diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,123016 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.9999024390243902,
+  "eval_steps": 500,
+  "global_step": 20498,
+  "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": 0.7517,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.2520325203252037e-07,
+      "loss": 73938.3203,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.878048780487805e-07,
+      "loss": 0.7489,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.504065040650407e-07,
+      "loss": 0.7362,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.130081300813009e-07,
+      "loss": 148561.9375,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.75609756097561e-07,
+      "loss": 158990.2031,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1382113821138213e-06,
+      "loss": 118955.5,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3008130081300815e-06,
+      "loss": 136134.0625,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4634146341463416e-06,
+      "loss": 147633.3594,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6260162601626018e-06,
+      "loss": 134917.2344,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.788617886178862e-06,
+      "loss": 167428.2812,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.951219512195122e-06,
+      "loss": 220828.9375,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.113821138211382e-06,
+      "loss": 151322.1094,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2764227642276426e-06,
+      "loss": 0.7924,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4390243902439027e-06,
+      "loss": 149858.1094,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.601626016260163e-06,
+      "loss": 0.7852,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7642276422764227e-06,
+      "loss": 138218.4688,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9268292682926833e-06,
+      "loss": 0.7684,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.0894308943089435e-06,
+      "loss": 120080.5234,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.2520325203252037e-06,
+      "loss": 122061.3438,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.414634146341464e-06,
+      "loss": 0.7826,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.577235772357724e-06,
+      "loss": 0.7968,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.739837398373984e-06,
+      "loss": 71727.1094,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.902439024390244e-06,
+      "loss": 65278.9375,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.0650406504065046e-06,
+      "loss": 87688.5625,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.227642276422764e-06,
+      "loss": 77673.2656,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.390243902439024e-06,
+      "loss": 0.7593,
+      "step": 27
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.552845528455285e-06,
+      "loss": 59035.793,
+      "step": 28
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.715447154471545e-06,
+      "loss": 56793.7188,
+      "step": 29
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.8780487804878055e-06,
+      "loss": 57031.7188,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.040650406504066e-06,
+      "loss": 0.804,
+      "step": 31
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.203252032520326e-06,
+      "loss": 28888.5,
+      "step": 32
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.365853658536585e-06,
+      "loss": 24802.4102,
+      "step": 33
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.528455284552845e-06,
+      "loss": 18743.9102,
+      "step": 34
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.6910569105691056e-06,
+      "loss": 0.7919,
+      "step": 35
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.853658536585367e-06,
+      "loss": 18437.6191,
+      "step": 36
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.016260162601627e-06,
+      "loss": 0.8216,
+      "step": 37
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.178861788617887e-06,
+      "loss": 0.8355,
+      "step": 38
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.341463414634147e-06,
+      "loss": 0.8168,
+      "step": 39
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.504065040650407e-06,
+      "loss": 16267.3027,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 12726.582,
+      "step": 41
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.829268292682928e-06,
+      "loss": 7291.1831,
+      "step": 42
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.991869918699187e-06,
+      "loss": 0.7382,
+      "step": 43
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.154471544715448e-06,
+      "loss": 3967.531,
+      "step": 44
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.317073170731707e-06,
+      "loss": 4358.8164,
+      "step": 45
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.479674796747968e-06,
+      "loss": 4753.0488,
+      "step": 46
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.642276422764228e-06,
+      "loss": 3412.7214,
+      "step": 47
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.804878048780489e-06,
+      "loss": 2155.0566,
+      "step": 48
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.967479674796748e-06,
+      "loss": 2098.1382,
+      "step": 49
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.130081300813009e-06,
+      "loss": 0.7845,
+      "step": 50
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.292682926829268e-06,
+      "loss": 2706.0708,
+      "step": 51
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.455284552845528e-06,
+      "loss": 0.7527,
+      "step": 52
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.617886178861789e-06,
+      "loss": 0.8494,
+      "step": 53
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.780487804878048e-06,
+      "loss": 0.7525,
+      "step": 54
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.94308943089431e-06,
+      "loss": 1113.9468,
+      "step": 55
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.10569105691057e-06,
+      "loss": 0.7957,
+      "step": 56
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.26829268292683e-06,
+      "loss": 818.5796,
+      "step": 57
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.43089430894309e-06,
+      "loss": 0.6774,
+      "step": 58
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.59349593495935e-06,
+      "loss": 0.7841,
+      "step": 59
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.756097560975611e-06,
+      "loss": 0.7937,
+      "step": 60
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.91869918699187e-06,
+      "loss": 490.9912,
+      "step": 61
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0081300813008131e-05,
+      "loss": 0.7511,
+      "step": 62
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.024390243902439e-05,
+      "loss": 456.3557,
+      "step": 63
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0406504065040652e-05,
+      "loss": 427.4767,
+      "step": 64
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0569105691056911e-05,
+      "loss": 471.9561,
+      "step": 65
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.073170731707317e-05,
+      "loss": 421.9743,
+      "step": 66
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0894308943089431e-05,
+      "loss": 0.7865,
+      "step": 67
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.105691056910569e-05,
+      "loss": 398.928,
+      "step": 68
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1219512195121952e-05,
+      "loss": 0.7716,
+      "step": 69
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1382113821138211e-05,
+      "loss": 0.7791,
+      "step": 70
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1544715447154472e-05,
+      "loss": 0.7531,
+      "step": 71
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1707317073170733e-05,
+      "loss": 386.9471,
+      "step": 72
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1869918699186992e-05,
+      "loss": 0.783,
+      "step": 73
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2032520325203254e-05,
+      "loss": 0.8053,
+      "step": 74
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2195121951219513e-05,
+      "loss": 0.7542,
+      "step": 75
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2357723577235774e-05,
+      "loss": 397.3582,
+      "step": 76
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2520325203252033e-05,
+      "loss": 379.2266,
+      "step": 77
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2682926829268294e-05,
+      "loss": 386.7066,
+      "step": 78
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2845528455284554e-05,
+      "loss": 0.7844,
+      "step": 79
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3008130081300815e-05,
+      "loss": 0.6771,
+      "step": 80
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3170731707317074e-05,
+      "loss": 354.114,
+      "step": 81
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 359.1651,
+      "step": 82
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3495934959349593e-05,
+      "loss": 0.822,
+      "step": 83
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3658536585365855e-05,
+      "loss": 367.7664,
+      "step": 84
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3821138211382115e-05,
+      "loss": 345.935,
+      "step": 85
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3983739837398374e-05,
+      "loss": 342.4774,
+      "step": 86
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4146341463414633e-05,
+      "loss": 0.8006,
+      "step": 87
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4308943089430896e-05,
+      "loss": 0.7073,
+      "step": 88
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4471544715447155e-05,
+      "loss": 0.7897,
+      "step": 89
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4634146341463415e-05,
+      "loss": 0.7847,
+      "step": 90
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4796747967479676e-05,
+      "loss": 350.6793,
+      "step": 91
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4959349593495937e-05,
+      "loss": 359.9893,
+      "step": 92
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5121951219512196e-05,
+      "loss": 0.7164,
+      "step": 93
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5284552845528455e-05,
+      "loss": 0.7841,
+      "step": 94
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5447154471544717e-05,
+      "loss": 0.7693,
+      "step": 95
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5609756097560978e-05,
+      "loss": 296.3467,
+      "step": 96
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5772357723577235e-05,
+      "loss": 0.7967,
+      "step": 97
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5934959349593496e-05,
+      "loss": 315.7057,
+      "step": 98
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6097560975609757e-05,
+      "loss": 0.7919,
+      "step": 99
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6260162601626018e-05,
+      "loss": 0.7916,
+      "step": 100
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6422764227642276e-05,
+      "loss": 312.6718,
+      "step": 101
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6585365853658537e-05,
+      "loss": 325.5378,
+      "step": 102
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6747967479674798e-05,
+      "loss": 294.8365,
+      "step": 103
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.6910569105691056e-05,
+      "loss": 288.1011,
+      "step": 104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.707317073170732e-05,
+      "loss": 0.7286,
+      "step": 105
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7235772357723578e-05,
+      "loss": 0.7681,
+      "step": 106
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.739837398373984e-05,
+      "loss": 0.8124,
+      "step": 107
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.7560975609756096e-05,
+      "loss": 271.641,
+      "step": 108
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.772357723577236e-05,
+      "loss": 288.5695,
+      "step": 109
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.788617886178862e-05,
+      "loss": 271.0004,
+      "step": 110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.804878048780488e-05,
+      "loss": 0.8112,
+      "step": 111
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.821138211382114e-05,
+      "loss": 251.8119,
+      "step": 112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.83739837398374e-05,
+      "loss": 239.3929,
+      "step": 113
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.853658536585366e-05,
+      "loss": 0.7755,
+      "step": 114
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.869918699186992e-05,
+      "loss": 233.29,
+      "step": 115
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.886178861788618e-05,
+      "loss": 0.8322,
+      "step": 116
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.902439024390244e-05,
+      "loss": 219.2343,
+      "step": 117
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.91869918699187e-05,
+      "loss": 0.7941,
+      "step": 118
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9349593495934958e-05,
+      "loss": 310.3843,
+      "step": 119
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.9512195121951222e-05,
+      "loss": 0.7871,
+      "step": 120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.967479674796748e-05,
+      "loss": 0.7208,
+      "step": 121
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1.983739837398374e-05,
+      "loss": 0.7348,
+      "step": 122
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2e-05,
+      "loss": 208.7521,
+      "step": 123
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0162601626016263e-05,
+      "loss": 0.7874,
+      "step": 124
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.032520325203252e-05,
+      "loss": 195.0904,
+      "step": 125
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.048780487804878e-05,
+      "loss": 202.7263,
+      "step": 126
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0650406504065042e-05,
+      "loss": 0.805,
+      "step": 127
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.0813008130081303e-05,
+      "loss": 195.6412,
+      "step": 128
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.097560975609756e-05,
+      "loss": 0.7747,
+      "step": 129
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1138211382113822e-05,
+      "loss": 0.7681,
+      "step": 130
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1300813008130083e-05,
+      "loss": 193.9508,
+      "step": 131
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.146341463414634e-05,
+      "loss": 0.7817,
+      "step": 132
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1626016260162602e-05,
+      "loss": 189.1655,
+      "step": 133
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1788617886178863e-05,
+      "loss": 182.4647,
+      "step": 134
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.1951219512195124e-05,
+      "loss": 175.6289,
+      "step": 135
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.211382113821138e-05,
+      "loss": 179.3954,
+      "step": 136
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2276422764227646e-05,
+      "loss": 0.7363,
+      "step": 137
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2439024390243904e-05,
+      "loss": 0.755,
+      "step": 138
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2601626016260165e-05,
+      "loss": 199.1576,
+      "step": 139
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2764227642276422e-05,
+      "loss": 257.6119,
+      "step": 140
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.2926829268292687e-05,
+      "loss": 0.7749,
+      "step": 141
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3089430894308944e-05,
+      "loss": 259.3569,
+      "step": 142
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3252032520325205e-05,
+      "loss": 0.7569,
+      "step": 143
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3414634146341466e-05,
+      "loss": 0.7495,
+      "step": 144
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3577235772357724e-05,
+      "loss": 234.8088,
+      "step": 145
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3739837398373985e-05,
+      "loss": 156.6951,
+      "step": 146
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.3902439024390243e-05,
+      "loss": 157.8555,
+      "step": 147
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4065040650406507e-05,
+      "loss": 0.7516,
+      "step": 148
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4227642276422765e-05,
+      "loss": 162.3036,
+      "step": 149
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4390243902439026e-05,
+      "loss": 0.7668,
+      "step": 150
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4552845528455283e-05,
+      "loss": 162.6857,
+      "step": 151
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4715447154471548e-05,
+      "loss": 159.0766,
+      "step": 152
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.4878048780487805e-05,
+      "loss": 156.3176,
+      "step": 153
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5040650406504066e-05,
+      "loss": 0.7201,
+      "step": 154
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5203252032520324e-05,
+      "loss": 0.7598,
+      "step": 155
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.536585365853659e-05,
+      "loss": 186.5637,
+      "step": 156
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.552845528455285e-05,
+      "loss": 0.8243,
+      "step": 157
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5691056910569107e-05,
+      "loss": 212.9759,
+      "step": 158
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.5853658536585368e-05,
+      "loss": 234.5794,
+      "step": 159
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.601626016260163e-05,
+      "loss": 0.8417,
+      "step": 160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6178861788617887e-05,
+      "loss": 154.0451,
+      "step": 161
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6341463414634148e-05,
+      "loss": 0.7362,
+      "step": 162
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6504065040650406e-05,
+      "loss": 0.772,
+      "step": 163
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6666666666666667e-05,
+      "loss": 156.6683,
+      "step": 164
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.682926829268293e-05,
+      "loss": 0.8081,
+      "step": 165
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.6991869918699185e-05,
+      "loss": 147.9996,
+      "step": 166
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.715447154471545e-05,
+      "loss": 152.466,
+      "step": 167
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.731707317073171e-05,
+      "loss": 146.7048,
+      "step": 168
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.747967479674797e-05,
+      "loss": 143.8514,
+      "step": 169
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.764227642276423e-05,
+      "loss": 136.8205,
+      "step": 170
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.780487804878049e-05,
+      "loss": 144.6079,
+      "step": 171
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.7967479674796748e-05,
+      "loss": 146.3627,
+      "step": 172
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.813008130081301e-05,
+      "loss": 0.767,
+      "step": 173
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8292682926829267e-05,
+      "loss": 0.8388,
+      "step": 174
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8455284552845528e-05,
+      "loss": 0.8243,
+      "step": 175
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8617886178861792e-05,
+      "loss": 142.1441,
+      "step": 176
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.8780487804878046e-05,
+      "loss": 150.0498,
+      "step": 177
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.894308943089431e-05,
+      "loss": 133.5219,
+      "step": 178
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9105691056910572e-05,
+      "loss": 138.1176,
+      "step": 179
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.926829268292683e-05,
+      "loss": 0.8059,
+      "step": 180
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.943089430894309e-05,
+      "loss": 0.7395,
+      "step": 181
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.959349593495935e-05,
+      "loss": 134.5042,
+      "step": 182
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.975609756097561e-05,
+      "loss": 136.6164,
+      "step": 183
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 2.9918699186991874e-05,
+      "loss": 135.7037,
+      "step": 184
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0081300813008135e-05,
+      "loss": 0.7795,
+      "step": 185
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0243902439024392e-05,
+      "loss": 124.528,
+      "step": 186
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.0406504065040653e-05,
+      "loss": 0.8204,
+      "step": 187
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.056910569105691e-05,
+      "loss": 133.4435,
+      "step": 188
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.073170731707317e-05,
+      "loss": 123.6474,
+      "step": 189
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.089430894308943e-05,
+      "loss": 119.4204,
+      "step": 190
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.105691056910569e-05,
+      "loss": 0.724,
+      "step": 191
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.1219512195121955e-05,
+      "loss": 116.9933,
+      "step": 192
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.1382113821138216e-05,
+      "loss": 122.7089,
+      "step": 193
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.154471544715447e-05,
+      "loss": 0.7592,
+      "step": 194
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.170731707317073e-05,
+      "loss": 0.7742,
+      "step": 195
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.186991869918699e-05,
+      "loss": 125.2852,
+      "step": 196
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2032520325203253e-05,
+      "loss": 125.0263,
+      "step": 197
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2195121951219514e-05,
+      "loss": 0.7572,
+      "step": 198
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.235772357723577e-05,
+      "loss": 0.7201,
+      "step": 199
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.2520325203252037e-05,
+      "loss": 0.765,
+      "step": 200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.26829268292683e-05,
+      "loss": 112.6193,
+      "step": 201
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.284552845528455e-05,
+      "loss": 122.9481,
+      "step": 202
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.300813008130081e-05,
+      "loss": 118.0847,
+      "step": 203
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3170731707317074e-05,
+      "loss": 121.0946,
+      "step": 204
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3333333333333335e-05,
+      "loss": 119.0715,
+      "step": 205
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3495934959349596e-05,
+      "loss": 0.788,
+      "step": 206
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.365853658536586e-05,
+      "loss": 112.5526,
+      "step": 207
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.382113821138211e-05,
+      "loss": 0.7884,
+      "step": 208
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.398373983739838e-05,
+      "loss": 112.6994,
+      "step": 209
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.414634146341464e-05,
+      "loss": 106.7715,
+      "step": 210
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4308943089430894e-05,
+      "loss": 0.7365,
+      "step": 211
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4471544715447155e-05,
+      "loss": 111.3168,
+      "step": 212
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4634146341463416e-05,
+      "loss": 105.1352,
+      "step": 213
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.479674796747968e-05,
+      "loss": 116.7501,
+      "step": 214
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.495934959349594e-05,
+      "loss": 107.1387,
+      "step": 215
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.512195121951219e-05,
+      "loss": 108.8388,
+      "step": 216
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.5284552845528454e-05,
+      "loss": 0.7191,
+      "step": 217
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.544715447154472e-05,
+      "loss": 0.7891,
+      "step": 218
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.5609756097560976e-05,
+      "loss": 106.6367,
+      "step": 219
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.577235772357724e-05,
+      "loss": 113.5296,
+      "step": 220
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.59349593495935e-05,
+      "loss": 104.5221,
+      "step": 221
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.609756097560976e-05,
+      "loss": 106.9317,
+      "step": 222
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.626016260162602e-05,
+      "loss": 0.7297,
+      "step": 223
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.642276422764228e-05,
+      "loss": 111.1957,
+      "step": 224
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.6585365853658535e-05,
+      "loss": 118.4118,
+      "step": 225
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.67479674796748e-05,
+      "loss": 108.2973,
+      "step": 226
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.691056910569106e-05,
+      "loss": 0.7483,
+      "step": 227
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.707317073170732e-05,
+      "loss": 0.8662,
+      "step": 228
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.723577235772358e-05,
+      "loss": 0.8403,
+      "step": 229
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.739837398373984e-05,
+      "loss": 102.8662,
+      "step": 230
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.75609756097561e-05,
+      "loss": 0.7277,
+      "step": 231
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.772357723577236e-05,
+      "loss": 111.8944,
+      "step": 232
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.788617886178862e-05,
+      "loss": 0.775,
+      "step": 233
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.804878048780488e-05,
+      "loss": 103.7253,
+      "step": 234
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8211382113821145e-05,
+      "loss": 99.6541,
+      "step": 235
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.83739837398374e-05,
+      "loss": 0.8081,
+      "step": 236
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.853658536585366e-05,
+      "loss": 0.831,
+      "step": 237
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8699186991869915e-05,
+      "loss": 102.3435,
+      "step": 238
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.886178861788618e-05,
+      "loss": 0.8023,
+      "step": 239
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.9024390243902444e-05,
+      "loss": 102.9021,
+      "step": 240
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.91869918699187e-05,
+      "loss": 0.7648,
+      "step": 241
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.934959349593496e-05,
+      "loss": 0.7504,
+      "step": 242
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.951219512195122e-05,
+      "loss": 100.1067,
+      "step": 243
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.967479674796748e-05,
+      "loss": 100.1942,
+      "step": 244
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.983739837398374e-05,
+      "loss": 0.818,
+      "step": 245
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4e-05,
+      "loss": 0.7334,
+      "step": 246
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.016260162601626e-05,
+      "loss": 0.8073,
+      "step": 247
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0325203252032525e-05,
+      "loss": 0.7548,
+      "step": 248
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0487804878048786e-05,
+      "loss": 107.6775,
+      "step": 249
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.065040650406504e-05,
+      "loss": 105.097,
+      "step": 250
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.08130081300813e-05,
+      "loss": 96.3359,
+      "step": 251
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.097560975609756e-05,
+      "loss": 99.4294,
+      "step": 252
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.1138211382113824e-05,
+      "loss": 0.8032,
+      "step": 253
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.1300813008130085e-05,
+      "loss": 0.8291,
+      "step": 254
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.146341463414634e-05,
+      "loss": 102.2682,
+      "step": 255
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.162601626016261e-05,
+      "loss": 0.7929,
+      "step": 256
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.178861788617887e-05,
+      "loss": 0.8092,
+      "step": 257
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.195121951219512e-05,
+      "loss": 0.7997,
+      "step": 258
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.211382113821138e-05,
+      "loss": 0.7753,
+      "step": 259
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.2276422764227644e-05,
+      "loss": 102.0557,
+      "step": 260
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.2439024390243905e-05,
+      "loss": 97.2704,
+      "step": 261
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.2601626016260166e-05,
+      "loss": 98.6474,
+      "step": 262
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.276422764227642e-05,
+      "loss": 99.3268,
+      "step": 263
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.292682926829268e-05,
+      "loss": 0.7572,
+      "step": 264
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.308943089430895e-05,
+      "loss": 0.8058,
+      "step": 265
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.3252032520325204e-05,
+      "loss": 0.8054,
+      "step": 266
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.3414634146341465e-05,
+      "loss": 102.8061,
+      "step": 267
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.3577235772357726e-05,
+      "loss": 0.756,
+      "step": 268
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.373983739837399e-05,
+      "loss": 99.3441,
+      "step": 269
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.390243902439025e-05,
+      "loss": 96.2682,
+      "step": 270
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.406504065040651e-05,
+      "loss": 0.8074,
+      "step": 271
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.422764227642276e-05,
+      "loss": 93.4931,
+      "step": 272
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4390243902439024e-05,
+      "loss": 92.9995,
+      "step": 273
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.455284552845529e-05,
+      "loss": 0.7548,
+      "step": 274
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.4715447154471546e-05,
+      "loss": 102.809,
+      "step": 275
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.487804878048781e-05,
+      "loss": 0.7649,
+      "step": 276
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.504065040650406e-05,
+      "loss": 97.6398,
+      "step": 277
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.520325203252033e-05,
+      "loss": 96.8278,
+      "step": 278
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.536585365853659e-05,
+      "loss": 0.8216,
+      "step": 279
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.5528455284552844e-05,
+      "loss": 0.7704,
+      "step": 280
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.5691056910569105e-05,
+      "loss": 114.6199,
+      "step": 281
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.585365853658537e-05,
+      "loss": 126.8513,
+      "step": 282
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.601626016260163e-05,
+      "loss": 115.2029,
+      "step": 283
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.617886178861789e-05,
+      "loss": 0.8075,
+      "step": 284
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.634146341463415e-05,
+      "loss": 0.7842,
+      "step": 285
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.650406504065041e-05,
+      "loss": 0.8234,
+      "step": 286
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.666666666666667e-05,
+      "loss": 92.2613,
+      "step": 287
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.682926829268293e-05,
+      "loss": 0.7525,
+      "step": 288
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.699186991869919e-05,
+      "loss": 0.8349,
+      "step": 289
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.715447154471545e-05,
+      "loss": 95.4324,
+      "step": 290
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.731707317073171e-05,
+      "loss": 94.6752,
+      "step": 291
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.747967479674797e-05,
+      "loss": 93.4538,
+      "step": 292
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.764227642276423e-05,
+      "loss": 88.3808,
+      "step": 293
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.7804878048780485e-05,
+      "loss": 0.7759,
+      "step": 294
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.796747967479675e-05,
+      "loss": 0.7853,
+      "step": 295
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.8130081300813014e-05,
+      "loss": 93.7523,
+      "step": 296
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.829268292682927e-05,
+      "loss": 89.4954,
+      "step": 297
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.845528455284553e-05,
+      "loss": 0.8177,
+      "step": 298
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.861788617886179e-05,
+      "loss": 92.8814,
+      "step": 299
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.878048780487805e-05,
+      "loss": 0.8233,
+      "step": 300
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.894308943089431e-05,
+      "loss": 0.7547,
+      "step": 301
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.910569105691057e-05,
+      "loss": 0.7624,
+      "step": 302
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.926829268292683e-05,
+      "loss": 0.6851,
+      "step": 303
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9430894308943096e-05,
+      "loss": 124.7233,
+      "step": 304
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.959349593495935e-05,
+      "loss": 0.7243,
+      "step": 305
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.975609756097561e-05,
+      "loss": 99.4604,
+      "step": 306
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.991869918699187e-05,
+      "loss": 92.9601,
+      "step": 307
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.008130081300813e-05,
+      "loss": 0.8383,
+      "step": 308
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.0243902439024394e-05,
+      "loss": 96.6387,
+      "step": 309
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.040650406504065e-05,
+      "loss": 0.8049,
+      "step": 310
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.0569105691056916e-05,
+      "loss": 95.5992,
+      "step": 311
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.073170731707318e-05,
+      "loss": 96.9858,
+      "step": 312
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.089430894308943e-05,
+      "loss": 96.4951,
+      "step": 313
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.10569105691057e-05,
+      "loss": 0.7652,
+      "step": 314
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.121951219512195e-05,
+      "loss": 98.1678,
+      "step": 315
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.1382113821138214e-05,
+      "loss": 99.0109,
+      "step": 316
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.154471544715448e-05,
+      "loss": 0.7824,
+      "step": 317
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.1707317073170736e-05,
+      "loss": 99.9446,
+      "step": 318
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.186991869918699e-05,
+      "loss": 92.0011,
+      "step": 319
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.203252032520326e-05,
+      "loss": 0.748,
+      "step": 320
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.219512195121952e-05,
+      "loss": 0.7385,
+      "step": 321
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.2357723577235774e-05,
+      "loss": 102.6258,
+      "step": 322
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.252032520325203e-05,
+      "loss": 102.5765,
+      "step": 323
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.2682926829268296e-05,
+      "loss": 0.758,
+      "step": 324
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.284552845528456e-05,
+      "loss": 0.7688,
+      "step": 325
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.300813008130081e-05,
+      "loss": 91.4879,
+      "step": 326
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.317073170731708e-05,
+      "loss": 0.7621,
+      "step": 327
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.333333333333333e-05,
+      "loss": 88.9684,
+      "step": 328
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.3495934959349594e-05,
+      "loss": 89.8101,
+      "step": 329
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.365853658536586e-05,
+      "loss": 89.8818,
+      "step": 330
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.3821138211382116e-05,
+      "loss": 93.2369,
+      "step": 331
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.398373983739837e-05,
+      "loss": 0.7788,
+      "step": 332
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.414634146341464e-05,
+      "loss": 0.8062,
+      "step": 333
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.43089430894309e-05,
+      "loss": 88.0875,
+      "step": 334
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.4471544715447154e-05,
+      "loss": 0.7436,
+      "step": 335
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.463414634146342e-05,
+      "loss": 0.784,
+      "step": 336
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.4796747967479676e-05,
+      "loss": 0.7236,
+      "step": 337
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.495934959349594e-05,
+      "loss": 91.4777,
+      "step": 338
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.5121951219512205e-05,
+      "loss": 0.7528,
+      "step": 339
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.528455284552846e-05,
+      "loss": 96.5415,
+      "step": 340
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.544715447154471e-05,
+      "loss": 85.9198,
+      "step": 341
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.560975609756098e-05,
+      "loss": 85.5737,
+      "step": 342
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.577235772357724e-05,
+      "loss": 84.9107,
+      "step": 343
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.5934959349593496e-05,
+      "loss": 87.4296,
+      "step": 344
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.6097560975609764e-05,
+      "loss": 87.2639,
+      "step": 345
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.626016260162602e-05,
+      "loss": 0.7652,
+      "step": 346
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.642276422764228e-05,
+      "loss": 0.7112,
+      "step": 347
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.6585365853658533e-05,
+      "loss": 0.8042,
+      "step": 348
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.67479674796748e-05,
+      "loss": 93.5885,
+      "step": 349
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.6910569105691056e-05,
+      "loss": 88.7945,
+      "step": 350
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.7073170731707317e-05,
+      "loss": 0.7957,
+      "step": 351
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.7235772357723584e-05,
+      "loss": 0.766,
+      "step": 352
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.739837398373984e-05,
+      "loss": 89.8735,
+      "step": 353
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.756097560975609e-05,
+      "loss": 0.8005,
+      "step": 354
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.772357723577236e-05,
+      "loss": 0.7936,
+      "step": 355
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.788617886178862e-05,
+      "loss": 90.8312,
+      "step": 356
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.8048780487804876e-05,
+      "loss": 0.7819,
+      "step": 357
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.8211382113821144e-05,
+      "loss": 0.7265,
+      "step": 358
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.83739837398374e-05,
+      "loss": 0.7572,
+      "step": 359
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.853658536585366e-05,
+      "loss": 0.7974,
+      "step": 360
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.869918699186993e-05,
+      "loss": 0.7308,
+      "step": 361
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.886178861788618e-05,
+      "loss": 91.8337,
+      "step": 362
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.902439024390244e-05,
+      "loss": 83.2892,
+      "step": 363
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.91869918699187e-05,
+      "loss": 0.7765,
+      "step": 364
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.9349593495934964e-05,
+      "loss": 89.2726,
+      "step": 365
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.951219512195122e-05,
+      "loss": 0.7124,
+      "step": 366
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.9674796747967486e-05,
+      "loss": 0.8091,
+      "step": 367
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 5.983739837398375e-05,
+      "loss": 0.7576,
+      "step": 368
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6e-05,
+      "loss": 83.0976,
+      "step": 369
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.016260162601627e-05,
+      "loss": 85.6402,
+      "step": 370
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.0325203252032524e-05,
+      "loss": 85.2141,
+      "step": 371
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.0487804878048785e-05,
+      "loss": 79.2895,
+      "step": 372
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.065040650406504e-05,
+      "loss": 0.746,
+      "step": 373
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.081300813008131e-05,
+      "loss": 0.7861,
+      "step": 374
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.097560975609756e-05,
+      "loss": 86.6769,
+      "step": 375
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.113821138211382e-05,
+      "loss": 91.8509,
+      "step": 376
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.130081300813008e-05,
+      "loss": 89.7941,
+      "step": 377
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.146341463414634e-05,
+      "loss": 84.5515,
+      "step": 378
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.16260162601626e-05,
+      "loss": 0.7942,
+      "step": 379
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.178861788617887e-05,
+      "loss": 0.7441,
+      "step": 380
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.195121951219513e-05,
+      "loss": 93.8158,
+      "step": 381
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.211382113821137e-05,
+      "loss": 0.7584,
+      "step": 382
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.227642276422765e-05,
+      "loss": 0.7251,
+      "step": 383
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.243902439024391e-05,
+      "loss": 90.5196,
+      "step": 384
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.260162601626016e-05,
+      "loss": 86.7979,
+      "step": 385
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.276422764227643e-05,
+      "loss": 0.8103,
+      "step": 386
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.292682926829268e-05,
+      "loss": 88.5195,
+      "step": 387
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.308943089430894e-05,
+      "loss": 87.0141,
+      "step": 388
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.325203252032522e-05,
+      "loss": 0.7731,
+      "step": 389
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.341463414634146e-05,
+      "loss": 87.9938,
+      "step": 390
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.357723577235772e-05,
+      "loss": 0.7748,
+      "step": 391
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.373983739837398e-05,
+      "loss": 90.5285,
+      "step": 392
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.390243902439025e-05,
+      "loss": 0.7968,
+      "step": 393
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.406504065040651e-05,
+      "loss": 96.7769,
+      "step": 394
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.422764227642277e-05,
+      "loss": 0.7492,
+      "step": 395
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.439024390243903e-05,
+      "loss": 85.1512,
+      "step": 396
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.455284552845529e-05,
+      "loss": 86.5623,
+      "step": 397
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.471544715447154e-05,
+      "loss": 83.268,
+      "step": 398
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.487804878048781e-05,
+      "loss": 87.6074,
+      "step": 399
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.504065040650407e-05,
+      "loss": 84.681,
+      "step": 400
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.520325203252032e-05,
+      "loss": 0.7393,
+      "step": 401
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.53658536585366e-05,
+      "loss": 86.4568,
+      "step": 402
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.552845528455284e-05,
+      "loss": 0.7905,
+      "step": 403
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.56910569105691e-05,
+      "loss": 85.2266,
+      "step": 404
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.585365853658538e-05,
+      "loss": 0.7555,
+      "step": 405
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.601626016260163e-05,
+      "loss": 89.1848,
+      "step": 406
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.617886178861789e-05,
+      "loss": 82.1218,
+      "step": 407
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.634146341463415e-05,
+      "loss": 0.764,
+      "step": 408
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.650406504065041e-05,
+      "loss": 80.4691,
+      "step": 409
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.666666666666667e-05,
+      "loss": 83.9419,
+      "step": 410
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.682926829268293e-05,
+      "loss": 0.8254,
+      "step": 411
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.699186991869919e-05,
+      "loss": 82.7363,
+      "step": 412
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.715447154471545e-05,
+      "loss": 87.4447,
+      "step": 413
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.731707317073171e-05,
+      "loss": 77.8148,
+      "step": 414
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.747967479674798e-05,
+      "loss": 83.1415,
+      "step": 415
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.764227642276422e-05,
+      "loss": 86.2474,
+      "step": 416
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.78048780487805e-05,
+      "loss": 82.0709,
+      "step": 417
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.796747967479676e-05,
+      "loss": 0.698,
+      "step": 418
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.8130081300813e-05,
+      "loss": 81.7057,
+      "step": 419
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.829268292682928e-05,
+      "loss": 0.8172,
+      "step": 420
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.845528455284553e-05,
+      "loss": 0.7587,
+      "step": 421
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.861788617886179e-05,
+      "loss": 0.7949,
+      "step": 422
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.878048780487805e-05,
+      "loss": 86.3426,
+      "step": 423
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.894308943089431e-05,
+      "loss": 0.7867,
+      "step": 424
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.910569105691057e-05,
+      "loss": 0.8162,
+      "step": 425
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.926829268292683e-05,
+      "loss": 0.7821,
+      "step": 426
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.94308943089431e-05,
+      "loss": 83.1586,
+      "step": 427
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.959349593495935e-05,
+      "loss": 0.8211,
+      "step": 428
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.97560975609756e-05,
+      "loss": 79.9211,
+      "step": 429
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 6.991869918699188e-05,
+      "loss": 0.7855,
+      "step": 430
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.008130081300814e-05,
+      "loss": 89.2547,
+      "step": 431
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.024390243902439e-05,
+      "loss": 84.1555,
+      "step": 432
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.040650406504066e-05,
+      "loss": 83.2166,
+      "step": 433
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.056910569105691e-05,
+      "loss": 0.7798,
+      "step": 434
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.073170731707317e-05,
+      "loss": 0.7675,
+      "step": 435
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.089430894308944e-05,
+      "loss": 0.7399,
+      "step": 436
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.105691056910569e-05,
+      "loss": 112.9201,
+      "step": 437
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.121951219512195e-05,
+      "loss": 0.7758,
+      "step": 438
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.138211382113821e-05,
+      "loss": 89.0947,
+      "step": 439
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.154471544715447e-05,
+      "loss": 0.7137,
+      "step": 440
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.170731707317073e-05,
+      "loss": 82.2047,
+      "step": 441
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.1869918699187e-05,
+      "loss": 80.5261,
+      "step": 442
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.203252032520326e-05,
+      "loss": 0.8079,
+      "step": 443
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.219512195121952e-05,
+      "loss": 0.7559,
+      "step": 444
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.235772357723578e-05,
+      "loss": 0.798,
+      "step": 445
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.252032520325204e-05,
+      "loss": 97.4162,
+      "step": 446
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.26829268292683e-05,
+      "loss": 89.6575,
+      "step": 447
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.284552845528456e-05,
+      "loss": 79.6773,
+      "step": 448
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.300813008130082e-05,
+      "loss": 0.7629,
+      "step": 449
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.317073170731707e-05,
+      "loss": 0.797,
+      "step": 450
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.333333333333333e-05,
+      "loss": 0.8088,
+      "step": 451
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.34959349593496e-05,
+      "loss": 89.1109,
+      "step": 452
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.365853658536585e-05,
+      "loss": 0.7463,
+      "step": 453
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.382113821138211e-05,
+      "loss": 96.8861,
+      "step": 454
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.398373983739838e-05,
+      "loss": 85.554,
+      "step": 455
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.414634146341464e-05,
+      "loss": 86.7738,
+      "step": 456
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.43089430894309e-05,
+      "loss": 0.7934,
+      "step": 457
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.447154471544716e-05,
+      "loss": 97.9769,
+      "step": 458
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.463414634146342e-05,
+      "loss": 0.7608,
+      "step": 459
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.479674796747968e-05,
+      "loss": 148.7433,
+      "step": 460
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.495934959349594e-05,
+      "loss": 0.8156,
+      "step": 461
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.51219512195122e-05,
+      "loss": 0.7681,
+      "step": 462
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.528455284552845e-05,
+      "loss": 114.4687,
+      "step": 463
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.544715447154472e-05,
+      "loss": 89.9569,
+      "step": 464
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.560975609756099e-05,
+      "loss": 0.7789,
+      "step": 465
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.577235772357723e-05,
+      "loss": 82.9912,
+      "step": 466
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.593495934959351e-05,
+      "loss": 87.7538,
+      "step": 467
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.609756097560976e-05,
+      "loss": 82.8786,
+      "step": 468
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.626016260162602e-05,
+      "loss": 0.7734,
+      "step": 469
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.642276422764229e-05,
+      "loss": 0.8298,
+      "step": 470
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.658536585365854e-05,
+      "loss": 84.391,
+      "step": 471
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.67479674796748e-05,
+      "loss": 0.7542,
+      "step": 472
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.691056910569106e-05,
+      "loss": 0.7777,
+      "step": 473
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.707317073170732e-05,
+      "loss": 0.7889,
+      "step": 474
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.723577235772358e-05,
+      "loss": 81.3893,
+      "step": 475
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.739837398373983e-05,
+      "loss": 0.7454,
+      "step": 476
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.75609756097561e-05,
+      "loss": 80.9911,
+      "step": 477
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.772357723577237e-05,
+      "loss": 80.0635,
+      "step": 478
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.788617886178861e-05,
+      "loss": 75.1018,
+      "step": 479
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.804878048780489e-05,
+      "loss": 0.7941,
+      "step": 480
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.821138211382114e-05,
+      "loss": 84.5774,
+      "step": 481
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.83739837398374e-05,
+      "loss": 79.5545,
+      "step": 482
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.853658536585367e-05,
+      "loss": 0.7935,
+      "step": 483
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.869918699186992e-05,
+      "loss": 79.654,
+      "step": 484
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.886178861788618e-05,
+      "loss": 79.9808,
+      "step": 485
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.902439024390244e-05,
+      "loss": 79.7054,
+      "step": 486
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.91869918699187e-05,
+      "loss": 79.1971,
+      "step": 487
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.934959349593496e-05,
+      "loss": 0.7232,
+      "step": 488
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.951219512195122e-05,
+      "loss": 0.791,
+      "step": 489
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.967479674796748e-05,
+      "loss": 77.8385,
+      "step": 490
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 7.983739837398375e-05,
+      "loss": 0.8009,
+      "step": 491
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8e-05,
+      "loss": 0.7779,
+      "step": 492
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.016260162601627e-05,
+      "loss": 73.7272,
+      "step": 493
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.032520325203252e-05,
+      "loss": 0.8412,
+      "step": 494
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.048780487804879e-05,
+      "loss": 0.7899,
+      "step": 495
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.065040650406505e-05,
+      "loss": 0.825,
+      "step": 496
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.08130081300813e-05,
+      "loss": 103.2765,
+      "step": 497
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.097560975609757e-05,
+      "loss": 0.7762,
+      "step": 498
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.113821138211382e-05,
+      "loss": 0.7434,
+      "step": 499
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.130081300813008e-05,
+      "loss": 79.4886,
+      "step": 500
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.146341463414634e-05,
+      "loss": 79.8968,
+      "step": 501
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.16260162601626e-05,
+      "loss": 74.9619,
+      "step": 502
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.178861788617886e-05,
+      "loss": 78.6073,
+      "step": 503
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.195121951219513e-05,
+      "loss": 76.924,
+      "step": 504
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.211382113821139e-05,
+      "loss": 73.6012,
+      "step": 505
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.227642276422765e-05,
+      "loss": 79.3721,
+      "step": 506
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.243902439024391e-05,
+      "loss": 0.7466,
+      "step": 507
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.260162601626017e-05,
+      "loss": 76.2062,
+      "step": 508
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.276422764227643e-05,
+      "loss": 0.7425,
+      "step": 509
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.292682926829268e-05,
+      "loss": 79.2716,
+      "step": 510
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.308943089430895e-05,
+      "loss": 81.6172,
+      "step": 511
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 8.325203252032521e-05,
+      "loss": 77.1114,
+      "step": 512
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.341463414634146e-05,
+      "loss": 0.7478,
+      "step": 513
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.357723577235774e-05,
+      "loss": 76.329,
+      "step": 514
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.373983739837398e-05,
+      "loss": 0.7648,
+      "step": 515
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.390243902439024e-05,
+      "loss": 0.8183,
+      "step": 516
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.406504065040652e-05,
+      "loss": 78.5997,
+      "step": 517
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.422764227642277e-05,
+      "loss": 0.773,
+      "step": 518
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.439024390243903e-05,
+      "loss": 84.1783,
+      "step": 519
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.455284552845529e-05,
+      "loss": 0.7711,
+      "step": 520
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.471544715447155e-05,
+      "loss": 0.7728,
+      "step": 521
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.487804878048781e-05,
+      "loss": 78.2558,
+      "step": 522
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.504065040650407e-05,
+      "loss": 78.9152,
+      "step": 523
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.520325203252033e-05,
+      "loss": 0.8354,
+      "step": 524
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.53658536585366e-05,
+      "loss": 104.3187,
+      "step": 525
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.552845528455284e-05,
+      "loss": 120.6488,
+      "step": 526
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.569105691056912e-05,
+      "loss": 88.0983,
+      "step": 527
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.585365853658536e-05,
+      "loss": 75.2004,
+      "step": 528
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.601626016260162e-05,
+      "loss": 0.8468,
+      "step": 529
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.61788617886179e-05,
+      "loss": 0.7429,
+      "step": 530
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.634146341463415e-05,
+      "loss": 85.4469,
+      "step": 531
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.650406504065041e-05,
+      "loss": 89.3407,
+      "step": 532
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.666666666666667e-05,
+      "loss": 87.8479,
+      "step": 533
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.682926829268293e-05,
+      "loss": 88.6581,
+      "step": 534
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.699186991869919e-05,
+      "loss": 0.7815,
+      "step": 535
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.715447154471545e-05,
+      "loss": 92.7441,
+      "step": 536
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.731707317073171e-05,
+      "loss": 90.4961,
+      "step": 537
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.747967479674797e-05,
+      "loss": 0.8271,
+      "step": 538
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.764227642276423e-05,
+      "loss": 80.0083,
+      "step": 539
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.78048780487805e-05,
+      "loss": 72.6806,
+      "step": 540
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.796747967479674e-05,
+      "loss": 0.7518,
+      "step": 541
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.813008130081302e-05,
+      "loss": 0.8039,
+      "step": 542
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.829268292682928e-05,
+      "loss": 76.4484,
+      "step": 543
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.845528455284553e-05,
+      "loss": 0.7864,
+      "step": 544
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.86178861788618e-05,
+      "loss": 74.9542,
+      "step": 545
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.878048780487805e-05,
+      "loss": 0.799,
