887345e-06,
+      "loss": 0.0221,
+      "step": 2427
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.423323317150975e-06,
+      "loss": 0.0191,
+      "step": 2428
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.415647064035397e-06,
+      "loss": 0.0226,
+      "step": 2429
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.407972442386527e-06,
+      "loss": 0.0265,
+      "step": 2430
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.400299457049252e-06,
+      "loss": 0.02,
+      "step": 2431
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.392628112867415e-06,
+      "loss": 0.0235,
+      "step": 2432
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.384958414683839e-06,
+      "loss": 0.0325,
+      "step": 2433
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.377290367340297e-06,
+      "loss": 0.0272,
+      "step": 2434
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.369623975677519e-06,
+      "loss": 0.0208,
+      "step": 2435
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.361959244535199e-06,
+      "loss": 0.0235,
+      "step": 2436
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.354296178751976e-06,
+      "loss": 0.0283,
+      "step": 2437
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.346634783165435e-06,
+      "loss": 0.0244,
+      "step": 2438
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 7.338975062612115e-06,
+      "loss": 0.0183,
+      "step": 2439
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.331317021927489e-06,
+      "loss": 0.0212,
+      "step": 2440
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.323660665945972e-06,
+      "loss": 0.0237,
+      "step": 2441
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.316005999500924e-06,
+      "loss": 0.021,
+      "step": 2442
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.308353027424629e-06,
+      "loss": 0.0256,
+      "step": 2443
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.300701754548295e-06,
+      "loss": 0.0165,
+      "step": 2444
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.293052185702079e-06,
+      "loss": 0.0283,
+      "step": 2445
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.285404325715044e-06,
+      "loss": 0.0173,
+      "step": 2446
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.277758179415175e-06,
+      "loss": 0.0319,
+      "step": 2447
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.270113751629388e-06,
+      "loss": 0.0202,
+      "step": 2448
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.262471047183501e-06,
+      "loss": 0.0217,
+      "step": 2449
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.2548300709022455e-06,
+      "loss": 0.022,
+      "step": 2450
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.247190827609273e-06,
+      "loss": 0.0188,
+      "step": 2451
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 7.239553322127131e-06,
+      "loss": 0.0255,
+      "step": 2452
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.231917559277266e-06,
+      "loss": 0.0218,
+      "step": 2453
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.224283543880041e-06,
+      "loss": 0.0197,
+      "step": 2454
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.216651280754701e-06,
+      "loss": 0.0231,
+      "step": 2455
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.209020774719386e-06,
+      "loss": 0.024,
+      "step": 2456
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.201392030591137e-06,
+      "loss": 0.0283,
+      "step": 2457
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.193765053185872e-06,
+      "loss": 0.0194,
+      "step": 2458
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.186139847318394e-06,
+      "loss": 0.0367,
+      "step": 2459
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.178516417802399e-06,
+      "loss": 0.0157,
+      "step": 2460
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.170894769450451e-06,
+      "loss": 0.0239,
+      "step": 2461
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.163274907073987e-06,
+      "loss": 0.0264,
+      "step": 2462
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.155656835483331e-06,
+      "loss": 0.0238,
+      "step": 2463
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.148040559487664e-06,
+      "loss": 0.0222,
+      "step": 2464
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.1404260838950344e-06,
+      "loss": 0.0272,
+      "step": 2465
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 7.132813413512361e-06,
+      "loss": 0.0185,
+      "step": 2466
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.125202553145416e-06,
+      "loss": 0.02,
+      "step": 2467
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.117593507598827e-06,
+      "loss": 0.0263,
+      "step": 2468
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.10998628167609e-06,
+      "loss": 0.0235,
+      "step": 2469
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.102380880179538e-06,
+      "loss": 0.022,
+      "step": 2470
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.0947773079103565e-06,
+      "loss": 0.024,
+      "step": 2471
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.087175569668576e-06,
+      "loss": 0.0232,
+      "step": 2472
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.079575670253074e-06,
+      "loss": 0.0182,
+      "step": 2473
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.071977614461559e-06,
+      "loss": 0.0187,
+      "step": 2474
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.064381407090584e-06,
+      "loss": 0.0257,
+      "step": 2475
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.056787052935528e-06,