+      "step": 546
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.894308943089431e-05,
+      "loss": 82.5349,
+      "step": 547
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.910569105691058e-05,
+      "loss": 77.0917,
+      "step": 548
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.926829268292683e-05,
+      "loss": 75.9369,
+      "step": 549
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.943089430894309e-05,
+      "loss": 78.4129,
+      "step": 550
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.959349593495935e-05,
+      "loss": 0.8285,
+      "step": 551
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.975609756097561e-05,
+      "loss": 0.7155,
+      "step": 552
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 8.991869918699188e-05,
+      "loss": 0.7733,
+      "step": 553
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.008130081300812e-05,
+      "loss": 0.7402,
+      "step": 554
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.02439024390244e-05,
+      "loss": 0.7973,
+      "step": 555
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.040650406504066e-05,
+      "loss": 90.6369,
+      "step": 556
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.05691056910569e-05,
+      "loss": 0.7753,
+      "step": 557
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.073170731707318e-05,
+      "loss": 0.7975,
+      "step": 558
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.089430894308944e-05,
+      "loss": 77.8253,
+      "step": 559
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.105691056910569e-05,
+      "loss": 0.7969,
+      "step": 560
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.121951219512196e-05,
+      "loss": 0.7549,
+      "step": 561
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.138211382113821e-05,
+      "loss": 0.794,
+      "step": 562
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.154471544715447e-05,
+      "loss": 0.7528,
+      "step": 563
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.170731707317075e-05,
+      "loss": 0.7743,
+      "step": 564
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.1869918699187e-05,
+      "loss": 0.7209,
+      "step": 565
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.203252032520325e-05,
+      "loss": 87.4018,
+      "step": 566
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.219512195121952e-05,
+      "loss": 75.9744,
+      "step": 567
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.235772357723578e-05,
+      "loss": 0.7582,
+      "step": 568
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.252032520325204e-05,
+      "loss": 85.0537,
+      "step": 569
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.26829268292683e-05,
+      "loss": 75.0047,
+      "step": 570
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.284552845528456e-05,
+      "loss": 70.2187,
+      "step": 571
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.300813008130082e-05,
+      "loss": 69.1689,
+      "step": 572
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.317073170731708e-05,
+      "loss": 0.7868,
+      "step": 573
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.333333333333334e-05,
+      "loss": 0.8024,
+      "step": 574
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.349593495934959e-05,
+      "loss": 0.8134,
+      "step": 575
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.365853658536587e-05,
+      "loss": 81.5576,
+      "step": 576
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.382113821138213e-05,
+      "loss": 74.127,
+      "step": 577
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.398373983739837e-05,
+      "loss": 75.9421,
+      "step": 578
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.414634146341463e-05,
+      "loss": 75.9159,
+      "step": 579
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.43089430894309e-05,
+      "loss": 0.7762,
+      "step": 580
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.447154471544716e-05,
+      "loss": 87.354,
+      "step": 581
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.463414634146342e-05,
+      "loss": 88.8371,
+      "step": 582
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.479674796747968e-05,
+      "loss": 0.7482,
+      "step": 583
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.495934959349594e-05,
+      "loss": 77.3781,
+      "step": 584
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.51219512195122e-05,
+      "loss": 70.6823,
+      "step": 585
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.528455284552846e-05,
+      "loss": 74.2894,
+      "step": 586
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.544715447154472e-05,
+      "loss": 0.7314,
+      "step": 587
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.560975609756097e-05,
+      "loss": 75.8039,
+      "step": 588
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.577235772357725e-05,
+      "loss": 78.3846,
+      "step": 589
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.59349593495935e-05,
+      "loss": 71.9321,
+      "step": 590
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.609756097560975e-05,
+      "loss": 70.052,
+      "step": 591
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.626016260162603e-05,
+      "loss": 68.6095,
+      "step": 592
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.642276422764228e-05,
+      "loss": 0.7886,
+      "step": 593
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.658536585365854e-05,
+      "loss": 0.8576,
+      "step": 594
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.674796747967481e-05,
+      "loss": 84.74,
+      "step": 595
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.691056910569106e-05,
+      "loss": 81.5393,
+      "step": 596
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.707317073170732e-05,
+      "loss": 0.787,
+      "step": 597
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.723577235772358e-05,
+      "loss": 75.5274,
+      "step": 598
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.739837398373984e-05,
+      "loss": 0.8339,
+      "step": 599
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.75609756097561e-05,
+      "loss": 71.4407,
+      "step": 600
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.772357723577236e-05,
+      "loss": 68.9694,
+      "step": 601
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.788617886178862e-05,
+      "loss": 71.2264,
+      "step": 602
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.804878048780489e-05,
+      "loss": 0.7556,
+      "step": 603
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.821138211382113e-05,
+      "loss": 74.4697,
+      "step": 604
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.837398373983741e-05,
+      "loss": 0.81,
+      "step": 605
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.853658536585366e-05,
+      "loss": 71.1234,
+      "step": 606
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.869918699186992e-05,
+      "loss": 72.6654,
+      "step": 607
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.886178861788619e-05,
+      "loss": 0.7418,
+      "step": 608
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.902439024390244e-05,
+      "loss": 0.8008,
+      "step": 609
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.91869918699187e-05,
+      "loss": 74.4627,
+      "step": 610
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.934959349593496e-05,
+      "loss": 72.2311,
+      "step": 611
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.951219512195122e-05,
+      "loss": 63.0526,
+      "step": 612
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.967479674796748e-05,
+      "loss": 73.6538,
+      "step": 613
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.983739837398374e-05,
+      "loss": 0.7387,
+      "step": 614
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0001,
+      "loss": 72.609,
+      "step": 615
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999999937599429e-05,
+      "loss": 0.7662,
+      "step": 616
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99999975039772e-05,
+      "loss": 69.8877,
+      "step": 617
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999999438394872e-05,
+      "loss": 0.7731,
+      "step": 618
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999999001590899e-05,
+      "loss": 0.7589,
+      "step": 619
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999998439985807e-05,
+      "loss": 0.7775,
+      "step": 620
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999997753579614e-05,
+      "loss": 68.5838,
+      "step": 621
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999996942372336e-05,
+      "loss": 0.7328,
+      "step": 622
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99999600636399e-05,
+      "loss": 0.8246,
+      "step": 623
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999994945554605e-05,
+      "loss": 0.7108,
+      "step": 624
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999993759944203e-05,
+      "loss": 66.3509,
+      "step": 625
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999992449532815e-05,
+      "loss": 70.4283,
+      "step": 626
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999991014320474e-05,
+      "loss": 63.4848,
+      "step": 627
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999989454307217e-05,
+      "loss": 0.7506,
+      "step": 628
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99998776949308e-05,
+      "loss": 62.1415,
+      "step": 629
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999985959878106e-05,
+      "loss": 0.8029,
+      "step": 630
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999984025462342e-05,
+      "loss": 0.8279,
+      "step": 631
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999981966245835e-05,
+      "loss": 71.924,
+      "step": 632
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999979782228636e-05,
+      "loss": 0.751,
+      "step": 633
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.9999774734108e-05,
+      "loss": 66.6791,
+      "step": 634
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999975039792387e-05,
+      "loss": 0.7006,
+      "step": 635
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999972481373452e-05,
+      "loss": 66.2688,
+      "step": 636
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999969798154064e-05,
+      "loss": 0.7653,
+      "step": 637
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999966990134287e-05,
+      "loss": 66.5854,
+      "step": 638
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999964057314194e-05,
+      "loss": 62.122,
+      "step": 639
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999960999693856e-05,
+      "loss": 0.7614,
+      "step": 640
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999957817273349e-05,
+      "loss": 64.54,
+      "step": 641
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999954510052753e-05,
+      "loss": 0.7425,
+      "step": 642
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999951078032151e-05,
+      "loss": 67.3583,
+      "step": 643
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99994752121163e-05,
+      "loss": 65.4687,
+      "step": 644
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999943839591274e-05,
+      "loss": 61.9655,
+      "step": 645
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999940033171182e-05,
+      "loss": 62.6926,
+      "step": 646
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999936101951443e-05,
+      "loss": 0.7715,
+      "step": 647
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999932045932157e-05,
+      "loss": 0.7264,
+      "step": 648
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999927865113426e-05,
+      "loss": 68.7846,
+      "step": 649
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999923559495354e-05,
+      "loss": 0.7715,
+      "step": 650
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999919129078049e-05,
+      "loss": 67.6888,
+      "step": 651
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999914573861619e-05,
+      "loss": 0.7542,
+      "step": 652
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999909893846182e-05,
+      "loss": 69.816,
+      "step": 653
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999905089031852e-05,
+      "loss": 0.787,
+      "step": 654
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999900159418747e-05,
+      "loss": 0.7662,
+      "step": 655
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999895105006994e-05,
+      "loss": 0.8072,
+      "step": 656
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999889925796718e-05,
+      "loss": 0.7552,
+      "step": 657
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999884621788048e-05,
+      "loss": 113.8962,
+      "step": 658
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999879192981114e-05,
+      "loss": 0.7566,
+      "step": 659
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999873639376056e-05,
+      "loss": 86.1744,
+      "step": 660
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999867960973009e-05,
+      "loss": 0.7445,
+      "step": 661
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999862157772118e-05,
+      "loss": 75.9596,
+      "step": 662
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999856229773523e-05,
+      "loss": 0.759,
+      "step": 663
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999850176977376e-05,
+      "loss": 0.745,
+      "step": 664
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99984399938383e-05,
+      "loss": 75.4387,
+      "step": 665
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999837696993033e-05,
+      "loss": 75.9379,
+      "step": 666
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999831269805145e-05,
+      "loss": 0.7847,
+      "step": 667
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999824717820328e-05,
+      "loss": 0.7325,
+      "step": 668
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999818041038745e-05,
+      "loss": 0.7776,
+      "step": 669
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999811239460561e-05,
+      "loss": 0.7645,
+      "step": 670
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999804313085948e-05,
+      "loss": 69.8016,
+      "step": 671
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999797261915077e-05,
+      "loss": 0.7538,
+      "step": 672
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999790085948124e-05,
+      "loss": 0.7544,
+      "step": 673
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99978278518527e-05,
+      "loss": 0.7451,
+      "step": 674
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999775359626696e-05,
+      "loss": 64.1471,
+      "step": 675
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999767809272588e-05,
+      "loss": 0.796,
+      "step": 676
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999760134123132e-05,
+      "loss": 0.7131,
+      "step": 677
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999752334178523e-05,
+      "loss": 65.8035,
+      "step": 678
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999744409438954e-05,
+      "loss": 0.7896,
+      "step": 679
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999736359904623e-05,
+      "loss": 0.7977,
+      "step": 680
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999728185575729e-05,
+      "loss": 0.8111,
+      "step": 681
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999719886452481e-05,
+      "loss": 0.733,
+      "step": 682
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999711462535082e-05,
+      "loss": 71.4202,
+      "step": 683
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999702913823742e-05,
+      "loss": 0.7651,
+      "step": 684
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999694240318676e-05,
+      "loss": 61.0656,
+      "step": 685
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999685442020101e-05,
+      "loss": 0.7645,
+      "step": 686
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999676518928236e-05,
+      "loss": 61.2966,
+      "step": 687
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999667471043303e-05,
+      "loss": 69.9346,
+      "step": 688
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999658298365528e-05,
+      "loss": 0.7216,
+      "step": 689
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99964900089514e-05,
+      "loss": 67.764,
+      "step": 690
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999639578632373e-05,
+      "loss": 0.7911,
+      "step": 691
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999630031577459e-05,
+      "loss": 0.7834,
+      "step": 692
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99962035973064e-05,
+      "loss": 0.7563,
+      "step": 693
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999610563092153e-05,
+      "loss": 68.7251,
+      "step": 694
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999600641662245e-05,
+      "loss": 62.9454,
+      "step": 695
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999590595441164e-05,
+      "loss": 60.9159,
+      "step": 696
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99958042442916e-05,
+      "loss": 0.7239,
+      "step": 697
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999570128626488e-05,
+      "loss": 65.1601,
+      "step": 698
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999559708033401e-05,
+      "loss": 0.7314,
+      "step": 699
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999549162650166e-05,
+      "loss": 0.7718,
+      "step": 700
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999538492477041e-05,
+      "loss": 0.7718,
+      "step": 701
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999527697514293e-05,
+      "loss": 0.7484,
+      "step": 702
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999516777762192e-05,
+      "loss": 98.4959,
+      "step": 703
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999505733221011e-05,
+      "loss": 0.8125,
+      "step": 704
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999494563891025e-05,
+      "loss": 0.747,
+      "step": 705
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999483269772514e-05,
+      "loss": 0.7511,
+      "step": 706
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999471850865758e-05,
+      "loss": 0.7516,
+      "step": 707
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999460307171044e-05,
+      "loss": 0.7503,
+      "step": 708
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99944863868866e-05,
+      "loss": 83.5887,
+      "step": 709
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999436845418894e-05,
+      "loss": 79.472,
+      "step": 710
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999424927362045e-05,
+      "loss": 0.7573,
+      "step": 711
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999412884518409e-05,
+      "loss": 0.761,
+      "step": 712
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999400716888283e-05,
+      "loss": 76.037,
+      "step": 713
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999388424471977e-05,
+      "loss": 0.7286,
+      "step": 714
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999376007269791e-05,
+      "loss": 72.0486,
+      "step": 715
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99936346528204e-05,
+      "loss": 0.7682,
+      "step": 716
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.999350798509036e-05,
+      "loss": 65.4268,
+      "step": 717
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999338006951093e-05,
+      "loss": 0.7848,
+      "step": 718
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999325090608533e-05,
+      "loss": 62.4401,
+      "step": 719
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999312049481678e-05,
+      "loss": 64.47,
+      "step": 720
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99929888357085e-05,
+      "loss": 0.7698,
+      "step": 721
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999285592876382e-05,
+      "loss": 62.0365,
+      "step": 722
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999272177398604e-05,
+      "loss": 0.7644,
+      "step": 723
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999258637137851e-05,
+      "loss": 65.418,
+      "step": 724
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999244972094461e-05,
+      "loss": 0.777,
+      "step": 725
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999231182268776e-05,
+      "loss": 0.7946,
+      "step": 726
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999217267661139e-05,
+      "loss": 63.9624,
+      "step": 727
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999203228271897e-05,
+      "loss": 0.7715,
+      "step": 728
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.9991890641014e-05,
+      "loss": 60.5511,
+      "step": 729
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999174775150003e-05,
+      "loss": 63.1611,
+      "step": 730
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999160361418062e-05,
+      "loss": 0.7828,
+      "step": 731
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999145822905938e-05,
+      "loss": 64.5694,
+      "step": 732
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999131159613991e-05,
+      "loss": 62.2431,
+      "step": 733
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999116371542591e-05,
+      "loss": 63.1838,
+      "step": 734
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999101458692104e-05,
+      "loss": 0.7715,
+      "step": 735
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999086421062903e-05,
+      "loss": 0.8102,
+      "step": 736
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999071258655363e-05,
+      "loss": 57.1001,
+      "step": 737
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999055971469864e-05,
+      "loss": 59.1836,
+      "step": 738
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999040559506785e-05,
+      "loss": 56.7024,
+      "step": 739
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999025022766514e-05,
+      "loss": 60.5593,
+      "step": 740
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.999009361249437e-05,
+      "loss": 0.7908,
+      "step": 741
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998993574955944e-05,
+      "loss": 61.8612,
+      "step": 742
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998977663886428e-05,
+      "loss": 62.3995,
+      "step": 743
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998961628041291e-05,
+      "loss": 61.7024,
+      "step": 744
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99894546742093e-05,
+      "loss": 59.6545,
+      "step": 745
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998929182025747e-05,
+      "loss": 0.8134,
+      "step": 746
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998912771856152e-05,
+      "loss": 70.7325,
+      "step": 747
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998896236912553e-05,
+      "loss": 69.1981,
+      "step": 748
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998879577195361e-05,
+      "loss": 0.748,
+      "step": 749
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998862792704995e-05,
+      "loss": 0.7703,
+      "step": 750
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998845883441872e-05,
+      "loss": 64.7346,
+      "step": 751
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998828849406413e-05,
+      "loss": 0.7711,
+      "step": 752
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998811690599045e-05,
+      "loss": 0.8106,
+      "step": 753
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998794407020197e-05,
+      "loss": 58.4023,
+      "step": 754
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998776998670298e-05,
+      "loss": 0.7974,
+      "step": 755
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998759465549783e-05,
+      "loss": 0.7613,
+      "step": 756
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998741807659093e-05,
+      "loss": 0.793,
+      "step": 757
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998724024998664e-05,
+      "loss": 0.7342,
+      "step": 758
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998706117568944e-05,
+      "loss": 75.8958,
+      "step": 759
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998688085370375e-05,
+      "loss": 0.7625,
+      "step": 760
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998669928403412e-05,
+      "loss": 64.9445,
+      "step": 761
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998651646668507e-05,
+      "loss": 0.7638,
+      "step": 762
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998633240166113e-05,
+      "loss": 0.7932,
+      "step": 763
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998614708896694e-05,
+      "loss": 0.757,
+      "step": 764
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99859605286071e-05,
+      "loss": 0.7616,
+      "step": 765
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998577272058627e-05,
+      "loss": 65.6017,
+      "step": 766
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998558366490914e-05,
+      "loss": 65.3418,
+      "step": 767
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998539336158044e-05,
+      "loss": 65.2811,
+      "step": 768
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998520181060489e-05,
+      "loss": 0.7253,
+      "step": 769
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998500901198731e-05,
+      "loss": 0.7523,
+      "step": 770
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998481496573247e-05,
+      "loss": 0.7515,
+      "step": 771
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998461967184525e-05,
+      "loss": 128.3291,
+      "step": 772
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99844231303305e-05,
+      "loss": 0.7825,
+      "step": 773
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998422534119314e-05,
+      "loss": 159.3056,
+      "step": 774
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99840263044381e-05,
+      "loss": 108.8061,
+      "step": 775
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998382602007036e-05,
+      "loss": 69.7557,
+      "step": 776
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998362448809491e-05,
+      "loss": 53.9984,
+      "step": 777
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998342170851676e-05,
+      "loss": 0.8119,
+      "step": 778
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998321768134102e-05,
+      "loss": 0.7731,
+      "step": 779
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998301240657273e-05,
+      "loss": 0.7591,
+      "step": 780
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998280588421704e-05,
+      "loss": 77.835,
+      "step": 781
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998259811427913e-05,
+      "loss": 80.4308,
+      "step": 782
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998238909676413e-05,
+      "loss": 0.7502,
+      "step": 783
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998217883167728e-05,
+      "loss": 75.4623,
+      "step": 784
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998196731902386e-05,
+      "loss": 0.7752,
+      "step": 785
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998175455880912e-05,
+      "loss": 0.7612,
+      "step": 786
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998154055103837e-05,
+      "loss": 71.6736,
+      "step": 787
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998132529571693e-05,
+      "loss": 0.7318,
+      "step": 788
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998110879285023e-05,
+      "loss": 0.7428,
+      "step": 789
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998089104244363e-05,
+      "loss": 0.8076,
+      "step": 790
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998067204450258e-05,
+      "loss": 0.7852,
+      "step": 791
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998045179903255e-05,
+      "loss": 65.4545,
+      "step": 792
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998023030603903e-05,
+      "loss": 0.7394,
+      "step": 793
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.998000756552754e-05,
+      "loss": 0.7839,
+      "step": 794
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997978357750364e-05,
+      "loss": 61.5173,
+      "step": 795
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997955834197294e-05,
+      "loss": 57.5499,
+      "step": 796
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997933185894106e-05,
+      "loss": 58.3055,
+      "step": 797
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997910412841364e-05,
+      "loss": 0.7691,
+      "step": 798
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997887515039636e-05,
+      "loss": 0.7745,
+      "step": 799
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997864492489497e-05,
+      "loss": 57.4281,
+      "step": 800
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997841345191517e-05,
+      "loss": 59.2338,
+      "step": 801
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997818073146277e-05,
+      "loss": 0.7806,
+      "step": 802
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997794676354355e-05,
+      "loss": 0.6998,
+      "step": 803
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997771154816338e-05,
+      "loss": 0.7922,
+      "step": 804
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997747508532811e-05,
+      "loss": 0.8024,
+      "step": 805
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997723737504366e-05,
+      "loss": 0.8098,
+      "step": 806
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997699841731597e-05,
+      "loss": 0.8174,
+      "step": 807
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997675821215096e-05,
+      "loss": 0.7958,
+      "step": 808
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997651675955466e-05,
+      "loss": 71.4525,
+      "step": 809
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997627405953309e-05,
+      "loss": 69.6205,
+      "step": 810
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997603011209231e-05,
+      "loss": 65.7036,
+      "step": 811
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997578491723841e-05,
+      "loss": 62.9618,
+      "step": 812
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997553847497752e-05,
+      "loss": 61.1015,
+      "step": 813
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997529078531575e-05,
+      "loss": 0.7735,
+      "step": 814
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997504184825935e-05,
+      "loss": 0.8,
+      "step": 815
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997479166381445e-05,
+      "loss": 118.0717,
+      "step": 816
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997454023198736e-05,
+      "loss": 0.7439,
+      "step": 817
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997428755278435e-05,
+      "loss": 0.7706,
+      "step": 818
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99740336262117e-05,
+      "loss": 0.8097,
+      "step": 819
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997377845227576e-05,
+      "loss": 0.79,
+      "step": 820
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99735220309829e-05,
+      "loss": 0.7534,
+      "step": 821
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99732643623395e-05,
+      "loss": 127.7609,
+      "step": 822
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997300544635204e-05,
+      "loss": 95.9523,
+      "step": 823
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997274528302694e-05,
+      "loss": 0.8062,
+      "step": 824
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99724838723707e-05,
+      "loss": 0.7957,
+      "step": 825
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997222121438988e-05,
+      "loss": 66.6171,
+      "step": 826
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997195730909098e-05,
+      "loss": 59.615,
+      "step": 827
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997169215648062e-05,
+      "loss": 0.739,
+      "step": 828
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997142575656542e-05,
+      "loss": 62.8531,
+      "step": 829
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997115810935202e-05,
+      "loss": 0.7611,
+      "step": 830
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997088921484712e-05,
+      "loss": 0.6894,
+      "step": 831
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997061907305739e-05,
+      "loss": 0.746,
+      "step": 832
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997034768398961e-05,
+      "loss": 0.8054,
+      "step": 833
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.997007504765054e-05,
+      "loss": 0.7561,
+      "step": 834
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.9969801164047e-05,
+      "loss": 0.8239,
+      "step": 835
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99695260331858e-05,
+      "loss": 126.9888,
+      "step": 836
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996924965507381e-05,
+      "loss": 111.0056,
+      "step": 837
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996897202971796e-05,
+      "loss": 78.3833,
+      "step": 838
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996869315712517e-05,
+      "loss": 0.7538,
+      "step": 839
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996841303730236e-05,
+      "loss": 0.7478,
+      "step": 840
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996813167025656e-05,
+      "loss": 0.8696,
+      "step": 841
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996784905599478e-05,
+      "loss": 67.8119,
+      "step": 842
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996756519452408e-05,
+      "loss": 66.1996,
+      "step": 843
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996728008585155e-05,
+      "loss": 62.7335,
+      "step": 844
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99669937299843e-05,
+      "loss": 62.324,
+      "step": 845
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996670612692948e-05,
+      "loss": 58.3722,
+      "step": 846
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996641727669427e-05,
+      "loss": 64.6481,
+      "step": 847
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996612717928586e-05,
+      "loss": 60.4162,
+      "step": 848
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996583583471152e-05,
+      "loss": 62.3403,
+      "step": 849
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99655432429785e-05,
+      "loss": 0.7775,
+      "step": 850
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996524940409412e-05,
+      "loss": 60.1692,
+      "step": 851
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99649543180657e-05,
+      "loss": 0.7356,
+      "step": 852
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996465798490061e-05,
+      "loss": 0.7193,
+      "step": 853
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996436040460626e-05,
+      "loss": 66.703,
+      "step": 854
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996406157719005e-05,
+      "loss": 63.8849,
+      "step": 855
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996376150265946e-05,
+      "loss": 62.0058,
+      "step": 856
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996346018102198e-05,
+      "loss": 0.7725,
+      "step": 857
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996315761228511e-05,
+      "loss": 0.7802,
+      "step": 858
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996285379645644e-05,
+      "loss": 0.7823,
+      "step": 859
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996254873354354e-05,
+      "loss": 126.804,
+      "step": 860
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996224242355399e-05,
+      "loss": 0.7651,
+      "step": 861
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996193486649548e-05,
+      "loss": 183.3675,
+      "step": 862
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996162606237565e-05,
+      "loss": 137.3225,
+      "step": 863
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996131601120222e-05,
+      "loss": 95.892,
+      "step": 864
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996100471298295e-05,
+      "loss": 59.5294,
+      "step": 865
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99606921677256e-05,
+      "loss": 56.8655,
+      "step": 866
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996037837543795e-05,
+      "loss": 53.9704,
+      "step": 867
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.996006333612786e-05,
+      "loss": 0.7642,
+      "step": 868
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995974704980317e-05,
+      "loss": 0.7455,
+      "step": 869
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995942951647181e-05,
+      "loss": 0.7863,
+      "step": 870
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995911073614166e-05,
+      "loss": 65.6919,
+      "step": 871
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99587907088207e-05,
+      "loss": 0.7815,
+      "step": 872
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995846943451693e-05,
+      "loss": 0.751,
+      "step": 873
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995814691323835e-05,
+      "loss": 63.4796,
+      "step": 874
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995782314499302e-05,
+      "loss": 65.8398,
+      "step": 875
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995749812978901e-05,
+      "loss": 62.9557,
+      "step": 876
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995717186763444e-05,
+      "loss": 53.1321,
+      "step": 877
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995684435853746e-05,
+      "loss": 51.7837,
+      "step": 878
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995651560250622e-05,
+      "loss": 56.5262,
+      "step": 879
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995618559954898e-05,
+      "loss": 57.3765,
+      "step": 880
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995585434967392e-05,
+      "loss": 62.3419,
+      "step": 881
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995552185288933e-05,
+      "loss": 58.1936,
+      "step": 882
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995518810920351e-05,
+      "loss": 57.5677,
+      "step": 883
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995485311862477e-05,
+      "loss": 0.8128,
+      "step": 884
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995451688116151e-05,
+      "loss": 0.8165,
+      "step": 885
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995417939682209e-05,
+      "loss": 0.7387,
+      "step": 886
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995384066561494e-05,
+      "loss": 59.589,
+      "step": 887
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995350068754855e-05,
+      "loss": 57.6129,
+      "step": 888
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995315946263136e-05,
+      "loss": 0.7742,
+      "step": 889
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995281699087191e-05,
+      "loss": 0.8002,
+      "step": 890
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995247327227873e-05,
+      "loss": 60.1013,
+      "step": 891
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995212830686042e-05,
+      "loss": 61.0053,
+      "step": 892
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995178209462559e-05,
+      "loss": 0.7145,
+      "step": 893
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995143463558285e-05,
+      "loss": 0.7744,
+      "step": 894
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995108592974092e-05,
+      "loss": 0.796,
+      "step": 895
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995073597710847e-05,
+      "loss": 61.3352,
+      "step": 896
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995038477769425e-05,
+      "loss": 63.2612,
+      "step": 897
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.995003233150701e-05,
+      "loss": 65.3131,
+      "step": 898
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994967863855556e-05,
+      "loss": 57.5135,
+      "step": 899
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994932369884873e-05,
+      "loss": 0.7356,
+      "step": 900
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994896751239538e-05,
+      "loss": 55.8259,
+      "step": 901
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99486100792044e-05,
+      "loss": 0.7508,
+      "step": 902
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99482513992847e-05,
+      "loss": 56.6854,
+      "step": 903
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994789147264523e-05,
+      "loss": 0.7481,
+      "step": 904
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.9947530299295e-05,
+      "loss": 0.7938,
+      "step": 905
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.9947167879243e-05,
+      "loss": 56.7883,
+      "step": 906
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99468042124983e-05,
+      "loss": 58.1589,
+      "step": 907
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994643929906993e-05,
+      "loss": 0.7913,
+      "step": 908
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994607313896707e-05,
+      "loss": 54.2129,
+      "step": 909
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994570573219879e-05,
+      "loss": 49.9096,
+      "step": 910
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994533707877431e-05,
+      "loss": 0.8067,
+      "step": 911
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994496717870281e-05,
+      "loss": 0.7533,
+      "step": 912
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994459603199351e-05,
+      "loss": 0.7797,
+      "step": 913
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.99442236386557e-05,
+      "loss": 0.7698,
+      "step": 914
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994384999869867e-05,
+      "loss": 71.5847,
+      "step": 915
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994347511213174e-05,
+      "loss": 64.211,
+      "step": 916
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994309897896425e-05,
+      "loss": 62.4127,
+      "step": 917
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994272159920563e-05,
+      "loss": 56.0463,
+      "step": 918
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994234297286526e-05,
+      "loss": 0.7715,
+      "step": 919
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994196309995261e-05,
+      "loss": 0.8036,
+      "step": 920
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994158198047717e-05,
+      "loss": 63.9622,
+      "step": 921
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.994119961444841e-05,
+      "loss": 0.7647,
+      "step": 922
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.994081600187593e-05,
+      "loss": 0.739,
+      "step": 923
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.994043114276928e-05,
+      "loss": 0.7884,
+      "step": 924
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.994004503713805e-05,
+      "loss": 68.3888,
+      "step": 925
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.99396576849919e-05,
+      "loss": 56.3071,
+      "step": 926
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.99392690863405e-05,
+      "loss": 58.1878,
+      "step": 927
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993887924119353e-05,
+      "loss": 56.0583,
+      "step": 928
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993848814956072e-05,
+      "loss": 0.7345,
+      "step": 929
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993809581145188e-05,
+      "loss": 64.9174,
+      "step": 930
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993770222687673e-05,
+      "loss": 67.8347,
+      "step": 931
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993730739584515e-05,
+      "loss": 0.73,
+      "step": 932
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993691131836697e-05,
+      "loss": 0.7849,
+      "step": 933
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993651399445209e-05,
+      "loss": 58.304,
+      "step": 934
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993611542411041e-05,
+      "loss": 0.7528,
+      "step": 935
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993571560735188e-05,
+      "loss": 62.1101,
+      "step": 936
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.99353145441865e-05,
+      "loss": 0.7094,
+      "step": 937
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993491223462426e-05,
+      "loss": 55.6776,
+      "step": 938
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993450867867521e-05,
+      "loss": 0.7972,
+      "step": 939
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993410387634943e-05,
+      "loss": 0.817,
+      "step": 940
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993369782765701e-05,
+      "loss": 0.8102,
+      "step": 941
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993329053260808e-05,
+      "loss": 0.782,
+      "step": 942
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993288199121283e-05,
+      "loss": 0.7696,
+      "step": 943
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993247220348143e-05,
+      "loss": 0.7882,
+      "step": 944
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993206116942414e-05,
+      "loss": 55.5624,
+      "step": 945
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.99316488890512e-05,
+      "loss": 53.3678,
+      "step": 946
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.99312353623729e-05,
+      "loss": 59.1876,
+      "step": 947
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993082058939956e-05,
+      "loss": 0.7884,
+      "step": 948
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.993040457014153e-05,
+      "loss": 0.7658,
+      "step": 949
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992998730460921e-05,
+      "loss": 0.804,
+      "step": 950
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.9929568792813e-05,
+      "loss": 0.7001,
+      "step": 951
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992914903476338e-05,
+      "loss": 0.7549,
+      "step": 952
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992872803047077e-05,
+      "loss": 52.4245,
+      "step": 953
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992830577994571e-05,
+      "loss": 57.3892,
+      "step": 954
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992788228319876e-05,
+      "loss": 54.1611,
+      "step": 955
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992745754024047e-05,
+      "loss": 50.3869,
+      "step": 956
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.99270315510814e-05,
+      "loss": 58.5763,
+      "step": 957
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992660431573227e-05,
+      "loss": 52.5997,
+      "step": 958
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992617583420366e-05,
+      "loss": 0.7752,
+      "step": 959
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992574610650632e-05,
+      "loss": 0.8287,
+      "step": 960
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992531513265095e-05,
+      "loss": 57.3127,
+      "step": 961
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992488291264832e-05,
+      "loss": 0.781,
+      "step": 962
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.99244494465092e-05,
+      "loss": 0.8042,
+      "step": 963
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992401473424444e-05,
+      "loss": 57.8599,
+      "step": 964
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992357877586487e-05,
+      "loss": 55.7778,
+      "step": 965
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992314157138137e-05,
+      "loss": 54.2093,
+      "step": 966
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992270312080485e-05,
+      "loss": 56.7433,
+      "step": 967
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992226342414628e-05,
+      "loss": 0.7781,
+      "step": 968
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.99218224814166e-05,
+      "loss": 0.8244,
+      "step": 969
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992138029262684e-05,
+      "loss": 52.3993,
+      "step": 970
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992093685778802e-05,
+      "loss": 0.7838,
+      "step": 971
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992049217691124e-05,
+      "loss": 0.7477,
+      "step": 972
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.992004625000756e-05,
+      "loss": 0.8198,
+      "step": 973
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991959907708812e-05,
+      "loss": 60.2148,
+      "step": 974
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.99191506581641e-05,
+      "loss": 0.7738,
+      "step": 975
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991870099324669e-05,
+      "loss": 60.6704,
+      "step": 976
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.99182500823471e-05,
+      "loss": 0.8346,
+      "step": 977
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991779792547657e-05,
+      "loss": 58.4476,
+      "step": 978
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991734452264643e-05,
+      "loss": 55.4139,
+      "step": 979
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991688987386796e-05,
+      "loss": 55.3904,
+      "step": 980
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991643397915254e-05,
+      "loss": 54.4955,
+      "step": 981
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991597683851151e-05,
+      "loss": 56.0593,
+      "step": 982
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.99155184519563e-05,
+      "loss": 0.7376,
+      "step": 983
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991505881949837e-05,
+      "loss": 53.3363,
+      "step": 984
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991459794114916e-05,
+      "loss": 54.9651,
+      "step": 985
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991413581692019e-05,
+      "loss": 0.7326,
+      "step": 986
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991367244682299e-05,
+      "loss": 53.0163,
+      "step": 987
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991320783086914e-05,
+      "loss": 0.7473,
+      "step": 988
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991274196907021e-05,
+      "loss": 0.8032,
+      "step": 989
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991227486143787e-05,
+      "loss": 54.1052,
+      "step": 990
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991180650798373e-05,
+      "loss": 52.3497,
+      "step": 991
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991133690871951e-05,
+      "loss": 53.8213,
+      "step": 992
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991086606365693e-05,
+      "loss": 54.0736,
+      "step": 993
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.991039397280773e-05,
+      "loss": 0.7918,
+      "step": 994
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.99099206361837e-05,
+      "loss": 0.7697,
+      "step": 995
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990944605379666e-05,
+      "loss": 56.9705,
+      "step": 996
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990897022565845e-05,
+      "loss": 0.7925,
+      "step": 997
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990849315178095e-05,
+      "loss": 0.7296,
+      "step": 998
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990801483217606e-05,
+      "loss": 0.7979,
+      "step": 999
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990753526685572e-05,
+      "loss": 54.16,
+      "step": 1000
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990705445583191e-05,
+      "loss": 0.7473,
+      "step": 1001
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990657239911663e-05,
+      "loss": 56.0132,
+      "step": 1002
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.99060890967219e-05,
+      "loss": 48.7915,
+      "step": 1003
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990560454865979e-05,
+      "loss": 0.7335,
+      "step": 1004
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990511875494239e-05,
+      "loss": 52.5048,
+      "step": 1005
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990463171558185e-05,
+      "loss": 52.6981,
+      "step": 1006
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.99041434305903e-05,
+      "loss": 54.7409,
+      "step": 1007
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990365389997994e-05,
+      "loss": 56.4671,
+      "step": 1008
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990316312376297e-05,
+      "loss": 0.7339,
+      "step": 1009
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990267110195166e-05,
+      "loss": 51.6223,
+      "step": 1010
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990217783455828e-05,
+      "loss": 0.8066,
+      "step": 1011
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990168332159517e-05,
+      "loss": 54.335,
+      "step": 1012
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990118756307463e-05,
+      "loss": 57.7289,
+      "step": 1013
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990069055900905e-05,
+      "loss": 0.7331,
+      "step": 1014
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.990019230941086e-05,
+      "loss": 52.4687,
+      "step": 1015
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989969281429245e-05,
+      "loss": 0.7655,
+      "step": 1016
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989919207366634e-05,
+      "loss": 0.7821,
+      "step": 1017
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.9898690087545e-05,
+      "loss": 49.8098,
+      "step": 1018
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989818685594094e-05,
+      "loss": 0.7857,
+      "step": 1019
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989768237886676e-05,
+      "loss": 0.7046,
+      "step": 1020
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989717665633504e-05,
+      "loss": 49.8289,
+      "step": 1021
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989666968835838e-05,
+      "loss": 0.7813,
+      "step": 1022
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989616147494947e-05,
+      "loss": 0.7796,
+      "step": 1023
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989565201612096e-05,
+      "loss": 0.8081,
+      "step": 1024
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989514131188559e-05,
+      "loss": 0.7843,
+      "step": 1025
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989462936225609e-05,
+      "loss": 52.3815,
+      "step": 1026
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989411616724527e-05,
+      "loss": 54.5425,
+      "step": 1027
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98936017268659e-05,
+      "loss": 0.7487,
+      "step": 1028
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989308604113086e-05,
+      "loss": 56.7302,
+      "step": 1029
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989256911005299e-05,
+      "loss": 49.8261,
+      "step": 1030
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989205093364519e-05,
+      "loss": 0.7535,
+      "step": 1031
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989153151192041e-05,
+      "loss": 0.8152,
+      "step": 1032
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.989101084489162e-05,
+      "loss": 55.7264,
+      "step": 1033
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98904889325718e-05,
+      "loss": 50.5484,
+      "step": 1034
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.988996577497401e-05,
+      "loss": 0.7826,
+      "step": 1035
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.988944137211124e-05,
+      "loss": 0.7614,
+      "step": 1036
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.988891572399666e-05,
+      "loss": 50.743,
+      "step": 1037
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.988838883064333e-05,
+      "loss": 54.1332,
+      "step": 1038
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.988786069206444e-05,
+      "loss": 52.4028,
+      "step": 1039
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.988733130827315e-05,
+      "loss": 51.9257,
+      "step": 1040
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.988680067928267e-05,
+      "loss": 0.7694,
+      "step": 1041
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.988626880510625e-05,
+      "loss": 0.7452,
+      "step": 1042
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.988573568575718e-05,
+      "loss": 0.7403,
+      "step": 1043
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.988520132124877e-05,
+      "loss": 65.5905,
+      "step": 1044
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.988466571159434e-05,
+      "loss": 59.3094,
+      "step": 1045
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.988412885680724e-05,
+      "loss": 54.8427,
+      "step": 1046
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.988359075690092e-05,
+      "loss": 47.8238,
+      "step": 1047
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.988305141188879e-05,
+      "loss": 50.0902,
+      "step": 1048
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.988251082178428e-05,
+      "loss": 52.072,
+      "step": 1049
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.988196898660092e-05,
+      "loss": 51.6118,
+      "step": 1050
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.988142590635223e-05,
+      "loss": 53.3547,
+      "step": 1051
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.988088158105176e-05,
+      "loss": 0.8106,
+      "step": 1052
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98803360107131e-05,
+      "loss": 0.7655,
+      "step": 1053
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987978919534986e-05,
+      "loss": 53.6532,
+      "step": 1054
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987924113497569e-05,
+      "loss": 55.3047,
+      "step": 1055
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987869182960426e-05,
+      "loss": 0.6999,
+      "step": 1056
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987814127924932e-05,
+      "loss": 0.7341,
+      "step": 1057
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987758948392456e-05,
+      "loss": 0.7556,
+      "step": 1058
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987703644364379e-05,
+      "loss": 0.7679,
+      "step": 1059
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98764821584208e-05,
+      "loss": 0.7886,
+      "step": 1060
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987592662826944e-05,
+      "loss": 50.446,
+      "step": 1061
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987536985320354e-05,
+      "loss": 51.0697,
+      "step": 1062
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987481183323704e-05,
+      "loss": 47.4853,
+      "step": 1063
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987425256838384e-05,
+      "loss": 51.1842,
+      "step": 1064
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987369205865791e-05,
+      "loss": 0.7816,
+      "step": 1065
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987313030407323e-05,
+      "loss": 53.8202,
+      "step": 1066
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987256730464385e-05,
+      "loss": 0.7932,
+      "step": 1067
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987200306038378e-05,
+      "loss": 48.2657,
+      "step": 1068
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987143757130713e-05,
+      "loss": 0.7956,
+      "step": 1069
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987087083742804e-05,
+      "loss": 0.7683,
+      "step": 1070
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.987030285876059e-05,
+      "loss": 50.8863,
+      "step": 1071
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.9869733635319e-05,
+      "loss": 0.8098,
+      "step": 1072
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98691631671175e-05,
+      "loss": 50.9774,
+      "step": 1073
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.986859145417027e-05,
+      "loss": 0.7561,
+      "step": 1074
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.986801849649162e-05,
+      "loss": 0.7939,
+      "step": 1075
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.986744429409584e-05,
+      "loss": 0.7842,
+      "step": 1076
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.986686884699727e-05,
+      "loss": 0.7842,
+      "step": 1077
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.986629215521026e-05,
+      "loss": 60.63,
+      "step": 1078
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.986571421874921e-05,
+      "loss": 50.3488,
+      "step": 1079
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.986513503762857e-05,
+      "loss": 0.7529,
+      "step": 1080
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.986455461186276e-05,
+      "loss": 49.9124,
+      "step": 1081
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.986397294146627e-05,
+      "loss": 53.667,
+      "step": 1082
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.986339002645364e-05,
+      "loss": 0.7881,
+      "step": 1083
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.986280586683942e-05,
+      "loss": 0.7769,
+      "step": 1084
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.986222046263816e-05,
+      "loss": 49.5058,
+      "step": 1085
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.986163381386451e-05,
+      "loss": 48.0262,
+      "step": 1086
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98610459205331e-05,
+      "loss": 48.4847,
+      "step": 1087
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.986045678265857e-05,
+      "loss": 0.7408,
+      "step": 1088
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985986640025566e-05,
+      "loss": 51.8765,
+      "step": 1089
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985927477333911e-05,
+      "loss": 0.8239,
+      "step": 1090
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985868190192369e-05,
+      "loss": 50.2889,
+      "step": 1091
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985808778602417e-05,
+      "loss": 0.757,
+      "step": 1092
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98574924256554e-05,
+      "loss": 0.7714,
+      "step": 1093
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985689582083223e-05,
+      "loss": 50.0201,
+      "step": 1094
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985629797156957e-05,
+      "loss": 48.0031,
+      "step": 1095
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985569887788231e-05,
+      "loss": 48.7777,
+      "step": 1096
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985509853978543e-05,
+      "loss": 50.8443,
+      "step": 1097
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98544969572939e-05,
+      "loss": 0.723,
+      "step": 1098
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985389413042276e-05,
+      "loss": 49.0258,
+      "step": 1099
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985329005918702e-05,
+      "loss": 49.2627,
+      "step": 1100
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985268474360179e-05,
+      "loss": 0.7825,
+      "step": 1101
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985207818368216e-05,
+      "loss": 0.7631,
+      "step": 1102
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985147037944328e-05,
+      "loss": 0.7933,
+      "step": 1103
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985086133090032e-05,
+      "loss": 0.7018,
+      "step": 1104
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.985025103806848e-05,
+      "loss": 55.582,
+      "step": 1105
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.984963950096298e-05,
+      "loss": 0.8068,
+      "step": 1106
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.984902671959911e-05,
+      "loss": 0.7138,
+      "step": 1107
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.984841269399215e-05,
+      "loss": 53.2553,
+      "step": 1108
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.984779742415741e-05,
+      "loss": 51.3823,
+      "step": 1109
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.984718091011028e-05,
+      "loss": 0.7873,
+      "step": 1110
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.984656315186613e-05,
+      "loss": 0.8263,
+      "step": 1111
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.984594414944037e-05,
+      "loss": 46.8953,
+      "step": 1112
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.984532390284848e-05,
+      "loss": 47.8754,
+      "step": 1113
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.984470241210591e-05,
+      "loss": 0.8053,
+      "step": 1114
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98440796772282e-05,
+      "loss": 52.1512,
+      "step": 1115
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.984345569823087e-05,
+      "loss": 52.9803,
+      "step": 1116
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.984283047512951e-05,
+      "loss": 52.1623,
+      "step": 1117
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.984220400793972e-05,
+      "loss": 0.745,
+      "step": 1118
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.984157629667714e-05,
+      "loss": 0.8231,
+      "step": 1119
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.984094734135743e-05,
+      "loss": 51.3033,
+      "step": 1120
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.984031714199629e-05,
+      "loss": 49.2951,
+      "step": 1121
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.983968569860947e-05,
+      "loss": 49.7831,
+      "step": 1122
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98390530112127e-05,
+      "loss": 0.7016,
+      "step": 1123
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98384190798218e-05,
+      "loss": 47.5578,
+      "step": 1124
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.983778390445258e-05,
+      "loss": 48.0402,
+      "step": 1125
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.983714748512088e-05,
+      "loss": 0.8095,
+      "step": 1126
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.98365098218426e-05,
+      "loss": 0.7724,
+      "step": 1127
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.983587091463367e-05,
+      "loss": 51.3284,
+      "step": 1128
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.983523076351001e-05,
+      "loss": 45.6961,
+      "step": 1129
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.983458936848761e-05,
+      "loss": 0.7569,
+      "step": 1130
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.983394672958251e-05,
+      "loss": 0.7952,
+      "step": 1131
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.983330284681069e-05,
+      "loss": 49.7239,
+      "step": 1132
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.983265772018827e-05,
+      "loss": 49.314,
+      "step": 1133
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.983201134973132e-05,
+      "loss": 49.9689,
+      "step": 1134
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.983136373545598e-05,
+      "loss": 0.7528,
+      "step": 1135
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.983071487737846e-05,
+      "loss": 0.728,
+      "step": 1136
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.983006477551489e-05,
+      "loss": 49.8614,
+      "step": 1137
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.982941342988153e-05,
+      "loss": 0.8003,
+      "step": 1138
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.982876084049465e-05,
+      "loss": 0.8183,
+      "step": 1139
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.98281070073705e-05,
+      "loss": 52.5187,
+      "step": 1140
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.982745193052542e-05,
+      "loss": 0.8038,
+      "step": 1141
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.982679560997577e-05,
+      "loss": 48.7873,
+      "step": 1142
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.982613804573793e-05,
+      "loss": 51.7795,
+      "step": 1143
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.98254792378283e-05,
+      "loss": 48.0285,
+      "step": 1144
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.982481918626334e-05,
+      "loss": 48.3229,
+      "step": 1145
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.982415789105952e-05,
+      "loss": 50.7267,
+      "step": 1146
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.982349535223334e-05,
+      "loss": 47.9919,
+      "step": 1147
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.982283156980132e-05,
+      "loss": 0.7875,
+      "step": 1148
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.982216654378008e-05,
+      "loss": 51.006,
+      "step": 1149
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.982150027418616e-05,
+      "loss": 47.8581,
+      "step": 1150
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.982083276103623e-05,
+      "loss": 0.8124,
+      "step": 1151
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.982016400434694e-05,
+      "loss": 0.7845,
+      "step": 1152
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.981949400413498e-05,
+      "loss": 0.8023,
+      "step": 1153
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.981882276041707e-05,
+      "loss": 0.7286,
+      "step": 1154
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.981815027320997e-05,
+      "loss": 0.7888,
+      "step": 1155
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.981747654253046e-05,
+      "loss": 51.7318,
+      "step": 1156
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.981680156839537e-05,
+      "loss": 55.0432,
+      "step": 1157
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.981612535082154e-05,
+      "loss": 0.8078,
+      "step": 1158
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.981544788982585e-05,
+      "loss": 0.7385,
+      "step": 1159
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.981476918542518e-05,
+      "loss": 49.2695,
+      "step": 1160
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.981408923763652e-05,
+      "loss": 0.7313,
+      "step": 1161
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.98134080464768e-05,
+      "loss": 0.8179,
+      "step": 1162
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.981272561196307e-05,
+      "loss": 47.9288,
+      "step": 1163
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.981204193411231e-05,
+      "loss": 0.7948,
+      "step": 1164
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.981135701294161e-05,
+      "loss": 45.9457,
+      "step": 1165
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.981067084846809e-05,
+      "loss": 0.7765,
+      "step": 1166
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.980998344070884e-05,
+      "loss": 48.3403,
+      "step": 1167
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.980929478968102e-05,
+      "loss": 47.0968,
+      "step": 1168
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.980860489540184e-05,
+      "loss": 50.0249,
+      "step": 1169
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.980791375788849e-05,
+      "loss": 0.7854,
+      "step": 1170
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.980722137715827e-05,
+      "loss": 53.468,
+      "step": 1171
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.98065277532284e-05,
+      "loss": 51.6094,
+      "step": 1172
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.980583288611626e-05,
+      "loss": 0.7734,
+      "step": 1173
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.980513677583913e-05,
+      "loss": 0.7638,
+      "step": 1174
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.980443942241443e-05,
+      "loss": 0.8172,
+      "step": 1175
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.980374082585955e-05,
+      "loss": 46.7276,
+      "step": 1176
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.980304098619192e-05,
+      "loss": 0.7557,
+      "step": 1177
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.980233990342902e-05,
+      "loss": 46.522,
+      "step": 1178
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.980163757758834e-05,
+      "loss": 51.868,
+      "step": 1179
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.980093400868742e-05,
+      "loss": 0.7644,
+      "step": 1180
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.98002291967438e-05,
+      "loss": 0.7785,
+      "step": 1181
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.979952314177512e-05,
+      "loss": 0.7989,
+      "step": 1182
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.979881584379895e-05,
+      "loss": 0.8114,
+      "step": 1183
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.979810730283297e-05,
+      "loss": 49.8811,
+      "step": 1184
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.979739751889486e-05,
+      "loss": 0.7611,
+      "step": 1185
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.979668649200233e-05,
+      "loss": 0.7352,
+      "step": 1186
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.979597422217315e-05,
+      "loss": 0.7801,
+      "step": 1187
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.979526070942509e-05,
+      "loss": 0.7888,
+      "step": 1188
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.979454595377594e-05,
+      "loss": 0.7671,
+      "step": 1189
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.979382995524354e-05,
+      "loss": 0.7021,
+      "step": 1190
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97931127138458e-05,
+      "loss": 0.7751,
+      "step": 1191
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.979239422960058e-05,
+      "loss": 0.6761,
+      "step": 1192
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.979167450252583e-05,
+      "loss": 60.8622,
+      "step": 1193
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.979095353263953e-05,
+      "loss": 51.948,
+      "step": 1194
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.979023131995967e-05,
+      "loss": 0.7877,
+      "step": 1195
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.978950786450422e-05,
+      "loss": 50.9894,
+      "step": 1196
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.978878316629133e-05,
+      "loss": 0.77,
+      "step": 1197
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.978805722533904e-05,
+      "loss": 0.7866,
+      "step": 1198
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.978733004166545e-05,
+      "loss": 0.7849,
+      "step": 1199
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.978660161528875e-05,
+      "loss": 49.0164,
+      "step": 1200
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97858719462271e-05,
+      "loss": 51.4917,
+      "step": 1201
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.978514103449873e-05,
+      "loss": 48.0787,
+      "step": 1202
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.978440888012186e-05,
+      "loss": 0.7498,
+      "step": 1203
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.978367548311479e-05,
+      "loss": 0.7944,
+      "step": 1204
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97829408434958e-05,
+      "loss": 49.6608,
+      "step": 1205
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.978220496128326e-05,
+      "loss": 52.5561,
+      "step": 1206
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97814678364955e-05,
+      "loss": 48.31,
+      "step": 1207
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.978072946915093e-05,
+      "loss": 0.7336,
+      "step": 1208
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.977998985926799e-05,
+      "loss": 45.0353,
+      "step": 1209
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.977924900686514e-05,
+      "loss": 47.142,
+      "step": 1210
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.977850691196087e-05,
+      "loss": 40.4243,
+      "step": 1211
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.977776357457372e-05,
+      "loss": 42.3217,
+      "step": 1212
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97770189947222e-05,
+      "loss": 48.4643,
+      "step": 1213
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.977627317242492e-05,
+      "loss": 45.0808,
+      "step": 1214
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.977552610770051e-05,
+      "loss": 45.457,
+      "step": 1215
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.977477780056759e-05,
+      "loss": 0.785,
+      "step": 1216
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.977402825104487e-05,
+      "loss": 0.7456,
+      "step": 1217
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.977327745915101e-05,
+      "loss": 46.4559,
+      "step": 1218
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97725254249048e-05,
+      "loss": 43.8251,
+      "step": 1219
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.977177214832499e-05,
+      "loss": 43.6789,
+      "step": 1220
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.977101762943039e-05,
+      "loss": 40.9766,
+      "step": 1221
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.977026186823981e-05,
+      "loss": 0.8035,
+      "step": 1222
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.976950486477214e-05,
+      "loss": 0.7736,
+      "step": 1223
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.976874661904626e-05,
+      "loss": 49.5574,
+      "step": 1224
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.976798713108111e-05,
+      "loss": 49.101,
+      "step": 1225
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.976722640089563e-05,
+      "loss": 0.8077,
+      "step": 1226
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.976646442850881e-05,
+      "loss": 0.7794,
+      "step": 1227
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.976570121393967e-05,
+      "loss": 48.5536,
+      "step": 1228
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.976493675720729e-05,
+      "loss": 0.8451,
+      "step": 1229
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97641710583307e-05,
+      "loss": 51.4284,
+      "step": 1230
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.976340411732905e-05,
+      "loss": 48.9184,
+      "step": 1231
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.976263593422147e-05,
+      "loss": 0.814,
+      "step": 1232
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.976186650902714e-05,
+      "loss": 0.7867,
+      "step": 1233
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.976109584176525e-05,
+      "loss": 47.8187,
+      "step": 1234
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.976032393245504e-05,
+      "loss": 52.1653,
+      "step": 1235
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97595507811158e-05,
+      "loss": 0.7413,
+      "step": 1236
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.975877638776681e-05,
+      "loss": 50.1341,
+      "step": 1237
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.975800075242738e-05,
+      "loss": 46.8478,
+      "step": 1238
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.975722387511692e-05,
+      "loss": 42.7641,
+      "step": 1239
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.975644575585477e-05,
+      "loss": 0.7735,
+      "step": 1240
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.975566639466037e-05,
+      "loss": 0.7698,
+      "step": 1241
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.975488579155319e-05,
+      "loss": 0.7521,
+      "step": 1242
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97541039465527e-05,
+      "loss": 50.6693,
+      "step": 1243
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97533208596784e-05,
+      "loss": 0.813,
+      "step": 1244
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.975253653094987e-05,
+      "loss": 46.3751,
+      "step": 1245
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.975175096038665e-05,
+      "loss": 51.2599,
+      "step": 1246
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.975096414800838e-05,
+      "loss": 45.4587,
+      "step": 1247
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.975017609383468e-05,
+      "loss": 0.7399,
+      "step": 1248
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.974938679788521e-05,
+      "loss": 51.5733,
+      "step": 1249
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.974859626017972e-05,
+      "loss": 0.7729,
+      "step": 1250
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.974780448073788e-05,
+      "loss": 49.1808,
+      "step": 1251
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97470114595795e-05,
+      "loss": 44.6392,
+      "step": 1252
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.974621719672434e-05,
+      "loss": 49.3617,
+      "step": 1253
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.974542169219225e-05,
+      "loss": 0.7529,
+      "step": 1254
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.974462494600307e-05,
+      "loss": 0.8062,
+      "step": 1255
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97438269581767e-05,
+      "loss": 50.3272,
+      "step": 1256
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.974302772873304e-05,
+      "loss": 50.4626,
+      "step": 1257
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.974222725769207e-05,
+      "loss": 53.0567,
+      "step": 1258
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.974142554507373e-05,
+      "loss": 0.6842,
+      "step": 1259
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.974062259089804e-05,
+      "loss": 50.8096,
+      "step": 1260
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.973981839518509e-05,
+      "loss": 0.728,
+      "step": 1261
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.973901295795489e-05,
+      "loss": 46.0117,
+      "step": 1262
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.973820627922759e-05,
+      "loss": 49.4732,
+      "step": 1263
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.973739835902327e-05,
+      "loss": 0.7759,
+      "step": 1264
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.973658919736216e-05,
+      "loss": 0.7753,
+      "step": 1265
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.973577879426441e-05,
+      "loss": 46.3658,
+      "step": 1266
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.973496714975028e-05,
+      "loss": 41.1845,
+      "step": 1267
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.973415426384e-05,
+      "loss": 0.754,
+      "step": 1268
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97333401365539e-05,
+      "loss": 42.8742,
+      "step": 1269
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.973252476791224e-05,
+      "loss": 0.7776,
+      "step": 1270
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.973170815793543e-05,
+      "loss": 0.8006,
+      "step": 1271
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.973089030664382e-05,
+      "loss": 0.8274,
+      "step": 1272
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.973007121405784e-05,
+      "loss": 0.7482,
+      "step": 1273
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.972925088019792e-05,
+      "loss": 0.775,
+      "step": 1274
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.972842930508454e-05,
+      "loss": 50.3994,
+      "step": 1275
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.972760648873822e-05,
+      "loss": 0.7797,
+      "step": 1276
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.972678243117947e-05,
+      "loss": 0.6999,
+      "step": 1277
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.972595713242891e-05,
+      "loss": 47.0042,
+      "step": 1278
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.972513059250708e-05,
+      "loss": 0.8043,
+      "step": 1279
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.972430281143464e-05,
+      "loss": 0.7123,
+      "step": 1280
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.972347378923226e-05,
+      "loss": 0.7884,
+      "step": 1281
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97226435259206e-05,
+      "loss": 45.8657,
+      "step": 1282
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.972181202152042e-05,
+      "loss": 43.2381,
+      "step": 1283
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.972097927605246e-05,
+      "loss": 47.5002,
+      "step": 1284
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.972014528953751e-05,
+      "loss": 44.8247,
+      "step": 1285
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.971931006199635e-05,
+      "loss": 42.7206,
+      "step": 1286
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.971847359344988e-05,
+      "loss": 48.4563,
+      "step": 1287
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.971763588391896e-05,
+      "loss": 0.8051,
+      "step": 1288
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.971679693342449e-05,
+      "loss": 0.7939,
+      "step": 1289
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.971595674198742e-05,
+      "loss": 0.7961,
+      "step": 1290
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.971511530962871e-05,
+      "loss": 43.9435,
+      "step": 1291
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.971427263636939e-05,
+      "loss": 0.7675,
+      "step": 1292
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.971342872223046e-05,
+      "loss": 41.2375,
+      "step": 1293
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.9712583567233e-05,
+      "loss": 46.8933,
+      "step": 1294
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97117371713981e-05,
+      "loss": 0.8474,
+      "step": 1295
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.971088953474689e-05,
+      "loss": 43.1106,
+      "step": 1296
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.971004065730054e-05,
+      "loss": 43.8724,
+      "step": 1297
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97091905390802e-05,
+      "loss": 43.4638,
+      "step": 1298
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.970833918010714e-05,
+      "loss": 0.7959,
+      "step": 1299
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.970748658040257e-05,
+      "loss": 47.3027,
+      "step": 1300
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.970663273998779e-05,
+      "loss": 0.7903,
+      "step": 1301
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97057776588841e-05,
+      "loss": 0.7213,
+      "step": 1302
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.970492133711285e-05,
+      "loss": 0.7277,
+      "step": 1303
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.970406377469544e-05,
+      "loss": 45.8067,
+      "step": 1304
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.970320497165322e-05,
+      "loss": 0.7942,
+      "step": 1305
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.970234492800765e-05,
+      "loss": 41.9408,
+      "step": 1306
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.970148364378023e-05,
+      "loss": 0.7859,
+      "step": 1307
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.97006211189924e-05,
+      "loss": 0.7991,
+      "step": 1308
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.969975735366575e-05,
+      "loss": 0.7564,
+      "step": 1309
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.96988923478218e-05,
+      "loss": 0.6911,
+      "step": 1310
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.969802610148213e-05,
+      "loss": 0.7482,
+      "step": 1311
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.96971586146684e-05,
+      "loss": 56.8531,
+      "step": 1312
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.969628988740223e-05,
+      "loss": 61.0739,
+      "step": 1313
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.969541991970532e-05,
+      "loss": 48.1009,
+      "step": 1314
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.969454871159939e-05,
+      "loss": 45.7131,
+      "step": 1315
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.969367626310618e-05,
+      "loss": 0.8238,
+      "step": 1316
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.969280257424745e-05,
+      "loss": 41.9916,
+      "step": 1317
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.969192764504503e-05,
+      "loss": 0.7618,
+      "step": 1318
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.969105147552075e-05,
+      "loss": 46.8112,
+      "step": 1319
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.969017406569647e-05,
+      "loss": 43.9028,
+      "step": 1320
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.968929541559411e-05,
+      "loss": 45.1557,
+      "step": 1321
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.96884155252356e-05,
+      "loss": 46.6749,
+      "step": 1322
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.968753439464287e-05,
+      "loss": 47.7583,
+      "step": 1323
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.968665202383795e-05,
+      "loss": 42.898,
+      "step": 1324
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.968576841284285e-05,
+      "loss": 0.776,
+      "step": 1325
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.968488356167962e-05,
+      "loss": 46.1194,
+      "step": 1326
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.968399747037037e-05,
+      "loss": 0.7595,
+      "step": 1327
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.968311013893717e-05,
+      "loss": 45.9595,
+      "step": 1328
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.968222156740221e-05,
+      "loss": 0.7737,
+      "step": 1329
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.968133175578766e-05,
+      "loss": 0.8306,
+      "step": 1330
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.968044070411573e-05,
+      "loss": 46.3298,
+      "step": 1331
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.967954841240866e-05,
+      "loss": 0.777,
+      "step": 1332
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.96786548806887e-05,
+      "loss": 0.7404,
+      "step": 1333
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.967776010897818e-05,
+      "loss": 0.7542,
+      "step": 1334
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.967686409729942e-05,
+      "loss": 0.7575,
+      "step": 1335
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.96759668456748e-05,
+      "loss": 0.7375,
+      "step": 1336
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.967506835412669e-05,
+      "loss": 41.2276,
+      "step": 1337
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.967416862267753e-05,
+      "loss": 0.778,
+      "step": 1338
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.96732676513498e-05,
+      "loss": 43.9074,
+      "step": 1339
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.967236544016597e-05,
+      "loss": 41.4331,
+      "step": 1340
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.967146198914853e-05,
+      "loss": 0.8052,
+      "step": 1341
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.967055729832007e-05,
+      "loss": 45.7676,
+      "step": 1342
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.966965136770317e-05,
+      "loss": 0.7701,
+      "step": 1343
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.966874419732041e-05,
+      "loss": 0.7879,
+      "step": 1344
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.966783578719446e-05,
+      "loss": 0.8291,
+      "step": 1345
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.9666926137348e-05,
+      "loss": 0.7646,
+      "step": 1346
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.966601524780372e-05,
+      "loss": 0.7553,
+      "step": 1347
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.966510311858435e-05,
+      "loss": 0.7969,
+      "step": 1348
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.966418974971266e-05,
+      "loss": 40.783,
+      "step": 1349
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.966327514121147e-05,
+      "loss": 0.7892,
+      "step": 1350
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.966235929310359e-05,
+      "loss": 46.3126,
+      "step": 1351
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.966144220541188e-05,
+      "loss": 0.7903,
+      "step": 1352
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.966052387815923e-05,
+      "loss": 43.0704,
+      "step": 1353
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.965960431136857e-05,
+      "loss": 0.7456,
+      "step": 1354
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.965868350506284e-05,
+      "loss": 0.7797,
+      "step": 1355
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.965776145926504e-05,
+      "loss": 39.0519,
+      "step": 1356
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.965683817399818e-05,
+      "loss": 41.6424,
+      "step": 1357
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.965591364928529e-05,
+      "loss": 0.8018,
+      "step": 1358
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.965498788514946e-05,
+      "loss": 41.9047,
+      "step": 1359
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.96540608816138e-05,
+      "loss": 44.2393,
+      "step": 1360
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.965313263870144e-05,
+      "loss": 0.8122,
+      "step": 1361
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.965220315643555e-05,
+      "loss": 0.7418,
+      "step": 1362
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.965127243483933e-05,
+      "loss": 41.4999,
+      "step": 1363
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.965034047393601e-05,
+      "loss": 40.29,
+      "step": 1364
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.964940727374887e-05,
+      "loss": 0.7745,
+      "step": 1365
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.964847283430119e-05,
+      "loss": 0.7524,
+      "step": 1366
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.964753715561628e-05,
+      "loss": 41.8357,
+      "step": 1367
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.964660023771751e-05,
+      "loss": 45.4741,
+      "step": 1368
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.964566208062825e-05,
+      "loss": 0.7347,
+      "step": 1369
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.964472268437194e-05,
+      "loss": 41.8841,
+      "step": 1370
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.964378204897202e-05,
+      "loss": 37.568,
+      "step": 1371
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.964284017445196e-05,
+      "loss": 38.1467,
+      "step": 1372
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.964189706083526e-05,
+      "loss": 43.1645,
+      "step": 1373
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.96409527081455e-05,
+      "loss": 0.7517,
+      "step": 1374
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.964000711640622e-05,
+      "loss": 46.0114,
+      "step": 1375
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.963906028564102e-05,
+      "loss": 0.7726,
+      "step": 1376
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.963811221587354e-05,
+      "loss": 38.6775,
+      "step": 1377
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.963716290712745e-05,
+      "loss": 40.6119,
+      "step": 1378
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.963621235942644e-05,
+      "loss": 0.7054,
+      "step": 1379
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.963526057279425e-05,
+      "loss": 0.7544,
+      "step": 1380
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.96343075472546e-05,
+      "loss": 0.8068,
+      "step": 1381
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.96333532828313e-05,
+      "loss": 43.1595,
+      "step": 1382
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.963239777954818e-05,
+      "loss": 40.4565,
+      "step": 1383
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.963144103742908e-05,
+      "loss": 42.4699,
+      "step": 1384
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.963048305649788e-05,
+      "loss": 0.777,
+      "step": 1385
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.962952383677848e-05,
+      "loss": 45.1097,
+      "step": 1386
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.962856337829485e-05,
+      "loss": 40.5762,
+      "step": 1387
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.962760168107093e-05,
+      "loss": 41.2139,
+      "step": 1388
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.962663874513076e-05,
+      "loss": 40.5507,
+      "step": 1389
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.962567457049835e-05,
+      "loss": 0.7896,
+      "step": 1390
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.962470915719775e-05,
+      "loss": 0.8187,
+      "step": 1391
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.96237425052531e-05,
+      "loss": 0.7607,
+      "step": 1392
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.96227746146885e-05,
+      "loss": 43.89,
+      "step": 1393
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.962180548552812e-05,
+      "loss": 0.7372,
+      "step": 1394
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.962083511779614e-05,
+      "loss": 44.1823,
+      "step": 1395
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.961986351151679e-05,
+      "loss": 0.8059,
+      "step": 1396
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.96188906667143e-05,
+      "loss": 0.7536,
+      "step": 1397
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.9617916583413e-05,
+      "loss": 0.7625,
+      "step": 1398
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.961694126163714e-05,
+      "loss": 42.1043,
+      "step": 1399
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.961596470141112e-05,
+      "loss": 0.7686,
+      "step": 1400
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.961498690275929e-05,
+      "loss": 42.686,
+      "step": 1401
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.961400786570605e-05,
+      "loss": 0.7494,
+      "step": 1402
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.961302759027584e-05,
+      "loss": 0.7195,
+      "step": 1403
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.961204607649315e-05,
+      "loss": 0.7299,
+      "step": 1404
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.961106332438246e-05,
+      "loss": 39.5519,
+      "step": 1405
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.961007933396829e-05,
+      "loss": 0.7493,
+      "step": 1406
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.960909410527521e-05,
+      "loss": 0.7681,
+      "step": 1407
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.960810763832783e-05,
+      "loss": 43.3199,
+      "step": 1408
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.960711993315076e-05,
+      "loss": 0.7669,
+      "step": 1409
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.960613098976863e-05,
+      "loss": 44.855,
+      "step": 1410
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.960514080820615e-05,
+      "loss": 0.7875,
+      "step": 1411
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.960414938848803e-05,
+      "loss": 41.5306,
+      "step": 1412
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.960315673063902e-05,
+      "loss": 38.966,
+      "step": 1413
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.960216283468389e-05,
+      "loss": 42.2335,
+      "step": 1414
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.960116770064745e-05,
+      "loss": 40.0992,
+      "step": 1415
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.960017132855453e-05,
+      "loss": 41.9211,
+      "step": 1416
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.959917371843001e-05,
+      "loss": 42.5637,
+      "step": 1417
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.959817487029881e-05,
+      "loss": 38.3712,
+      "step": 1418
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.959717478418582e-05,
+      "loss": 42.466,
+      "step": 1419
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.959617346011602e-05,
+      "loss": 36.7205,
+      "step": 1420
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.959517089811442e-05,
+      "loss": 40.0621,
+      "step": 1421
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.959416709820602e-05,
+      "loss": 39.1434,
+      "step": 1422
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.959316206041589e-05,
+      "loss": 0.7306,
+      "step": 1423
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.959215578476912e-05,
+      "loss": 39.6461,
+      "step": 1424
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.959114827129081e-05,
+      "loss": 37.4302,
+      "step": 1425
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.95901395200061e-05,
+      "loss": 0.7387,
+      "step": 1426
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.958912953094021e-05,
+      "loss": 38.9341,
+      "step": 1427
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.958811830411832e-05,
+      "loss": 42.7583,
+      "step": 1428
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.958710583956568e-05,
+      "loss": 42.3134,
+      "step": 1429
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.958609213730754e-05,
+      "loss": 36.8157,
+      "step": 1430
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.958507719736923e-05,
+      "loss": 0.7989,
+      "step": 1431
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.958406101977605e-05,
+      "loss": 39.714,
+      "step": 1432
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.958304360455341e-05,
+      "loss": 40.9845,
+      "step": 1433
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.958202495172666e-05,
+      "loss": 40.6677,
+      "step": 1434
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.958100506132127e-05,
+      "loss": 0.7462,
+      "step": 1435
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.957998393336266e-05,
+      "loss": 0.7512,
+      "step": 1436
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.957896156787632e-05,
+      "loss": 0.7829,
+      "step": 1437
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.957793796488779e-05,
+      "loss": 0.7203,
+      "step": 1438
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.957691312442262e-05,
+      "loss": 0.773,
+      "step": 1439
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.957588704650635e-05,
+      "loss": 43.3317,
+      "step": 1440
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.957485973116464e-05,
+      "loss": 0.7848,
+      "step": 1441
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.95738311784231e-05,
+      "loss": 0.7253,
+      "step": 1442
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.957280138830743e-05,
+      "loss": 0.7676,
+      "step": 1443
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.957177036084331e-05,
+      "loss": 0.7425,
+      "step": 1444
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.957073809605648e-05,
+      "loss": 0.7775,
+      "step": 1445
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.956970459397272e-05,
+      "loss": 45.0945,
+      "step": 1446
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.95686698546178e-05,
+      "loss": 41.3117,
+      "step": 1447
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.956763387801757e-05,
+      "loss": 36.8647,
+      "step": 1448
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.956659666419787e-05,
+      "loss": 41.9878,
+      "step": 1449
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.956555821318463e-05,
+      "loss": 0.7618,
+      "step": 1450
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.956451852500371e-05,
+      "loss": 40.7211,
+      "step": 1451
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.95634775996811e-05,
+      "loss": 0.7741,
+      "step": 1452
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.956243543724277e-05,
+      "loss": 38.7813,
+      "step": 1453
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.956139203771476e-05,
+      "loss": 40.6963,
+      "step": 1454
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.956034740112306e-05,
+      "loss": 37.6954,
+      "step": 1455
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.955930152749377e-05,
+      "loss": 42.2049,
+      "step": 1456
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.955825441685301e-05,
+      "loss": 40.6766,
+      "step": 1457
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.955720606922689e-05,
+      "loss": 0.7691,
+      "step": 1458
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.95561564846416e-05,
+      "loss": 0.7957,
+      "step": 1459
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.955510566312334e-05,
+      "loss": 45.5703,
+      "step": 1460
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.955405360469832e-05,
+      "loss": 0.7309,
+      "step": 1461
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.955300030939281e-05,
+      "loss": 0.7867,
+      "step": 1462
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.95519457772331e-05,
+      "loss": 0.7405,
+      "step": 1463
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.95508900082455e-05,
+      "loss": 0.8039,
+      "step": 1464
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.954983300245637e-05,
+      "loss": 0.7998,
+      "step": 1465
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.954877475989211e-05,
+      "loss": 0.7296,
+      "step": 1466
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.954771528057908e-05,
+      "loss": 0.7658,
+      "step": 1467
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.95466545645438e-05,
+      "loss": 0.7305,
+      "step": 1468
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.954559261181271e-05,
+      "loss": 0.7725,
+      "step": 1469
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.95445294224123e-05,
+      "loss": 0.7556,
+      "step": 1470
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.954346499636913e-05,
+      "loss": 40.4684,
+      "step": 1471
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.954239933370977e-05,
+      "loss": 0.7419,
+      "step": 1472
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.954133243446079e-05,
+      "loss": 41.7572,
+      "step": 1473
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.954026429864886e-05,
+      "loss": 41.5149,
+      "step": 1474
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.953919492630062e-05,
+      "loss": 37.0747,
+      "step": 1475
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.953812431744276e-05,
+      "loss": 0.7985,
+      "step": 1476
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.953705247210199e-05,
+      "loss": 40.4627,
+      "step": 1477
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.953597939030509e-05,
+      "loss": 44.2422,
+      "step": 1478
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.953490507207883e-05,
+      "loss": 0.7734,
+      "step": 1479
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.953382951745005e-05,
+      "loss": 40.0426,
+      "step": 1480
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.953275272644555e-05,
+      "loss": 39.8957,
+      "step": 1481
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.953167469909224e-05,
+      "loss": 39.3777,
+      "step": 1482
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.953059543541703e-05,
+      "loss": 0.7905,
+      "step": 1483
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.952951493544684e-05,
+      "loss": 0.7803,
+      "step": 1484
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.952843319920864e-05,
+      "loss": 42.5974,
+      "step": 1485
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.952735022672945e-05,
+      "loss": 0.7344,
+      "step": 1486
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.95262660180363e-05,
+      "loss": 42.5255,
+      "step": 1487
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.952518057315623e-05,
+      "loss": 0.7709,
+      "step": 1488
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.952409389211635e-05,
+      "loss": 0.7505,
+      "step": 1489
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.952300597494377e-05,
+      "loss": 0.7938,
+      "step": 1490
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.952191682166565e-05,
+      "loss": 0.7418,
+      "step": 1491
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.952082643230919e-05,
+      "loss": 0.7498,
+      "step": 1492
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.951973480690158e-05,
+      "loss": 41.1852,
+      "step": 1493
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.95186419454701e-05,
+      "loss": 0.8306,
+      "step": 1494
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.9517547848042e-05,
+      "loss": 0.7743,
+      "step": 1495
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.951645251464461e-05,
+      "loss": 39.63,
+      "step": 1496
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.951535594530525e-05,
+      "loss": 37.7984,
+      "step": 1497
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.95142581400513e-05,
+      "loss": 39.4186,
+      "step": 1498
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.951315909891017e-05,
+      "loss": 41.1414,
+      "step": 1499
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.951205882190928e-05,
+      "loss": 41.4263,
+      "step": 1500
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.951095730907609e-05,
+      "loss": 37.6228,
+      "step": 1501
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.95098545604381e-05,
+      "loss": 0.7568,
+      "step": 1502
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.950875057602284e-05,
+      "loss": 39.6349,
+      "step": 1503
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.950764535585786e-05,
+      "loss": 37.8543,
+      "step": 1504
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.950653889997074e-05,
+      "loss": 37.1801,
+      "step": 1505
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.950543120838912e-05,
+      "loss": 37.231,
+      "step": 1506
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.950432228114063e-05,
+      "loss": 0.7386,
+      "step": 1507
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.950321211825295e-05,
+      "loss": 40.6839,
+      "step": 1508
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.950210071975378e-05,
+      "loss": 0.7988,
+      "step": 1509
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.950098808567089e-05,
+      "loss": 0.7177,
+      "step": 1510
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.949987421603203e-05,
+      "loss": 37.2988,
+      "step": 1511
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.949875911086502e-05,
+      "loss": 38.3978,
+      "step": 1512
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.949764277019766e-05,
+      "loss": 0.7412,
+      "step": 1513
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.949652519405784e-05,
+      "loss": 36.9412,
+      "step": 1514
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.949540638247346e-05,
+      "loss": 0.7871,
+      "step": 1515
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.949428633547242e-05,
+      "loss": 41.8018,
+      "step": 1516
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.949316505308271e-05,
+      "loss": 0.7377,
+      "step": 1517
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.949204253533229e-05,
+      "loss": 0.766,
+      "step": 1518
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.949091878224919e-05,
+      "loss": 40.4901,
+      "step": 1519
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.948979379386145e-05,
+      "loss": 0.8143,
+      "step": 1520
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.948866757019715e-05,
+      "loss": 0.7525,
+      "step": 1521
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.948754011128442e-05,
+      "loss": 41.2174,
+      "step": 1522
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.948641141715141e-05,
+      "loss": 0.7852,
+      "step": 1523
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.948528148782623e-05,
+      "loss": 36.5842,
+      "step": 1524
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.948415032333715e-05,
+      "loss": 39.3511,
+      "step": 1525
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.948301792371237e-05,
+      "loss": 36.6555,
+      "step": 1526
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.948188428898018e-05,
+      "loss": 0.7422,
+      "step": 1527
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.948074941916886e-05,
+      "loss": 0.7087,
+      "step": 1528
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.94796133143067e-05,
+      "loss": 0.7633,
+      "step": 1529
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.947847597442212e-05,
+      "loss": 43.2449,
+      "step": 1530
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.94773373995435e-05,
+      "loss": 0.7149,
+      "step": 1531
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.947619758969922e-05,
+      "loss": 0.7367,
+      "step": 1532
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.947505654491776e-05,
+      "loss": 37.6756,
+      "step": 1533
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.947391426522757e-05,
+      "loss": 0.7235,
+      "step": 1534
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.947277075065721e-05,
+      "loss": 41.3455,
+      "step": 1535
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.947162600123518e-05,
+      "loss": 39.4562,
+      "step": 1536
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.947048001699007e-05,
+      "loss": 36.7899,
+      "step": 1537
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.94693327979505e-05,
+      "loss": 36.5274,
+      "step": 1538
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.946818434414507e-05,
+      "loss": 38.9297,
+      "step": 1539
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.946703465560247e-05,
+      "loss": 38.227,
+      "step": 1540
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.94658837323514e-05,
+      "loss": 35.5599,
+      "step": 1541
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.946473157442056e-05,
+      "loss": 0.7613,
+      "step": 1542
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.946357818183873e-05,
+      "loss": 0.7749,
+      "step": 1543
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.94624235546347e-05,
+      "loss": 0.7269,
+      "step": 1544
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.946126769283727e-05,
+      "loss": 39.2856,
+      "step": 1545
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.946011059647534e-05,
+      "loss": 37.9671,
+      "step": 1546
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.945895226557772e-05,
+      "loss": 0.7473,
+      "step": 1547
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.945779270017337e-05,
+      "loss": 0.8061,
+      "step": 1548
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.945663190029123e-05,
+      "loss": 37.5904,
+      "step": 1549
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.945546986596026e-05,
+      "loss": 38.4142,
+      "step": 1550
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.945430659720947e-05,
+      "loss": 39.1612,
+      "step": 1551
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.945314209406789e-05,
+      "loss": 40.3296,
+      "step": 1552
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.94519763565646e-05,
+      "loss": 38.088,
+      "step": 1553
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.945080938472868e-05,
+      "loss": 37.9208,
+      "step": 1554
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.944964117858927e-05,
+      "loss": 36.4605,
+      "step": 1555
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.944847173817553e-05,
+      "loss": 36.2108,
+      "step": 1556
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.944730106351664e-05,
+      "loss": 36.4851,
+      "step": 1557
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.944612915464183e-05,
+      "loss": 0.7417,
+      "step": 1558
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.944495601158034e-05,
+      "loss": 35.5371,
+      "step": 1559
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.944378163436146e-05,
+      "loss": 0.7952,
+      "step": 1560
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.94426060230145e-05,
+      "loss": 0.7747,
+      "step": 1561
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.944142917756881e-05,
+      "loss": 40.8921,
+      "step": 1562
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.944025109805375e-05,
+      "loss": 39.495,
+      "step": 1563
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.943907178449874e-05,
+      "loss": 42.0788,
+      "step": 1564
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.943789123693322e-05,
+      "loss": 36.9065,
+      "step": 1565
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.943670945538664e-05,
+      "loss": 0.753,
+      "step": 1566
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.94355264398885e-05,
+      "loss": 36.0496,
+      "step": 1567
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.943434219046833e-05,
+      "loss": 0.7611,
+      "step": 1568
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.94331567071557e-05,
+      "loss": 38.134,
+      "step": 1569
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.943196998998017e-05,
+      "loss": 0.7969,
+      "step": 1570
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.94307820389714e-05,
+      "loss": 37.2282,
+      "step": 1571
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.942959285415902e-05,
+      "loss": 0.7509,
+      "step": 1572
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.942840243557272e-05,
+      "loss": 39.9829,
+      "step": 1573
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.942721078324221e-05,
+      "loss": 33.342,
+      "step": 1574
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.942601789719723e-05,
+      "loss": 0.7725,
+      "step": 1575
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.942482377746756e-05,
+      "loss": 34.7051,
+      "step": 1576
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.9423628424083e-05,
+      "loss": 0.7731,
+      "step": 1577
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.942243183707338e-05,
+      "loss": 0.7572,
+      "step": 1578
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.942123401646859e-05,
+      "loss": 43.5342,
+      "step": 1579
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.942003496229853e-05,
+      "loss": 35.6302,
+      "step": 1580
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.941883467459309e-05,
+      "loss": 35.9017,
+      "step": 1581
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.941763315338225e-05,
+      "loss": 0.8454,
+      "step": 1582
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.941643039869601e-05,
+      "loss": 39.8769,
+      "step": 1583
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.941522641056439e-05,
+      "loss": 35.8899,
+      "step": 1584
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.941402118901744e-05,
+      "loss": 0.7355,
+      "step": 1585
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.941281473408521e-05,
+      "loss": 37.0924,
+      "step": 1586
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.941160704579787e-05,
+      "loss": 0.7551,
+      "step": 1587
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.941039812418553e-05,
+      "loss": 34.4716,
+      "step": 1588
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.940918796927836e-05,
+      "loss": 0.7562,
+      "step": 1589
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.940797658110658e-05,
+      "loss": 0.7233,
+      "step": 1590
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.940676395970045e-05,
+      "loss": 0.6964,
+      "step": 1591
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.940555010509018e-05,
+      "loss": 35.9302,
+      "step": 1592
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.94043350173061e-05,
+      "loss": 38.2215,
+      "step": 1593
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.940311869637856e-05,
+      "loss": 0.7776,
+      "step": 1594
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.940190114233787e-05,
+      "loss": 36.4611,
+      "step": 1595
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.940068235521446e-05,
+      "loss": 36.4812,
+      "step": 1596
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.939946233503874e-05,
+      "loss": 0.7687,
+      "step": 1597
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.939824108184118e-05,
+      "loss": 36.721,
+      "step": 1598
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.93970185956522e-05,
+      "loss": 37.437,
+      "step": 1599
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.93957948765024e-05,
+      "loss": 39.3823,
+      "step": 1600
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.939456992442225e-05,
+      "loss": 39.8221,
+      "step": 1601
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.939334373944236e-05,
+      "loss": 33.154,
+      "step": 1602
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.939211632159335e-05,
+      "loss": 0.7526,
+      "step": 1603
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.939088767090582e-05,
+      "loss": 0.7943,
+      "step": 1604
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.938965778741047e-05,
+      "loss": 34.8462,
+      "step": 1605
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.938842667113797e-05,
+      "loss": 38.1451,
+      "step": 1606
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.938719432211908e-05,
+      "loss": 34.8656,
+      "step": 1607
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.938596074038453e-05,
+      "loss": 33.8443,
+      "step": 1608
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.938472592596514e-05,
+      "loss": 0.7775,
+      "step": 1609
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.93834898788917e-05,
+      "loss": 0.7532,
+      "step": 1610
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.938225259919509e-05,
+      "loss": 37.7847,
+      "step": 1611
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.938101408690616e-05,
+      "loss": 0.7618,
+      "step": 1612
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.937977434205585e-05,
+      "loss": 40.0179,
+      "step": 1613
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.937853336467511e-05,
+      "loss": 36.8058,
+      "step": 1614
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.93772911547949e-05,
+      "loss": 34.6592,
+      "step": 1615
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.937604771244622e-05,
+      "loss": 37.262,
+      "step": 1616
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.937480303766012e-05,
+      "loss": 0.8068,
+      "step": 1617
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.937355713046765e-05,
+      "loss": 0.7902,
+      "step": 1618
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.937230999089994e-05,
+      "loss": 33.6053,
+      "step": 1619
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.937106161898809e-05,
+      "loss": 0.7658,
+      "step": 1620
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.936981201476328e-05,
+      "loss": 0.7459,
+      "step": 1621
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.936856117825668e-05,
+      "loss": 0.7798,
+      "step": 1622
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.936730910949951e-05,
+      "loss": 36.2467,
+      "step": 1623
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.936605580852306e-05,
+      "loss": 38.1621,
+      "step": 1624
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.936480127535856e-05,
+      "loss": 0.7671,
+      "step": 1625
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.936354551003735e-05,
+      "loss": 39.5153,
+      "step": 1626
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.936228851259079e-05,
+      "loss": 40.5052,
+      "step": 1627
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.936103028305021e-05,
+      "loss": 0.8152,
+      "step": 1628
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.935977082144707e-05,
+      "loss": 0.7287,
+      "step": 1629
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.935851012781276e-05,
+      "loss": 0.7903,
+      "step": 1630
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.93572482021788e-05,
+      "loss": 0.7713,
+      "step": 1631
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.935598504457661e-05,
+      "loss": 43.8216,