+      "loss": 0.0204,
+      "step": 2476
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.049194556790601e-06,
+      "loss": 0.0179,
+      "step": 2477
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.041603923448847e-06,
+      "loss": 0.0214,
+      "step": 2478
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.034015157702127e-06,
+      "loss": 0.0282,
+      "step": 2479
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 7.026428264341125e-06,
+      "loss": 0.027,
+      "step": 2480
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.018843248155345e-06,
+      "loss": 0.0321,
+      "step": 2481
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.0112601139331005e-06,
+      "loss": 0.0187,
+      "step": 2482
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 7.003678866461518e-06,
+      "loss": 0.0339,
+      "step": 2483
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.996099510526546e-06,
+      "loss": 0.0196,
+      "step": 2484
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.988522050912921e-06,
+      "loss": 0.0217,
+      "step": 2485
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.980946492404188e-06,
+      "loss": 0.0253,
+      "step": 2486
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.973372839782699e-06,
+      "loss": 0.0268,
+      "step": 2487
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.965801097829596e-06,
+      "loss": 0.0187,
+      "step": 2488
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.9582312713248144e-06,
+      "loss": 0.0298,
+      "step": 2489
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.950663365047083e-06,
+      "loss": 0.0231,
+      "step": 2490
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.9430973837739204e-06,
+      "loss": 0.0243,
+      "step": 2491
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.935533332281619e-06,
+      "loss": 0.0182,
+      "step": 2492
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 6.927971215345271e-06,
+      "loss": 0.0199,
+      "step": 2493
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.920411037738735e-06,
+      "loss": 0.0252,
+      "step": 2494
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.91285280423464e-06,
+      "loss": 0.0272,
+      "step": 2495
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.905296519604407e-06,
+      "loss": 0.0261,
+      "step": 2496
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.89774218861821e-06,
+      "loss": 0.0328,
+      "step": 2497
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.890189816044992e-06,
+      "loss": 0.0236,
+      "step": 2498
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.8826394066524695e-06,
+      "loss": 0.0223,
+      "step": 2499
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.875090965207108e-06,
+      "loss": 0.022,
+      "step": 2500
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.867544496474132e-06,
+      "loss": 0.02,
+      "step": 2501
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.860000005217533e-06,
+      "loss": 0.0337,
+      "step": 2502
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.852457496200042e-06,
+      "loss": 0.0244,
+      "step": 2503
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.8449169741831356e-06,
+      "loss": 0.0236,
+      "step": 2504
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.837378443927052e-06,
+      "loss": 0.0294,
+      "step": 2505
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.829841910190759e-06,
+      "loss": 0.0218,
+      "step": 2506
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 6.8223073777319625e-06,
+      "loss": 0.0273,
+      "step": 2507
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.814774851307118e-06,
+      "loss": 0.0252,
+      "step": 2508
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.807244335671401e-06,
+      "loss": 0.0247,
+      "step": 2509
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.799715835578721e-06,
+      "loss": 0.0206,
+      "step": 2510
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7921893557817246e-06,
+      "loss": 0.0244,
+      "step": 2511
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.784664901031772e-06,
+      "loss": 0.0193,
+      "step": 2512
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.777142476078944e-06,
+      "loss": 0.0274,
+      "step": 2513
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.769622085672054e-06,
+      "loss": 0.0231,
+      "step": 2514
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.762103734558616e-06,
+      "loss": 0.0194,
+      "step": 2515
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.754587427484862e-06,
+      "loss": 0.0279,
+      "step": 2516
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.747073169195739e-06,
+      "loss": 0.0247,
+      "step": 2517
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.7395609644348905e-06,
+      "loss": 0.0299,
+      "step": 2518
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.732050817944666e-06,
+      "loss": 0.0274,
+      "step": 2519
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 6.724542734466127e-06,
+      "loss": 0.0239,
+      "step": 2520
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.717036718739021e-06,
+      "loss": 0.0237,
+      "step": 2521
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.7095327755017856e-06,
+      "loss": 0.0183,
+      "step": 2522
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.70203090949157e-06,
+      "loss": 0.0222,
+      "step": 2523
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.694531125444193e-06,
+      "loss": 0.0179,
+      "step": 2524