+      "step": 1632
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.93547206550378e-05,
+      "loss": 40.7035,
+      "step": 1633
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.935345503359388e-05,
+      "loss": 0.8113,
+      "step": 1634
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.935218818027645e-05,
+      "loss": 40.0671,
+      "step": 1635
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.935092009511715e-05,
+      "loss": 34.2045,
+      "step": 1636
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.93496507781476e-05,
+      "loss": 0.7944,
+      "step": 1637
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.934838022939949e-05,
+      "loss": 37.7672,
+      "step": 1638
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.934710844890457e-05,
+      "loss": 34.1777,
+      "step": 1639
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.934583543669453e-05,
+      "loss": 0.7742,
+      "step": 1640
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.934456119280119e-05,
+      "loss": 37.1595,
+      "step": 1641
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.934328571725633e-05,
+      "loss": 0.7242,
+      "step": 1642
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.934200901009178e-05,
+      "loss": 35.1013,
+      "step": 1643
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.934073107133943e-05,
+      "loss": 0.7256,
+      "step": 1644
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.933945190103117e-05,
+      "loss": 0.7328,
+      "step": 1645
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.933817149919892e-05,
+      "loss": 33.9182,
+      "step": 1646
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.933688986587464e-05,
+      "loss": 37.8973,
+      "step": 1647
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.933560700109033e-05,
+      "loss": 0.7811,
+      "step": 1648
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.9334322904878e-05,
+      "loss": 0.7546,
+      "step": 1649
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.933303757726971e-05,
+      "loss": 38.5632,
+      "step": 1650
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.933175101829752e-05,
+      "loss": 38.8648,
+      "step": 1651
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.93304632279936e-05,
+      "loss": 0.7764,
+      "step": 1652
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.932917420639002e-05,
+      "loss": 0.7829,
+      "step": 1653
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.932788395351899e-05,
+      "loss": 0.8067,
+      "step": 1654
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.932659246941274e-05,
+      "loss": 0.8534,
+      "step": 1655
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.932529975410346e-05,
+      "loss": 41.5701,
+      "step": 1656
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.932400580762342e-05,
+      "loss": 42.4263,
+      "step": 1657
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.932271063000495e-05,
+      "loss": 38.4852,
+      "step": 1658
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.932141422128036e-05,
+      "loss": 0.7406,
+      "step": 1659
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.932011658148201e-05,
+      "loss": 36.3574,
+      "step": 1660
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.93188177106423e-05,
+      "loss": 37.0735,
+      "step": 1661
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.931751760879363e-05,
+      "loss": 33.7426,
+      "step": 1662
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.931621627596846e-05,
+      "loss": 0.7607,
+      "step": 1663
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.931491371219926e-05,
+      "loss": 0.82,
+      "step": 1664
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.931360991751857e-05,
+      "loss": 36.351,
+      "step": 1665
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.93123048919589e-05,
+      "loss": 0.7339,
+      "step": 1666
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.931099863555287e-05,
+      "loss": 0.8186,
+      "step": 1667
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.930969114833302e-05,
+      "loss": 34.983,
+      "step": 1668
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.930838243033205e-05,
+      "loss": 0.7613,
+      "step": 1669
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.930707248158257e-05,
+      "loss": 34.4885,
+      "step": 1670
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.930576130211732e-05,
+      "loss": 0.7555,
+      "step": 1671
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.930444889196899e-05,
+      "loss": 36.9254,
+      "step": 1672
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.930313525117039e-05,
+      "loss": 36.0803,
+      "step": 1673
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.930182037975425e-05,
+      "loss": 35.513,
+      "step": 1674
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.930050427775343e-05,
+      "loss": 0.789,
+      "step": 1675
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.929918694520075e-05,
+      "loss": 0.8251,
+      "step": 1676
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.929786838212913e-05,
+      "loss": 34.0681,
+      "step": 1677
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.929654858857145e-05,
+      "loss": 35.3753,
+      "step": 1678
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.929522756456065e-05,
+      "loss": 38.0109,
+      "step": 1679
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.929390531012972e-05,
+      "loss": 36.8064,
+      "step": 1680
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.929258182531167e-05,
+      "loss": 34.3174,
+      "step": 1681
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.929125711013952e-05,
+      "loss": 33.0404,
+      "step": 1682
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.928993116464632e-05,
+      "loss": 36.0616,
+      "step": 1683
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.928860398886519e-05,
+      "loss": 34.8395,
+      "step": 1684
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.928727558282926e-05,
+      "loss": 33.7801,
+      "step": 1685
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.928594594657168e-05,
+      "loss": 35.4657,
+      "step": 1686
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.928461508012563e-05,
+      "loss": 0.7332,
+      "step": 1687
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.928328298352434e-05,
+      "loss": 39.7395,
+      "step": 1688
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.928194965680105e-05,
+      "loss": 36.4024,
+      "step": 1689
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.928061509998902e-05,
+      "loss": 0.7562,
+      "step": 1690
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.927927931312162e-05,
+      "loss": 37.0438,
+      "step": 1691
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.927794229623215e-05,
+      "loss": 33.256,
+      "step": 1692
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.927660404935397e-05,
+      "loss": 36.0237,
+      "step": 1693
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.927526457252052e-05,
+      "loss": 0.6993,
+      "step": 1694
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.927392386576522e-05,
+      "loss": 0.7578,
+      "step": 1695
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.927258192912152e-05,
+      "loss": 0.7926,
+      "step": 1696
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.927123876262292e-05,
+      "loss": 37.2664,
+      "step": 1697
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.926989436630294e-05,
+      "loss": 0.7725,
+      "step": 1698
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.926854874019517e-05,
+      "loss": 35.6433,
+      "step": 1699
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.926720188433315e-05,
+      "loss": 34.2552,
+      "step": 1700
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.926585379875055e-05,
+      "loss": 0.7254,
+      "step": 1701
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.926450448348097e-05,
+      "loss": 32.4762,
+      "step": 1702
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.926315393855812e-05,
+      "loss": 0.7975,
+      "step": 1703
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.926180216401571e-05,
+      "loss": 0.788,
+      "step": 1704
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.926044915988743e-05,
+      "loss": 0.7631,
+      "step": 1705
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.925909492620713e-05,
+      "loss": 35.807,
+      "step": 1706
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.925773946300855e-05,
+      "loss": 0.81,
+      "step": 1707
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.925638277032557e-05,
+      "loss": 34.9909,
+      "step": 1708
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.925502484819201e-05,
+      "loss": 31.0533,
+      "step": 1709
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.925366569664178e-05,
+      "loss": 0.7224,
+      "step": 1710
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.925230531570882e-05,
+      "loss": 0.7971,
+      "step": 1711
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.925094370542708e-05,
+      "loss": 38.1067,
+      "step": 1712
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.924958086583054e-05,
+      "loss": 37.264,
+      "step": 1713
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.92482167969532e-05,
+      "loss": 31.2824,
+      "step": 1714
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.924685149882913e-05,
+      "loss": 0.7448,
+      "step": 1715
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.92454849714924e-05,
+      "loss": 0.7692,
+      "step": 1716
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.924411721497713e-05,
+      "loss": 36.5103,
+      "step": 1717
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.924274822931744e-05,
+      "loss": 0.7559,
+      "step": 1718
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.924137801454751e-05,
+      "loss": 34.0172,
+      "step": 1719
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.924000657070155e-05,
+      "loss": 0.7838,
+      "step": 1720
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.923863389781378e-05,
+      "loss": 0.7519,
+      "step": 1721
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.923725999591847e-05,
+      "loss": 0.8075,
+      "step": 1722
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.923588486504989e-05,
+      "loss": 0.7771,
+      "step": 1723
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.923450850524241e-05,
+      "loss": 0.8166,
+      "step": 1724
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.923313091653032e-05,
+      "loss": 39.6367,
+      "step": 1725
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.923175209894809e-05,
+      "loss": 0.7628,
+      "step": 1726
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.923037205253006e-05,
+      "loss": 35.9494,
+      "step": 1727
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.922899077731069e-05,
+      "loss": 33.8289,
+      "step": 1728
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.922760827332447e-05,
+      "loss": 34.1235,
+      "step": 1729
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.922622454060593e-05,
+      "loss": 35.1437,
+      "step": 1730
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.922483957918956e-05,
+      "loss": 35.4558,
+      "step": 1731
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.922345338910998e-05,
+      "loss": 0.8064,
+      "step": 1732
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.922206597040176e-05,
+      "loss": 31.798,
+      "step": 1733
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.922067732309952e-05,
+      "loss": 35.8492,
+      "step": 1734
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.921928744723794e-05,
+      "loss": 32.4734,
+      "step": 1735
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.92178963428517e-05,
+      "loss": 31.9823,
+      "step": 1736
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.921650400997555e-05,
+      "loss": 0.7198,
+      "step": 1737
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.921511044864421e-05,
+      "loss": 0.746,
+      "step": 1738
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.921371565889249e-05,
+      "loss": 38.2273,
+      "step": 1739
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.921231964075517e-05,
+      "loss": 0.762,
+      "step": 1740
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.921092239426712e-05,
+      "loss": 0.8033,
+      "step": 1741
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.920952391946321e-05,
+      "loss": 40.5314,
+      "step": 1742
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.920812421637836e-05,
+      "loss": 35.9787,
+      "step": 1743
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.920672328504748e-05,
+      "loss": 33.4805,
+      "step": 1744
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.920532112550555e-05,
+      "loss": 0.7923,
+      "step": 1745
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.920391773778757e-05,
+      "loss": 0.7272,
+      "step": 1746
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.920251312192859e-05,
+      "loss": 0.7331,
+      "step": 1747
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.920110727796363e-05,
+      "loss": 0.7604,
+      "step": 1748
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.91997002059278e-05,
+      "loss": 50.5739,
+      "step": 1749
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.919829190585621e-05,
+      "loss": 0.7469,
+      "step": 1750
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.919688237778404e-05,
+      "loss": 46.4117,
+      "step": 1751
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.919547162174643e-05,
+      "loss": 0.7562,
+      "step": 1752
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.919405963777864e-05,
+      "loss": 0.7477,
+      "step": 1753
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.919264642591587e-05,
+      "loss": 40.1069,
+      "step": 1754
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.919123198619342e-05,
+      "loss": 0.7496,
+      "step": 1755
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.918981631864659e-05,
+      "loss": 39.2104,
+      "step": 1756
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.918839942331071e-05,
+      "loss": 0.7086,
+      "step": 1757
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.918698130022114e-05,
+      "loss": 34.4087,
+      "step": 1758
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.91855619494133e-05,
+      "loss": 43.8158,
+      "step": 1759
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.91841413709226e-05,
+      "loss": 0.824,
+      "step": 1760
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.918271956478451e-05,
+      "loss": 0.7656,
+      "step": 1761
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.91812965310345e-05,
+      "loss": 33.6669,
+      "step": 1762
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.91798722697081e-05,
+      "loss": 31.8166,
+      "step": 1763
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.917844678084084e-05,
+      "loss": 0.7529,
+      "step": 1764
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.917702006446836e-05,
+      "loss": 33.8829,
+      "step": 1765
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.917559212062621e-05,
+      "loss": 0.7718,
+      "step": 1766
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.917416294935004e-05,
+      "loss": 32.4053,
+      "step": 1767
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.917273255067555e-05,
+      "loss": 0.7846,
+      "step": 1768
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.917130092463843e-05,
+      "loss": 0.7619,
+      "step": 1769
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.91698680712744e-05,
+      "loss": 0.8184,
+      "step": 1770
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.916843399061924e-05,
+      "loss": 34.4206,
+      "step": 1771
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.916699868270875e-05,
+      "loss": 34.6268,
+      "step": 1772
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.916556214757873e-05,
+      "loss": 33.6949,
+      "step": 1773
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.916412438526507e-05,
+      "loss": 36.5211,
+      "step": 1774
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.916268539580363e-05,
+      "loss": 0.7278,
+      "step": 1775
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.916124517923033e-05,
+      "loss": 33.9439,
+      "step": 1776
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.915980373558113e-05,
+      "loss": 34.8321,
+      "step": 1777
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.915836106489203e-05,
+      "loss": 34.1042,
+      "step": 1778
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.915691716719898e-05,
+      "loss": 31.5257,
+      "step": 1779
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.915547204253807e-05,
+      "loss": 35.5906,
+      "step": 1780
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.915402569094534e-05,
+      "loss": 33.9264,
+      "step": 1781
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.915257811245693e-05,
+      "loss": 0.7767,
+      "step": 1782
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.915112930710892e-05,
+      "loss": 0.8062,
+      "step": 1783
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.914967927493752e-05,
+      "loss": 33.4634,
+      "step": 1784
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.91482280159789e-05,
+      "loss": 35.2066,
+      "step": 1785
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.914677553026928e-05,
+      "loss": 0.7481,
+      "step": 1786
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.914532181784492e-05,
+      "loss": 34.9376,
+      "step": 1787
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.914386687874211e-05,
+      "loss": 0.7898,
+      "step": 1788
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.914241071299716e-05,
+      "loss": 32.691,
+      "step": 1789
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.914095332064641e-05,
+      "loss": 29.8564,
+      "step": 1790
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.913949470172625e-05,
+      "loss": 37.1838,
+      "step": 1791
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.91380348562731e-05,
+      "loss": 0.764,
+      "step": 1792
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.913657378432337e-05,
+      "loss": 34.8299,
+      "step": 1793
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.913511148591352e-05,
+      "loss": 28.6245,
+      "step": 1794
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.913364796108008e-05,
+      "loss": 32.2786,
+      "step": 1795
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.913218320985958e-05,
+      "loss": 0.6973,
+      "step": 1796
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.913071723228856e-05,
+      "loss": 0.7841,
+      "step": 1797
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.912925002840363e-05,
+      "loss": 0.7726,
+      "step": 1798
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.912778159824138e-05,
+      "loss": 0.7882,
+      "step": 1799
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.91263119418385e-05,
+      "loss": 33.5849,
+      "step": 1800
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.912484105923165e-05,
+      "loss": 36.0585,
+      "step": 1801
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.912336895045755e-05,
+      "loss": 38.1705,
+      "step": 1802
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.912189561555297e-05,
+      "loss": 34.5071,
+      "step": 1803
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.912042105455463e-05,
+      "loss": 0.686,
+      "step": 1804
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.911894526749936e-05,
+      "loss": 0.8029,
+      "step": 1805
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.911746825442401e-05,
+      "loss": 38.3079,
+      "step": 1806
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.911599001536544e-05,
+      "loss": 0.7581,
+      "step": 1807
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.911451055036056e-05,
+      "loss": 0.769,
+      "step": 1808
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.911302985944627e-05,
+      "loss": 0.7115,
+      "step": 1809
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.911154794265953e-05,
+      "loss": 37.4198,
+      "step": 1810
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.911006480003734e-05,
+      "loss": 0.7639,
+      "step": 1811
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.910858043161674e-05,
+      "loss": 35.6944,
+      "step": 1812
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.910709483743474e-05,
+      "loss": 35.2569,
+      "step": 1813
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.910560801752845e-05,
+      "loss": 0.7337,
+      "step": 1814
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.910411997193497e-05,
+      "loss": 35.4138,
+      "step": 1815
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.910263070069144e-05,
+      "loss": 31.4748,
+      "step": 1816
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.910114020383504e-05,
+      "loss": 0.7923,
+      "step": 1817
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.909964848140296e-05,
+      "loss": 31.4916,
+      "step": 1818
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.909815553343245e-05,
+      "loss": 32.8661,
+      "step": 1819
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.909666135996078e-05,
+      "loss": 0.7798,
+      "step": 1820
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.909516596102522e-05,
+      "loss": 33.2087,
+      "step": 1821
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.90936693366631e-05,
+      "loss": 0.7839,
+      "step": 1822
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.90921714869118e-05,
+      "loss": 0.7718,
+      "step": 1823
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.909067241180868e-05,
+      "loss": 34.3529,
+      "step": 1824
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.908917211139116e-05,
+      "loss": 0.7976,
+      "step": 1825
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.908767058569671e-05,
+      "loss": 37.5117,
+      "step": 1826
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.908616783476279e-05,
+      "loss": 34.5298,
+      "step": 1827
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.908466385862691e-05,
+      "loss": 0.8211,
+      "step": 1828
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.908315865732663e-05,
+      "loss": 31.4581,
+      "step": 1829
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.908165223089947e-05,
+      "loss": 35.986,
+      "step": 1830
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.908014457938309e-05,
+      "loss": 0.7637,
+      "step": 1831
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.907863570281508e-05,
+      "loss": 35.7706,
+      "step": 1832
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.907712560123312e-05,
+      "loss": 0.7867,
+      "step": 1833
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.90756142746749e-05,
+      "loss": 33.2537,
+      "step": 1834
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.907410172317815e-05,
+      "loss": 0.7859,
+      "step": 1835
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.907258794678062e-05,
+      "loss": 0.7945,
+      "step": 1836
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.907107294552007e-05,
+      "loss": 34.3009,
+      "step": 1837
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.906955671943436e-05,
+      "loss": 0.7783,
+      "step": 1838
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.906803926856128e-05,
+      "loss": 0.7907,
+      "step": 1839
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.906652059293876e-05,
+      "loss": 34.5141,
+      "step": 1840
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.906500069260467e-05,
+      "loss": 0.7418,
+      "step": 1841
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.906347956759696e-05,
+      "loss": 0.7158,
+      "step": 1842
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.906195721795362e-05,
+      "loss": 34.4526,
+      "step": 1843
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.90604336437126e-05,
+      "loss": 0.7846,
+      "step": 1844
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.905890884491195e-05,
+      "loss": 0.8046,
+      "step": 1845
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.905738282158975e-05,
+      "loss": 0.8413,
+      "step": 1846
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.905585557378407e-05,
+      "loss": 35.4448,
+      "step": 1847
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.905432710153304e-05,
+      "loss": 0.7834,
+      "step": 1848
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.905279740487478e-05,
+      "loss": 0.7787,
+      "step": 1849
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.905126648384753e-05,
+      "loss": 33.5267,
+      "step": 1850
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.904973433848945e-05,
+      "loss": 0.7991,
+      "step": 1851
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.90482009688388e-05,
+      "loss": 0.7311,
+      "step": 1852
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.904666637493386e-05,
+      "loss": 31.6761,
+      "step": 1853
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.904513055681293e-05,
+      "loss": 0.7232,
+      "step": 1854
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.904359351451433e-05,
+      "loss": 33.0196,
+      "step": 1855
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.904205524807646e-05,
+      "loss": 31.6054,
+      "step": 1856
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.904051575753767e-05,
+      "loss": 31.2541,
+      "step": 1857
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.903897504293642e-05,
+      "loss": 0.7617,
+      "step": 1858
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.903743310431117e-05,
+      "loss": 34.3431,
+      "step": 1859
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.903588994170037e-05,
+      "loss": 0.7658,
+      "step": 1860
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.903434555514257e-05,
+      "loss": 33.7826,
+      "step": 1861
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.903279994467632e-05,
+      "loss": 0.7825,
+      "step": 1862
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.903125311034018e-05,
+      "loss": 31.8535,
+      "step": 1863
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.902970505217275e-05,
+      "loss": 0.7443,
+      "step": 1864
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.90281557702127e-05,
+      "loss": 32.5019,
+      "step": 1865
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.90266052644987e-05,
+      "loss": 31.6793,
+      "step": 1866
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.902505353506942e-05,
+      "loss": 34.9978,
+      "step": 1867
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.902350058196361e-05,
+      "loss": 0.8724,
+      "step": 1868
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.902194640522005e-05,
+      "loss": 0.8032,
+      "step": 1869
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.90203910048775e-05,
+      "loss": 32.9667,
+      "step": 1870
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.90188343809748e-05,
+      "loss": 32.4098,
+      "step": 1871
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.90172765335508e-05,
+      "loss": 0.7307,
+      "step": 1872
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.90157174626444e-05,
+      "loss": 35.0111,
+      "step": 1873
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.901415716829448e-05,
+      "loss": 0.7919,
+      "step": 1874
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.901259565054002e-05,
+      "loss": 0.7224,
+      "step": 1875
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.901103290941997e-05,
+      "loss": 0.779,
+      "step": 1876
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.900946894497335e-05,
+      "loss": 36.8074,
+      "step": 1877
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.90079037572392e-05,
+      "loss": 0.7507,
+      "step": 1878
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.900633734625657e-05,
+      "loss": 0.7538,
+      "step": 1879
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.900476971206458e-05,
+      "loss": 0.7686,
+      "step": 1880
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.900320085470236e-05,
+      "loss": 34.8178,
+      "step": 1881
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.900163077420905e-05,
+      "loss": 34.85,
+      "step": 1882
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.900005947062384e-05,
+      "loss": 0.8351,
+      "step": 1883
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.899848694398596e-05,
+      "loss": 31.9465,
+      "step": 1884
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.899691319433466e-05,
+      "loss": 33.9069,
+      "step": 1885
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.899533822170922e-05,
+      "loss": 0.8162,
+      "step": 1886
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.899376202614895e-05,
+      "loss": 33.0807,
+      "step": 1887
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.899218460769319e-05,
+      "loss": 33.0079,
+      "step": 1888
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.899060596638132e-05,
+      "loss": 33.5571,
+      "step": 1889
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.898902610225273e-05,
+      "loss": 0.7968,
+      "step": 1890
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.898744501534689e-05,
+      "loss": 34.0783,
+      "step": 1891
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.898586270570321e-05,
+      "loss": 0.7773,
+      "step": 1892
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.898427917336121e-05,
+      "loss": 0.7816,
+      "step": 1893
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.898269441836043e-05,
+      "loss": 0.723,
+      "step": 1894
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.898110844074042e-05,
+      "loss": 0.7522,
+      "step": 1895
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.897952124054075e-05,
+      "loss": 33.4455,
+      "step": 1896
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.897793281780104e-05,
+      "loss": 34.8105,
+      "step": 1897
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.897634317256095e-05,
+      "loss": 32.5035,
+      "step": 1898
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.897475230486015e-05,
+      "loss": 35.3663,
+      "step": 1899
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.897316021473836e-05,
+      "loss": 0.7533,
+      "step": 1900
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.89715669022353e-05,
+      "loss": 0.7092,
+      "step": 1901
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.896997236739076e-05,
+      "loss": 30.5666,
+      "step": 1902
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.89683766102445e-05,
+      "loss": 0.7562,
+      "step": 1903
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.89667796308364e-05,
+      "loss": 0.7841,
+      "step": 1904
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.89651814292063e-05,
+      "loss": 37.0478,
+      "step": 1905
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.896358200539408e-05,
+      "loss": 0.7421,
+      "step": 1906
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.896198135943968e-05,
+      "loss": 31.6907,
+      "step": 1907
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.896037949138306e-05,
+      "loss": 33.7504,
+      "step": 1908
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.895877640126416e-05,
+      "loss": 0.7975,
+      "step": 1909
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.895717208912306e-05,
+      "loss": 33.6536,
+      "step": 1910
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.895556655499974e-05,
+      "loss": 30.4596,
+      "step": 1911
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.89539597989343e-05,
+      "loss": 30.2534,
+      "step": 1912
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.895235182096685e-05,
+      "loss": 0.7736,
+      "step": 1913
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.895074262113752e-05,
+      "loss": 0.7506,
+      "step": 1914
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.894913219948647e-05,
+      "loss": 0.7696,
+      "step": 1915
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.894752055605392e-05,
+      "loss": 34.1521,
+      "step": 1916
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.894590769088008e-05,
+      "loss": 0.7533,
+      "step": 1917
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.89442936040052e-05,
+      "loss": 0.7792,
+      "step": 1918
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.894267829546957e-05,
+      "loss": 29.8644,
+      "step": 1919
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.894106176531353e-05,
+      "loss": 35.9889,
+      "step": 1920
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.893944401357739e-05,
+      "loss": 31.6011,
+      "step": 1921
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.893782504030159e-05,
+      "loss": 0.7823,
+      "step": 1922
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.893620484552648e-05,
+      "loss": 33.4447,
+      "step": 1923
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.893458342929252e-05,
+      "loss": 0.7499,
+      "step": 1924
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.89329607916402e-05,
+      "loss": 33.6306,
+      "step": 1925
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.893133693260999e-05,
+      "loss": 32.9238,
+      "step": 1926
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.892971185224245e-05,
+      "loss": 36.8969,
+      "step": 1927
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.892808555057812e-05,
+      "loss": 32.4273,
+      "step": 1928
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.892645802765761e-05,
+      "loss": 0.7955,
+      "step": 1929
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.892482928352154e-05,
+      "loss": 0.7642,
+      "step": 1930
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.892319931821055e-05,
+      "loss": 33.1123,
+      "step": 1931
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.892156813176533e-05,
+      "loss": 31.8347,
+      "step": 1932
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.891993572422661e-05,
+      "loss": 34.867,
+      "step": 1933
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.891830209563512e-05,
+      "loss": 32.5788,
+      "step": 1934
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.891666724603163e-05,
+      "loss": 0.7198,
+      "step": 1935
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.891503117545698e-05,
+      "loss": 31.6252,
+      "step": 1936
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.891339388395198e-05,
+      "loss": 36.8898,
+      "step": 1937
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.891175537155747e-05,
+      "loss": 0.7532,
+      "step": 1938
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.891011563831441e-05,
+      "loss": 0.8146,
+      "step": 1939
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.890847468426368e-05,
+      "loss": 0.8251,
+      "step": 1940
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.890683250944627e-05,
+      "loss": 31.3236,
+      "step": 1941
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.890518911390314e-05,
+      "loss": 0.7321,
+      "step": 1942
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.890354449767533e-05,
+      "loss": 0.6772,
+      "step": 1943
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.890189866080389e-05,
+      "loss": 32.711,
+      "step": 1944
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.890025160332988e-05,
+      "loss": 0.7853,
+      "step": 1945
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.889860332529441e-05,
+      "loss": 0.7623,
+      "step": 1946
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.889695382673867e-05,
+      "loss": 0.7583,
+      "step": 1947
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.889530310770378e-05,
+      "loss": 30.9809,
+      "step": 1948
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.889365116823095e-05,
+      "loss": 30.9986,
+      "step": 1949
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.889199800836144e-05,
+      "loss": 31.9832,
+      "step": 1950
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.88903436281365e-05,
+      "loss": 0.7648,
+      "step": 1951
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.888868802759741e-05,
+      "loss": 31.5453,
+      "step": 1952
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.88870312067855e-05,
+      "loss": 32.0544,
+      "step": 1953
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.888537316574213e-05,
+      "loss": 32.6548,
+      "step": 1954
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.88837139045087e-05,
+      "loss": 28.6802,
+      "step": 1955
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.888205342312659e-05,
+      "loss": 0.7332,
+      "step": 1956
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.888039172163728e-05,
+      "loss": 32.489,
+      "step": 1957
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.887872880008223e-05,
+      "loss": 28.378,
+      "step": 1958
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.887706465850293e-05,
+      "loss": 31.9723,
+      "step": 1959
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.887539929694095e-05,
+      "loss": 30.6159,
+      "step": 1960
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.887373271543784e-05,
+      "loss": 31.7866,
+      "step": 1961
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.887206491403522e-05,
+      "loss": 32.1848,
+      "step": 1962
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.887039589277467e-05,
+      "loss": 30.3134,
+      "step": 1963
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.88687256516979e-05,
+      "loss": 34.9186,
+      "step": 1964
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.886705419084658e-05,
+      "loss": 0.8063,
+      "step": 1965
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.886538151026242e-05,
+      "loss": 0.7921,
+      "step": 1966
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.886370760998718e-05,
+      "loss": 0.7397,
+      "step": 1967
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.886203249006265e-05,
+      "loss": 28.7607,
+      "step": 1968
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.886035615053064e-05,
+      "loss": 31.6585,
+      "step": 1969
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.885867859143296e-05,
+      "loss": 0.8067,
+      "step": 1970
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.885699981281152e-05,
+      "loss": 0.6834,
+      "step": 1971
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.885531981470822e-05,
+      "loss": 32.6875,
+      "step": 1972
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.885363859716497e-05,
+      "loss": 33.6138,
+      "step": 1973
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.885195616022373e-05,
+      "loss": 0.781,
+      "step": 1974
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.885027250392653e-05,
+      "loss": 33.4669,
+      "step": 1975
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.884858762831538e-05,
+      "loss": 32.8699,
+      "step": 1976
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.884690153343232e-05,
+      "loss": 0.8072,
+      "step": 1977
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.884521421931945e-05,
+      "loss": 30.5782,
+      "step": 1978
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.884352568601887e-05,
+      "loss": 30.8974,
+      "step": 1979
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.884183593357274e-05,
+      "loss": 0.7844,
+      "step": 1980
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.884014496202323e-05,
+      "loss": 0.7481,
+      "step": 1981
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.883845277141256e-05,
+      "loss": 29.2424,
+      "step": 1982
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.883675936178295e-05,
+      "loss": 28.7904,
+      "step": 1983
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.883506473317668e-05,
+      "loss": 0.765,
+      "step": 1984
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.883336888563603e-05,
+      "loss": 0.729,
+      "step": 1985
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.883167181920337e-05,
+      "loss": 33.2553,
+      "step": 1986
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.882997353392101e-05,
+      "loss": 0.753,
+      "step": 1987
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.882827402983137e-05,
+      "loss": 32.2699,
+      "step": 1988
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.882657330697685e-05,
+      "loss": 0.7639,
+      "step": 1989
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.882487136539993e-05,
+      "loss": 30.9132,
+      "step": 1990
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.882316820514306e-05,
+      "loss": 0.8035,
+      "step": 1991
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.882146382624877e-05,
+      "loss": 33.7957,
+      "step": 1992
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.881975822875958e-05,
+      "loss": 28.8747,
+      "step": 1993
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.881805141271809e-05,
+      "loss": 32.3321,
+      "step": 1994
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.88163433781669e-05,
+      "loss": 32.0685,
+      "step": 1995
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.881463412514862e-05,
+      "loss": 30.0311,
+      "step": 1996
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.881292365370593e-05,
+      "loss": 30.5017,
+      "step": 1997
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.881121196388152e-05,
+      "loss": 0.7266,
+      "step": 1998
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.88094990557181e-05,
+      "loss": 0.7712,
+      "step": 1999
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.880778492925846e-05,
+      "loss": 33.8183,
+      "step": 2000
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.880606958454536e-05,
+      "loss": 34.3507,
+      "step": 2001
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.88043530216216e-05,
+      "loss": 37.3859,
+      "step": 2002
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.880263524053007e-05,
+      "loss": 33.3788,
+      "step": 2003
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.880091624131362e-05,
+      "loss": 0.7685,
+      "step": 2004
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.879919602401515e-05,
+      "loss": 0.7799,
+      "step": 2005
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.879747458867761e-05,
+      "loss": 34.249,
+      "step": 2006
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.879575193534395e-05,
+      "loss": 0.7492,
+      "step": 2007
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.879402806405718e-05,
+      "loss": 33.5236,
+      "step": 2008
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.879230297486034e-05,
+      "loss": 30.7372,
+      "step": 2009
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.879057666779648e-05,
+      "loss": 30.6607,
+      "step": 2010
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.878884914290867e-05,
+      "loss": 0.7453,
+      "step": 2011
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.878712040024006e-05,
+      "loss": 31.8322,
+      "step": 2012
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.878539043983377e-05,
+      "loss": 31.0064,
+      "step": 2013
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.8783659261733e-05,
+      "loss": 0.7683,
+      "step": 2014
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.878192686598098e-05,
+      "loss": 0.7611,
+      "step": 2015
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.87801932526209e-05,
+      "loss": 29.4209,
+      "step": 2016
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.877845842169607e-05,
+      "loss": 0.7615,
+      "step": 2017
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.877672237324978e-05,
+      "loss": 0.764,
+      "step": 2018
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.877498510732534e-05,
+      "loss": 0.7315,
+      "step": 2019
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.877324662396616e-05,
+      "loss": 32.7965,
+      "step": 2020
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.877150692321562e-05,
+      "loss": 28.9509,
+      "step": 2021
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.87697660051171e-05,
+      "loss": 0.7937,
+      "step": 2022
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.87680238697141e-05,
+      "loss": 0.7669,
+      "step": 2023
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.876628051705008e-05,
+      "loss": 0.7495,
+      "step": 2024
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.876453594716858e-05,
+      "loss": 32.0462,
+      "step": 2025
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.876279016011312e-05,
+      "loss": 35.1168,
+      "step": 2026
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.876104315592728e-05,
+      "loss": 33.8326,
+      "step": 2027
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.875929493465468e-05,
+      "loss": 33.8059,
+      "step": 2028
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.875754549633893e-05,
+      "loss": 0.7956,
+      "step": 2029
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.875579484102371e-05,
+      "loss": 0.7983,
+      "step": 2030
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.875404296875274e-05,
+      "loss": 33.4294,
+      "step": 2031
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.875228987956971e-05,
+      "loss": 32.669,
+      "step": 2032
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.875053557351838e-05,
+      "loss": 0.7522,
+      "step": 2033
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.874878005064259e-05,
+      "loss": 0.7812,
+      "step": 2034
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.874702331098608e-05,
+      "loss": 31.2431,
+      "step": 2035
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.874526535459277e-05,
+      "loss": 34.1619,
+      "step": 2036
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.874350618150647e-05,
+      "loss": 0.7594,
+      "step": 2037
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.874174579177116e-05,
+      "loss": 35.0114,
+      "step": 2038
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.873998418543073e-05,
+      "loss": 27.0103,
+      "step": 2039
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.873822136252917e-05,
+      "loss": 29.8325,
+      "step": 2040
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.873645732311048e-05,
+      "loss": 0.7404,
+      "step": 2041
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.873469206721868e-05,
+      "loss": 29.9427,
+      "step": 2042
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.873292559489784e-05,
+      "loss": 0.7344,
+      "step": 2043
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.873115790619206e-05,
+      "loss": 28.0189,
+      "step": 2044
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.872938900114543e-05,
+      "loss": 0.7743,
+      "step": 2045
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.872761887980213e-05,
+      "loss": 0.8059,
+      "step": 2046
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.872584754220635e-05,
+      "loss": 0.7326,
+      "step": 2047
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.872407498840229e-05,
+      "loss": 0.7295,
+      "step": 2048
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.872230121843418e-05,
+      "loss": 33.5095,
+      "step": 2049
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.872052623234632e-05,
+      "loss": 33.0546,
+      "step": 2050
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.871875003018298e-05,
+      "loss": 30.3808,
+      "step": 2051
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.871697261198853e-05,
+      "loss": 0.7068,
+      "step": 2052
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.871519397780732e-05,
+      "loss": 32.3577,
+      "step": 2053
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.871341412768374e-05,
+      "loss": 0.7988,
+      "step": 2054
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.871163306166223e-05,
+      "loss": 31.3149,
+      "step": 2055
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.870985077978723e-05,
+      "loss": 30.2827,
+      "step": 2056
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.870806728210323e-05,
+      "loss": 0.777,
+      "step": 2057
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.870628256865475e-05,
+      "loss": 0.7645,
+      "step": 2058
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.870449663948635e-05,
+      "loss": 0.7707,
+      "step": 2059
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.870270949464258e-05,
+      "loss": 36.2369,
+      "step": 2060
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.870092113416807e-05,
+      "loss": 33.1551,
+      "step": 2061
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.869913155810744e-05,
+      "loss": 33.3207,
+      "step": 2062
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.869734076650537e-05,
+      "loss": 0.7175,
+      "step": 2063
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.869554875940654e-05,
+      "loss": 29.2793,
+      "step": 2064
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.869375553685572e-05,
+      "loss": 33.4068,
+      "step": 2065
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.869196109889762e-05,
+      "loss": 30.616,
+      "step": 2066
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.869016544557707e-05,
+      "loss": 31.7637,
+      "step": 2067
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.868836857693887e-05,
+      "loss": 35.5194,
+      "step": 2068
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.868657049302787e-05,
+      "loss": 0.7985,
+      "step": 2069
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.868477119388896e-05,
+      "loss": 30.4892,
+      "step": 2070
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.868297067956704e-05,
+      "loss": 31.5105,
+      "step": 2071
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.868116895010706e-05,
+      "loss": 30.7937,
+      "step": 2072
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.867936600555398e-05,
+      "loss": 0.7591,
+      "step": 2073
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.867756184595281e-05,
+      "loss": 32.8292,
+      "step": 2074
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.867575647134859e-05,
+      "loss": 0.7261,
+      "step": 2075
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.867394988178637e-05,
+      "loss": 0.7972,
+      "step": 2076
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.867214207731124e-05,
+      "loss": 0.8038,
+      "step": 2077
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.867033305796833e-05,
+      "loss": 0.7608,
+      "step": 2078
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.866852282380282e-05,
+      "loss": 0.7204,
+      "step": 2079
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.866671137485985e-05,
+      "loss": 33.6412,
+      "step": 2080
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.866489871118464e-05,
+      "loss": 31.4743,
+      "step": 2081
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.866308483282246e-05,
+      "loss": 0.747,
+      "step": 2082
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.866126973981857e-05,
+      "loss": 31.3435,
+      "step": 2083
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.865945343221828e-05,
+      "loss": 0.7525,
+      "step": 2084
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.865763591006691e-05,
+      "loss": 30.3355,
+      "step": 2085
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.865581717340984e-05,
+      "loss": 35.9908,
+      "step": 2086
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.865399722229247e-05,
+      "loss": 0.773,
+      "step": 2087
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.865217605676022e-05,
+      "loss": 0.7351,
+      "step": 2088
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.865035367685855e-05,
+      "loss": 31.8928,
+      "step": 2089
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.864853008263293e-05,
+      "loss": 0.7242,
+      "step": 2090
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.864670527412891e-05,
+      "loss": 30.4375,
+      "step": 2091
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.8644879251392e-05,
+      "loss": 33.3542,
+      "step": 2092
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.864305201446782e-05,
+      "loss": 29.8907,
+      "step": 2093
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.864122356340194e-05,
+      "loss": 0.7731,
+      "step": 2094
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.863939389824002e-05,
+      "loss": 29.663,
+      "step": 2095
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.863756301902772e-05,
+      "loss": 31.5798,
+      "step": 2096
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.863573092581073e-05,
+      "loss": 28.7693,
+      "step": 2097
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.86338976186348e-05,
+      "loss": 0.827,
+      "step": 2098
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.863206309754569e-05,
+      "loss": 32.064,
+      "step": 2099
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.863022736258917e-05,
+      "loss": 0.8428,
+      "step": 2100
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.862839041381108e-05,
+      "loss": 29.6618,
+      "step": 2101
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.862655225125726e-05,
+      "loss": 30.0229,
+      "step": 2102
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.862471287497358e-05,
+      "loss": 27.5246,
+      "step": 2103
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.862287228500598e-05,
+      "loss": 29.7722,
+      "step": 2104
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.862103048140037e-05,
+      "loss": 0.7434,
+      "step": 2105
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.861918746420276e-05,
+      "loss": 0.7658,
+      "step": 2106
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.86173432334591e-05,
+      "loss": 29.896,
+      "step": 2107
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.861549778921548e-05,
+      "loss": 0.7329,
+      "step": 2108
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.861365113151791e-05,
+      "loss": 29.3952,
+      "step": 2109
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.861180326041253e-05,
+      "loss": 0.7268,
+      "step": 2110
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.860995417594543e-05,
+      "loss": 0.7935,
+      "step": 2111
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.860810387816277e-05,
+      "loss": 33.6471,
+      "step": 2112
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.860625236711075e-05,
+      "loss": 30.7128,
+      "step": 2113
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.860439964283557e-05,
+      "loss": 0.7475,
+      "step": 2114
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.860254570538347e-05,
+      "loss": 0.7529,
+      "step": 2115
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.860069055480074e-05,
+      "loss": 0.7871,
+      "step": 2116
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.859883419113367e-05,
+      "loss": 30.6358,
+      "step": 2117
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.859697661442861e-05,
+      "loss": 30.4615,
+      "step": 2118
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.859511782473192e-05,
+      "loss": 0.8013,
+      "step": 2119
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.859325782208998e-05,
+      "loss": 32.5328,
+      "step": 2120
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.859139660654923e-05,
+      "loss": 0.8013,
+      "step": 2121
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.858953417815612e-05,
+      "loss": 32.8867,
+      "step": 2122
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.858767053695717e-05,
+      "loss": 0.8201,
+      "step": 2123
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.858580568299885e-05,
+      "loss": 0.7387,
+      "step": 2124
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.858393961632774e-05,
+      "loss": 27.3914,
+      "step": 2125
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.85820723369904e-05,
+      "loss": 29.6757,
+      "step": 2126
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.858020384503343e-05,
+      "loss": 0.7914,
+      "step": 2127
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.857833414050348e-05,
+      "loss": 0.7384,
+      "step": 2128
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.857646322344723e-05,
+      "loss": 32.0593,
+      "step": 2129
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.857459109391136e-05,
+      "loss": 0.7704,
+      "step": 2130
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.85727177519426e-05,
+      "loss": 31.0122,
+      "step": 2131
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.857084319758772e-05,
+      "loss": 30.311,
+      "step": 2132
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.85689674308935e-05,
+      "loss": 0.766,
+      "step": 2133
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.856709045190677e-05,
+      "loss": 0.7762,
+      "step": 2134
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.856521226067437e-05,
+      "loss": 31.5879,
+      "step": 2135
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.856333285724319e-05,
+      "loss": 32.4425,
+      "step": 2136
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.856145224166012e-05,
+      "loss": 29.8547,
+      "step": 2137
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.855957041397212e-05,
+      "loss": 28.7539,
+      "step": 2138
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.855768737422615e-05,
+      "loss": 27.0091,
+      "step": 2139
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.855580312246923e-05,
+      "loss": 29.4432,
+      "step": 2140
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.855391765874836e-05,
+      "loss": 34.0669,
+      "step": 2141
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.855203098311064e-05,
+      "loss": 0.772,
+      "step": 2142
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.855014309560313e-05,
+      "loss": 0.7656,
+      "step": 2143
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.854825399627294e-05,
+      "loss": 0.778,
+      "step": 2144
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.854636368516727e-05,
+      "loss": 28.2425,
+      "step": 2145
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.854447216233328e-05,
+      "loss": 0.8014,
+      "step": 2146
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.854257942781818e-05,
+      "loss": 30.3549,
+      "step": 2147
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.85406854816692e-05,
+      "loss": 30.8021,
+      "step": 2148
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.853879032393364e-05,
+      "loss": 0.7319,
+      "step": 2149
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.853689395465879e-05,
+      "loss": 0.7881,
+      "step": 2150
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.853499637389198e-05,
+      "loss": 31.8872,
+      "step": 2151
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.853309758168058e-05,
+      "loss": 30.4814,
+      "step": 2152
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.853119757807198e-05,
+      "loss": 0.762,
+      "step": 2153
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.85292963631136e-05,
+      "loss": 0.7323,
+      "step": 2154
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.852739393685292e-05,
+      "loss": 30.3821,
+      "step": 2155
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.85254902993374e-05,
+      "loss": 27.0681,
+      "step": 2156
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.852358545061455e-05,
+      "loss": 0.7602,
+      "step": 2157
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.852167939073195e-05,
+      "loss": 0.8086,
+      "step": 2158
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.851977211973713e-05,
+      "loss": 35.3627,
+      "step": 2159
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.851786363767775e-05,
+      "loss": 30.7393,
+      "step": 2160
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.851595394460139e-05,
+      "loss": 33.6401,
+      "step": 2161
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.851404304055575e-05,
+      "loss": 31.4034,
+      "step": 2162
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.851213092558851e-05,
+      "loss": 29.5,
+      "step": 2163
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.851021759974742e-05,
+      "loss": 30.0446,
+      "step": 2164
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.85083030630802e-05,
+      "loss": 0.8138,
+      "step": 2165
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.850638731563468e-05,
+      "loss": 31.0697,
+      "step": 2166
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.850447035745866e-05,
+      "loss": 28.4006,
+      "step": 2167
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.850255218859998e-05,
+      "loss": 27.5065,
+      "step": 2168
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.850063280910652e-05,
+      "loss": 29.2836,
+      "step": 2169
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.849871221902618e-05,
+      "loss": 0.7776,
+      "step": 2170
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.849679041840693e-05,
+      "loss": 0.797,
+      "step": 2171
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.84948674072967e-05,
+      "loss": 33.5916,
+      "step": 2172
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.849294318574351e-05,
+      "loss": 0.7628,
+      "step": 2173
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.84910177537954e-05,
+      "loss": 35.1992,
+      "step": 2174
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.848909111150041e-05,
+      "loss": 0.8159,
+      "step": 2175
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.848716325890662e-05,
+      "loss": 0.7778,
+      "step": 2176
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.848523419606218e-05,
+      "loss": 29.1197,
+      "step": 2177
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.848330392301521e-05,
+      "loss": 31.0803,
+      "step": 2178
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.848137243981393e-05,
+      "loss": 0.7746,
+      "step": 2179
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.847943974650649e-05,
+      "loss": 0.7401,
+      "step": 2180
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.847750584314118e-05,
+      "loss": 0.7645,
+      "step": 2181
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.847557072976626e-05,
+      "loss": 33.1623,
+      "step": 2182
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.847363440643002e-05,
+      "loss": 29.0592,
+      "step": 2183
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.847169687318081e-05,
+      "loss": 27.573,
+      "step": 2184
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.846975813006695e-05,
+      "loss": 27.46,
+      "step": 2185
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.846781817713689e-05,
+      "loss": 0.7479,
+      "step": 2186
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.846587701443902e-05,
+      "loss": 29.4872,
+      "step": 2187
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.846393464202177e-05,
+      "loss": 32.5409,
+      "step": 2188
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.846199105993365e-05,
+      "loss": 0.7618,
+      "step": 2189
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.846004626822318e-05,
+      "loss": 30.3995,
+      "step": 2190
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.845810026693889e-05,
+      "loss": 0.7994,
+      "step": 2191
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.845615305612934e-05,
+      "loss": 0.7477,
+      "step": 2192
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.845420463584316e-05,
+      "loss": 30.6685,
+      "step": 2193
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.845225500612896e-05,
+      "loss": 27.2634,
+      "step": 2194
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.845030416703542e-05,
+      "loss": 0.7256,
+      "step": 2195
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.844835211861124e-05,
+      "loss": 29.9402,
+      "step": 2196
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.844639886090509e-05,
+      "loss": 0.765,
+      "step": 2197
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.844444439396578e-05,
+      "loss": 32.4246,
+      "step": 2198
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.844248871784209e-05,
+      "loss": 30.7934,
+      "step": 2199
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.84405318325828e-05,
+      "loss": 28.6379,
+      "step": 2200
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.843857373823679e-05,
+      "loss": 29.2965,
+      "step": 2201
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.843661443485291e-05,
+      "loss": 28.6213,
+      "step": 2202
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.843465392248008e-05,
+      "loss": 27.7642,
+      "step": 2203
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.843269220116723e-05,
+      "loss": 30.6894,
+      "step": 2204
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.843072927096332e-05,
+      "loss": 28.4315,
+      "step": 2205
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.842876513191735e-05,
+      "loss": 0.7971,
+      "step": 2206
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.842679978407834e-05,
+      "loss": 0.7575,
+      "step": 2207
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.842483322749536e-05,
+      "loss": 0.7872,
+      "step": 2208
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.842286546221749e-05,
+      "loss": 0.7739,
+      "step": 2209
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.842089648829382e-05,
+      "loss": 29.4751,
+      "step": 2210
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.841892630577355e-05,
+      "loss": 30.0086,
+      "step": 2211
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.84169549147058e-05,
+      "loss": 0.7973,
+      "step": 2212
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.84149823151398e-05,
+      "loss": 28.8681,
+      "step": 2213
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.84130085071248e-05,
+      "loss": 0.7214,
+      "step": 2214
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.841103349071003e-05,
+      "loss": 0.8148,
+      "step": 2215
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.840905726594483e-05,
+      "loss": 0.8242,
+      "step": 2216
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.840707983287852e-05,
+      "loss": 0.7266,
+      "step": 2217
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.840510119156041e-05,
+      "loss": 30.1689,
+      "step": 2218
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.840312134203994e-05,
+      "loss": 0.75,
+      "step": 2219
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.840114028436652e-05,
+      "loss": 0.7599,
+      "step": 2220
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.839915801858958e-05,
+      "loss": 0.7814,
+      "step": 2221
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.83971745447586e-05,
+      "loss": 27.4393,
+      "step": 2222
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.839518986292308e-05,
+      "loss": 0.7572,
+      "step": 2223
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.839320397313259e-05,
+      "loss": 0.7884,
+      "step": 2224
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.839121687543669e-05,
+      "loss": 0.7688,
+      "step": 2225
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.838922856988495e-05,
+      "loss": 26.1783,
+      "step": 2226
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.838723905652701e-05,
+      "loss": 29.1896,
+      "step": 2227
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.838524833541254e-05,
+      "loss": 31.1365,
+      "step": 2228
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.838325640659123e-05,
+      "loss": 0.7364,
+      "step": 2229
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.838126327011279e-05,
+      "loss": 0.7844,
+      "step": 2230
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.837926892602698e-05,
+      "loss": 0.7043,
+      "step": 2231
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.837727337438357e-05,
+      "loss": 0.8028,
+      "step": 2232
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.837527661523235e-05,
+      "loss": 31.7541,
+      "step": 2233
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.837327864862319e-05,
+      "loss": 31.2678,
+      "step": 2234
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.837127947460596e-05,
+      "loss": 29.9996,
+      "step": 2235
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.836927909323053e-05,
+      "loss": 31.5019,
+      "step": 2236
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.836727750454688e-05,
+      "loss": 0.7645,
+      "step": 2237
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.836527470860491e-05,
+      "loss": 28.1645,
+      "step": 2238
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.836327070545465e-05,
+      "loss": 32.9744,
+      "step": 2239
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.836126549514612e-05,
+      "loss": 30.5269,
+      "step": 2240
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.835925907772936e-05,
+      "loss": 0.775,
+      "step": 2241
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.835725145325444e-05,
+      "loss": 28.7364,
+      "step": 2242
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.83552426217715e-05,
+      "loss": 0.7805,
+      "step": 2243
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.835323258333065e-05,
+      "loss": 0.7265,
+      "step": 2244
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.835122133798208e-05,
+      "loss": 28.7397,
+      "step": 2245
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.834920888577596e-05,
+      "loss": 0.7708,
+      "step": 2246
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.834719522676257e-05,
+      "loss": 0.7649,
+      "step": 2247
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.834518036099215e-05,
+      "loss": 0.8344,
+      "step": 2248
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.834316428851498e-05,
+      "loss": 28.8369,
+      "step": 2249
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.834114700938139e-05,
+      "loss": 0.7335,
+      "step": 2250
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.833912852364176e-05,
+      "loss": 0.7581,
+      "step": 2251
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.833710883134642e-05,
+      "loss": 31.1058,
+      "step": 2252
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.833508793254579e-05,
+      "loss": 28.1824,
+      "step": 2253
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.833306582729036e-05,
+      "loss": 31.6678,
+      "step": 2254
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.833104251563056e-05,
+      "loss": 0.7597,
+      "step": 2255
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.83290179976169e-05,
+      "loss": 25.3179,
+      "step": 2256
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.832699227329994e-05,
+      "loss": 0.7398,
+      "step": 2257
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.83249653427302e-05,
+      "loss": 29.3582,
+      "step": 2258
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.832293720595828e-05,
+      "loss": 30.1797,
+      "step": 2259
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.832090786303482e-05,
+      "loss": 27.1495,
+      "step": 2260
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.831887731401048e-05,
+      "loss": 0.7455,
+      "step": 2261
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.831684555893591e-05,
+      "loss": 32.5487,
+      "step": 2262
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.831481259786185e-05,
+      "loss": 31.837,
+      "step": 2263
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.831277843083904e-05,
+      "loss": 0.7969,
+      "step": 2264
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.831074305791825e-05,
+      "loss": 0.7439,
+      "step": 2265
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.830870647915027e-05,
+      "loss": 29.3754,
+      "step": 2266
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.830666869458595e-05,
+      "loss": 32.1956,
+      "step": 2267
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.830462970427616e-05,
+      "loss": 30.7406,
+      "step": 2268
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.830258950827178e-05,
+      "loss": 28.759,
+      "step": 2269
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.830054810662374e-05,
+      "loss": 0.7552,
+      "step": 2270
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.8298505499383e-05,
+      "loss": 0.7758,
+      "step": 2271
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.829646168660051e-05,
+      "loss": 0.765,
+      "step": 2272
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.829441666832731e-05,
+      "loss": 30.0432,
+      "step": 2273
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.829237044461445e-05,
+      "loss": 29.5155,
+      "step": 2274
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.829032301551301e-05,
+      "loss": 0.7631,
+      "step": 2275
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.828827438107406e-05,
+      "loss": 30.8883,
+      "step": 2276
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.828622454134877e-05,
+      "loss": 0.7607,
+      "step": 2277
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.82841734963883e-05,
+      "loss": 29.1754,
+      "step": 2278
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.828212124624382e-05,
+      "loss": 0.7688,
+      "step": 2279
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.828006779096657e-05,
+      "loss": 0.8741,
+      "step": 2280
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.82780131306078e-05,
+      "loss": 26.3335,
+      "step": 2281
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.827595726521881e-05,
+      "loss": 26.0369,
+      "step": 2282
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.82739001948509e-05,
+      "loss": 30.0152,
+      "step": 2283
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.82718419195554e-05,
+      "loss": 29.1155,
+      "step": 2284
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.826978243938374e-05,
+      "loss": 0.7824,
+      "step": 2285
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.826772175438727e-05,
+      "loss": 0.746,
+      "step": 2286
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.826565986461744e-05,
+      "loss": 28.6215,
+      "step": 2287
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.826359677012572e-05,
+      "loss": 27.6213,
+      "step": 2288
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.826153247096362e-05,
+      "loss": 28.4823,
+      "step": 2289
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.825946696718262e-05,
+      "loss": 0.8278,
+      "step": 2290
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.825740025883433e-05,
+      "loss": 27.138,
+      "step": 2291
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.825533234597031e-05,
+      "loss": 0.7389,
+      "step": 2292
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.825326322864217e-05,
+      "loss": 0.7935,
+      "step": 2293
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.825119290690156e-05,
+      "loss": 0.7628,
+      "step": 2294
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.824912138080016e-05,
+      "loss": 27.0031,
+      "step": 2295
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.824704865038968e-05,
+      "loss": 27.076,
+      "step": 2296
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.824497471572183e-05,
+      "loss": 25.0563,
+      "step": 2297
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.824289957684842e-05,
+      "loss": 0.7451,
+      "step": 2298
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.82408232338212e-05,
+      "loss": 29.0683,
+      "step": 2299
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.823874568669203e-05,
+      "loss": 30.6255,
+      "step": 2300
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.823666693551276e-05,
+      "loss": 28.7325,
+      "step": 2301
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.823458698033525e-05,
+      "loss": 25.3146,
+      "step": 2302
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.823250582121144e-05,
+      "loss": 29.7091,
+      "step": 2303
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.823042345819329e-05,
+      "loss": 27.9586,
+      "step": 2304
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.822833989133274e-05,
+      "loss": 30.8927,
+      "step": 2305
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.822625512068181e-05,
+      "loss": 0.7495,
+      "step": 2306
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.822416914629257e-05,
+      "loss": 31.7838,
+      "step": 2307
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.822208196821703e-05,
+      "loss": 28.8998,
+      "step": 2308
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.821999358650733e-05,
+      "loss": 27.3173,
+      "step": 2309
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.821790400121557e-05,
+      "loss": 0.77,
+      "step": 2310
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.821581321239392e-05,
+      "loss": 33.4774,
+      "step": 2311
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.821372122009455e-05,
+      "loss": 0.7841,
+      "step": 2312
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.82116280243697e-05,
+      "loss": 0.7305,
+      "step": 2313
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.820953362527161e-05,
+      "loss": 0.7258,
+      "step": 2314
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.820743802285254e-05,
+      "loss": 0.6858,
+      "step": 2315
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.820534121716483e-05,
+      "loss": 29.7741,
+      "step": 2316
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.82032432082608e-05,
+      "loss": 30.8667,
+      "step": 2317
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.82011439961928e-05,
+      "loss": 28.7641,
+      "step": 2318
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.819904358101324e-05,
+      "loss": 33.1713,
+      "step": 2319
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.819694196277454e-05,
+      "loss": 0.7967,
+      "step": 2320
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.819483914152917e-05,
+      "loss": 0.799,
+      "step": 2321
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.819273511732961e-05,
+      "loss": 29.156,
+      "step": 2322
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.819062989022838e-05,
+      "loss": 0.7225,
+      "step": 2323
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.818852346027803e-05,
+      "loss": 0.8224,
+      "step": 2324
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.818641582753112e-05,
+      "loss": 0.7397,
+      "step": 2325
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.818430699204028e-05,
+      "loss": 0.7938,
+      "step": 2326
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.818219695385813e-05,
+      "loss": 0.8156,
+      "step": 2327
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.818008571303733e-05,
+      "loss": 0.744,
+      "step": 2328
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.817797326963061e-05,
+      "loss": 30.0168,
+      "step": 2329
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.817585962369067e-05,
+      "loss": 30.6054,
+      "step": 2330
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.817374477527027e-05,
+      "loss": 30.3599,
+      "step": 2331
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.817162872442221e-05,
+      "loss": 28.2586,
+      "step": 2332
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.816951147119928e-05,
+      "loss": 31.1123,
+      "step": 2333
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.816739301565436e-05,
+      "loss": 31.7018,
+      "step": 2334
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.816527335784032e-05,
+      "loss": 29.2669,
+      "step": 2335
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.816315249781004e-05,
+      "loss": 30.1935,
+      "step": 2336
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.816103043561648e-05,
+      "loss": 28.7705,
+      "step": 2337
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.81589071713126e-05,
+      "loss": 0.8013,
+      "step": 2338
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.815678270495142e-05,
+      "loss": 30.2926,
+      "step": 2339
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.815465703658593e-05,
+      "loss": 26.5219,
+      "step": 2340
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.815253016626922e-05,
+      "loss": 26.6806,
+      "step": 2341
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.815040209405436e-05,
+      "loss": 0.7,
+      "step": 2342
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.814827281999446e-05,
+      "loss": 25.9049,
+      "step": 2343
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.814614234414268e-05,
+      "loss": 0.7115,
+      "step": 2344
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.814401066655218e-05,
+      "loss": 0.7676,
+      "step": 2345
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.814187778727621e-05,
+      "loss": 0.7493,
+      "step": 2346
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.813974370636796e-05,
+      "loss": 0.7906,
+      "step": 2347
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.813760842388073e-05,
+      "loss": 0.8109,
+      "step": 2348
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.813547193986778e-05,
+      "loss": 0.7831,
+      "step": 2349
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.813333425438248e-05,
+      "loss": 27.9629,
+      "step": 2350
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.813119536747816e-05,
+      "loss": 0.8121,
+      "step": 2351
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.812905527920822e-05,
+      "loss": 0.7751,
+      "step": 2352
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.812691398962605e-05,
+      "loss": 0.7837,
+      "step": 2353
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.812477149878514e-05,
+      "loss": 25.8607,
+      "step": 2354
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.812262780673893e-05,
+      "loss": 0.7455,
+      "step": 2355
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.812048291354095e-05,
+      "loss": 29.9999,
+      "step": 2356
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.811833681924472e-05,
+      "loss": 0.7627,
+      "step": 2357
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.811618952390383e-05,
+      "loss": 28.1105,
+      "step": 2358
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.811404102757187e-05,
+      "loss": 29.6627,
+      "step": 2359
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.811189133030244e-05,
+      "loss": 0.7758,
+      "step": 2360
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.810974043214922e-05,
+      "loss": 0.7643,
+      "step": 2361
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.81075883331659e-05,
+      "loss": 0.8149,
+      "step": 2362
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.810543503340619e-05,
+      "loss": 33.0096,
+      "step": 2363
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.810328053292384e-05,
+      "loss": 0.7294,
+      "step": 2364
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.810112483177262e-05,
+      "loss": 27.4798,
+      "step": 2365
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.809896793000635e-05,
+      "loss": 0.7597,
+      "step": 2366
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.809680982767884e-05,
+      "loss": 0.7431,
+      "step": 2367
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.8094650524844e-05,
+      "loss": 28.0573,
+      "step": 2368
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.809249002155568e-05,
+      "loss": 26.4432,
+      "step": 2369
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.809032831786782e-05,
+      "loss": 28.0976,
+      "step": 2370
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.80881654138344e-05,
+      "loss": 29.0354,
+      "step": 2371
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.80860013095094e-05,
+      "loss": 27.7329,
+      "step": 2372
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.808383600494682e-05,
+      "loss": 0.7997,
+      "step": 2373
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.80816695002007e-05,
+      "loss": 29.5742,
+      "step": 2374
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.807950179532515e-05,
+      "loss": 29.6727,
+      "step": 2375
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.807733289037426e-05,
+      "loss": 0.8052,
+      "step": 2376
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.807516278540214e-05,
+      "loss": 27.0366,
+      "step": 2377
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.8072991480463e-05,
+      "loss": 27.7265,
+      "step": 2378
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.807081897561102e-05,
+      "loss": 28.3746,
+      "step": 2379
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.80686452709004e-05,
+      "loss": 26.546,
+      "step": 2380
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.806647036638543e-05,
+      "loss": 0.741,
+      "step": 2381
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.806429426212041e-05,
+      "loss": 0.8579,
+      "step": 2382
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.806211695815959e-05,
+      "loss": 26.1732,
+      "step": 2383
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.805993845455738e-05,
+      "loss": 0.7811,
+      "step": 2384
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.805775875136812e-05,
+      "loss": 28.7911,
+      "step": 2385
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.805557784864623e-05,
+      "loss": 0.721,
+      "step": 2386
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.805339574644617e-05,
+      "loss": 0.749,
+      "step": 2387
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.805121244482235e-05,
+      "loss": 0.7893,
+      "step": 2388
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.804902794382931e-05,
+      "loss": 27.0903,
+      "step": 2389
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.804684224352154e-05,
+      "loss": 25.5187,
+      "step": 2390
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.804465534395364e-05,
+      "loss": 0.7758,
+      "step": 2391
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.804246724518016e-05,
+      "loss": 0.7844,
+      "step": 2392
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.804027794725574e-05,
+      "loss": 27.2447,
+      "step": 2393
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.803808745023501e-05,
+      "loss": 29.2274,
+      "step": 2394
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.803589575417265e-05,
+      "loss": 0.7501,
+      "step": 2395
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.803370285912336e-05,
+      "loss": 28.074,
+      "step": 2396
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.803150876514188e-05,
+      "loss": 0.7261,
+      "step": 2397
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.802931347228297e-05,
+      "loss": 28.3414,
+      "step": 2398
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.802711698060144e-05,
+      "loss": 0.752,
+      "step": 2399
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.80249192901521e-05,
+      "loss": 0.7851,
+      "step": 2400
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.802272040098982e-05,
+      "loss": 0.7261,
+      "step": 2401
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.802052031316944e-05,
+      "loss": 0.7411,
+      "step": 2402
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.801831902674593e-05,
+      "loss": 30.8216,
+      "step": 2403
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.80161165417742e-05,
+      "loss": 30.21,
+      "step": 2404
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.801391285830926e-05,
+      "loss": 0.7646,
+      "step": 2405
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.801170797640609e-05,
+      "loss": 0.7569,
+      "step": 2406
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.80095018961197e-05,
+      "loss": 0.7941,
+      "step": 2407
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.800729461750521e-05,
+      "loss": 28.1499,
+      "step": 2408
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.800508614061768e-05,
+      "loss": 0.7996,
+      "step": 2409
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.800287646551223e-05,
+      "loss": 0.7928,
+      "step": 2410
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.800066559224401e-05,
+      "loss": 0.752,
+      "step": 2411
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.799845352086825e-05,
+      "loss": 30.6911,
+      "step": 2412
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.79962402514401e-05,
+      "loss": 28.5383,
+      "step": 2413
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.799402578401485e-05,
+      "loss": 28.8519,
+      "step": 2414
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.799181011864775e-05,
+      "loss": 0.8016,
+      "step": 2415
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.798959325539412e-05,
+      "loss": 0.7861,
+      "step": 2416
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.798737519430928e-05,
+      "loss": 0.7673,
+      "step": 2417
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.79851559354486e-05,
+      "loss": 27.7073,
+      "step": 2418
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.798293547886748e-05,
+      "loss": 0.7939,
+      "step": 2419
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.798071382462133e-05,
+      "loss": 0.7785,
+      "step": 2420
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.797849097276559e-05,
+      "loss": 28.6325,
+      "step": 2421
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.797626692335577e-05,
+      "loss": 26.2916,
+      "step": 2422
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.797404167644739e-05,
+      "loss": 31.5463,
+      "step": 2423
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.797181523209596e-05,
+      "loss": 27.0309,
+      "step": 2424
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.796958759035706e-05,
+      "loss": 29.3028,
+      "step": 2425
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.796735875128632e-05,
+      "loss": 0.7564,
+      "step": 2426
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.796512871493934e-05,
+      "loss": 28.7035,
+      "step": 2427
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.796289748137181e-05,
+      "loss": 0.7458,
+      "step": 2428
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.796066505063939e-05,
+      "loss": 0.7619,
+      "step": 2429
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.795843142279784e-05,
+      "loss": 0.7704,
+      "step": 2430
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.795619659790287e-05,
+      "loss": 0.7923,
+      "step": 2431
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.79539605760103e-05,
+      "loss": 26.968,
+      "step": 2432
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.79517233571759e-05,
+      "loss": 28.2478,
+      "step": 2433
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.794948494145556e-05,
+      "loss": 28.6335,
+      "step": 2434
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.794724532890513e-05,
+      "loss": 0.7722,
+      "step": 2435
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.794500451958048e-05,
+      "loss": 0.8109,
+      "step": 2436
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.79427625135376e-05,
+      "loss": 29.0284,
+      "step": 2437
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.794051931083242e-05,
+      "loss": 0.7411,
+      "step": 2438
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.793827491152091e-05,
+      "loss": 0.7157,
+      "step": 2439
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.793602931565913e-05,
+      "loss": 27.0124,
+      "step": 2440
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.79337825233031e-05,
+      "loss": 0.7839,
+      "step": 2441
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.793153453450892e-05,
+      "loss": 0.7814,
+      "step": 2442
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.79292853493327e-05,
+      "loss": 28.9625,
+      "step": 2443
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.792703496783058e-05,
+      "loss": 29.4159,
+      "step": 2444
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.792478339005871e-05,
+      "loss": 28.8864,
+      "step": 2445
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.79225306160733e-05,
+      "loss": 25.7797,
+      "step": 2446
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.792027664593058e-05,
+      "loss": 30.9538,
+      "step": 2447
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.791802147968682e-05,
+      "loss": 0.8046,
+      "step": 2448
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.791576511739831e-05,
+      "loss": 0.7217,
+      "step": 2449
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.791350755912137e-05,
+      "loss": 0.8162,
+      "step": 2450
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.791124880491232e-05,
+      "loss": 30.3944,
+      "step": 2451
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.790898885482758e-05,
+      "loss": 0.7307,
+      "step": 2452
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.790672770892352e-05,
+      "loss": 0.7345,
+      "step": 2453
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.790446536725661e-05,
+      "loss": 0.7884,
+      "step": 2454
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.79022018298833e-05,
+      "loss": 0.7745,
+      "step": 2455
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.78999370968601e-05,
+      "loss": 28.7104,
+      "step": 2456
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.789767116824353e-05,
+      "loss": 26.0839,
+      "step": 2457
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.789540404409016e-05,
+      "loss": 27.3705,
+      "step": 2458
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.789313572445656e-05,
+      "loss": 0.7343,
+      "step": 2459
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.789086620939936e-05,
+      "loss": 27.2427,
+      "step": 2460
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.78885954989752e-05,
+      "loss": 0.7967,
+      "step": 2461
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.788632359324078e-05,
+      "loss": 27.7389,
+      "step": 2462
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.788405049225277e-05,
+      "loss": 28.9986,
+      "step": 2463
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.788177619606793e-05,
+      "loss": 0.7819,
+      "step": 2464
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.787950070474301e-05,
+      "loss": 30.079,
+      "step": 2465
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.787722401833483e-05,
+      "loss": 28.0004,
+      "step": 2466
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.787494613690022e-05,
+      "loss": 30.4912,
+      "step": 2467
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.787266706049602e-05,
+      "loss": 27.3937,
+      "step": 2468
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.78703867891791e-05,
+      "loss": 0.7319,
+      "step": 2469
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.786810532300643e-05,
+      "loss": 0.7851,
+      "step": 2470
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.786582266203489e-05,
+      "loss": 29.1059,
+      "step": 2471
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.78635388063215e-05,
+      "loss": 27.1773,
+      "step": 2472
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.786125375592325e-05,
+      "loss": 0.8162,
+      "step": 2473
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.785896751089719e-05,
+      "loss": 0.8117,
+      "step": 2474
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.785668007130036e-05,
+      "loss": 25.4953,
+      "step": 2475
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.785439143718986e-05,
+      "loss": 0.72,
+      "step": 2476
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.785210160862284e-05,
+      "loss": 29.4747,
+      "step": 2477
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.784981058565642e-05,
+      "loss": 0.7102,
+      "step": 2478
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.784751836834781e-05,
+      "loss": 0.7851,
+      "step": 2479
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.784522495675422e-05,
+      "loss": 27.3263,
+      "step": 2480
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.784293035093289e-05,
+      "loss": 26.7242,
+      "step": 2481
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.78406345509411e-05,
+      "loss": 0.805,
+      "step": 2482
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.783833755683613e-05,
+      "loss": 29.9422,
+      "step": 2483
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.783603936867536e-05,
+      "loss": 27.8199,
+      "step": 2484
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.783373998651611e-05,
+      "loss": 0.7528,
+      "step": 2485
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.783143941041579e-05,
+      "loss": 27.6865,
+      "step": 2486
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.782913764043181e-05,
+      "loss": 28.5604,
+      "step": 2487
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.782683467662164e-05,
+      "loss": 0.7778,
+      "step": 2488
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.782453051904276e-05,
+      "loss": 32.3446,
+      "step": 2489
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.782222516775267e-05,
+      "loss": 0.7633,
+      "step": 2490
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.781991862280893e-05,
+      "loss": 27.9749,
+      "step": 2491
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.78176108842691e-05,
+      "loss": 28.2574,
+      "step": 2492
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.78153019521908e-05,
+      "loss": 0.813,
+      "step": 2493
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.781299182663161e-05,
+      "loss": 26.2159,
+      "step": 2494
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.781068050764925e-05,
+      "loss": 27.6483,
+      "step": 2495
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.78083679953014e-05,
+      "loss": 26.0543,
+      "step": 2496
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.780605428964575e-05,
+      "loss": 28.6855,
+      "step": 2497
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.780373939074007e-05,
+      "loss": 26.54,
+      "step": 2498
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.780142329864214e-05,
+      "loss": 24.8658,
+      "step": 2499
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.779910601340978e-05,
+      "loss": 26.8812,
+      "step": 2500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.779678753510081e-05,
+      "loss": 0.7751,
+      "step": 2501
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.77944678637731e-05,
+      "loss": 31.4599,
+      "step": 2502
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.779214699948458e-05,
+      "loss": 31.5294,
+      "step": 2503
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.778982494229315e-05,
+      "loss": 27.1014,
+      "step": 2504
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.778750169225678e-05,
+      "loss": 0.8222,
+      "step": 2505
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.778517724943346e-05,
+      "loss": 28.4197,
+      "step": 2506
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.778285161388119e-05,
+      "loss": 26.724,
+      "step": 2507
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.778052478565805e-05,
+      "loss": 0.7845,
+      "step": 2508
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.77781967648221e-05,
+      "loss": 30.3945,
+      "step": 2509
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.777586755143143e-05,
+      "loss": 0.756,
+      "step": 2510
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.777353714554422e-05,
+      "loss": 27.9638,
+      "step": 2511
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.777120554721859e-05,
+      "loss": 27.6354,
+      "step": 2512
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.776887275651279e-05,
+      "loss": 0.8065,
+      "step": 2513
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.776653877348501e-05,
+      "loss": 0.7837,
+      "step": 2514
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.77642035981935e-05,
+      "loss": 29.508,
+      "step": 2515
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.776186723069659e-05,
+      "loss": 27.4877,
+      "step": 2516
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.775952967105255e-05,
+      "loss": 27.519,
+      "step": 2517
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.775719091931975e-05,
+      "loss": 0.7905,
+      "step": 2518
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.775485097555654e-05,
+      "loss": 26.7912,
+      "step": 2519
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.775250983982137e-05,
+      "loss": 29.6073,
+      "step": 2520
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.775016751217265e-05,
+      "loss": 30.6578,
+      "step": 2521
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.774782399266884e-05,
+      "loss": 29.3762,
+      "step": 2522
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.774547928136844e-05,
+      "loss": 28.554,
+      "step": 2523
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.774313337832998e-05,
+      "loss": 27.18,
+      "step": 2524
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.7740786283612e-05,
+      "loss": 0.7499,
+      "step": 2525
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.773843799727311e-05,
+      "loss": 0.7861,
+      "step": 2526
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.773608851937189e-05,
+      "loss": 27.4576,
+      "step": 2527
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.7733737849967e-05,
+      "loss": 0.7319,
+      "step": 2528
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.773138598911714e-05,
+      "loss": 0.8295,
+      "step": 2529
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.772903293688096e-05,
+      "loss": 0.7849,
+      "step": 2530
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.772667869331724e-05,
+      "loss": 28.2677,
+      "step": 2531
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.77243232584847e-05,
+      "loss": 30.3846,
+      "step": 2532
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.772196663244218e-05,
+      "loss": 30.3032,
+      "step": 2533
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.771960881524846e-05,
+      "loss": 24.4003,
+      "step": 2534
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.771724980696241e-05,
+      "loss": 0.7152,
+      "step": 2535
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.771488960764293e-05,
+      "loss": 27.5575,
+      "step": 2536
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.771252821734888e-05,
+      "loss": 29.0657,
+      "step": 2537
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.771016563613924e-05,
+      "loss": 0.7995,
+      "step": 2538
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.770780186407298e-05,
+      "loss": 29.0231,
+      "step": 2539
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.770543690120909e-05,
+      "loss": 30.2232,
+      "step": 2540
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.77030707476066e-05,
+      "loss": 28.7482,
+      "step": 2541
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.770070340332456e-05,
+      "loss": 0.7618,
+      "step": 2542
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.76983348684221e-05,
+      "loss": 23.4433,
+      "step": 2543
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.769596514295829e-05,
+      "loss": 0.7166,
+      "step": 2544
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.769359422699231e-05,
+      "loss": 0.803,
+      "step": 2545
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.769122212058333e-05,
+      "loss": 23.7603,
+      "step": 2546
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.768884882379053e-05,
+      "loss": 25.832,
+      "step": 2547
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.76864743366732e-05,
+      "loss": 28.064,
+      "step": 2548
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.768409865929058e-05,
+      "loss": 0.8069,
+      "step": 2549
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.768172179170196e-05,
+      "loss": 0.7303,
+      "step": 2550
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.767934373396669e-05,
+      "loss": 0.7906,
+      "step": 2551
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.767696448614411e-05,
+      "loss": 0.7159,
+      "step": 2552
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.76745840482936e-05,
+      "loss": 28.39,
+      "step": 2553
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.767220242047458e-05,
+      "loss": 26.8836,
+      "step": 2554
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.766981960274653e-05,
+      "loss": 0.7806,
+      "step": 2555
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.766743559516889e-05,
+      "loss": 0.8422,
+      "step": 2556
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.766505039780116e-05,
+      "loss": 29.6544,
+      "step": 2557
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.76626640107029e-05,
+      "loss": 30.0046,
+      "step": 2558
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.766027643393366e-05,
+      "loss": 0.7921,
+      "step": 2559
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.765788766755303e-05,
+      "loss": 26.9361,
+      "step": 2560
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.765549771162066e-05,
+      "loss": 0.8131,
+      "step": 2561
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.765310656619617e-05,
+      "loss": 0.8075,
+      "step": 2562
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.765071423133928e-05,
+      "loss": 29.5411,
+      "step": 2563
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.764832070710967e-05,
+      "loss": 30.79,
+      "step": 2564
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.76459259935671e-05,
+      "loss": 29.2243,
+      "step": 2565
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.764353009077131e-05,
+      "loss": 0.7616,
+      "step": 2566
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.764113299878218e-05,
+      "loss": 0.7596,
+      "step": 2567
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.763873471765945e-05,
+      "loss": 0.7489,
+      "step": 2568
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.763633524746304e-05,
+      "loss": 0.7018,
+      "step": 2569
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.763393458825284e-05,
+      "loss": 28.2627,
+      "step": 2570
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.763153274008873e-05,
+      "loss": 0.7327,
+      "step": 2571
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.76291297030307e-05,
+      "loss": 0.7658,
+      "step": 2572
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.762672547713871e-05,
+      "loss": 0.7665,
+      "step": 2573
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.762432006247278e-05,
+      "loss": 0.7693,
+      "step": 2574
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.762191345909295e-05,
+      "loss": 28.86,
+      "step": 2575
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.761950566705929e-05,
+      "loss": 0.8043,
+      "step": 2576
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.76170966864319e-05,
+      "loss": 0.7688,
+      "step": 2577
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.76146865172709e-05,
+      "loss": 0.8087,
+      "step": 2578
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.761227515963645e-05,
+      "loss": 0.8035,
+      "step": 2579
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.760986261358873e-05,
+      "loss": 0.7498,
+      "step": 2580
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.760744887918797e-05,
+      "loss": 0.814,
+      "step": 2581
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.760503395649445e-05,
+      "loss": 0.8261,
+      "step": 2582
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.760261784556839e-05,
+      "loss": 24.8639,
+      "step": 2583
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.760020054647011e-05,
+      "loss": 0.752,
+      "step": 2584
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.759778205925996e-05,
+      "loss": 0.8104,
+      "step": 2585
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.759536238399831e-05,
+      "loss": 0.8041,
+      "step": 2586
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.759294152074555e-05,
+      "loss": 29.139,
+      "step": 2587
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.759051946956208e-05,
+      "loss": 0.7244,
+      "step": 2588
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.75880962305084e-05,
+      "loss": 0.7259,
+      "step": 2589
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.758567180364496e-05,
+      "loss": 30.4123,
+      "step": 2590
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.75832461890323e-05,
+      "loss": 26.7814,
+      "step": 2591
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.758081938673094e-05,
+      "loss": 0.7944,
+      "step": 2592
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.757839139680148e-05,
+      "loss": 0.7707,
+      "step": 2593
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.757596221930448e-05,
+      "loss": 0.8284,
+      "step": 2594
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.75735318543006e-05,
+      "loss": 0.7563,
+      "step": 2595
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.757110030185053e-05,
+      "loss": 28.7279,
+      "step": 2596
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.756866756201492e-05,
+      "loss": 0.7911,
+      "step": 2597
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.756623363485451e-05,
+      "loss": 0.8507,
+      "step": 2598
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.756379852043004e-05,
+      "loss": 29.5311,
+      "step": 2599
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.75613622188023e-05,
+      "loss": 0.7659,
+      "step": 2600
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.75589247300321e-05,
+      "loss": 24.3811,
+      "step": 2601
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.755648605418029e-05,
+      "loss": 24.2666,
+      "step": 2602
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.755404619130772e-05,
+      "loss": 0.7835,
+      "step": 2603
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.75516051414753e-05,
+      "loss": 25.5973,
+      "step": 2604
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.754916290474396e-05,
+      "loss": 24.6607,
+      "step": 2605
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.754671948117465e-05,
+      "loss": 23.975,
+      "step": 2606
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.754427487082838e-05,
+      "loss": 0.7642,
+      "step": 2607
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.754182907376615e-05,
+      "loss": 29.0998,
+      "step": 2608
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.7539382090049e-05,
+      "loss": 0.7593,
+      "step": 2609
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.753693391973802e-05,
+      "loss": 29.3221,
+      "step": 2610
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.753448456289432e-05,
+      "loss": 0.805,
+      "step": 2611
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.753203401957904e-05,
+      "loss": 0.8498,
+      "step": 2612
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.752958228985332e-05,
+      "loss": 24.8914,
+      "step": 2613
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.752712937377839e-05,
+      "loss": 25.8239,
+      "step": 2614
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.752467527141544e-05,
+      "loss": 0.737,
+      "step": 2615
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.752221998282576e-05,
+      "loss": 25.5564,
+      "step": 2616
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.751976350807062e-05,
+      "loss": 25.1715,
+      "step": 2617
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.751730584721132e-05,
+      "loss": 0.7157,
+      "step": 2618
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.751484700030923e-05,
+      "loss": 22.961,
+      "step": 2619
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.751238696742571e-05,
+      "loss": 26.9525,
+      "step": 2620
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.750992574862215e-05,
+      "loss": 0.7941,
+      "step": 2621
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.750746334395999e-05,
+      "loss": 0.7344,
+      "step": 2622
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.750499975350071e-05,
+      "loss": 0.7887,
+      "step": 2623
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.75025349773058e-05,
+      "loss": 0.8044,
+      "step": 2624
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.750006901543676e-05,
+      "loss": 28.0206,
+      "step": 2625
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.749760186795515e-05,
+      "loss": 0.8017,
+      "step": 2626
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.749513353492254e-05,
+      "loss": 28.7738,
+      "step": 2627
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.749266401640057e-05,
+      "loss": 0.7982,
+      "step": 2628
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.749019331245086e-05,
+      "loss": 0.7802,
+      "step": 2629
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.74877214231351e-05,
+      "loss": 27.1256,
+      "step": 2630
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.748524834851494e-05,
+      "loss": 27.3091,
+      "step": 2631
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.748277408865215e-05,
+      "loss": 28.4266,
+      "step": 2632
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.748029864360847e-05,
+      "loss": 0.8292,
+      "step": 2633
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.74778220134457e-05,
+      "loss": 0.7142,
+      "step": 2634
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.747534419822565e-05,
+      "loss": 0.7827,
+      "step": 2635
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.747286519801018e-05,
+      "loss": 0.7319,
+      "step": 2636
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.747038501286114e-05,
+      "loss": 0.741,
+      "step": 2637
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.746790364284047e-05,
+      "loss": 0.7304,
+      "step": 2638
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.746542108801007e-05,
+      "loss": 0.7658,
+      "step": 2639
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.746293734843192e-05,
+      "loss": 0.7773,
+      "step": 2640
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.746045242416801e-05,
+      "loss": 0.7354,
+      "step": 2641
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.74579663152804e-05,
+      "loss": 30.8574,
+      "step": 2642
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.74554790218311e-05,
+      "loss": 0.7755,
+      "step": 2643
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.745299054388219e-05,
+      "loss": 0.7752,
+      "step": 2644
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.745050088149584e-05,
+      "loss": 0.7456,
+      "step": 2645
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.744801003473412e-05,
+      "loss": 0.7961,
+      "step": 2646
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.744551800365924e-05,
+      "loss": 25.91,
+      "step": 2647
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.744302478833341e-05,
+      "loss": 28.1403,
+      "step": 2648
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.744053038881884e-05,
+      "loss": 25.9968,
+      "step": 2649
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.743803480517781e-05,
+      "loss": 29.0407,
+      "step": 2650
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.743553803747258e-05,
+      "loss": 0.7182,
+      "step": 2651
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.74330400857655e-05,
+      "loss": 26.6811,
+      "step": 2652
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.743054095011892e-05,
+      "loss": 0.7369,
+      "step": 2653
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.742804063059519e-05,
+      "loss": 23.9653,
+      "step": 2654
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.742553912725676e-05,
+      "loss": 0.7369,
+      "step": 2655
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.742303644016602e-05,
+      "loss": 29.8578,
+      "step": 2656
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.742053256938548e-05,
+      "loss": 0.7856,
+      "step": 2657
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.741802751497761e-05,
+      "loss": 0.8286,
+      "step": 2658
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.741552127700495e-05,
+      "loss": 27.7005,
+      "step": 2659
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.741301385553006e-05,
+      "loss": 0.7559,
+      "step": 2660
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.74105052506155e-05,
+      "loss": 26.8702,
+      "step": 2661
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.740799546232391e-05,
+      "loss": 28.1892,
+      "step": 2662
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.740548449071794e-05,
+      "loss": 0.7547,
+      "step": 2663
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.740297233586024e-05,
+      "loss": 0.7728,
+      "step": 2664
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.740045899781352e-05,
+      "loss": 23.064,
+      "step": 2665
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.739794447664055e-05,
+      "loss": 0.7621,
+      "step": 2666
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.739542877240404e-05,
+      "loss": 0.7363,
+      "step": 2667
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.739291188516682e-05,
+      "loss": 0.7458,
+      "step": 2668
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.739039381499167e-05,
+      "loss": 28.0943,
+      "step": 2669
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.73878745619415e-05,
+      "loss": 0.7573,
+      "step": 2670
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.738535412607915e-05,
+      "loss": 27.3032,
+      "step": 2671
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.738283250746754e-05,
+      "loss": 29.1953,
+      "step": 2672
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.738030970616961e-05,
+      "loss": 0.7194,
+      "step": 2673
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.737778572224833e-05,
+      "loss": 25.5861,
+      "step": 2674
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.737526055576671e-05,
+      "loss": 27.1075,
+      "step": 2675
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.737273420678776e-05,
+      "loss": 26.3697,
+      "step": 2676
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.737020667537457e-05,
+      "loss": 0.7441,
+      "step": 2677
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.736767796159019e-05,
+      "loss": 0.7593,
+      "step": 2678
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.736514806549774e-05,
+      "loss": 25.9218,
+      "step": 2679
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.73626169871604e-05,
+      "loss": 0.7771,
+      "step": 2680
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.736008472664131e-05,
+      "loss": 0.7633,
+      "step": 2681
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.73575512840037e-05,
+      "loss": 0.7806,
+      "step": 2682
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.735501665931079e-05,
+      "loss": 27.693,
+      "step": 2683
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.735248085262585e-05,
+      "loss": 0.8324,
+      "step": 2684
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.73499438640122e-05,
+      "loss": 0.8526,
+      "step": 2685
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.734740569353312e-05,
+      "loss": 0.8116,
+      "step": 2686
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.734486634125197e-05,
+      "loss": 26.9443,
+      "step": 2687
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.734232580723218e-05,
+      "loss": 26.6709,
+      "step": 2688
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.73397840915371e-05,
+      "loss": 25.9041,
+      "step": 2689
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.73372411942302e-05,
+      "loss": 0.7728,
+      "step": 2690
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.733469711537496e-05,
+      "loss": 25.1557,
+      "step": 2691
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.733215185503488e-05,
+      "loss": 27.5372,
+      "step": 2692
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.732960541327347e-05,
+      "loss": 26.2348,
+      "step": 2693
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.73270577901543e-05,
+      "loss": 0.7834,
+      "step": 2694
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.732450898574097e-05,
+      "loss": 0.7816,
+      "step": 2695
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.732195900009708e-05,
+      "loss": 0.772,
+      "step": 2696
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.73194078332863e-05,
+      "loss": 28.0614,
+      "step": 2697
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.731685548537231e-05,
+      "loss": 0.7462,
+      "step": 2698
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.731430195641877e-05,
+      "loss": 27.6549,
+      "step": 2699
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.731174724648947e-05,
+      "loss": 0.7135,
+      "step": 2700
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.730919135564815e-05,
+      "loss": 0.7187,
+      "step": 2701
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.730663428395861e-05,
+      "loss": 24.8941,
+      "step": 2702
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.730407603148469e-05,
+      "loss": 0.7186,
+      "step": 2703
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.730151659829022e-05,
+      "loss": 0.7447,
+      "step": 2704
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.72989559844391e-05,
+      "loss": 0.7315,
+      "step": 2705
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.729639418999523e-05,
+      "loss": 26.9489,
+      "step": 2706
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.729383121502257e-05,
+      "loss": 25.2915,
+      "step": 2707
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.729126705958509e-05,
+      "loss": 26.3848,
+      "step": 2708
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.728870172374677e-05,
+      "loss": 0.7607,
+      "step": 2709
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.728613520757166e-05,
+      "loss": 0.761,
+      "step": 2710