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.687033428094164e-06,
+      "loss": 0.0285,
+      "step": 2525
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.679537822174682e-06,
+      "loss": 0.0236,
+      "step": 2526
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.672044312417618e-06,
+      "loss": 0.0247,
+      "step": 2527
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.664552903553519e-06,
+      "loss": 0.0171,
+      "step": 2528
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.657063600311616e-06,
+      "loss": 0.0209,
+      "step": 2529
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.649576407419801e-06,
+      "loss": 0.0259,
+      "step": 2530
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.642091329604632e-06,
+      "loss": 0.0191,
+      "step": 2531
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.634608371591343e-06,
+      "loss": 0.022,
+      "step": 2532
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.627127538103821e-06,
+      "loss": 0.018,
+      "step": 2533
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 6.619648833864614e-06,
+      "loss": 0.0238,
+      "step": 2534
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.6121722635949244e-06,
+      "loss": 0.0265,
+      "step": 2535
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.604697832014612e-06,
+      "loss": 0.0244,
+      "step": 2536
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.597225543842176e-06,
+      "loss": 0.0215,
+      "step": 2537
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.58975540379478e-06,
+      "loss": 0.0238,
+      "step": 2538
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.582287416588216e-06,
+      "loss": 0.0239,
+      "step": 2539
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.574821586936921e-06,
+      "loss": 0.0228,
+      "step": 2540
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.567357919553973e-06,
+      "loss": 0.0196,
+      "step": 2541
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.559896419151085e-06,
+      "loss": 0.0202,
+      "step": 2542
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.552437090438597e-06,
+      "loss": 0.0234,
+      "step": 2543
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.544979938125485e-06,
+      "loss": 0.0158,
+      "step": 2544
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.537524966919346e-06,
+      "loss": 0.0347,
+      "step": 2545
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.5300721815263956e-06,
+      "loss": 0.0276,
+      "step": 2546
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.522621586651485e-06,
+      "loss": 0.0244,
+      "step": 2547
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 6.51517318699807e-06,
+      "loss": 0.0179,
+      "step": 2548
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.507726987268219e-06,
+      "loss": 0.0241,
+      "step": 2549
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.5002829921626206e-06,
+      "loss": 0.0228,
+      "step": 2550
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.492841206380566e-06,
+      "loss": 0.0243,
+      "step": 2551
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.485401634619952e-06,
+      "loss": 0.0237,
+      "step": 2552
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.477964281577282e-06,
+      "loss": 0.0257,
+      "step": 2553
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.470529151947652e-06,
+      "loss": 0.017,
+      "step": 2554
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.463096250424756e-06,
+      "loss": 0.0221,
+      "step": 2555
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.4556655817008895e-06,
+      "loss": 0.0264,
+      "step": 2556
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.4482371504669276e-06,
+      "loss": 0.0267,
+      "step": 2557
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.440810961412338e-06,
+      "loss": 0.0208,
+      "step": 2558
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.433387019225175e-06,
+      "loss": 0.0223,
+      "step": 2559
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.425965328592068e-06,
+      "loss": 0.021,
+      "step": 2560
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 6.418545894198228e-06,
+      "loss": 0.0309,
+      "step": 2561
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.411128720727448e-06,
+      "loss": 0.0234,
+      "step": 2562
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.403713812862085e-06,
+      "loss": 0.0171,
+      "step": 2563
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.3963011752830706e-06,
+      "loss": 0.0223,
+      "step": 2564
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.3888908126699015e-06,
+      "loss": 0.019,
+      "step": 2565
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.381482729700636e-06,
+      "loss": 0.0199,
+      "step": 2566
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.374076931051899e-06,
+      "loss": 0.0203,
+      "step": 2567
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.366673421398869e-06,
+      "loss": 0.0241,
+      "step": 2568
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.359272205415274e-06,
+      "loss": 0.0269,
+      "step": 2569
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.351873287773413e-06,
+      "loss": 0.0195,
+      "step": 2570
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.344476673144113e-06,
+      "loss": 0.026,
+      "step": 2571
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.337082366196756e-06,
+      "loss": 0.026,
+      "step": 2572
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.329690371599271e-06,
+      "loss": 0.021,
+      "step": 2573