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.728356751112383e-05,
+      "loss": 27.128,
+      "step": 2711
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.728099863446736e-05,
+      "loss": 0.7462,
+      "step": 2712
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.727842857766636e-05,
+      "loss": 26.3846,
+      "step": 2713
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.727585734078499e-05,
+      "loss": 24.2597,
+      "step": 2714
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.727328492388742e-05,
+      "loss": 0.8061,
+      "step": 2715
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.727071132703786e-05,
+      "loss": 0.7881,
+      "step": 2716
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.726813655030057e-05,
+      "loss": 0.7454,
+      "step": 2717
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.726556059373979e-05,
+      "loss": 27.0075,
+      "step": 2718
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.726298345741982e-05,
+      "loss": 0.738,
+      "step": 2719
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.7260405141405e-05,
+      "loss": 0.7397,
+      "step": 2720
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.725782564575967e-05,
+      "loss": 26.967,
+      "step": 2721
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.725524497054823e-05,
+      "loss": 26.7737,
+      "step": 2722
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.725266311583508e-05,
+      "loss": 23.8699,
+      "step": 2723
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.725008008168468e-05,
+      "loss": 0.7533,
+      "step": 2724
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.724749586816147e-05,
+      "loss": 28.0788,
+      "step": 2725
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.724491047533e-05,
+      "loss": 0.7401,
+      "step": 2726
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.724232390325476e-05,
+      "loss": 30.578,
+      "step": 2727
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.723973615200034e-05,
+      "loss": 27.8675,
+      "step": 2728
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.72371472216313e-05,
+      "loss": 30.2453,
+      "step": 2729
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.72345571122123e-05,
+      "loss": 27.6925,
+      "step": 2730
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.723196582380795e-05,
+      "loss": 0.8363,
+      "step": 2731
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.722937335648298e-05,
+      "loss": 0.7792,
+      "step": 2732
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.722677971030205e-05,
+      "loss": 28.3452,
+      "step": 2733
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.722418488532992e-05,
+      "loss": 28.0052,
+      "step": 2734
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.722158888163133e-05,
+      "loss": 0.7834,
+      "step": 2735
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.721899169927113e-05,
+      "loss": 32.153,
+      "step": 2736
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.72163933383141e-05,
+      "loss": 28.7527,
+      "step": 2737
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.72137937988251e-05,
+      "loss": 26.0371,
+      "step": 2738
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.721119308086904e-05,
+      "loss": 0.7455,
+      "step": 2739
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.720859118451082e-05,
+      "loss": 0.7159,
+      "step": 2740
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.720598810981538e-05,
+      "loss": 0.8047,
+      "step": 2741
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.720338385684771e-05,
+      "loss": 0.796,
+      "step": 2742
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.720077842567279e-05,
+      "loss": 28.748,
+      "step": 2743
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.719817181635567e-05,
+      "loss": 31.0293,
+      "step": 2744
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.71955640289614e-05,
+      "loss": 0.7363,
+      "step": 2745
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.719295506355508e-05,
+      "loss": 0.7413,
+      "step": 2746
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.719034492020183e-05,
+      "loss": 0.7911,
+      "step": 2747
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.718773359896677e-05,
+      "loss": 0.7943,
+      "step": 2748
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.718512109991514e-05,
+      "loss": 27.0518,
+      "step": 2749
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.71825074231121e-05,
+      "loss": 28.4557,
+      "step": 2750
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.717989256862289e-05,
+      "loss": 26.3776,
+      "step": 2751
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.717727653651281e-05,
+      "loss": 0.7597,
+      "step": 2752
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.717465932684712e-05,
+      "loss": 27.0843,
+      "step": 2753
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.717204093969117e-05,
+      "loss": 25.387,
+      "step": 2754
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.71694213751103e-05,
+      "loss": 0.7974,
+      "step": 2755
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.716680063316991e-05,
+      "loss": 33.2931,
+      "step": 2756
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.71641787139354e-05,
+      "loss": 0.8079,
+      "step": 2757
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.716155561747222e-05,
+      "loss": 25.6761,
+      "step": 2758
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.715893134384584e-05,
+      "loss": 31.1064,
+      "step": 2759
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.715630589312179e-05,
+      "loss": 0.7739,
+      "step": 2760
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.715367926536556e-05,
+      "loss": 0.7879,
+      "step": 2761
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.715105146064272e-05,
+      "loss": 0.7807,
+      "step": 2762
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.714842247901888e-05,
+      "loss": 30.3448,
+      "step": 2763
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.714579232055964e-05,
+      "loss": 0.8456,
+      "step": 2764
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.714316098533067e-05,
+      "loss": 0.7868,
+      "step": 2765
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.714052847339765e-05,
+      "loss": 0.8144,
+      "step": 2766
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.713789478482625e-05,
+      "loss": 0.7444,
+      "step": 2767
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.713525991968224e-05,
+      "loss": 0.7181,
+      "step": 2768
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.71326238780314e-05,
+      "loss": 35.3647,
+      "step": 2769
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.712998665993947e-05,
+      "loss": 36.0009,
+      "step": 2770
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.712734826547234e-05,
+      "loss": 0.7731,
+      "step": 2771
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.712470869469584e-05,
+      "loss": 0.7666,
+      "step": 2772
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.712206794767582e-05,
+      "loss": 0.7696,
+      "step": 2773
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.711942602447826e-05,
+      "loss": 28.6556,
+      "step": 2774
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.711678292516904e-05,
+      "loss": 0.7293,
+      "step": 2775
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.711413864981418e-05,
+      "loss": 26.556,
+      "step": 2776
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.711149319847965e-05,
+      "loss": 25.7706,
+      "step": 2777
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.71088465712315e-05,
+      "loss": 0.7325,
+      "step": 2778
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.710619876813577e-05,
+      "loss": 0.7172,
+      "step": 2779
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.710354978925858e-05,
+      "loss": 28.6003,
+      "step": 2780
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.710089963466603e-05,
+      "loss": 26.756,
+      "step": 2781
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.709824830442426e-05,
+      "loss": 0.7586,
+      "step": 2782
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.709559579859945e-05,
+      "loss": 0.7755,
+      "step": 2783
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.709294211725784e-05,
+      "loss": 25.6619,
+      "step": 2784
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.709028726046563e-05,
+      "loss": 24.7875,
+      "step": 2785
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.70876312282891e-05,
+      "loss": 23.4657,
+      "step": 2786
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.708497402079454e-05,
+      "loss": 0.744,
+      "step": 2787
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.708231563804828e-05,
+      "loss": 0.7858,
+      "step": 2788
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.707965608011666e-05,
+      "loss": 27.3811,
+      "step": 2789
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.70769953470661e-05,
+      "loss": 28.8705,
+      "step": 2790
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.707433343896298e-05,
+      "loss": 23.8123,
+      "step": 2791
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.707167035587372e-05,
+      "loss": 28.4936,
+      "step": 2792
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.706900609786485e-05,
+      "loss": 23.6118,
+      "step": 2793
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.706634066500283e-05,
+      "loss": 0.8257,
+      "step": 2794
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.70636740573542e-05,
+      "loss": 26.3978,
+      "step": 2795
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.706100627498553e-05,
+      "loss": 24.2708,
+      "step": 2796
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.705833731796339e-05,
+      "loss": 0.7686,
+      "step": 2797
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.70556671863544e-05,
+      "loss": 24.7117,
+      "step": 2798
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.705299588022521e-05,
+      "loss": 25.7577,
+      "step": 2799
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.705032339964252e-05,
+      "loss": 25.857,
+      "step": 2800
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.7047649744673e-05,
+      "loss": 0.7604,
+      "step": 2801
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.70449749153834e-05,
+      "loss": 27.5831,
+      "step": 2802
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.70422989118405e-05,
+      "loss": 0.7345,
+      "step": 2803
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.703962173411106e-05,
+      "loss": 0.8111,
+      "step": 2804
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.703694338226193e-05,
+      "loss": 0.777,
+      "step": 2805
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.703426385635994e-05,
+      "loss": 29.9017,
+      "step": 2806
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.703158315647199e-05,
+      "loss": 0.8056,
+      "step": 2807
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.7028901282665e-05,
+      "loss": 28.6072,
+      "step": 2808
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.702621823500589e-05,
+      "loss": 24.5249,
+      "step": 2809
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.702353401356161e-05,
+      "loss": 29.0598,
+      "step": 2810
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.702084861839921e-05,
+      "loss": 24.5835,
+      "step": 2811
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.701816204958569e-05,
+      "loss": 0.7934,
+      "step": 2812
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.701547430718811e-05,
+      "loss": 25.9236,
+      "step": 2813
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.701278539127355e-05,
+      "loss": 23.7794,
+      "step": 2814
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.701009530190914e-05,
+      "loss": 0.724,
+      "step": 2815
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.7007404039162e-05,
+      "loss": 0.7559,
+      "step": 2816
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.700471160309934e-05,
+      "loss": 0.7891,
+      "step": 2817
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.700201799378833e-05,
+      "loss": 0.7613,
+      "step": 2818
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.699932321129624e-05,
+      "loss": 29.4162,
+      "step": 2819
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.69966272556903e-05,
+      "loss": 27.5425,
+      "step": 2820
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.699393012703781e-05,
+      "loss": 0.7794,
+      "step": 2821
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.69912318254061e-05,
+      "loss": 0.7933,
+      "step": 2822
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.698853235086252e-05,
+      "loss": 0.7106,
+      "step": 2823
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.698583170347444e-05,
+      "loss": 0.8525,
+      "step": 2824
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.698312988330927e-05,
+      "loss": 0.7523,
+      "step": 2825
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.698042689043446e-05,
+      "loss": 27.0819,
+      "step": 2826
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.697772272491743e-05,
+      "loss": 30.7256,
+      "step": 2827
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.697501738682576e-05,
+      "loss": 0.7662,
+      "step": 2828
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.697231087622691e-05,
+      "loss": 26.3185,
+      "step": 2829
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.696960319318844e-05,
+      "loss": 26.3684,
+      "step": 2830
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.696689433777797e-05,
+      "loss": 0.761,
+      "step": 2831
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.696418431006309e-05,
+      "loss": 0.7625,
+      "step": 2832
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.696147311011145e-05,
+      "loss": 25.548,
+      "step": 2833
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.695876073799071e-05,
+      "loss": 0.7531,
+      "step": 2834
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.695604719376859e-05,
+      "loss": 28.1438,
+      "step": 2835
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.69533324775128e-05,
+      "loss": 0.7721,
+      "step": 2836
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.695061658929112e-05,
+      "loss": 0.7502,
+      "step": 2837
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.694789952917132e-05,
+      "loss": 0.7501,
+      "step": 2838
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.694518129722123e-05,
+      "loss": 0.8245,
+      "step": 2839
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.69424618935087e-05,
+      "loss": 25.8764,
+      "step": 2840
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.69397413181016e-05,
+      "loss": 26.2473,
+      "step": 2841
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.693701957106784e-05,
+      "loss": 0.7606,
+      "step": 2842
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.693429665247535e-05,
+      "loss": 25.1953,
+      "step": 2843
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.693157256239211e-05,
+      "loss": 0.7596,
+      "step": 2844
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.69288473008861e-05,
+      "loss": 0.7484,
+      "step": 2845
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.692612086802535e-05,
+      "loss": 26.9798,
+      "step": 2846
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.692339326387789e-05,
+      "loss": 26.1561,
+      "step": 2847
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.692066448851184e-05,
+      "loss": 0.7588,
+      "step": 2848
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.691793454199528e-05,
+      "loss": 27.227,
+      "step": 2849
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.691520342439638e-05,
+      "loss": 27.2905,
+      "step": 2850
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.691247113578326e-05,
+      "loss": 25.9705,
+      "step": 2851
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.690973767622417e-05,
+      "loss": 0.7878,
+      "step": 2852
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.690700304578731e-05,
+      "loss": 0.7158,
+      "step": 2853
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.690426724454094e-05,
+      "loss": 0.8274,
+      "step": 2854
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.690153027255335e-05,
+      "loss": 0.7865,
+      "step": 2855
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.689879212989285e-05,
+      "loss": 27.2662,
+      "step": 2856
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.689605281662781e-05,
+      "loss": 0.7593,
+      "step": 2857
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.689331233282656e-05,
+      "loss": 26.3145,
+      "step": 2858
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.689057067855755e-05,
+      "loss": 26.9829,
+      "step": 2859
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.688782785388918e-05,
+      "loss": 0.7376,
+      "step": 2860
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.688508385888993e-05,
+      "loss": 23.7988,
+      "step": 2861
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.688233869362826e-05,
+      "loss": 27.0286,
+      "step": 2862
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.687959235817272e-05,
+      "loss": 24.9231,
+      "step": 2863
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.687684485259187e-05,
+      "loss": 25.4951,
+      "step": 2864
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.687409617695424e-05,
+      "loss": 0.8094,
+      "step": 2865
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.687134633132849e-05,
+      "loss": 28.3124,
+      "step": 2866
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.686859531578322e-05,
+      "loss": 0.8023,
+      "step": 2867
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.686584313038711e-05,
+      "loss": 0.7951,
+      "step": 2868
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.686308977520886e-05,
+      "loss": 0.798,
+      "step": 2869
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.686033525031719e-05,
+      "loss": 24.305,
+      "step": 2870
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.685757955578085e-05,
+      "loss": 26.4776,
+      "step": 2871
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.68548226916686e-05,
+      "loss": 0.7585,
+      "step": 2872
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.685206465804933e-05,
+      "loss": 24.4732,
+      "step": 2873
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.684930545499179e-05,
+      "loss": 0.7257,
+      "step": 2874
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.68465450825649e-05,
+      "loss": 27.4148,
+      "step": 2875
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.684378354083754e-05,
+      "loss": 26.9961,
+      "step": 2876
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.684102082987866e-05,
+      "loss": 26.2508,
+      "step": 2877
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.68382569497572e-05,
+      "loss": 0.7864,
+      "step": 2878
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.683549190054215e-05,
+      "loss": 25.3953,
+      "step": 2879
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.683272568230252e-05,
+      "loss": 0.7828,
+      "step": 2880
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.682995829510739e-05,
+      "loss": 0.7958,
+      "step": 2881
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.682718973902579e-05,
+      "loss": 27.0966,
+      "step": 2882
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.682442001412684e-05,
+      "loss": 26.044,
+      "step": 2883
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.682164912047969e-05,
+      "loss": 26.5631,
+      "step": 2884
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.681887705815347e-05,
+      "loss": 0.788,
+      "step": 2885
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.681610382721739e-05,
+      "loss": 0.7811,
+      "step": 2886
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.681332942774066e-05,
+      "loss": 0.7645,
+      "step": 2887
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.681055385979256e-05,
+      "loss": 27.5586,
+      "step": 2888
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.680777712344234e-05,
+      "loss": 26.5494,
+      "step": 2889
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.680499921875932e-05,
+      "loss": 0.7497,
+      "step": 2890
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.680222014581282e-05,
+      "loss": 0.7221,
+      "step": 2891
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.679943990467222e-05,
+      "loss": 0.7946,
+      "step": 2892
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.679665849540694e-05,
+      "loss": 0.7794,
+      "step": 2893
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.679387591808635e-05,
+      "loss": 0.7958,
+      "step": 2894
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.679109217277994e-05,
+      "loss": 30.1229,
+      "step": 2895
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.678830725955719e-05,
+      "loss": 26.4547,
+      "step": 2896
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.67855211784876e-05,
+      "loss": 0.7423,
+      "step": 2897
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.678273392964072e-05,
+      "loss": 0.8422,
+      "step": 2898
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.677994551308612e-05,
+      "loss": 25.3481,
+      "step": 2899
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.67771559288934e-05,
+      "loss": 26.3265,
+      "step": 2900
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.67743651771322e-05,
+      "loss": 0.7518,
+      "step": 2901
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.677157325787215e-05,
+      "loss": 27.5009,
+      "step": 2902
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.676878017118297e-05,
+      "loss": 0.7654,
+      "step": 2903
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.676598591713432e-05,
+      "loss": 27.178,
+      "step": 2904
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.676319049579601e-05,
+      "loss": 25.1375,
+      "step": 2905
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.676039390723779e-05,
+      "loss": 0.7637,
+      "step": 2906
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.675759615152946e-05,
+      "loss": 0.7304,
+      "step": 2907
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.675479722874086e-05,
+      "loss": 26.8832,
+      "step": 2908
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.675199713894183e-05,
+      "loss": 0.7535,
+      "step": 2909
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.674919588220228e-05,
+      "loss": 0.7725,
+      "step": 2910
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.674639345859214e-05,
+      "loss": 26.6013,
+      "step": 2911
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.674358986818132e-05,
+      "loss": 26.9535,
+      "step": 2912
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.674078511103986e-05,
+      "loss": 0.7773,
+      "step": 2913
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.67379791872377e-05,
+      "loss": 26.9894,
+      "step": 2914
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.673517209684491e-05,
+      "loss": 0.7826,
+      "step": 2915
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.673236383993157e-05,
+      "loss": 0.7502,
+      "step": 2916
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.672955441656775e-05,
+      "loss": 24.6281,
+      "step": 2917
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.672674382682359e-05,
+      "loss": 0.7637,
+      "step": 2918
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.672393207076921e-05,
+      "loss": 0.7881,
+      "step": 2919
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.672111914847484e-05,
+      "loss": 0.7445,
+      "step": 2920
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.671830506001068e-05,
+      "loss": 26.4979,
+      "step": 2921
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.671548980544693e-05,
+      "loss": 25.1515,
+      "step": 2922
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.67126733848539e-05,
+      "loss": 25.045,
+      "step": 2923
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.670985579830188e-05,
+      "loss": 0.776,
+      "step": 2924
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.670703704586118e-05,
+      "loss": 0.7601,
+      "step": 2925
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.670421712760218e-05,
+      "loss": 0.7982,
+      "step": 2926
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.670139604359525e-05,
+      "loss": 27.4429,
+      "step": 2927
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.669857379391082e-05,
+      "loss": 26.7156,
+      "step": 2928
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.669575037861933e-05,
+      "loss": 0.7438,
+      "step": 2929
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.669292579779123e-05,
+      "loss": 0.7636,
+      "step": 2930
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.669010005149706e-05,
+      "loss": 23.8623,
+      "step": 2931
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.668727313980733e-05,
+      "loss": 27.2161,
+      "step": 2932
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.66844450627926e-05,
+      "loss": 24.1423,
+      "step": 2933
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.668161582052345e-05,
+      "loss": 0.7677,
+      "step": 2934
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.667878541307052e-05,
+      "loss": 26.2559,
+      "step": 2935
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.667595384050446e-05,
+      "loss": 0.7721,
+      "step": 2936
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.667312110289591e-05,
+      "loss": 0.7688,
+      "step": 2937
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.667028720031561e-05,
+      "loss": 26.1712,
+      "step": 2938
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.666745213283429e-05,
+      "loss": 0.7826,
+      "step": 2939
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.666461590052272e-05,
+      "loss": 26.5945,
+      "step": 2940
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.666177850345166e-05,
+      "loss": 25.8436,
+      "step": 2941
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.665893994169197e-05,
+      "loss": 23.3106,
+      "step": 2942
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.665610021531447e-05,
+      "loss": 0.7927,
+      "step": 2943
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.665325932439007e-05,
+      "loss": 27.187,
+      "step": 2944
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.665041726898966e-05,
+      "loss": 25.6713,
+      "step": 2945
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.664757404918418e-05,
+      "loss": 0.726,
+      "step": 2946
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.66447296650446e-05,
+      "loss": 0.7341,
+      "step": 2947
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.664188411664193e-05,
+      "loss": 26.3617,
+      "step": 2948
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.663903740404717e-05,
+      "loss": 28.7246,
+      "step": 2949
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.663618952733138e-05,
+      "loss": 0.7477,
+      "step": 2950
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.663334048656567e-05,
+      "loss": 24.5174,
+      "step": 2951
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.663049028182111e-05,
+      "loss": 26.6991,
+      "step": 2952
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.662763891316888e-05,
+      "loss": 25.0217,
+      "step": 2953
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.662478638068015e-05,
+      "loss": 22.0022,
+      "step": 2954
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.662193268442609e-05,
+      "loss": 26.7065,
+      "step": 2955
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.661907782447795e-05,
+      "loss": 25.5723,
+      "step": 2956
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.661622180090698e-05,
+      "loss": 0.7698,
+      "step": 2957
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.661336461378447e-05,
+      "loss": 0.7924,
+      "step": 2958
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.661050626318174e-05,
+      "loss": 0.7627,
+      "step": 2959
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.660764674917013e-05,
+      "loss": 30.3221,
+      "step": 2960
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.660478607182101e-05,
+      "loss": 0.7761,
+      "step": 2961
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.660192423120579e-05,
+      "loss": 0.737,
+      "step": 2962
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.65990612273959e-05,
+      "loss": 0.6672,
+      "step": 2963
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.659619706046282e-05,
+      "loss": 24.2592,
+      "step": 2964
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.6593331730478e-05,
+      "loss": 26.483,
+      "step": 2965
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.659046523751298e-05,
+      "loss": 25.6024,
+      "step": 2966
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.658759758163931e-05,
+      "loss": 25.7898,
+      "step": 2967
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.65847287629286e-05,
+      "loss": 0.7797,
+      "step": 2968
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.658185878145238e-05,
+      "loss": 0.8683,
+      "step": 2969
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.657898763728235e-05,
+      "loss": 25.1803,
+      "step": 2970
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.657611533049014e-05,
+      "loss": 0.8043,
+      "step": 2971
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.657324186114747e-05,
+      "loss": 26.2759,
+      "step": 2972
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.657036722932604e-05,
+      "loss": 0.7556,
+      "step": 2973
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.656749143509763e-05,
+      "loss": 0.7985,
+      "step": 2974
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.656461447853398e-05,
+      "loss": 0.7623,
+      "step": 2975
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.656173635970693e-05,
+      "loss": 0.7479,
+      "step": 2976
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.65588570786883e-05,
+      "loss": 25.6058,
+      "step": 2977
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.655597663554998e-05,
+      "loss": 26.3757,
+      "step": 2978
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.655309503036385e-05,
+      "loss": 0.797,
+      "step": 2979
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.655021226320182e-05,
+      "loss": 27.9953,
+      "step": 2980
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.654732833413589e-05,
+      "loss": 0.7657,
+      "step": 2981
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.6544443243238e-05,
+      "loss": 25.8325,
+      "step": 2982
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.654155699058019e-05,
+      "loss": 25.4863,
+      "step": 2983
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.653866957623448e-05,
+      "loss": 24.5589,
+      "step": 2984
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.653578100027297e-05,
+      "loss": 27.3013,
+      "step": 2985
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.653289126276771e-05,
+      "loss": 0.7714,
+      "step": 2986
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.653000036379088e-05,
+      "loss": 26.9558,
+      "step": 2987
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.652710830341461e-05,
+      "loss": 25.3357,
+      "step": 2988
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.65242150817111e-05,
+      "loss": 0.7472,
+      "step": 2989
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.652132069875255e-05,
+      "loss": 0.7912,
+      "step": 2990
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.651842515461121e-05,
+      "loss": 0.753,
+      "step": 2991
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.651552844935937e-05,
+      "loss": 0.8073,
+      "step": 2992
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.651263058306932e-05,
+      "loss": 27.0602,
+      "step": 2993
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.650973155581337e-05,
+      "loss": 0.7977,
+      "step": 2994
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.650683136766392e-05,
+      "loss": 0.7536,
+      "step": 2995
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.650393001869333e-05,
+      "loss": 28.4971,
+      "step": 2996
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.650102750897404e-05,
+      "loss": 24.8541,
+      "step": 2997
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.649812383857848e-05,
+      "loss": 26.202,
+      "step": 2998
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.649521900757912e-05,
+      "loss": 0.7851,
+      "step": 2999
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.64923130160485e-05,
+      "loss": 0.7152,
+      "step": 3000
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.648940586405912e-05,
+      "loss": 28.4788,
+      "step": 3001
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.648649755168355e-05,
+      "loss": 0.767,
+      "step": 3002
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.648358807899439e-05,
+      "loss": 27.1193,
+      "step": 3003
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.648067744606426e-05,
+      "loss": 25.3222,
+      "step": 3004
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.64777656529658e-05,
+      "loss": 0.799,
+      "step": 3005
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.647485269977171e-05,
+      "loss": 0.7337,
+      "step": 3006
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.647193858655467e-05,
+      "loss": 0.8636,
+      "step": 3007
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.646902331338745e-05,
+      "loss": 24.698,
+      "step": 3008
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.646610688034279e-05,
+      "loss": 0.793,
+      "step": 3009
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.646318928749349e-05,
+      "loss": 0.7727,
+      "step": 3010
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.646027053491238e-05,
+      "loss": 27.6641,
+      "step": 3011
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.645735062267231e-05,
+      "loss": 0.7353,
+      "step": 3012
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.645442955084615e-05,
+      "loss": 0.7558,
+      "step": 3013
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.645150731950684e-05,
+      "loss": 0.7084,
+      "step": 3014
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.64485839287273e-05,
+      "loss": 27.1945,
+      "step": 3015
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.644565937858048e-05,
+      "loss": 25.4943,
+      "step": 3016
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.644273366913941e-05,
+      "loss": 0.7251,
+      "step": 3017
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.643980680047711e-05,
+      "loss": 0.7671,
+      "step": 3018
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.643687877266661e-05,
+      "loss": 0.7709,
+      "step": 3019
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.643394958578102e-05,
+      "loss": 23.1629,
+      "step": 3020
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.643101923989345e-05,
+      "loss": 25.3449,
+      "step": 3021
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.642808773507704e-05,
+      "loss": 26.6875,
+      "step": 3022
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.642515507140495e-05,
+      "loss": 23.9446,
+      "step": 3023
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.64222212489504e-05,
+      "loss": 23.1071,
+      "step": 3024
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.64192862677866e-05,
+      "loss": 26.0864,
+      "step": 3025
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.641635012798682e-05,
+      "loss": 27.1896,
+      "step": 3026
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.641341282962435e-05,
+      "loss": 27.6235,
+      "step": 3027
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.641047437277249e-05,
+      "loss": 23.9827,
+      "step": 3028
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.64075347575046e-05,
+      "loss": 0.7708,
+      "step": 3029
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.640459398389402e-05,
+      "loss": 0.7649,
+      "step": 3030
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.64016520520142e-05,
+      "loss": 28.6267,
+      "step": 3031
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.639870896193855e-05,
+      "loss": 0.822,
+      "step": 3032
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.639576471374052e-05,
+      "loss": 25.8382,
+      "step": 3033
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.639281930749362e-05,
+      "loss": 0.7955,
+      "step": 3034
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.638987274327136e-05,
+      "loss": 0.811,
+      "step": 3035
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.638692502114728e-05,
+      "loss": 0.75,
+      "step": 3036
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.638397614119495e-05,
+      "loss": 0.754,
+      "step": 3037
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.6381026103488e-05,
+      "loss": 26.0092,
+      "step": 3038
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.637807490810003e-05,
+      "loss": 24.5662,
+      "step": 3039
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.637512255510475e-05,
+      "loss": 25.0763,
+      "step": 3040
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.63721690445758e-05,
+      "loss": 0.7866,
+      "step": 3041
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.636921437658693e-05,
+      "loss": 0.7974,
+      "step": 3042
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.636625855121186e-05,
+      "loss": 0.7644,
+      "step": 3043
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.636330156852442e-05,
+      "loss": 29.0655,
+      "step": 3044
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.636034342859836e-05,
+      "loss": 27.2184,
+      "step": 3045
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.635738413150757e-05,
+      "loss": 25.0974,
+      "step": 3046
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.635442367732588e-05,
+      "loss": 0.7701,
+      "step": 3047
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.635146206612717e-05,
+      "loss": 28.6182,
+      "step": 3048
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.634849929798541e-05,
+      "loss": 26.5648,
+      "step": 3049
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.634553537297451e-05,
+      "loss": 24.1217,
+      "step": 3050
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.634257029116846e-05,
+      "loss": 0.7888,
+      "step": 3051
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.633960405264128e-05,
+      "loss": 0.7447,
+      "step": 3052
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.633663665746701e-05,
+      "loss": 27.0083,
+      "step": 3053
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.633366810571972e-05,
+      "loss": 0.8059,
+      "step": 3054
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.633069839747346e-05,
+      "loss": 27.0385,
+      "step": 3055
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.632772753280241e-05,
+      "loss": 26.2451,
+      "step": 3056
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.632475551178071e-05,
+      "loss": 24.0705,
+      "step": 3057
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.632178233448253e-05,
+      "loss": 27.2823,
+      "step": 3058
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.631880800098208e-05,
+      "loss": 25.2172,
+      "step": 3059
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.631583251135361e-05,
+      "loss": 26.5842,
+      "step": 3060
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.631285586567138e-05,
+      "loss": 25.0561,
+      "step": 3061
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.63098780640097e-05,
+      "loss": 23.3912,
+      "step": 3062
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.630689910644288e-05,
+      "loss": 0.7587,
+      "step": 3063
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.63039189930453e-05,
+      "loss": 25.307,
+      "step": 3064
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.630093772389132e-05,
+      "loss": 0.7424,
+      "step": 3065
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.629795529905536e-05,
+      "loss": 0.7903,
+      "step": 3066
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.629497171861188e-05,
+      "loss": 29.5724,
+      "step": 3067
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.629198698263531e-05,
+      "loss": 0.8087,
+      "step": 3068
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.628900109120018e-05,
+      "loss": 29.831,
+      "step": 3069
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.628601404438103e-05,
+      "loss": 0.804,
+      "step": 3070
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.62830258422524e-05,
+      "loss": 0.733,
+      "step": 3071
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.628003648488886e-05,
+      "loss": 0.7802,
+      "step": 3072
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.627704597236506e-05,
+      "loss": 25.2225,
+      "step": 3073
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.627405430475562e-05,
+      "loss": 23.286,
+      "step": 3074
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.627106148213522e-05,
+      "loss": 26.9389,
+      "step": 3075
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.626806750457855e-05,
+      "loss": 24.3954,
+      "step": 3076
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.626507237216036e-05,
+      "loss": 0.693,
+      "step": 3077
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.62620760849554e-05,
+      "loss": 0.7782,
+      "step": 3078
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.625907864303845e-05,
+      "loss": 25.6823,
+      "step": 3079
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.625608004648434e-05,
+      "loss": 27.2675,
+      "step": 3080
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.625308029536793e-05,
+      "loss": 27.1817,
+      "step": 3081
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.625007938976404e-05,
+      "loss": 24.3329,
+      "step": 3082
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.624707732974763e-05,
+      "loss": 23.705,
+      "step": 3083
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.62440741153936e-05,
+      "loss": 0.8071,
+      "step": 3084
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.624106974677693e-05,
+      "loss": 0.8526,
+      "step": 3085
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.623806422397258e-05,
+      "loss": 0.7935,
+      "step": 3086
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.623505754705562e-05,
+      "loss": 26.3877,
+      "step": 3087
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.623204971610106e-05,
+      "loss": 0.7885,
+      "step": 3088
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.622904073118397e-05,
+      "loss": 0.7721,
+      "step": 3089
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.622603059237947e-05,
+      "loss": 25.3305,
+      "step": 3090
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.622301929976269e-05,
+      "loss": 0.7633,
+      "step": 3091
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.622000685340882e-05,
+      "loss": 25.5941,
+      "step": 3092
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.6216993253393e-05,
+      "loss": 0.7545,
+      "step": 3093
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.621397849979048e-05,
+      "loss": 24.2677,
+      "step": 3094
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.62109625926765e-05,
+      "loss": 0.77,
+      "step": 3095
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.620794553212636e-05,
+      "loss": 23.5085,
+      "step": 3096
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.620492731821533e-05,
+      "loss": 26.7057,
+      "step": 3097
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.620190795101878e-05,
+      "loss": 25.748,
+      "step": 3098
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.619888743061204e-05,
+      "loss": 0.769,
+      "step": 3099
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.619586575707054e-05,
+      "loss": 25.3808,
+      "step": 3100
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.619284293046966e-05,
+      "loss": 0.7866,
+      "step": 3101
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.618981895088489e-05,
+      "loss": 0.8225,
+      "step": 3102
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.618679381839168e-05,
+      "loss": 24.483,
+      "step": 3103
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.618376753306557e-05,
+      "loss": 24.8735,
+      "step": 3104
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.618074009498206e-05,
+      "loss": 25.1864,
+      "step": 3105
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.617771150421675e-05,
+      "loss": 26.0741,
+      "step": 3106
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.617468176084521e-05,
+      "loss": 0.7643,
+      "step": 3107
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.617165086494306e-05,
+      "loss": 25.3528,
+      "step": 3108
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.616861881658596e-05,
+      "loss": 0.7554,
+      "step": 3109
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.616558561584961e-05,
+      "loss": 0.7564,
+      "step": 3110
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.616255126280968e-05,
+      "loss": 26.1953,
+      "step": 3111
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.615951575754197e-05,
+      "loss": 0.7547,
+      "step": 3112
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.615647910012217e-05,
+      "loss": 25.5929,
+      "step": 3113
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.615344129062612e-05,
+      "loss": 0.7764,
+      "step": 3114
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.615040232912964e-05,
+      "loss": 23.6766,
+      "step": 3115
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.61473622157086e-05,
+      "loss": 0.8195,
+      "step": 3116
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.614432095043883e-05,
+      "loss": 25.7697,
+      "step": 3117
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.614127853339628e-05,
+      "loss": 26.8677,
+      "step": 3118
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.61382349646569e-05,
+      "loss": 27.0936,
+      "step": 3119
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.613519024429662e-05,
+      "loss": 0.7776,
+      "step": 3120
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.613214437239146e-05,
+      "loss": 0.7718,
+      "step": 3121
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.612909734901747e-05,
+      "loss": 25.0608,
+      "step": 3122
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.612604917425066e-05,
+      "loss": 0.7739,
+      "step": 3123
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.612299984816712e-05,
+      "loss": 0.7526,
+      "step": 3124
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.611994937084297e-05,
+      "loss": 27.3592,
+      "step": 3125
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.611689774235437e-05,
+      "loss": 25.6435,
+      "step": 3126
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.611384496277746e-05,
+      "loss": 29.5667,
+      "step": 3127
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.611079103218847e-05,
+      "loss": 0.7814,
+      "step": 3128
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.610773595066358e-05,
+      "loss": 0.8126,
+      "step": 3129
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.61046797182791e-05,
+      "loss": 25.0125,
+      "step": 3130
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.610162233511127e-05,
+      "loss": 0.7848,
+      "step": 3131
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.609856380123642e-05,
+      "loss": 23.6238,
+      "step": 3132
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.60955041167309e-05,
+      "loss": 0.7471,
+      "step": 3133
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.609244328167107e-05,
+      "loss": 0.8048,
+      "step": 3134
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.608938129613333e-05,
+      "loss": 0.8021,
+      "step": 3135
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.608631816019411e-05,
+      "loss": 24.7045,
+      "step": 3136
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.608325387392986e-05,
+      "loss": 0.7032,
+      "step": 3137
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.608018843741708e-05,
+      "loss": 0.7549,
+      "step": 3138
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.607712185073227e-05,
+      "loss": 0.7568,
+      "step": 3139
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.607405411395198e-05,
+      "loss": 0.8161,
+      "step": 3140
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.607098522715278e-05,
+      "loss": 24.2651,
+      "step": 3141
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.606791519041128e-05,
+      "loss": 0.7616,
+      "step": 3142
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.606484400380408e-05,
+      "loss": 25.4119,
+      "step": 3143
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.606177166740786e-05,
+      "loss": 0.8147,
+      "step": 3144
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.60586981812993e-05,
+      "loss": 21.3146,
+      "step": 3145
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.605562354555513e-05,
+      "loss": 24.6595,
+      "step": 3146
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.605254776025207e-05,
+      "loss": 22.7314,
+      "step": 3147
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.604947082546689e-05,
+      "loss": 25.2006,
+      "step": 3148
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.604639274127643e-05,
+      "loss": 0.7506,
+      "step": 3149
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.604331350775748e-05,
+      "loss": 24.6523,
+      "step": 3150
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.604023312498691e-05,
+      "loss": 25.3274,
+      "step": 3151
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.60371515930416e-05,
+      "loss": 27.4023,
+      "step": 3152
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.60340689119985e-05,
+      "loss": 0.7045,
+      "step": 3153
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.603098508193451e-05,
+      "loss": 20.8296,
+      "step": 3154
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.602790010292662e-05,
+      "loss": 0.7278,
+      "step": 3155
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.602481397505186e-05,
+      "loss": 0.7221,
+      "step": 3156
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.602172669838721e-05,
+      "loss": 0.7446,
+      "step": 3157
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.601863827300975e-05,
+      "loss": 25.0717,
+      "step": 3158
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.601554869899658e-05,
+      "loss": 0.7846,
+      "step": 3159
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.601245797642482e-05,
+      "loss": 24.4388,
+      "step": 3160
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.600936610537159e-05,
+      "loss": 25.2805,
+      "step": 3161
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.600627308591407e-05,
+      "loss": 0.7867,
+      "step": 3162
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.600317891812948e-05,
+      "loss": 23.8116,
+      "step": 3163
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.600008360209503e-05,
+      "loss": 0.7666,
+      "step": 3164
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.5996987137888e-05,
+      "loss": 0.7714,
+      "step": 3165
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.599388952558566e-05,
+      "loss": 30.4414,
+      "step": 3166
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.599079076526534e-05,
+      "loss": 24.6899,
+      "step": 3167
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.598769085700439e-05,
+      "loss": 0.7732,
+      "step": 3168
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.598458980088018e-05,
+      "loss": 25.5374,
+      "step": 3169
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.59814875969701e-05,
+      "loss": 0.7906,
+      "step": 3170
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.597838424535158e-05,
+      "loss": 21.5482,
+      "step": 3171
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.59752797461021e-05,
+      "loss": 25.8721,
+      "step": 3172
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.597217409929914e-05,
+      "loss": 0.7811,
+      "step": 3173
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.596906730502022e-05,
+      "loss": 0.7627,
+      "step": 3174
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.596595936334287e-05,
+      "loss": 0.7535,
+      "step": 3175
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.59628502743447e-05,
+      "loss": 0.7637,
+      "step": 3176
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.595974003810329e-05,
+      "loss": 27.0791,
+      "step": 3177
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.595662865469625e-05,
+      "loss": 0.7629,
+      "step": 3178
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.595351612420129e-05,
+      "loss": 23.5901,
+      "step": 3179
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.595040244669606e-05,
+      "loss": 24.8708,
+      "step": 3180
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.594728762225831e-05,
+      "loss": 0.7841,
+      "step": 3181
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.594417165096577e-05,
+      "loss": 25.1135,
+      "step": 3182
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.594105453289619e-05,
+      "loss": 0.7281,
+      "step": 3183
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.593793626812741e-05,
+      "loss": 24.0701,
+      "step": 3184
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.593481685673726e-05,
+      "loss": 0.7527,
+      "step": 3185
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.593169629880358e-05,
+      "loss": 0.7172,
+      "step": 3186
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.592857459440429e-05,
+      "loss": 26.9461,
+      "step": 3187
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.592545174361729e-05,
+      "loss": 24.2582,
+      "step": 3188
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.592232774652052e-05,
+      "loss": 0.8443,
+      "step": 3189
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.591920260319197e-05,
+      "loss": 27.9703,
+      "step": 3190
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.591607631370963e-05,
+      "loss": 25.5903,
+      "step": 3191
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.591294887815155e-05,
+      "loss": 25.2304,
+      "step": 3192
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.590982029659578e-05,
+      "loss": 0.8017,
+      "step": 3193
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.59066905691204e-05,
+      "loss": 24.9876,
+      "step": 3194
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.590355969580356e-05,
+      "loss": 24.8725,
+      "step": 3195
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.590042767672338e-05,
+      "loss": 25.4132,
+      "step": 3196
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.589729451195805e-05,
+      "loss": 0.7673,
+      "step": 3197
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.589416020158578e-05,
+      "loss": 24.4775,
+      "step": 3198
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.589102474568477e-05,
+      "loss": 0.7266,
+      "step": 3199
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.58878881443333e-05,
+      "loss": 27.32,
+      "step": 3200
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.588475039760968e-05,
+      "loss": 0.7801,
+      "step": 3201
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.58816115055922e-05,
+      "loss": 26.7238,
+      "step": 3202
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.587847146835925e-05,
+      "loss": 0.7605,
+      "step": 3203
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.587533028598915e-05,
+      "loss": 0.7509,
+      "step": 3204
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.587218795856034e-05,
+      "loss": 0.754,
+      "step": 3205
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.586904448615124e-05,
+      "loss": 25.3023,
+      "step": 3206
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.586589986884031e-05,
+      "loss": 26.2735,
+      "step": 3207
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.586275410670606e-05,
+      "loss": 24.8154,
+      "step": 3208
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.585960719982699e-05,
+      "loss": 22.8377,
+      "step": 3209
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.585645914828166e-05,
+      "loss": 22.9605,
+      "step": 3210
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.585330995214863e-05,
+      "loss": 0.7767,
+      "step": 3211
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.585015961150652e-05,
+      "loss": 29.6564,
+      "step": 3212
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.584700812643395e-05,
+      "loss": 27.4723,
+      "step": 3213
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.58438554970096e-05,
+      "loss": 26.2343,
+      "step": 3214
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.584070172331214e-05,
+      "loss": 26.9809,
+      "step": 3215
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.58375468054203e-05,
+      "loss": 0.7759,
+      "step": 3216
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.583439074341282e-05,
+      "loss": 0.7684,
+      "step": 3217
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.58312335373685e-05,
+      "loss": 23.7096,
+      "step": 3218
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.582807518736612e-05,
+      "loss": 26.2815,
+      "step": 3219
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.582491569348451e-05,
+      "loss": 24.4341,
+      "step": 3220
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.582175505580255e-05,
+      "loss": 27.2649,
+      "step": 3221
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.581859327439911e-05,
+      "loss": 0.7022,
+      "step": 3222
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.581543034935313e-05,
+      "loss": 25.5329,
+      "step": 3223
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.581226628074356e-05,
+      "loss": 0.7793,
+      "step": 3224
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.580910106864933e-05,
+      "loss": 25.2443,
+      "step": 3225
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.58059347131495e-05,
+      "loss": 26.1482,
+      "step": 3226
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.580276721432308e-05,
+      "loss": 0.7638,
+      "step": 3227
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.579959857224912e-05,
+      "loss": 0.7321,
+      "step": 3228
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.579642878700674e-05,
+      "loss": 25.9347,
+      "step": 3229
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.579325785867503e-05,
+      "loss": 26.8728,
+      "step": 3230
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.579008578733315e-05,
+      "loss": 0.75,
+      "step": 3231
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.578691257306027e-05,
+      "loss": 0.8279,
+      "step": 3232
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.57837382159356e-05,
+      "loss": 0.7495,
+      "step": 3233
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.578056271603837e-05,
+      "loss": 0.7601,
+      "step": 3234
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.577738607344784e-05,
+      "loss": 25.0492,
+      "step": 3235
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.577420828824331e-05,
+      "loss": 0.7835,
+      "step": 3236
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.577102936050409e-05,
+      "loss": 0.7798,
+      "step": 3237
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.576784929030951e-05,
+      "loss": 24.9479,
+      "step": 3238
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.576466807773899e-05,
+      "loss": 22.4476,
+      "step": 3239
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.576148572287187e-05,
+      "loss": 0.7354,
+      "step": 3240
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.575830222578764e-05,
+      "loss": 0.7553,
+      "step": 3241
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.575511758656572e-05,
+      "loss": 26.2434,
+      "step": 3242
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.575193180528564e-05,
+      "loss": 24.1558,
+      "step": 3243
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.574874488202689e-05,
+      "loss": 0.7647,
+      "step": 3244
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.5745556816869e-05,
+      "loss": 0.8068,
+      "step": 3245
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.57423676098916e-05,
+      "loss": 24.4348,
+      "step": 3246
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.573917726117426e-05,
+      "loss": 0.7415,
+      "step": 3247
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.573598577079659e-05,
+      "loss": 23.609,
+      "step": 3248
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.573279313883827e-05,
+      "loss": 27.5903,
+      "step": 3249
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.5729599365379e-05,
+      "loss": 0.7619,
+      "step": 3250
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.57264044504985e-05,
+      "loss": 0.7977,
+      "step": 3251
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.572320839427649e-05,
+      "loss": 25.1321,
+      "step": 3252
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.572001119679276e-05,
+      "loss": 0.7858,
+      "step": 3253
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.57168128581271e-05,
+      "loss": 0.7749,
+      "step": 3254
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.571361337835939e-05,
+      "loss": 21.9159,
+      "step": 3255
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.57104127575694e-05,
+      "loss": 24.5643,
+      "step": 3256
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.57072109958371e-05,
+      "loss": 26.3563,
+      "step": 3257
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.570400809324238e-05,
+      "loss": 23.2665,
+      "step": 3258
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.570080404986519e-05,
+      "loss": 25.9947,
+      "step": 3259
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.569759886578548e-05,
+      "loss": 25.4486,
+      "step": 3260
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.569439254108328e-05,
+      "loss": 0.7285,
+      "step": 3261
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.569118507583858e-05,
+      "loss": 0.7867,
+      "step": 3262
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.56879764701315e-05,
+      "loss": 0.791,
+      "step": 3263
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.568476672404209e-05,
+      "loss": 0.8014,
+      "step": 3264
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.568155583765046e-05,
+      "loss": 24.2327,
+      "step": 3265
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.567834381103677e-05,
+      "loss": 23.017,
+      "step": 3266
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.56751306442812e-05,
+      "loss": 0.7516,
+      "step": 3267
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.567191633746391e-05,
+      "loss": 24.7424,
+      "step": 3268
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.56687008906652e-05,
+      "loss": 0.7479,
+      "step": 3269
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.566548430396526e-05,
+      "loss": 24.9001,
+      "step": 3270
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.566226657744441e-05,
+      "loss": 0.7283,
+      "step": 3271
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.565904771118297e-05,
+      "loss": 0.8217,
+      "step": 3272
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.565582770526127e-05,
+      "loss": 27.1571,
+      "step": 3273
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.565260655975967e-05,
+      "loss": 0.7184,
+      "step": 3274
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.56493842747586e-05,
+      "loss": 0.7643,
+      "step": 3275
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.564616085033849e-05,
+      "loss": 26.0059,
+      "step": 3276
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.564293628657976e-05,
+      "loss": 0.7519,
+      "step": 3277
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.563971058356293e-05,
+      "loss": 25.9032,
+      "step": 3278
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.56364837413685e-05,
+      "loss": 0.8091,
+      "step": 3279
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.563325576007701e-05,
+      "loss": 26.3892,
+      "step": 3280
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.563002663976905e-05,
+      "loss": 0.7172,
+      "step": 3281
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.562679638052519e-05,
+      "loss": 25.0949,
+      "step": 3282
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.562356498242608e-05,
+      "loss": 0.8007,
+      "step": 3283
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.562033244555238e-05,
+      "loss": 22.9732,
+      "step": 3284
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.561709876998476e-05,
+      "loss": 0.6973,
+      "step": 3285
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.561386395580395e-05,
+      "loss": 23.7729,
+      "step": 3286
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.561062800309067e-05,
+      "loss": 25.0365,
+      "step": 3287
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.560739091192568e-05,
+      "loss": 0.8301,
+      "step": 3288
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.560415268238983e-05,
+      "loss": 0.7609,
+      "step": 3289
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.56009133145639e-05,
+      "loss": 24.2427,
+      "step": 3290
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.559767280852877e-05,
+      "loss": 25.8628,
+      "step": 3291
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.55944311643653e-05,
+      "loss": 24.5662,
+      "step": 3292
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.559118838215444e-05,
+      "loss": 24.3505,
+      "step": 3293
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.55879444619771e-05,
+      "loss": 23.8013,
+      "step": 3294
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.558469940391426e-05,
+      "loss": 0.8196,
+      "step": 3295
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.55814532080469e-05,
+      "loss": 0.7794,
+      "step": 3296
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.557820587445607e-05,
+      "loss": 22.6628,
+      "step": 3297
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.557495740322283e-05,
+      "loss": 0.758,
+      "step": 3298
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.557170779442821e-05,
+      "loss": 22.5297,
+      "step": 3299
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.556845704815338e-05,
+      "loss": 24.3551,
+      "step": 3300
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.556520516447946e-05,
+      "loss": 24.8375,
+      "step": 3301
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.556195214348762e-05,
+      "loss": 0.8265,
+      "step": 3302
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.555869798525905e-05,
+      "loss": 25.8716,
+      "step": 3303
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.555544268987497e-05,
+      "loss": 0.6873,
+      "step": 3304
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.555218625741663e-05,
+      "loss": 27.2951,
+      "step": 3305
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.554892868796532e-05,
+      "loss": 0.7813,
+      "step": 3306
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.554566998160238e-05,
+      "loss": 0.798,
+      "step": 3307
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.554241013840908e-05,
+      "loss": 21.7971,
+      "step": 3308
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.553914915846685e-05,
+      "loss": 0.7702,
+      "step": 3309
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.553588704185704e-05,
+      "loss": 28.3157,
+      "step": 3310
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.55326237886611e-05,
+      "loss": 24.1215,
+      "step": 3311
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.552935939896049e-05,
+      "loss": 24.8032,
+      "step": 3312
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.552609387283665e-05,
+      "loss": 24.1636,
+      "step": 3313
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.552282721037112e-05,
+      "loss": 23.8967,
+      "step": 3314
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.551955941164543e-05,
+      "loss": 24.9928,
+      "step": 3315
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.551629047674114e-05,
+      "loss": 26.4187,
+      "step": 3316
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.551302040573986e-05,
+      "loss": 0.7799,
+      "step": 3317
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.550974919872318e-05,
+      "loss": 0.736,
+      "step": 3318
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.550647685577278e-05,
+      "loss": 0.782,
+      "step": 3319
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.550320337697033e-05,
+      "loss": 0.7914,
+      "step": 3320
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.549992876239753e-05,
+      "loss": 22.0482,
+      "step": 3321
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.549665301213611e-05,
+      "loss": 24.0562,
+      "step": 3322
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.549337612626786e-05,
+      "loss": 24.3812,
+      "step": 3323
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.549009810487455e-05,
+      "loss": 25.2254,
+      "step": 3324
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.548681894803799e-05,
+      "loss": 23.7542,
+      "step": 3325
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.548353865584006e-05,
+      "loss": 27.379,
+      "step": 3326
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.548025722836261e-05,
+      "loss": 0.737,
+      "step": 3327
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.547697466568756e-05,
+      "loss": 23.9227,
+      "step": 3328
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.547369096789682e-05,
+      "loss": 25.194,
+      "step": 3329
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.54704061350724e-05,
+      "loss": 23.8898,
+      "step": 3330
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.546712016729624e-05,
+      "loss": 0.7491,
+      "step": 3331
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.546383306465039e-05,
+      "loss": 0.7007,
+      "step": 3332
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.54605448272169e-05,
+      "loss": 0.7782,
+      "step": 3333
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.54572554550778e-05,
+      "loss": 26.6114,
+      "step": 3334
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.545396494831524e-05,
+      "loss": 25.8684,
+      "step": 3335
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.545067330701133e-05,
+      "loss": 0.8377,
+      "step": 3336
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.544738053124825e-05,
+      "loss": 0.6966,
+      "step": 3337
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.544408662110816e-05,
+      "loss": 24.4554,
+      "step": 3338
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.544079157667329e-05,
+      "loss": 0.7579,
+      "step": 3339
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.54374953980259e-05,
+      "loss": 26.2739,
+      "step": 3340
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.543419808524824e-05,
+      "loss": 26.3349,
+      "step": 3341
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.543089963842264e-05,
+      "loss": 24.1474,
+      "step": 3342
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.542760005763138e-05,
+      "loss": 25.9793,
+      "step": 3343
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.542429934295688e-05,
+      "loss": 27.5227,
+      "step": 3344
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.542099749448148e-05,
+      "loss": 25.6422,
+      "step": 3345
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.541769451228763e-05,
+      "loss": 0.7525,
+      "step": 3346
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.541439039645775e-05,
+      "loss": 25.0407,
+      "step": 3347
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.541108514707429e-05,
+      "loss": 23.9566,
+      "step": 3348
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.540777876421981e-05,
+      "loss": 23.5433,
+      "step": 3349
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.540447124797678e-05,
+      "loss": 22.8452,
+      "step": 3350
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.540116259842778e-05,
+      "loss": 0.763,
+      "step": 3351
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.539785281565541e-05,
+      "loss": 0.7993,
+      "step": 3352
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.539454189974226e-05,
+      "loss": 0.7346,
+      "step": 3353
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.539122985077098e-05,
+      "loss": 0.8067,
+      "step": 3354
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.538791666882423e-05,
+      "loss": 0.7798,
+      "step": 3355
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.538460235398475e-05,
+      "loss": 0.7763,
+      "step": 3356
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.538128690633518e-05,
+      "loss": 0.8015,
+      "step": 3357
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.537797032595837e-05,
+      "loss": 26.813,
+      "step": 3358
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.537465261293704e-05,
+      "loss": 0.7316,
+      "step": 3359
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.537133376735404e-05,
+      "loss": 23.7865,
+      "step": 3360
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.536801378929216e-05,
+      "loss": 0.7464,
+      "step": 3361
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.536469267883433e-05,
+      "loss": 26.5911,
+      "step": 3362
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.536137043606339e-05,
+      "loss": 0.7484,
+      "step": 3363
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.535804706106229e-05,
+      "loss": 23.7284,
+      "step": 3364
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.535472255391398e-05,
+      "loss": 23.7464,
+      "step": 3365
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.535139691470143e-05,
+      "loss": 26.856,
+      "step": 3366
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.534807014350766e-05,
+      "loss": 21.8734,
+      "step": 3367
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.534474224041572e-05,
+      "loss": 25.1786,
+      "step": 3368
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.534141320550866e-05,
+      "loss": 0.7242,
+      "step": 3369
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.533808303886955e-05,
+      "loss": 0.7926,
+      "step": 3370
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.533475174058156e-05,
+      "loss": 24.6833,
+      "step": 3371
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.533141931072781e-05,
+      "loss": 0.7594,
+      "step": 3372
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.532808574939147e-05,
+      "loss": 0.7957,
+      "step": 3373
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.532475105665578e-05,
+      "loss": 27.309,
+      "step": 3374
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.532141523260393e-05,
+      "loss": 22.3094,
+      "step": 3375
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.531807827731924e-05,
+      "loss": 24.6401,
+      "step": 3376
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.531474019088495e-05,
+      "loss": 25.673,
+      "step": 3377
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.53114009733844e-05,
+      "loss": 22.7213,
+      "step": 3378
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.530806062490094e-05,
+      "loss": 24.4545,
+      "step": 3379
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.530471914551792e-05,
+      "loss": 0.8307,
+      "step": 3380
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.530137653531878e-05,
+      "loss": 27.0753,
+      "step": 3381
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.529803279438695e-05,
+      "loss": 0.7417,
+      "step": 3382
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.529468792280586e-05,
+      "loss": 0.7918,
+      "step": 3383
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.529134192065903e-05,
+      "loss": 22.0046,
+      "step": 3384
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.528799478802997e-05,
+      "loss": 24.1244,
+      "step": 3385
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.52846465250022e-05,
+      "loss": 23.7926,
+      "step": 3386
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.528129713165932e-05,
+      "loss": 23.0582,
+      "step": 3387
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.527794660808493e-05,
+      "loss": 0.7096,
+      "step": 3388
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.527459495436265e-05,
+      "loss": 23.6413,
+      "step": 3389
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.527124217057614e-05,
+      "loss": 0.7082,
+      "step": 3390
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.52678882568091e-05,
+      "loss": 0.7897,
+      "step": 3391
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.526453321314523e-05,
+      "loss": 26.4708,
+      "step": 3392
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.526117703966827e-05,
+      "loss": 0.7904,
+      "step": 3393
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.525781973646199e-05,
+      "loss": 27.0224,
+      "step": 3394
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.525446130361021e-05,
+      "loss": 0.7929,
+      "step": 3395
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.525110174119674e-05,
+      "loss": 22.81,
+      "step": 3396
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.524774104930543e-05,
+      "loss": 0.7605,
+      "step": 3397
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.524437922802019e-05,
+      "loss": 0.8211,
+      "step": 3398
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.52410162774249e-05,
+      "loss": 24.2334,
+      "step": 3399
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.52376521976035e-05,
+      "loss": 22.3714,
+      "step": 3400
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.523428698864e-05,
+      "loss": 0.7731,
+      "step": 3401
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.523092065061834e-05,
+      "loss": 24.5496,
+      "step": 3402
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.52275531836226e-05,
+      "loss": 0.7235,
+      "step": 3403
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.522418458773679e-05,
+      "loss": 22.6934,
+      "step": 3404
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.522081486304501e-05,
+      "loss": 26.5621,
+      "step": 3405
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.521744400963137e-05,
+      "loss": 0.7475,
+      "step": 3406
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.521407202757999e-05,
+      "loss": 0.8287,
+      "step": 3407
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.521069891697507e-05,
+      "loss": 23.6243,
+      "step": 3408
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.520732467790076e-05,
+      "loss": 20.0686,
+      "step": 3409
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.520394931044132e-05,
+      "loss": 26.5203,
+      "step": 3410
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.520057281468098e-05,
+      "loss": 25.1395,
+      "step": 3411
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.519719519070403e-05,
+      "loss": 0.836,
+      "step": 3412
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.519381643859476e-05,
+      "loss": 25.0934,
+      "step": 3413
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.519043655843749e-05,
+      "loss": 0.7854,
+      "step": 3414
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.518705555031663e-05,
+      "loss": 0.8233,
+      "step": 3415
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.518367341431655e-05,
+      "loss": 22.5953,
+      "step": 3416
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.518029015052166e-05,
+      "loss": 0.7896,
+      "step": 3417
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.51769057590164e-05,
+      "loss": 23.565,
+      "step": 3418
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.517352023988526e-05,
+      "loss": 25.358,
+      "step": 3419
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.517013359321276e-05,
+      "loss": 22.9713,
+      "step": 3420
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.516674581908338e-05,
+      "loss": 0.8025,
+      "step": 3421
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.516335691758174e-05,
+      "loss": 0.7171,
+      "step": 3422
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.515996688879238e-05,
+      "loss": 25.5397,
+      "step": 3423
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.515657573279994e-05,
+      "loss": 0.8052,
+      "step": 3424
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.515318344968906e-05,
+      "loss": 0.7071,
+      "step": 3425
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.514979003954443e-05,
+      "loss": 0.7635,
+      "step": 3426
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.514639550245071e-05,
+      "loss": 0.7601,
+      "step": 3427
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.514299983849266e-05,
+      "loss": 22.8648,
+      "step": 3428
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.513960304775503e-05,
+      "loss": 25.7508,
+      "step": 3429
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.513620513032258e-05,
+      "loss": 21.2233,
+      "step": 3430
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.513280608628016e-05,
+      "loss": 26.2512,
+      "step": 3431
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.51294059157126e-05,
+      "loss": 21.4182,
+      "step": 3432
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.512600461870475e-05,
+      "loss": 0.7766,
+      "step": 3433
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.512260219534153e-05,
+      "loss": 24.5153,
+      "step": 3434
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.511919864570784e-05,
+      "loss": 0.736,
+      "step": 3435
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.511579396988868e-05,
+      "loss": 0.7496,
+      "step": 3436
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.511238816796897e-05,
+      "loss": 26.2455,
+      "step": 3437
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.510898124003376e-05,
+      "loss": 23.6107,
+      "step": 3438
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.510557318616809e-05,
+      "loss": 26.0128,
+      "step": 3439
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.510216400645701e-05,
+      "loss": 24.4093,
+      "step": 3440
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.509875370098562e-05,
+      "loss": 22.5178,
+      "step": 3441
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.509534226983903e-05,
+      "loss": 0.7252,
+      "step": 3442
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.509192971310242e-05,
+      "loss": 0.7771,
+      "step": 3443
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.508851603086093e-05,
+      "loss": 22.7361,
+      "step": 3444
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.508510122319977e-05,
+      "loss": 25.0717,
+      "step": 3445
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.508168529020423e-05,
+      "loss": 26.5576,
+      "step": 3446
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.507826823195951e-05,
+      "loss": 26.3232,
+      "step": 3447
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.507485004855092e-05,
+      "loss": 22.9332,
+      "step": 3448
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.50714307400638e-05,
+      "loss": 0.7699,
+      "step": 3449
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.506801030658346e-05,
+      "loss": 0.7805,
+      "step": 3450
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.506458874819528e-05,
+      "loss": 0.8143,
+      "step": 3451
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.50611660649847e-05,
+      "loss": 23.8587,
+      "step": 3452
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.50577422570371e-05,
+      "loss": 25.3822,
+      "step": 3453
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.5054317324438e-05,
+      "loss": 0.7412,
+      "step": 3454
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.505089126727283e-05,
+      "loss": 0.7901,
+      "step": 3455
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.504746408562712e-05,
+      "loss": 0.8165,
+      "step": 3456
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.504403577958643e-05,
+      "loss": 24.8408,
+      "step": 3457
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.504060634923632e-05,
+      "loss": 27.3235,
+      "step": 3458
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.50371757946624e-05,
+      "loss": 0.7627,
+      "step": 3459
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.503374411595029e-05,
+      "loss": 0.7375,
+      "step": 3460
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.503031131318564e-05,
+      "loss": 0.6986,
+      "step": 3461
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.502687738645413e-05,
+      "loss": 0.7246,
+      "step": 3462
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.502344233584149e-05,
+      "loss": 0.8205,
+      "step": 3463
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.502000616143343e-05,
+      "loss": 0.7587,
+      "step": 3464
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.501656886331576e-05,
+      "loss": 26.4174,
+      "step": 3465
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.501313044157426e-05,
+      "loss": 26.3693,
+      "step": 3466
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.500969089629472e-05,
+      "loss": 26.4228,
+      "step": 3467
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.500625022756303e-05,
+      "loss": 24.9231,
+      "step": 3468
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.500280843546505e-05,
+      "loss": 0.7512,
+      "step": 3469
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.499936552008671e-05,
+      "loss": 24.6414,
+      "step": 3470
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.499592148151393e-05,
+      "loss": 0.761,
+      "step": 3471
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.499247631983266e-05,
+      "loss": 0.7484,
+      "step": 3472
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.49890300351289e-05,
+      "loss": 24.663,
+      "step": 3473
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.498558262748871e-05,
+      "loss": 0.7209,
+      "step": 3474
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.498213409699809e-05,
+      "loss": 26.3112,
+      "step": 3475
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.497868444374313e-05,
+      "loss": 0.8013,
+      "step": 3476
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.497523366780996e-05,
+      "loss": 0.7332,
+      "step": 3477
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.497178176928467e-05,
+      "loss": 22.6434,
+      "step": 3478
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.496832874825344e-05,
+      "loss": 23.6321,
+      "step": 3479
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.496487460480245e-05,
+      "loss": 27.6408,
+      "step": 3480
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.496141933901792e-05,
+      "loss": 25.2837,
+      "step": 3481
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.495796295098611e-05,
+      "loss": 22.4417,
+      "step": 3482
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.495450544079328e-05,
+      "loss": 0.8464,
+      "step": 3483
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.495104680852572e-05,
+      "loss": 25.6525,
+      "step": 3484
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.494758705426978e-05,
+      "loss": 0.7728,
+      "step": 3485
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.494412617811179e-05,
+      "loss": 22.8343,
+      "step": 3486
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.494066418013816e-05,
+      "loss": 24.6469,
+      "step": 3487
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.493720106043527e-05,
+      "loss": 25.2618,
+      "step": 3488
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.493373681908961e-05,
+      "loss": 0.7869,
+      "step": 3489
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.493027145618761e-05,
+      "loss": 24.1272,
+      "step": 3490
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.492680497181576e-05,
+      "loss": 22.9898,
+      "step": 3491
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.492333736606061e-05,
+      "loss": 0.7826,
+      "step": 3492
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.491986863900871e-05,
+      "loss": 20.9688,
+      "step": 3493
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.491639879074662e-05,
+      "loss": 0.7162,
+      "step": 3494
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.491292782136096e-05,
+      "loss": 26.446,
+      "step": 3495
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.490945573093839e-05,
+      "loss": 24.4397,
+      "step": 3496
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.490598251956553e-05,
+      "loss": 0.8145,
+      "step": 3497
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.49025081873291e-05,
+      "loss": 26.1773,
+      "step": 3498
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.48990327343158e-05,
+      "loss": 0.7359,
+      "step": 3499
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.489555616061242e-05,
+      "loss": 25.0208,
+      "step": 3500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.489207846630568e-05,
+      "loss": 23.9179,
+      "step": 3501
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.488859965148242e-05,
+      "loss": 0.7131,
+      "step": 3502
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.488511971622947e-05,
+      "loss": 0.7557,
+      "step": 3503
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.488163866063367e-05,
+      "loss": 0.7517,
+      "step": 3504
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.487815648478193e-05,
+      "loss": 23.5458,
+      "step": 3505
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.487467318876115e-05,
+      "loss": 0.8462,
+      "step": 3506
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.487118877265828e-05,
+      "loss": 25.9679,
+      "step": 3507
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.48677032365603e-05,
+      "loss": 0.7265,
+      "step": 3508
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.48642165805542e-05,
+      "loss": 0.7377,
+      "step": 3509
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.4860728804727e-05,
+      "loss": 0.7676,
+      "step": 3510
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.485723990916579e-05,
+      "loss": 0.7877,
+      "step": 3511
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.485374989395761e-05,
+      "loss": 24.6959,
+      "step": 3512
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.48502587591896e-05,
+      "loss": 24.9741,
+      "step": 3513
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.484676650494889e-05,
+      "loss": 24.1422,
+      "step": 3514
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.484327313132265e-05,
+      "loss": 23.4768,
+      "step": 3515
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.483977863839807e-05,
+      "loss": 24.8416,
+      "step": 3516
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.483628302626238e-05,
+      "loss": 24.0068,
+      "step": 3517
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.483278629500283e-05,
+      "loss": 0.7702,
+      "step": 3518
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.482928844470669e-05,
+      "loss": 0.7644,
+      "step": 3519
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.482578947546128e-05,
+      "loss": 24.1253,
+      "step": 3520
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.482228938735392e-05,
+      "loss": 24.5684,
+      "step": 3521
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.481878818047198e-05,
+      "loss": 0.8024,
+      "step": 3522
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.481528585490286e-05,
+      "loss": 24.2144,
+      "step": 3523
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.481178241073399e-05,
+      "loss": 25.091,
+      "step": 3524
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.480827784805278e-05,
+      "loss": 23.2496,
+      "step": 3525
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.480477216694673e-05,
+      "loss": 0.7552,
+      "step": 3526
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.480126536750334e-05,
+      "loss": 23.1362,
+      "step": 3527
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.479775744981013e-05,
+      "loss": 27.0607,
+      "step": 3528
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.479424841395468e-05,
+      "loss": 0.8026,
+      "step": 3529
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.479073826002453e-05,
+      "loss": 0.8038,
+      "step": 3530
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.478722698810736e-05,
+      "loss": 26.9529,
+      "step": 3531
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.478371459829077e-05,
+      "loss": 0.7106,
+      "step": 3532
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.478020109066242e-05,
+      "loss": 0.7782,
+      "step": 3533
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.477668646531004e-05,
+      "loss": 25.3591,
+      "step": 3534
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.477317072232134e-05,
+      "loss": 0.8095,
+      "step": 3535
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.476965386178407e-05,
+      "loss": 27.3689,
+      "step": 3536
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.476613588378603e-05,
+      "loss": 0.7294,
+      "step": 3537
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.4762616788415e-05,
+      "loss": 0.7852,
+      "step": 3538
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.475909657575886e-05,
+      "loss": 0.8074,
+      "step": 3539
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.475557524590543e-05,
+      "loss": 23.6924,
+      "step": 3540
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.475205279894262e-05,
+      "loss": 27.0275,
+      "step": 3541
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.474852923495836e-05,
+      "loss": 28.0205,
+      "step": 3542
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.474500455404059e-05,
+      "loss": 28.3593,
+      "step": 3543
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.47414787562773e-05,
+      "loss": 24.834,
+      "step": 3544
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.473795184175648e-05,
+      "loss": 26.0211,
+      "step": 3545
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.473442381056617e-05,
+      "loss": 0.7446,
+      "step": 3546
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.47308946627944e-05,
+      "loss": 0.7468,
+      "step": 3547
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.472736439852931e-05,
+      "loss": 0.8151,
+      "step": 3548
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.4723833017859e-05,
+      "loss": 28.8725,
+      "step": 3549
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.47203005208716e-05,
+      "loss": 25.9015,
+      "step": 3550
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.471676690765528e-05,
+      "loss": 25.4066,
+      "step": 3551
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.471323217829826e-05,
+      "loss": 24.5409,
+      "step": 3552
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.470969633288875e-05,
+      "loss": 22.0247,
+      "step": 3553
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.470615937151499e-05,
+      "loss": 0.8314,
+      "step": 3554
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.47026212942653e-05,
+      "loss": 0.7673,
+      "step": 3555
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.469908210122799e-05,
+      "loss": 0.7361,
+      "step": 3556
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.469554179249136e-05,
+      "loss": 0.7296,
+      "step": 3557
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.469200036814382e-05,
+      "loss": 0.7674,
+      "step": 3558
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.468845782827374e-05,
+      "loss": 30.0015,
+      "step": 3559
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.468491417296956e-05,
+      "loss": 31.1803,
+      "step": 3560
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.468136940231971e-05,
+      "loss": 29.2151,
+      "step": 3561
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.467782351641268e-05,
+      "loss": 0.7578,
+      "step": 3562
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.467427651533696e-05,
+      "loss": 26.6387,
+      "step": 3563
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.467072839918111e-05,
+      "loss": 25.6121,
+      "step": 3564
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.466717916803369e-05,
+      "loss": 28.639,
+      "step": 3565
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.466362882198325e-05,
+      "loss": 24.9837,
+      "step": 3566
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.466007736111847e-05,
+      "loss": 0.7806,
+      "step": 3567
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.465652478552794e-05,
+      "loss": 24.3383,
+      "step": 3568
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.465297109530035e-05,
+      "loss": 0.8529,
+      "step": 3569
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.464941629052441e-05,
+      "loss": 0.8159,
+      "step": 3570
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.464586037128886e-05,
+      "loss": 28.46,
+      "step": 3571
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.464230333768244e-05,
+      "loss": 27.0427,
+      "step": 3572
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.463874518979392e-05,
+      "loss": 0.7581,
+      "step": 3573
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.463518592771214e-05,
+      "loss": 24.192,
+      "step": 3574
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.463162555152591e-05,
+      "loss": 24.6228,
+      "step": 3575
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.462806406132415e-05,
+      "loss": 29.8673,
+      "step": 3576
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.462450145719568e-05,
+      "loss": 28.1006,
+      "step": 3577
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.462093773922949e-05,
+      "loss": 26.188,
+      "step": 3578
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.46173729075145e-05,
+      "loss": 0.7635,
+      "step": 3579
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.46138069621397e-05,
+      "loss": 23.7533,
+      "step": 3580
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.461023990319409e-05,
+      "loss": 0.7835,
+      "step": 3581
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.46066717307667e-05,
+      "loss": 23.9953,
+      "step": 3582
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.460310244494658e-05,
+      "loss": 25.3878,
+      "step": 3583
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.459953204582288e-05,
+      "loss": 0.7756,
+      "step": 3584
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.459596053348464e-05,
+      "loss": 0.7574,
+      "step": 3585
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.459238790802107e-05,
+      "loss": 0.7924,
+      "step": 3586
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.45888141695213e-05,
+      "loss": 22.511,
+      "step": 3587
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.458523931807454e-05,
+      "loss": 0.7363,
+      "step": 3588
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.458166335377006e-05,
+      "loss": 27.4157,
+      "step": 3589
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.457808627669705e-05,
+      "loss": 0.7248,
+      "step": 3590
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.457450808694485e-05,
+      "loss": 0.8149,
+      "step": 3591
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.457092878460275e-05,
+      "loss": 23.3283,
+      "step": 3592
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.456734836976007e-05,
+      "loss": 0.7683,
+      "step": 3593
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.456376684250621e-05,
+      "loss": 23.5342,
+      "step": 3594
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.456018420293055e-05,
+      "loss": 0.772,
+      "step": 3595
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.455660045112252e-05,
+      "loss": 0.7208,
+      "step": 3596
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.455301558717158e-05,
+      "loss": 26.4908,
+      "step": 3597
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.454942961116719e-05,
+      "loss": 20.4382,
+      "step": 3598
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.454584252319886e-05,
+      "loss": 24.979,
+      "step": 3599
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.454225432335613e-05,
+      "loss": 26.2195,
+      "step": 3600
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.453866501172857e-05,
+      "loss": 20.2034,
+      "step": 3601
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.453507458840575e-05,
+      "loss": 0.7972,
+      "step": 3602
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.453148305347732e-05,
+      "loss": 0.7613,
+      "step": 3603
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.452789040703288e-05,
+      "loss": 0.7417,
+      "step": 3604
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.452429664916215e-05,
+      "loss": 23.5306,
+      "step": 3605
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.45207017799548e-05,
+      "loss": 0.7508,
+      "step": 3606
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.451710579950057e-05,
+      "loss": 24.1798,
+      "step": 3607
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.45135087078892e-05,
+      "loss": 22.4041,
+      "step": 3608
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.45099105052105e-05,
+      "loss": 0.7618,
+      "step": 3609
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.450631119155429e-05,
+      "loss": 25.0904,
+      "step": 3610
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.450271076701036e-05,
+      "loss": 0.701,
+      "step": 3611
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.449910923166862e-05,
+      "loss": 26.0729,
+      "step": 3612
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.449550658561896e-05,
+      "loss": 0.8124,
+      "step": 3613
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.449190282895129e-05,
+      "loss": 23.6817,
+      "step": 3614
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.448829796175556e-05,
+      "loss": 0.7436,
+      "step": 3615
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.448469198412178e-05,
+      "loss": 0.7953,
+      "step": 3616
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.448108489613989e-05,
+      "loss": 0.6928,
+      "step": 3617
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.44774766979e-05,
+      "loss": 23.8216,
+      "step": 3618
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.447386738949212e-05,
+      "loss": 22.1628,
+      "step": 3619
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.447025697100636e-05,
+      "loss": 0.7538,
+      "step": 3620
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.446664544253284e-05,
+      "loss": 22.5288,
+      "step": 3621
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.446303280416168e-05,
+      "loss": 0.783,
+      "step": 3622
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.445941905598308e-05,
+      "loss": 0.8231,
+      "step": 3623
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.445580419808722e-05,
+      "loss": 0.8095,
+      "step": 3624
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.445218823056434e-05,
+      "loss": 0.7335,
+      "step": 3625
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.44485711535047e-05,
+      "loss": 0.7636,
+      "step": 3626
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.444495296699857e-05,
+      "loss": 0.7694,
+      "step": 3627
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.444133367113625e-05,
+      "loss": 25.8387,
+      "step": 3628
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.443771326600812e-05,
+      "loss": 0.8284,
+      "step": 3629
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.44340917517045e-05,
+      "loss": 21.0689,
+      "step": 3630
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.443046912831583e-05,
+      "loss": 22.739,
+      "step": 3631
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.442684539593248e-05,
+      "loss": 21.8625,
+      "step": 3632
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.442322055464494e-05,
+      "loss": 25.8846,
+      "step": 3633
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.441959460454365e-05,
+      "loss": 24.4341,
+      "step": 3634
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.441596754571916e-05,
+      "loss": 23.4692,
+      "step": 3635
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.441233937826197e-05,
+      "loss": 0.7445,
+      "step": 3636
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.440871010226267e-05,
+      "loss": 0.7752,
+      "step": 3637
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.44050797178118e-05,
+      "loss": 0.7356,
+      "step": 3638
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.4401448225e-05,
+      "loss": 25.4085,
+      "step": 3639
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.439781562391792e-05,
+      "loss": 23.7237,
+      "step": 3640
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.439418191465624e-05,
+      "loss": 0.7616,
+      "step": 3641
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.439054709730562e-05,
+      "loss": 25.3411,
+      "step": 3642
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.438691117195682e-05,
+      "loss": 23.0274,
+      "step": 3643
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.438327413870057e-05,
+      "loss": 0.7603,
+      "step": 3644
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.437963599762766e-05,
+      "loss": 24.4892,
+      "step": 3645
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.437599674882893e-05,
+      "loss": 23.3999,
+      "step": 3646
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.437235639239518e-05,
+      "loss": 23.2909,
+      "step": 3647
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.436871492841728e-05,
+      "loss": 23.3261,
+      "step": 3648
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.436507235698612e-05,
+      "loss": 23.5208,
+      "step": 3649
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.436142867819262e-05,
+      "loss": 22.913,
+      "step": 3650
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.435778389212775e-05,
+      "loss": 0.811,
+      "step": 3651
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.435413799888245e-05,
+      "loss": 0.7692,
+      "step": 3652
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.435049099854773e-05,
+      "loss": 23.1902,
+      "step": 3653
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.434684289121465e-05,
+      "loss": 0.7963,
+      "step": 3654
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.434319367697423e-05,
+      "loss": 0.7447,
+      "step": 3655
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.433954335591757e-05,
+      "loss": 0.6757,
+      "step": 3656
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.433589192813578e-05,
+      "loss": 23.0462,
+      "step": 3657
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.433223939372002e-05,
+      "loss": 0.7609,
+      "step": 3658
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.432858575276145e-05,
+      "loss": 0.773,
+      "step": 3659
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.432493100535125e-05,
+      "loss": 23.2135,
+      "step": 3660
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.432127515158064e-05,
+      "loss": 22.6095,
+      "step": 3661
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.431761819154088e-05,
+      "loss": 0.7847,
+      "step": 3662
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.431396012532327e-05,
+      "loss": 22.8869,
+      "step": 3663
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.431030095301908e-05,
+      "loss": 0.7577,
+      "step": 3664
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.430664067471967e-05,
+      "loss": 0.8147,
+      "step": 3665
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.430297929051639e-05,
+      "loss": 0.7549,
+      "step": 3666
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.429931680050064e-05,
+      "loss": 0.7672,
+      "step": 3667
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.429565320476382e-05,
+      "loss": 23.9939,
+      "step": 3668
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.429198850339738e-05,
+      "loss": 0.7658,
+      "step": 3669
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.42883226964928e-05,
+      "loss": 22.7607,
+      "step": 3670
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.428465578414156e-05,
+      "loss": 19.6052,
+      "step": 3671
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.42809877664352e-05,
+      "loss": 19.7536,
+      "step": 3672
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.427731864346527e-05,
+      "loss": 21.007,
+      "step": 3673
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.427364841532338e-05,
+      "loss": 21.7877,
+      "step": 3674
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.42699770821011e-05,
+      "loss": 24.9819,
+      "step": 3675
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.42663046438901e-05,
+      "loss": 0.7854,
+      "step": 3676
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.426263110078201e-05,
+      "loss": 0.7878,
+      "step": 3677
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.425895645286856e-05,
+      "loss": 23.6081,
+      "step": 3678
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.425528070024144e-05,
+      "loss": 21.743,
+      "step": 3679
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.42516038429924e-05,
+      "loss": 0.6934,
+      "step": 3680
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.424792588121325e-05,
+      "loss": 0.8723,
+      "step": 3681
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.424424681499574e-05,
+      "loss": 0.7509,
+      "step": 3682
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.424056664443174e-05,
+      "loss": 0.8058,
+      "step": 3683
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.42368853696131e-05,
+      "loss": 0.698,
+      "step": 3684
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.423320299063169e-05,
+      "loss": 0.7829,
+      "step": 3685
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.422951950757943e-05,
+      "loss": 23.9332,
+      "step": 3686
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.422583492054827e-05,
+      "loss": 23.3611,
+      "step": 3687
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.422214922963018e-05,
+      "loss": 23.367,
+      "step": 3688
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.421846243491714e-05,
+      "loss": 0.7977,
+      "step": 3689
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.421477453650118e-05,
+      "loss": 0.7306,
+      "step": 3690
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.421108553447436e-05,
+      "loss": 0.8008,
+      "step": 3691
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.420739542892875e-05,
+      "loss": 23.4844,
+      "step": 3692
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.420370421995644e-05,
+      "loss": 0.7229,
+      "step": 3693
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.42000119076496e-05,
+      "loss": 23.6382,
+      "step": 3694
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.419631849210036e-05,
+      "loss": 0.7399,
+      "step": 3695
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.419262397340091e-05,
+      "loss": 26.1743,
+      "step": 3696
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.41889283516435e-05,
+      "loss": 25.3069,
+      "step": 3697
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.418523162692033e-05,
+      "loss": 23.5434,
+      "step": 3698
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.41815337993237e-05,
+      "loss": 21.4,
+      "step": 3699
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.41778348689459e-05,
+      "loss": 0.7774,
+      "step": 3700
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.417413483587925e-05,
+      "loss": 0.7799,
+      "step": 3701
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.41704337002161e-05,
+      "loss": 0.7614,
+      "step": 3702
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.416673146204884e-05,
+      "loss": 25.6003,
+      "step": 3703
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.416302812146989e-05,
+      "loss": 0.7602,
+      "step": 3704
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.415932367857166e-05,
+      "loss": 20.85,
+      "step": 3705
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.415561813344665e-05,
+      "loss": 0.7641,
+      "step": 3706
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.415191148618731e-05,
+      "loss": 0.7505,
+      "step": 3707
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.414820373688617e-05,
+      "loss": 21.6561,
+      "step": 3708
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.41444948856358e-05,
+      "loss": 22.628,
+      "step": 3709
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.414078493252875e-05,
+      "loss": 21.9746,
+      "step": 3710
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.413707387765765e-05,
+      "loss": 0.7332,
+      "step": 3711
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.413336172111508e-05,
+      "loss": 22.2067,
+      "step": 3712
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.412964846299374e-05,
+      "loss": 25.4385,
+      "step": 3713
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.41259341033863e-05,
+      "loss": 23.154,
+      "step": 3714
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.412221864238548e-05,
+      "loss": 22.7361,
+      "step": 3715
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.411850208008399e-05,
+      "loss": 23.6602,
+      "step": 3716
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.411478441657462e-05,
+      "loss": 22.4834,
+      "step": 3717
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.411106565195017e-05,
+      "loss": 0.7592,
+      "step": 3718
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.410734578630343e-05,
+      "loss": 0.793,
+      "step": 3719
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.410362481972727e-05,
+      "loss": 22.4645,
+      "step": 3720
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.409990275231456e-05,
+      "loss": 0.7749,
+      "step": 3721
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.409617958415821e-05,
+      "loss": 23.4302,
+      "step": 3722
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.409245531535116e-05,
+      "loss": 24.984,
+      "step": 3723
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.408872994598637e-05,
+      "loss": 0.7621,
+      "step": 3724
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.408500347615679e-05,
+      "loss": 24.1102,
+      "step": 3725
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.408127590595545e-05,
+      "loss": 0.7396,
+      "step": 3726
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.407754723547543e-05,
+      "loss": 26.1993,
+      "step": 3727
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.407381746480974e-05,
+      "loss": 0.7536,
+      "step": 3728
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.407008659405152e-05,
+      "loss": 0.7287,
+      "step": 3729
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.406635462329387e-05,
+      "loss": 0.8145,
+      "step": 3730
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.406262155262994e-05,
+      "loss": 24.923,
+      "step": 3731
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.405888738215293e-05,
+      "loss": 0.8,
+      "step": 3732
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.405515211195601e-05,
+      "loss": 24.8512,
+      "step": 3733
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.405141574213245e-05,
+      "loss": 0.7612,
+      "step": 3734
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.40476782727755e-05,
+      "loss": 0.782,
+      "step": 3735
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.404393970397843e-05,
+      "loss": 0.7558,
+      "step": 3736
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.404020003583456e-05,
+      "loss": 24.1513,
+      "step": 3737
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.403645926843726e-05,
+      "loss": 22.7623,
+      "step": 3738
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.403271740187988e-05,
+      "loss": 22.649,
+      "step": 3739
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.402897443625582e-05,
+      "loss": 22.8349,
+      "step": 3740
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.402523037165849e-05,
+      "loss": 0.6984,
+      "step": 3741
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.402148520818137e-05,
+      "loss": 22.9735,
+      "step": 3742
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.401773894591793e-05,
+      "loss": 26.0482,
+      "step": 3743
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.401399158496167e-05,
+      "loss": 21.499,
+      "step": 3744
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.401024312540614e-05,
+      "loss": 0.7897,
+      "step": 3745
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.400649356734488e-05,
+      "loss": 22.0254,
+      "step": 3746
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.40027429108715e-05,
+      "loss": 22.7252,
+      "step": 3747
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.399899115607959e-05,
+      "loss": 20.8258,
+      "step": 3748
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.399523830306285e-05,
+      "loss": 22.39,
+      "step": 3749
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.399148435191489e-05,
+      "loss": 20.103,
+      "step": 3750
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.398772930272945e-05,
+      "loss": 19.9445,
+      "step": 3751
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.398397315560024e-05,
+      "loss": 0.7902,
+      "step": 3752
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.398021591062101e-05,
+      "loss": 25.4241,
+      "step": 3753
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.397645756788554e-05,
+      "loss": 22.8664,
+      "step": 3754
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.397269812748766e-05,
+      "loss": 0.8174,
+      "step": 3755
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.39689375895212e-05,
+      "loss": 0.7801,
+      "step": 3756
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.396517595408001e-05,
+      "loss": 22.5641,
+      "step": 3757
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.3961413221258e-05,
+      "loss": 24.2855,
+      "step": 3758
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.395764939114906e-05,
+      "loss": 0.7694,
+      "step": 3759
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.395388446384717e-05,
+      "loss": 0.7732,
+      "step": 3760
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.395011843944629e-05,
+      "loss": 23.3381,
+      "step": 3761
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.394635131804041e-05,
+      "loss": 21.1407,
+      "step": 3762
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.394258309972356e-05,
+      "loss": 22.6674,
+      "step": 3763
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.393881378458981e-05,
+      "loss": 0.7418,
+      "step": 3764
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.393504337273325e-05,
+      "loss": 24.7624,
+      "step": 3765
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.393127186424796e-05,
+      "loss": 0.7904,
+      "step": 3766
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.39274992592281e-05,
+      "loss": 25.2765,
+      "step": 3767
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.392372555776783e-05,
+      "loss": 0.7711,
+      "step": 3768
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.391995075996133e-05,
+      "loss": 0.7463,
+      "step": 3769
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.391617486590283e-05,
+      "loss": 0.7362,
+      "step": 3770
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.391239787568659e-05,
+      "loss": 22.7941,
+      "step": 3771
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.390861978940686e-05,
+      "loss": 26.1846,
+      "step": 3772
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.390484060715797e-05,
+      "loss": 0.7454,
+      "step": 3773
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.390106032903423e-05,
+      "loss": 23.541,
+      "step": 3774
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.389727895512999e-05,
+      "loss": 21.3284,
+      "step": 3775
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.389349648553964e-05,
+      "loss": 0.8036,
+      "step": 3776
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.38897129203576e-05,
+      "loss": 0.8902,
+      "step": 3777
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.388592825967832e-05,
+      "loss": 0.7639,
+      "step": 3778
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.388214250359624e-05,
+      "loss": 0.7838,
+      "step": 3779
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.387835565220586e-05,
+      "loss": 0.7357,
+      "step": 3780
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.38745677056017e-05,
+      "loss": 23.491,
+      "step": 3781
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.387077866387832e-05,
+      "loss": 22.5605,
+      "step": 3782
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.38669885271303e-05,
+      "loss": 0.7944,
+      "step": 3783
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.386319729545221e-05,
+      "loss": 25.4767,
+      "step": 3784
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.38594049689387e-05,
+      "loss": 25.1075,
+      "step": 3785
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.385561154768446e-05,
+      "loss": 24.3253,
+      "step": 3786
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.38518170317841e-05,
+      "loss": 0.7576,
+      "step": 3787
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.38480214213324e-05,
+      "loss": 23.6195,
+      "step": 3788
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.384422471642406e-05,