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.322300694018122e-06,
+      "loss": 0.0239,
+      "step": 2574
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 6.314913338118308e-06,
+      "loss": 0.0175,
+      "step": 2575
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.307528308563371e-06,
+      "loss": 0.0222,
+      "step": 2576
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.3001456100153754e-06,
+      "loss": 0.0166,
+      "step": 2577
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.292765247134916e-06,
+      "loss": 0.0168,
+      "step": 2578
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.285387224581123e-06,
+      "loss": 0.0164,
+      "step": 2579
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.278011547011638e-06,
+      "loss": 0.017,
+      "step": 2580
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.270638219082617e-06,
+      "loss": 0.025,
+      "step": 2581
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.263267245448754e-06,
+      "loss": 0.0232,
+      "step": 2582
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.255898630763238e-06,
+      "loss": 0.0207,
+      "step": 2583
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.248532379677773e-06,
+      "loss": 0.0246,
+      "step": 2584
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.241168496842578e-06,
+      "loss": 0.024,
+      "step": 2585
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.233806986906367e-06,
+      "loss": 0.0307,
+      "step": 2586
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.226447854516358e-06,
+      "loss": 0.0156,
+      "step": 2587
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 6.21909110431828e-06,
+      "loss": 0.0303,
+      "step": 2588
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.211736740956343e-06,
+      "loss": 0.0202,
+      "step": 2589
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.204384769073253e-06,
+      "loss": 0.0245,
+      "step": 2590
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.197035193310217e-06,
+      "loss": 0.0258,
+      "step": 2591
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.189688018306919e-06,
+      "loss": 0.0215,
+      "step": 2592
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.1823432487015265e-06,
+      "loss": 0.0232,
+      "step": 2593
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.1750008891306975e-06,
+      "loss": 0.0216,
+      "step": 2594
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.167660944229561e-06,
+      "loss": 0.0195,
+      "step": 2595
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.160323418631721e-06,
+      "loss": 0.0308,
+      "step": 2596
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.152988316969263e-06,
+      "loss": 0.0193,
+      "step": 2597
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.145655643872733e-06,
+      "loss": 0.0276,
+      "step": 2598
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.138325403971144e-06,
+      "loss": 0.0261,
+      "step": 2599
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.130997601891986e-06,
+      "loss": 0.0229,
+      "step": 2600
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.123672242261191e-06,
+      "loss": 0.024,
+      "step": 2601
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 6.116349329703164e-06,
+      "loss": 0.03,
+      "step": 2602
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.1090288688407585e-06,
+      "loss": 0.0264,
+      "step": 2603
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.101710864295279e-06,
+      "loss": 0.0269,
+      "step": 2604
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.094395320686482e-06,
+      "loss": 0.0279,
+      "step": 2605
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.087082242632574e-06,
+      "loss": 0.0165,
+      "step": 2606
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0797716347502e-06,
+      "loss": 0.0262,
+      "step": 2607
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.0724635016544434e-06,
+      "loss": 0.0245,
+      "step": 2608
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.065157847958831e-06,
+      "loss": 0.0191,
+      "step": 2609
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.057854678275326e-06,
+      "loss": 0.0287,
+      "step": 2610
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.050553997214312e-06,
+      "loss": 0.0142,
+      "step": 2611
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.043255809384617e-06,
+      "loss": 0.0266,
+      "step": 2612
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.035960119393483e-06,
+      "loss": 0.0199,
+      "step": 2613
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.028666931846576e-06,
+      "loss": 0.0164,
+      "step": 2614
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.021376251347994e-06,
+      "loss": 0.0301,
+      "step": 2615
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 6.014088082500241e-06,
+      "loss": 0.019,
+      "step": 2616
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 6.006802429904235e-06,
+      "loss": 0.0198,
+      "step": 2617
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.999519298159313e-06,
+      "loss": 0.0262,
+      "step": 2618
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9922386918632145e-06,
+      "loss": 0.0246,
+      "step": 2619
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.984960615612083e-06,
+      "loss": 0.0181,
+      "step": 2620
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.977685074000478e-06,
+      "loss": 0.0233,
+      "step": 2621
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9704120716213435e-06,
+      "loss": 0.0132,
+      "step": 2622
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9631416130660235e-06,
+      "loss": 0.0211,
+      "step": 2623
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.955873702924266e-06,
+      "loss": 0.03,
+      "step": 2624
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.948608345784201e-06,
+      "loss": 0.0215,
+      "step": 2625
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.941345546232348e-06,
+      "loss": 0.028,
+      "step": 2626
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.9340853088536145e-06,
+      "loss": 0.026,
+      "step": 2627
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.926827638231289e-06,
+      "loss": 0.0224,
+      "step": 2628
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 5.919572538947036e-06,
+      "loss": 0.0182,
+      "step": 2629
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.912320015580907e-06,
+      "loss": 0.0231,
+      "step": 2630
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.905070072711318e-06,
+      "loss": 0.0271,
+      "step": 2631
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8978227149150545e-06,
+      "loss": 0.0233,
+      "step": 2632
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.890577946767284e-06,
+      "loss": 0.0281,
+      "step": 2633
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.883335772841523e-06,
+      "loss": 0.0282,
+      "step": 2634
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.876096197709658e-06,
+      "loss": 0.0243,
+      "step": 2635
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.868859225941934e-06,
+      "loss": 0.0228,
+      "step": 2636
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8616248621069545e-06,
+      "loss": 0.0227,
+      "step": 2637
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.854393110771668e-06,
+      "loss": 0.0241,
+      "step": 2638
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.847163976501391e-06,
+      "loss": 0.0244,
+      "step": 2639
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.83993746385977e-06,
+      "loss": 0.02,
+      "step": 2640
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.8327135774088e-06,
+      "loss": 0.0285,
+      "step": 2641
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.825492321708834e-06,
+      "loss": 0.0228,
+      "step": 2642
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 5.818273701318542e-06,
+      "loss": 0.0208,
+      "step": 2643
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.811057720794941e-06,
+      "loss": 0.0208,
+      "step": 2644
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.803844384693386e-06,
+      "loss": 0.0246,
+      "step": 2645
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.796633697567557e-06,
+      "loss": 0.0305,
+      "step": 2646
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.789425663969458e-06,
+      "loss": 0.0227,
+      "step": 2647
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.782220288449422e-06,
+      "loss": 0.0162,
+      "step": 2648
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.77501757555611e-06,
+      "loss": 0.0273,
+      "step": 2649
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.76781752983649e-06,
+      "loss": 0.032,
+      "step": 2650
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.760620155835861e-06,
+      "loss": 0.0208,
+      "step": 2651
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.753425458097817e-06,
+      "loss": 0.0265,
+      "step": 2652
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.746233441164275e-06,
+      "loss": 0.024,
+      "step": 2653
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.73904410957546e-06,
+      "loss": 0.0241,
+      "step": 2654
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.731857467869902e-06,
+      "loss": 0.0254,
+      "step": 2655
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 5.724673520584421e-06,
+      "loss": 0.0185,
+      "step": 2656
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.717492272254148e-06,
+      "loss": 0.0301,
+      "step": 2657
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.710313727412513e-06,
+      "loss": 0.0251,
+      "step": 2658
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.703137890591222e-06,
+      "loss": 0.0199,
+      "step": 2659
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.695964766320294e-06,
+      "loss": 0.0188,
+      "step": 2660
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.688794359128018e-06,
+      "loss": 0.0245,
+      "step": 2661
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.681626673540974e-06,
+      "loss": 0.0143,
+      "step": 2662
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.674461714084026e-06,
+      "loss": 0.0329,
+      "step": 2663
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.6672994852803184e-06,
+      "loss": 0.0153,
+      "step": 2664
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.660139991651261e-06,
+      "loss": 0.0231,
+      "step": 2665
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.652983237716549e-06,
+      "loss": 0.0301,
+      "step": 2666
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.645829227994146e-06,
+      "loss": 0.0171,
+      "step": 2667
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.638677967000268e-06,
+      "loss": 0.0207,
+      "step": 2668
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.631529459249426e-06,
+      "loss": 0.0179,
+      "step": 2669
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 5.624383709254363e-06,
+      "loss": 0.0312,
+      "step": 2670
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.617240721526096e-06,
+      "loss": 0.0222,
+      "step": 2671
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.610100500573898e-06,
+      "loss": 0.0191,
+      "step": 2672
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.602963050905296e-06,
+      "loss": 0.0258,
+      "step": 2673
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.5958283770260545e-06,
+      "loss": 0.0228,
+      "step": 2674
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.588696483440203e-06,
+      "loss": 0.0237,
+      "step": 2675
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.58156737465001e-06,
+      "loss": 0.0195,
+      "step": 2676
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.5744410551559725e-06,
+      "loss": 0.0172,
+      "step": 2677
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.567317529456856e-06,
+      "loss": 0.0239,
+      "step": 2678
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.560196802049633e-06,
+      "loss": 0.0177,
+      "step": 2679
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.553078877429524e-06,
+      "loss": 0.0161,
+      "step": 2680
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.545963760089978e-06,
+      "loss": 0.0263,
+      "step": 2681
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.538851454522678e-06,
+      "loss": 0.0163,
+      "step": 2682
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.531741965217514e-06,
+      "loss": 0.0269,
+      "step": 2683
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 5.524635296662615e-06,
+      "loss": 0.0332,
+      "step": 2684
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.517531453344327e-06,
+      "loss": 0.0219,
+      "step": 2685
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.510430439747197e-06,
+      "loss": 0.0218,
+      "step": 2686
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.503332260354017e-06,
+      "loss": 0.0274,
+      "step": 2687
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.496236919645754e-06,
+      "loss": 0.0223,
+      "step": 2688
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.489144422101607e-06,
+      "loss": 0.0204,
+      "step": 2689
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.482054772198972e-06,
+      "loss": 0.0159,
+      "step": 2690
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.474967974413451e-06,
+      "loss": 0.0146,
+      "step": 2691
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.467884033218837e-06,
+      "loss": 0.0203,
+      "step": 2692
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.460802953087126e-06,
+      "loss": 0.022,
+      "step": 2693
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.453724738488511e-06,
+      "loss": 0.0213,
+      "step": 2694
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.4466493938913705e-06,
+      "loss": 0.0183,
+      "step": 2695
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.439576923762275e-06,
+      "loss": 0.0193,
+      "step": 2696
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 5.432507332565968e-06,
+      "loss": 0.0229,
+      "step": 2697
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.425440624765403e-06,
+      "loss": 0.0222,
+      "step": 2698
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.418376804821682e-06,
+      "loss": 0.0167,
+      "step": 2699
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.411315877194104e-06,
+      "loss": 0.0172,
+      "step": 2700
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.404257846340138e-06,
+      "loss": 0.0198,
+      "step": 2701
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.397202716715417e-06,
+      "loss": 0.0185,
+      "step": 2702
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.390150492773749e-06,
+      "loss": 0.019,
+      "step": 2703
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.383101178967109e-06,
+      "loss": 0.0164,
+      "step": 2704
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.376054779745637e-06,
+      "loss": 0.0245,
+      "step": 2705
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.369011299557617e-06,
+      "loss": 0.0185,
+      "step": 2706
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.361970742849509e-06,
+      "loss": 0.0238,
+      "step": 2707
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.35493311406592e-06,
+      "loss": 0.0226,
+      "step": 2708
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.347898417649609e-06,
+      "loss": 0.0225,
+      "step": 2709
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.340866658041486e-06,
+      "loss": 0.0224,
+      "step": 2710
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 5.333837839680598e-06,
+      "loss": 0.0278,
+      "step": 2711
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.3268119670041465e-06,
+      "loss": 0.0247,
+      "step": 2712
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.319789044447468e-06,
+      "loss": 0.0236,
+      "step": 2713
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.312769076444041e-06,
+      "loss": 0.022,
+      "step": 2714
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.305752067425469e-06,
+      "loss": 0.0196,
+      "step": 2715
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.298738021821498e-06,
+      "loss": 0.0227,
+      "step": 2716
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.291726944059999e-06,
+      "loss": 0.0301,
+      "step": 2717
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 5.284718838566968e-06,
+      "loss": 0.0287,
+      "step": 2718
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 4077,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 2718,
+  "total_flos": 158065489870848.0,
+  "train_batch_size": 16,
+  "trial_name": null,
+  "trial_params": null
+}