+      "loss": 23.5678,
+      "step": 3789
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.384042691715387e-05,
+      "loss": 0.791,
+      "step": 3790
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.38366280236166e-05,
+      "loss": 0.7931,
+      "step": 3791
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.38328280359071e-05,
+      "loss": 0.7961,
+      "step": 3792
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.38290269541202e-05,
+      "loss": 0.8094,
+      "step": 3793
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.382522477835078e-05,
+      "loss": 23.7518,
+      "step": 3794
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.382142150869373e-05,
+      "loss": 0.8015,
+      "step": 3795
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.3817617145244e-05,
+      "loss": 23.5894,
+      "step": 3796
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.381381168809656e-05,
+      "loss": 24.0614,
+      "step": 3797
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.381000513734636e-05,
+      "loss": 23.6021,
+      "step": 3798
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.380619749308843e-05,
+      "loss": 0.7759,
+      "step": 3799
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.38023887554178e-05,
+      "loss": 0.7635,
+      "step": 3800
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.379857892442955e-05,
+      "loss": 23.8607,
+      "step": 3801
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.379476800021878e-05,
+      "loss": 23.9671,
+      "step": 3802
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.37909559828806e-05,
+      "loss": 26.4561,
+      "step": 3803
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.378714287251016e-05,
+      "loss": 0.7468,
+      "step": 3804
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.378332866920263e-05,
+      "loss": 0.7828,
+      "step": 3805
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.377951337305321e-05,
+      "loss": 0.8548,
+      "step": 3806
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.377569698415716e-05,
+      "loss": 22.0226,
+      "step": 3807
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.37718795026097e-05,
+      "loss": 0.7988,
+      "step": 3808
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.376806092850614e-05,
+      "loss": 21.5468,
+      "step": 3809
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.376424126194179e-05,
+      "loss": 0.8272,
+      "step": 3810
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.376042050301198e-05,
+      "loss": 0.75,
+      "step": 3811
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.375659865181208e-05,
+      "loss": 23.3879,
+      "step": 3812
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.37527757084375e-05,
+      "loss": 23.4876,
+      "step": 3813
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.374895167298363e-05,
+      "loss": 21.0445,
+      "step": 3814
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.374512654554595e-05,
+      "loss": 24.9503,
+      "step": 3815
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.374130032621993e-05,
+      "loss": 22.1093,
+      "step": 3816
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.373747301510105e-05,
+      "loss": 0.7748,
+      "step": 3817
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.373364461228486e-05,
+      "loss": 25.1902,
+      "step": 3818
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.372981511786692e-05,
+      "loss": 22.7824,
+      "step": 3819
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.372598453194282e-05,
+      "loss": 25.5521,
+      "step": 3820
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.372215285460814e-05,
+      "loss": 0.7342,
+      "step": 3821
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.371832008595855e-05,
+      "loss": 0.7984,
+      "step": 3822
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.371448622608971e-05,
+      "loss": 25.0842,
+      "step": 3823
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.371065127509731e-05,
+      "loss": 0.7695,
+      "step": 3824
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.370681523307708e-05,
+      "loss": 22.3018,
+      "step": 3825
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.370297810012476e-05,
+      "loss": 0.7698,
+      "step": 3826
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.369913987633613e-05,
+      "loss": 0.7708,
+      "step": 3827
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.369530056180699e-05,
+      "loss": 0.7567,
+      "step": 3828
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.369146015663317e-05,
+      "loss": 0.7519,
+      "step": 3829
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.368761866091052e-05,
+      "loss": 22.5638,
+      "step": 3830
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.368377607473494e-05,
+      "loss": 0.7874,
+      "step": 3831
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.367993239820234e-05,
+      "loss": 23.0761,
+      "step": 3832
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.367608763140864e-05,
+      "loss": 22.5728,
+      "step": 3833
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.367224177444983e-05,
+      "loss": 0.7648,
+      "step": 3834
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.36683948274219e-05,
+      "loss": 0.806,
+      "step": 3835
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.366454679042085e-05,
+      "loss": 0.786,
+      "step": 3836
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.366069766354275e-05,
+      "loss": 23.1663,
+      "step": 3837
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.365684744688366e-05,
+      "loss": 20.3844,
+      "step": 3838
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.36529961405397e-05,
+      "loss": 24.7674,
+      "step": 3839
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.364914374460698e-05,
+      "loss": 22.9552,
+      "step": 3840
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.364529025918167e-05,
+      "loss": 23.0666,
+      "step": 3841
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.364143568435998e-05,
+      "loss": 0.7645,
+      "step": 3842
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.363758002023804e-05,
+      "loss": 0.7852,
+      "step": 3843
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.363372326691216e-05,
+      "loss": 0.7375,
+      "step": 3844
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.362986542447857e-05,
+      "loss": 26.3337,
+      "step": 3845
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.362600649303361e-05,
+      "loss": 23.2643,
+      "step": 3846
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.362214647267354e-05,
+      "loss": 25.597,
+      "step": 3847
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.361828536349475e-05,
+      "loss": 0.7709,
+      "step": 3848
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.361442316559359e-05,
+      "loss": 0.7241,
+      "step": 3849
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.361055987906646e-05,
+      "loss": 0.8112,
+      "step": 3850
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.360669550400982e-05,
+      "loss": 0.7339,
+      "step": 3851
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.360283004052008e-05,
+      "loss": 24.0059,
+      "step": 3852
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.359896348869377e-05,
+      "loss": 24.2382,
+      "step": 3853
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.359509584862736e-05,
+      "loss": 21.3762,
+      "step": 3854
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.35912271204174e-05,
+      "loss": 25.44,
+      "step": 3855
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.358735730416046e-05,
+      "loss": 22.7111,
+      "step": 3856
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.358348639995314e-05,
+      "loss": 0.7118,
+      "step": 3857
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.357961440789204e-05,
+      "loss": 22.8735,
+      "step": 3858
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.357574132807384e-05,
+      "loss": 23.7101,
+      "step": 3859
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.357186716059514e-05,
+      "loss": 0.7811,
+      "step": 3860
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.356799190555272e-05,
+      "loss": 0.7374,
+      "step": 3861
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.356411556304325e-05,
+      "loss": 23.2862,
+      "step": 3862
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.356023813316353e-05,
+      "loss": 23.2494,
+      "step": 3863
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.355635961601032e-05,
+      "loss": 0.8088,
+      "step": 3864
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.355248001168042e-05,
+      "loss": 0.7946,
+      "step": 3865
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.354859932027066e-05,
+      "loss": 0.8222,
+      "step": 3866
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.354471754187795e-05,
+      "loss": 23.3065,
+      "step": 3867
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.354083467659913e-05,
+      "loss": 20.7912,
+      "step": 3868
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.353695072453113e-05,
+      "loss": 0.8135,
+      "step": 3869
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.353306568577091e-05,
+      "loss": 0.7825,
+      "step": 3870
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.352917956041542e-05,
+      "loss": 23.7152,
+      "step": 3871
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.352529234856168e-05,
+      "loss": 26.1999,
+      "step": 3872
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.35214040503067e-05,
+      "loss": 21.0359,
+      "step": 3873
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.351751466574754e-05,
+      "loss": 23.2445,
+      "step": 3874
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.351362419498128e-05,
+      "loss": 22.2835,
+      "step": 3875
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.350973263810501e-05,
+      "loss": 0.7724,
+      "step": 3876
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.350583999521589e-05,
+      "loss": 23.5332,
+      "step": 3877
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.350194626641107e-05,
+      "loss": 0.7622,
+      "step": 3878
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.349805145178774e-05,
+      "loss": 20.9211,
+      "step": 3879
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.34941555514431e-05,
+      "loss": 22.8222,
+      "step": 3880
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.349025856547443e-05,
+      "loss": 23.8498,
+      "step": 3881
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.348636049397897e-05,
+      "loss": 0.8097,
+      "step": 3882
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.3482461337054e-05,
+      "loss": 22.7487,
+      "step": 3883
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.347856109479689e-05,
+      "loss": 0.7723,
+      "step": 3884
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.347465976730496e-05,
+      "loss": 0.7383,
+      "step": 3885
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.34707573546756e-05,
+      "loss": 22.1334,
+      "step": 3886
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.34668538570062e-05,
+      "loss": 0.7602,
+      "step": 3887
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.34629492743942e-05,
+      "loss": 0.7438,
+      "step": 3888
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.345904360693709e-05,
+      "loss": 0.743,
+      "step": 3889
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.34551368547323e-05,
+      "loss": 0.7568,
+      "step": 3890
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.34512290178774e-05,
+      "loss": 0.7486,
+      "step": 3891
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.344732009646988e-05,
+      "loss": 0.7434,
+      "step": 3892
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.344341009060733e-05,
+      "loss": 0.8087,
+      "step": 3893
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.343949900038735e-05,
+      "loss": 24.7341,
+      "step": 3894
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.343558682590756e-05,
+      "loss": 0.7248,
+      "step": 3895
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.34316735672656e-05,
+      "loss": 23.7441,
+      "step": 3896
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.342775922455916e-05,
+      "loss": 23.3595,
+      "step": 3897
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.342384379788591e-05,
+      "loss": 20.7549,
+      "step": 3898
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.341992728734364e-05,
+      "loss": 18.7576,
+      "step": 3899
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.341600969303004e-05,
+      "loss": 23.5626,
+      "step": 3900
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.341209101504293e-05,
+      "loss": 0.7607,
+      "step": 3901
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.340817125348013e-05,
+      "loss": 24.6005,
+      "step": 3902
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.340425040843946e-05,
+      "loss": 22.4375,
+      "step": 3903
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.340032848001879e-05,
+      "loss": 24.7645,
+      "step": 3904
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.339640546831601e-05,
+      "loss": 23.8063,
+      "step": 3905
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.339248137342902e-05,
+      "loss": 26.4923,
+      "step": 3906
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.338855619545582e-05,
+      "loss": 0.8282,
+      "step": 3907
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.338462993449432e-05,
+      "loss": 0.797,
+      "step": 3908
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.338070259064258e-05,
+      "loss": 0.7703,
+      "step": 3909
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.337677416399858e-05,
+      "loss": 23.7068,
+      "step": 3910
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.33728446546604e-05,
+      "loss": 0.7982,
+      "step": 3911
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.336891406272611e-05,
+      "loss": 23.4979,
+      "step": 3912
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.336498238829384e-05,
+      "loss": 0.7911,
+      "step": 3913
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.336104963146169e-05,
+      "loss": 23.8801,
+      "step": 3914
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.335711579232781e-05,
+      "loss": 24.3651,
+      "step": 3915
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.335318087099048e-05,
+      "loss": 22.0599,
+      "step": 3916
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.334924486754782e-05,
+      "loss": 22.7342,
+      "step": 3917
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.334530778209812e-05,
+      "loss": 22.2696,
+      "step": 3918
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.334136961473965e-05,
+      "loss": 23.9471,
+      "step": 3919
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.333743036557067e-05,
+      "loss": 0.8256,
+      "step": 3920
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.333349003468957e-05,
+      "loss": 23.8695,
+      "step": 3921
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.332954862219464e-05,
+      "loss": 0.7619,
+      "step": 3922
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.332560612818428e-05,
+      "loss": 0.7148,
+      "step": 3923
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.332166255275692e-05,
+      "loss": 25.0979,
+      "step": 3924
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.331771789601096e-05,
+      "loss": 21.7617,
+      "step": 3925
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.331377215804487e-05,
+      "loss": 21.7058,
+      "step": 3926
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.330982533895714e-05,
+      "loss": 19.7508,
+      "step": 3927
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.330587743884627e-05,
+      "loss": 0.7571,
+      "step": 3928
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.330192845781082e-05,
+      "loss": 23.0423,
+      "step": 3929
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.329797839594934e-05,
+      "loss": 0.7992,
+      "step": 3930
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.329402725336045e-05,
+      "loss": 0.7463,
+      "step": 3931
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.329007503014273e-05,
+      "loss": 0.7871,
+      "step": 3932
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.328612172639489e-05,
+      "loss": 0.7719,
+      "step": 3933
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.328216734221553e-05,
+      "loss": 0.7206,
+      "step": 3934
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.327821187770341e-05,
+      "loss": 0.744,
+      "step": 3935
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.327425533295724e-05,
+      "loss": 0.7288,
+      "step": 3936
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.327029770807575e-05,
+      "loss": 0.7886,
+      "step": 3937
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.326633900315778e-05,
+      "loss": 0.7344,
+      "step": 3938
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.32623792183021e-05,
+      "loss": 0.7099,
+      "step": 3939
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.325841835360756e-05,
+      "loss": 0.7655,
+      "step": 3940
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.325445640917301e-05,
+      "loss": 0.7974,
+      "step": 3941
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.325049338509738e-05,
+      "loss": 24.9532,
+      "step": 3942
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.324652928147952e-05,
+      "loss": 0.7703,
+      "step": 3943
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.324256409841843e-05,
+      "loss": 24.9324,
+      "step": 3944
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.323859783601308e-05,
+      "loss": 0.7841,
+      "step": 3945
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.323463049436246e-05,
+      "loss": 26.8408,
+      "step": 3946
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.323066207356557e-05,
+      "loss": 0.7558,
+      "step": 3947
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.322669257372148e-05,
+      "loss": 22.1189,
+      "step": 3948
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.322272199492929e-05,
+      "loss": 0.8209,
+      "step": 3949
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.321875033728809e-05,
+      "loss": 23.5753,
+      "step": 3950
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.321477760089701e-05,
+      "loss": 0.8009,
+      "step": 3951
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.32108037858552e-05,
+      "loss": 0.773,
+      "step": 3952
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.320682889226188e-05,
+      "loss": 21.2621,
+      "step": 3953
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.320285292021625e-05,
+      "loss": 0.7408,
+      "step": 3954
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.319887586981752e-05,
+      "loss": 0.8197,
+      "step": 3955
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.3194897741165e-05,
+      "loss": 0.7387,
+      "step": 3956
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.319091853435797e-05,
+      "loss": 23.281,
+      "step": 3957
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.318693824949574e-05,
+      "loss": 24.5682,
+      "step": 3958
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.318295688667769e-05,
+      "loss": 23.4472,
+      "step": 3959
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.317897444600316e-05,
+      "loss": 0.7237,
+      "step": 3960
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.317499092757158e-05,
+      "loss": 22.6626,
+      "step": 3961
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.317100633148237e-05,
+      "loss": 23.9568,
+      "step": 3962
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.316702065783498e-05,
+      "loss": 19.4786,
+      "step": 3963
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.31630339067289e-05,
+      "loss": 0.7686,
+      "step": 3964
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.315904607826362e-05,
+      "loss": 21.4398,
+      "step": 3965
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.315505717253871e-05,
+      "loss": 22.7865,
+      "step": 3966
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.315106718965375e-05,
+      "loss": 25.7438,
+      "step": 3967
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.314707612970826e-05,
+      "loss": 0.8398,
+      "step": 3968
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.31430839928019e-05,
+      "loss": 21.9168,
+      "step": 3969
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.313909077903432e-05,
+      "loss": 22.6721,
+      "step": 3970
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.31350964885052e-05,
+      "loss": 0.7735,
+      "step": 3971
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.31311011213142e-05,
+      "loss": 25.1884,
+      "step": 3972
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.312710467756107e-05,
+      "loss": 0.7574,
+      "step": 3973
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.312310715734557e-05,
+      "loss": 0.7707,
+      "step": 3974
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.311910856076746e-05,
+      "loss": 21.9044,
+      "step": 3975
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.311510888792657e-05,
+      "loss": 23.31,
+      "step": 3976
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.31111081389227e-05,
+      "loss": 0.7933,
+      "step": 3977
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.310710631385574e-05,
+      "loss": 24.244,
+      "step": 3978
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.310310341282555e-05,
+      "loss": 24.5131,
+      "step": 3979
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.309909943593207e-05,
+      "loss": 0.7556,
+      "step": 3980
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.309509438327523e-05,
+      "loss": 0.7006,
+      "step": 3981
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.309108825495499e-05,
+      "loss": 24.3134,
+      "step": 3982
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.308708105107134e-05,
+      "loss": 24.0001,
+      "step": 3983
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.308307277172432e-05,
+      "loss": 0.7744,
+      "step": 3984
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.307906341701396e-05,
+      "loss": 0.7865,
+      "step": 3985
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.307505298704033e-05,
+      "loss": 0.7764,
+      "step": 3986
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.307104148190357e-05,
+      "loss": 21.0187,
+      "step": 3987
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.306702890170375e-05,
+      "loss": 25.8073,
+      "step": 3988
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.306301524654109e-05,
+      "loss": 21.5684,
+      "step": 3989
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.305900051651571e-05,
+      "loss": 0.7648,
+      "step": 3990
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.305498471172784e-05,
+      "loss": 19.1395,
+      "step": 3991
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.305096783227774e-05,
+      "loss": 22.169,
+      "step": 3992
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.304694987826566e-05,
+      "loss": 0.8193,
+      "step": 3993
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.304293084979185e-05,
+      "loss": 0.7844,
+      "step": 3994
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.303891074695668e-05,
+      "loss": 0.7495,
+      "step": 3995
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.303488956986046e-05,
+      "loss": 0.7861,
+      "step": 3996
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.303086731860358e-05,
+      "loss": 0.7834,
+      "step": 3997
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.302684399328643e-05,
+      "loss": 0.8305,
+      "step": 3998
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.302281959400942e-05,
+      "loss": 22.7447,
+      "step": 3999
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.301879412087302e-05,
+      "loss": 23.6832,
+      "step": 4000
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.301476757397769e-05,
+      "loss": 23.7517,
+      "step": 4001
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.301073995342392e-05,
+      "loss": 20.2615,
+      "step": 4002
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.300671125931228e-05,
+      "loss": 22.0075,
+      "step": 4003
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.300268149174332e-05,
+      "loss": 0.8509,
+      "step": 4004
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.29986506508176e-05,
+      "loss": 0.7476,
+      "step": 4005
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.299461873663574e-05,
+      "loss": 23.3777,
+      "step": 4006
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.299058574929838e-05,
+      "loss": 0.7311,
+      "step": 4007
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.298655168890618e-05,
+      "loss": 20.7129,
+      "step": 4008
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.298251655555983e-05,
+      "loss": 0.7397,
+      "step": 4009
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.297848034936006e-05,
+      "loss": 23.3092,
+      "step": 4010
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.297444307040761e-05,
+      "loss": 0.8165,
+      "step": 4011
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.297040471880326e-05,
+      "loss": 0.7831,
+      "step": 4012
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.296636529464778e-05,
+      "loss": 24.0104,
+      "step": 4013
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.2962324798042e-05,
+      "loss": 23.9749,
+      "step": 4014
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.295828322908679e-05,
+      "loss": 21.6647,
+      "step": 4015
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.295424058788305e-05,
+      "loss": 0.8009,
+      "step": 4016
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.295019687453162e-05,
+      "loss": 21.8364,
+      "step": 4017
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.294615208913348e-05,
+      "loss": 0.7538,
+      "step": 4018
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.294210623178959e-05,
+      "loss": 0.7549,
+      "step": 4019
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.293805930260093e-05,
+      "loss": 0.7742,
+      "step": 4020
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.293401130166849e-05,
+      "loss": 0.7743,
+      "step": 4021
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.292996222909332e-05,
+      "loss": 26.1366,
+      "step": 4022
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.29259120849765e-05,
+      "loss": 22.1983,
+      "step": 4023
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.292186086941913e-05,
+      "loss": 23.1345,
+      "step": 4024
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.291780858252229e-05,
+      "loss": 0.8108,
+      "step": 4025
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.291375522438715e-05,
+      "loss": 0.7353,
+      "step": 4026
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.290970079511488e-05,
+      "loss": 0.7341,
+      "step": 4027
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.290564529480668e-05,
+      "loss": 27.0257,
+      "step": 4028
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.290158872356378e-05,
+      "loss": 0.767,
+      "step": 4029
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.289753108148744e-05,
+      "loss": 23.0689,
+      "step": 4030
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.289347236867891e-05,
+      "loss": 21.8524,
+      "step": 4031
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.288941258523953e-05,
+      "loss": 23.2306,
+      "step": 4032
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.28853517312706e-05,
+      "loss": 19.4671,
+      "step": 4033
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.288128980687352e-05,
+      "loss": 0.8013,
+      "step": 4034
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.287722681214965e-05,
+      "loss": 0.7543,
+      "step": 4035
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.287316274720042e-05,
+      "loss": 22.517,
+      "step": 4036
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.286909761212724e-05,
+      "loss": 0.7973,
+      "step": 4037
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.28650314070316e-05,
+      "loss": 26.7432,
+      "step": 4038
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.286096413201499e-05,
+      "loss": 21.4447,
+      "step": 4039
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.285689578717894e-05,
+      "loss": 23.4533,
+      "step": 4040
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.285282637262496e-05,
+      "loss": 25.2406,
+      "step": 4041
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.284875588845467e-05,
+      "loss": 0.8072,
+      "step": 4042
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.284468433476963e-05,
+      "loss": 23.2316,
+      "step": 4043
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.284061171167151e-05,
+      "loss": 22.6717,
+      "step": 4044
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.283653801926193e-05,
+      "loss": 20.76,
+      "step": 4045
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.283246325764258e-05,
+      "loss": 0.7882,
+      "step": 4046
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.282838742691516e-05,
+      "loss": 20.9386,
+      "step": 4047
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.282431052718142e-05,
+      "loss": 0.8128,
+      "step": 4048
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.282023255854311e-05,
+      "loss": 0.7723,
+      "step": 4049
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.281615352110203e-05,
+      "loss": 22.6715,
+      "step": 4050
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.281207341495996e-05,
+      "loss": 23.2796,
+      "step": 4051
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.280799224021876e-05,
+      "loss": 23.5547,
+      "step": 4052
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.280390999698032e-05,
+      "loss": 0.7632,
+      "step": 4053
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.27998266853465e-05,
+      "loss": 0.7289,
+      "step": 4054
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.279574230541925e-05,
+      "loss": 22.7618,
+      "step": 4055
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.279165685730049e-05,
+      "loss": 0.7406,
+      "step": 4056
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.278757034109219e-05,
+      "loss": 0.7825,
+      "step": 4057
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.278348275689638e-05,
+      "loss": 20.6891,
+      "step": 4058
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.277939410481507e-05,
+      "loss": 0.7567,
+      "step": 4059
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.27753043849503e-05,
+      "loss": 0.7599,
+      "step": 4060
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.277121359740419e-05,
+      "loss": 0.7934,
+      "step": 4061
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.276712174227882e-05,
+      "loss": 0.7089,
+      "step": 4062
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.27630288196763e-05,
+      "loss": 0.7535,
+      "step": 4063
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.275893482969883e-05,
+      "loss": 20.7207,
+      "step": 4064
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.275483977244861e-05,
+      "loss": 0.8012,
+      "step": 4065
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.27507436480278e-05,
+      "loss": 0.7336,
+      "step": 4066
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.274664645653867e-05,
+      "loss": 0.7739,
+      "step": 4067
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.274254819808347e-05,
+      "loss": 21.9428,
+      "step": 4068
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.273844887276455e-05,
+      "loss": 0.7548,
+      "step": 4069
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.273434848068415e-05,
+      "loss": 24.3179,
+      "step": 4070
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.273024702194468e-05,
+      "loss": 24.0867,
+      "step": 4071
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.272614449664847e-05,
+      "loss": 0.7849,
+      "step": 4072
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.272204090489793e-05,
+      "loss": 0.7477,
+      "step": 4073
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.271793624679552e-05,
+      "loss": 21.5596,
+      "step": 4074
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.271383052244364e-05,
+      "loss": 0.7568,
+      "step": 4075
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.270972373194483e-05,
+      "loss": 22.0738,
+      "step": 4076
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.270561587540153e-05,
+      "loss": 22.2413,
+      "step": 4077
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.270150695291632e-05,
+      "loss": 21.5629,
+      "step": 4078
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.269739696459176e-05,
+      "loss": 23.1289,
+      "step": 4079
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.26932859105304e-05,
+      "loss": 22.6924,
+      "step": 4080
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.268917379083489e-05,
+      "loss": 20.5737,
+      "step": 4081
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.268506060560784e-05,
+      "loss": 0.7614,
+      "step": 4082
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.268094635495195e-05,
+      "loss": 24.0301,
+      "step": 4083
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.267683103896989e-05,
+      "loss": 21.8628,
+      "step": 4084
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.267271465776437e-05,
+      "loss": 0.7428,
+      "step": 4085
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.266859721143817e-05,
+      "loss": 0.7846,
+      "step": 4086
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.266447870009403e-05,
+      "loss": 0.7679,
+      "step": 4087
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.266035912383475e-05,
+      "loss": 0.7418,
+      "step": 4088
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.265623848276318e-05,
+      "loss": 0.8165,
+      "step": 4089
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.265211677698216e-05,
+      "loss": 26.5849,
+      "step": 4090
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.264799400659455e-05,
+      "loss": 0.7781,
+      "step": 4091
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.264387017170329e-05,
+      "loss": 23.1715,
+      "step": 4092
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.263974527241128e-05,
+      "loss": 22.1758,
+      "step": 4093
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.263561930882148e-05,
+      "loss": 19.9117,
+      "step": 4094
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.263149228103691e-05,
+      "loss": 0.7058,
+      "step": 4095
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.262736418916057e-05,
+      "loss": 21.5194,
+      "step": 4096
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.262323503329545e-05,
+      "loss": 24.0003,
+      "step": 4097
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.261910481354467e-05,
+      "loss": 0.7934,
+      "step": 4098
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.26149735300113e-05,
+      "loss": 22.0748,
+      "step": 4099
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.261084118279847e-05,
+      "loss": 21.6246,
+      "step": 4100
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.260670777200931e-05,
+      "loss": 23.1733,
+      "step": 4101
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.2602573297747e-05,
+      "loss": 21.3633,
+      "step": 4102
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.259843776011472e-05,
+      "loss": 0.8274,
+      "step": 4103
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.259430115921572e-05,
+      "loss": 0.7275,
+      "step": 4104
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.259016349515321e-05,
+      "loss": 0.7824,
+      "step": 4105
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.258602476803053e-05,
+      "loss": 22.8948,
+      "step": 4106
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.258188497795093e-05,
+      "loss": 23.1245,
+      "step": 4107
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.257774412501777e-05,
+      "loss": 0.77,
+      "step": 4108
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.257360220933437e-05,
+      "loss": 20.3537,
+      "step": 4109
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.256945923100416e-05,
+      "loss": 22.0582,
+      "step": 4110
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.256531519013053e-05,
+      "loss": 0.7684,
+      "step": 4111
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.25611700868169e-05,
+      "loss": 22.5362,
+      "step": 4112
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.255702392116677e-05,
+      "loss": 0.8022,
+      "step": 4113
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.255287669328358e-05,
+      "loss": 24.5589,
+      "step": 4114
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.254872840327087e-05,
+      "loss": 21.4332,
+      "step": 4115
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.25445790512322e-05,
+      "loss": 21.4582,
+      "step": 4116
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.254042863727112e-05,
+      "loss": 0.8368,
+      "step": 4117
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.253627716149122e-05,
+      "loss": 0.7064,
+      "step": 4118
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.253212462399614e-05,
+      "loss": 0.7218,
+      "step": 4119
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.25279710248895e-05,
+      "loss": 0.752,
+      "step": 4120
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.252381636427503e-05,
+      "loss": 25.7406,
+      "step": 4121
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.251966064225635e-05,
+      "loss": 21.5447,
+      "step": 4122
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.251550385893726e-05,
+      "loss": 0.7671,
+      "step": 4123
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.251134601442148e-05,
+      "loss": 26.4839,
+      "step": 4124
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.250718710881279e-05,
+      "loss": 18.4803,
+      "step": 4125
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.250302714221502e-05,
+      "loss": 22.9183,
+      "step": 4126
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.249886611473197e-05,
+      "loss": 23.2359,
+      "step": 4127
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.249470402646752e-05,
+      "loss": 23.0018,
+      "step": 4128
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.249054087752557e-05,
+      "loss": 22.8531,
+      "step": 4129
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.248637666801001e-05,
+      "loss": 0.7846,
+      "step": 4130
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.248221139802478e-05,
+      "loss": 0.7928,
+      "step": 4131
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.247804506767385e-05,
+      "loss": 23.8368,
+      "step": 4132
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.247387767706124e-05,
+      "loss": 22.5044,
+      "step": 4133
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.246970922629092e-05,
+      "loss": 0.7325,
+      "step": 4134
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.246553971546697e-05,
+      "loss": 22.1329,
+      "step": 4135
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.246136914469346e-05,
+      "loss": 21.1715,
+      "step": 4136
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.245719751407447e-05,
+      "loss": 0.7949,
+      "step": 4137
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.245302482371415e-05,
+      "loss": 22.1524,
+      "step": 4138
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.244885107371661e-05,
+      "loss": 0.8026,
+      "step": 4139
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.244467626418609e-05,
+      "loss": 0.7947,
+      "step": 4140
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.244050039522672e-05,
+      "loss": 0.7509,
+      "step": 4141
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.24363234669428e-05,
+      "loss": 22.2262,
+      "step": 4142
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.243214547943855e-05,
+      "loss": 0.8066,
+      "step": 4143
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.242796643281825e-05,
+      "loss": 0.6973,
+      "step": 4144
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.242378632718623e-05,
+      "loss": 0.7499,
+      "step": 4145
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.241960516264681e-05,
+      "loss": 0.7926,
+      "step": 4146
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.241542293930436e-05,
+      "loss": 0.7785,
+      "step": 4147
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.241123965726328e-05,
+      "loss": 21.6491,
+      "step": 4148
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.240705531662797e-05,
+      "loss": 23.4945,
+      "step": 4149
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.240286991750287e-05,
+      "loss": 23.9674,
+      "step": 4150
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.239868345999246e-05,
+      "loss": 25.2083,
+      "step": 4151
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.239449594420123e-05,
+      "loss": 0.7329,
+      "step": 4152
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.239030737023369e-05,
+      "loss": 25.0042,
+      "step": 4153
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.238611773819442e-05,
+      "loss": 0.796,
+      "step": 4154
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.238192704818795e-05,
+      "loss": 21.1035,
+      "step": 4155
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.237773530031892e-05,
+      "loss": 25.5569,
+      "step": 4156
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.237354249469194e-05,
+      "loss": 22.1139,
+      "step": 4157
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.236934863141167e-05,
+      "loss": 0.7817,
+      "step": 4158
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.236515371058277e-05,
+      "loss": 23.3313,
+      "step": 4159
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.236095773230997e-05,
+      "loss": 0.7695,
+      "step": 4160
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.235676069669798e-05,
+      "loss": 0.8179,
+      "step": 4161
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.23525626038516e-05,
+      "loss": 0.7004,
+      "step": 4162
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.234836345387555e-05,
+      "loss": 23.8645,
+      "step": 4163
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.23441632468747e-05,
+      "loss": 0.7051,
+      "step": 4164
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.233996198295386e-05,
+      "loss": 20.6702,
+      "step": 4165
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.23357596622179e-05,
+      "loss": 25.188,
+      "step": 4166
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.233155628477173e-05,
+      "loss": 23.2853,
+      "step": 4167
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.232735185072023e-05,
+      "loss": 0.7128,
+      "step": 4168
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.232314636016837e-05,
+      "loss": 22.4085,
+      "step": 4169
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.231893981322111e-05,
+      "loss": 20.7299,
+      "step": 4170
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.231473220998345e-05,
+      "loss": 0.8307,
+      "step": 4171
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.231052355056042e-05,
+      "loss": 21.4383,
+      "step": 4172
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.230631383505706e-05,
+      "loss": 23.207,
+      "step": 4173
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.230210306357842e-05,
+      "loss": 21.9395,
+      "step": 4174
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.229789123622967e-05,
+      "loss": 0.7521,
+      "step": 4175
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.229367835311587e-05,
+      "loss": 0.7327,
+      "step": 4176
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.228946441434221e-05,
+      "loss": 0.7755,
+      "step": 4177
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.228524942001386e-05,
+      "loss": 0.7269,
+      "step": 4178
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.228103337023604e-05,
+      "loss": 0.7605,
+      "step": 4179
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.227681626511397e-05,
+      "loss": 23.8237,
+      "step": 4180
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.22725981047529e-05,
+      "loss": 0.7008,
+      "step": 4181
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.226837888925813e-05,
+      "loss": 0.802,
+      "step": 4182
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.226415861873499e-05,
+      "loss": 0.7574,
+      "step": 4183
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.225993729328879e-05,
+      "loss": 22.2296,
+      "step": 4184
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.225571491302489e-05,
+      "loss": 21.3435,
+      "step": 4185
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.225149147804872e-05,
+      "loss": 21.9855,
+      "step": 4186
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.224726698846566e-05,
+      "loss": 21.4595,
+      "step": 4187
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.224304144438119e-05,
+      "loss": 0.774,
+      "step": 4188
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.223881484590075e-05,
+      "loss": 0.7533,
+      "step": 4189
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.223458719312984e-05,
+      "loss": 22.4584,
+      "step": 4190
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.223035848617399e-05,
+      "loss": 0.7757,
+      "step": 4191
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.222612872513876e-05,
+      "loss": 20.3266,
+      "step": 4192
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.222189791012972e-05,
+      "loss": 24.6542,
+      "step": 4193
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.221766604125245e-05,
+      "loss": 20.6155,
+      "step": 4194
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.22134331186126e-05,
+      "loss": 20.3784,
+      "step": 4195
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.220919914231582e-05,
+      "loss": 0.8089,
+      "step": 4196
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.220496411246781e-05,
+      "loss": 0.7599,
+      "step": 4197
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.220072802917424e-05,
+      "loss": 21.2872,
+      "step": 4198
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.219649089254087e-05,
+      "loss": 0.7612,
+      "step": 4199
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.219225270267344e-05,
+      "loss": 27.0851,
+      "step": 4200
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.218801345967777e-05,
+      "loss": 0.7643,
+      "step": 4201
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.218377316365965e-05,
+      "loss": 0.7624,
+      "step": 4202
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.21795318147249e-05,
+      "loss": 0.8052,
+      "step": 4203
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.217528941297941e-05,
+      "loss": 25.5433,
+      "step": 4204
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.217104595852908e-05,
+      "loss": 22.8525,
+      "step": 4205
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.21668014514798e-05,
+      "loss": 21.6324,
+      "step": 4206
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.216255589193755e-05,
+      "loss": 0.7475,
+      "step": 4207
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.215830928000825e-05,
+      "loss": 22.0141,
+      "step": 4208
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.215406161579794e-05,
+      "loss": 0.7544,
+      "step": 4209
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.214981289941262e-05,
+      "loss": 0.7408,
+      "step": 4210
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.214556313095836e-05,
+      "loss": 0.7257,
+      "step": 4211
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.214131231054122e-05,
+      "loss": 21.7895,
+      "step": 4212
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.213706043826728e-05,
+      "loss": 0.7919,
+      "step": 4213
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.213280751424272e-05,
+      "loss": 20.9755,
+      "step": 4214
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.212855353857363e-05,
+      "loss": 0.8338,
+      "step": 4215
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.212429851136625e-05,
+      "loss": 0.7283,
+      "step": 4216
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.212004243272676e-05,
+      "loss": 23.8277,
+      "step": 4217
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.211578530276139e-05,
+      "loss": 20.8563,
+      "step": 4218
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.21115271215764e-05,
+      "loss": 0.7445,
+      "step": 4219
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.210726788927807e-05,
+      "loss": 0.8201,
+      "step": 4220
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.210300760597272e-05,
+      "loss": 22.926,
+      "step": 4221
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.20987462717667e-05,
+      "loss": 0.7767,
+      "step": 4222
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.209448388676635e-05,
+      "loss": 21.5277,
+      "step": 4223
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.209022045107808e-05,
+      "loss": 18.5962,
+      "step": 4224
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.208595596480829e-05,
+      "loss": 21.874,
+      "step": 4225
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.208169042806342e-05,
+      "loss": 0.766,
+      "step": 4226
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.207742384094997e-05,
+      "loss": 0.8149,
+      "step": 4227
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.20731562035744e-05,
+      "loss": 20.4795,
+      "step": 4228
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.206888751604326e-05,
+      "loss": 0.7863,
+      "step": 4229
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.206461777846307e-05,
+      "loss": 0.7723,
+      "step": 4230
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.206034699094041e-05,
+      "loss": 23.5213,
+      "step": 4231
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.20560751535819e-05,
+      "loss": 0.7563,
+      "step": 4232
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.205180226649414e-05,
+      "loss": 0.7902,
+      "step": 4233
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.20475283297838e-05,
+      "loss": 0.7358,
+      "step": 4234
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.204325334355754e-05,
+      "loss": 0.788,
+      "step": 4235
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.203897730792208e-05,
+      "loss": 0.8148,
+      "step": 4236
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.203470022298416e-05,
+      "loss": 19.5378,
+      "step": 4237
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.203042208885052e-05,
+      "loss": 0.7595,
+      "step": 4238
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.202614290562794e-05,
+      "loss": 0.7893,
+      "step": 4239
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.202186267342325e-05,
+      "loss": 21.9587,
+      "step": 4240
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.201758139234327e-05,
+      "loss": 22.147,
+      "step": 4241
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.201329906249485e-05,
+      "loss": 0.7602,
+      "step": 4242
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.20090156839849e-05,
+      "loss": 20.0727,
+      "step": 4243
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.200473125692033e-05,
+      "loss": 0.7268,
+      "step": 4244
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.200044578140808e-05,
+      "loss": 0.766,
+      "step": 4245
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.199615925755512e-05,
+      "loss": 22.1703,
+      "step": 4246
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.199187168546842e-05,
+      "loss": 20.9509,
+      "step": 4247
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.198758306525503e-05,
+      "loss": 23.4961,
+      "step": 4248
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.198329339702196e-05,
+      "loss": 0.752,
+      "step": 4249
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.197900268087631e-05,
+      "loss": 0.7369,
+      "step": 4250
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.197471091692518e-05,
+      "loss": 0.7394,
+      "step": 4251
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.197041810527566e-05,
+      "loss": 22.9368,
+      "step": 4252
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.196612424603494e-05,
+      "loss": 0.7627,
+      "step": 4253
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.196182933931017e-05,
+      "loss": 0.7502,
+      "step": 4254
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.195753338520855e-05,
+      "loss": 21.3248,
+      "step": 4255
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.195323638383733e-05,
+      "loss": 22.5109,
+      "step": 4256
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.194893833530374e-05,
+      "loss": 18.6084,
+      "step": 4257
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.194463923971507e-05,
+      "loss": 23.1594,
+      "step": 4258
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.194033909717863e-05,
+      "loss": 21.1904,
+      "step": 4259
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.193603790780177e-05,
+      "loss": 0.782,
+      "step": 4260
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.19317356716918e-05,
+      "loss": 23.7386,
+      "step": 4261
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.192743238895613e-05,
+      "loss": 0.7692,
+      "step": 4262
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.19231280597022e-05,
+      "loss": 0.7422,
+      "step": 4263
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.191882268403743e-05,
+      "loss": 21.5461,
+      "step": 4264
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.191451626206924e-05,
+      "loss": 0.8056,
+      "step": 4265
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.191020879390517e-05,
+      "loss": 0.7415,
+      "step": 4266
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.190590027965272e-05,
+      "loss": 0.7598,
+      "step": 4267
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.190159071941941e-05,
+      "loss": 25.6426,
+      "step": 4268
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.189728011331285e-05,
+      "loss": 20.6827,
+      "step": 4269
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.189296846144062e-05,
+      "loss": 0.8288,
+      "step": 4270
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.188865576391032e-05,
+      "loss": 22.96,
+      "step": 4271
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.18843420208296e-05,
+      "loss": 21.8761,
+      "step": 4272
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.188002723230615e-05,
+      "loss": 0.7462,
+      "step": 4273
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.187571139844766e-05,
+      "loss": 21.9164,
+      "step": 4274
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.187139451936184e-05,
+      "loss": 21.7755,
+      "step": 4275
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.186707659515647e-05,
+      "loss": 0.7242,
+      "step": 4276
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.186275762593929e-05,
+      "loss": 22.469,
+      "step": 4277
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.185843761181813e-05,
+      "loss": 20.246,
+      "step": 4278
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.185411655290081e-05,
+      "loss": 0.7109,
+      "step": 4279
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.184979444929518e-05,
+      "loss": 24.4271,
+      "step": 4280
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.184547130110913e-05,
+      "loss": 20.9203,
+      "step": 4281
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.184114710845056e-05,
+      "loss": 21.9576,
+      "step": 4282
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.183682187142742e-05,
+      "loss": 22.0138,
+      "step": 4283
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.183249559014763e-05,
+      "loss": 22.0365,
+      "step": 4284
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.18281682647192e-05,
+      "loss": 21.1433,
+      "step": 4285
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.182383989525014e-05,
+      "loss": 0.7759,
+      "step": 4286
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.181951048184851e-05,
+      "loss": 23.5649,
+      "step": 4287
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.181518002462233e-05,
+      "loss": 0.7765,
+      "step": 4288
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.181084852367971e-05,
+      "loss": 0.7737,
+      "step": 4289
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.180651597912878e-05,
+      "loss": 23.2311,
+      "step": 4290
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.180218239107762e-05,
+      "loss": 0.7485,
+      "step": 4291
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.17978477596345e-05,
+      "loss": 19.0384,
+      "step": 4292
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.179351208490751e-05,
+      "loss": 0.7831,
+      "step": 4293
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.178917536700496e-05,
+      "loss": 0.8171,
+      "step": 4294
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.178483760603501e-05,
+      "loss": 22.8231,
+      "step": 4295
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.1780498802106e-05,
+      "loss": 21.7582,
+      "step": 4296
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.177615895532618e-05,
+      "loss": 0.7335,
+      "step": 4297
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.17718180658039e-05,
+      "loss": 24.4724,
+      "step": 4298
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.176747613364751e-05,
+      "loss": 22.1971,
+      "step": 4299
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.176313315896539e-05,
+      "loss": 0.7404,
+      "step": 4300
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.17587891418659e-05,
+      "loss": 21.9124,
+      "step": 4301
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.175444408245752e-05,
+      "loss": 21.3169,
+      "step": 4302
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.175009798084868e-05,
+      "loss": 20.8384,
+      "step": 4303
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.174575083714786e-05,
+      "loss": 0.7973,
+      "step": 4304
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.174140265146356e-05,
+      "loss": 21.5414,
+      "step": 4305
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.173705342390433e-05,
+      "loss": 0.7335,
+      "step": 4306
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.173270315457872e-05,
+      "loss": 0.83,
+      "step": 4307
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.17283518435953e-05,
+      "loss": 20.953,
+      "step": 4308
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.172399949106268e-05,
+      "loss": 0.7356,
+      "step": 4309
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.171964609708952e-05,
+      "loss": 21.957,
+      "step": 4310
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.171529166178447e-05,
+      "loss": 0.766,
+      "step": 4311
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.171093618525622e-05,
+      "loss": 23.9121,
+      "step": 4312
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.170657966761346e-05,
+      "loss": 0.7934,
+      "step": 4313
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.170222210896495e-05,
+      "loss": 0.7499,
+      "step": 4314
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.169786350941947e-05,
+      "loss": 0.7725,
+      "step": 4315
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.16935038690858e-05,
+      "loss": 22.8427,
+      "step": 4316
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.168914318807274e-05,
+      "loss": 0.8281,
+      "step": 4317
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.168478146648915e-05,
+      "loss": 0.7587,
+      "step": 4318
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.168041870444388e-05,
+      "loss": 0.7803,
+      "step": 4319
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.167605490204588e-05,
+      "loss": 22.9907,
+      "step": 4320
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.1671690059404e-05,
+      "loss": 0.7249,
+      "step": 4321
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.166732417662723e-05,
+      "loss": 0.7844,
+      "step": 4322
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.166295725382454e-05,
+      "loss": 21.5827,
+      "step": 4323
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.165858929110491e-05,
+      "loss": 22.487,
+      "step": 4324
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.165422028857739e-05,
+      "loss": 0.7435,
+      "step": 4325
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.164985024635103e-05,
+      "loss": 0.7861,
+      "step": 4326
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.16454791645349e-05,
+      "loss": 25.2022,
+      "step": 4327
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.16411070432381e-05,
+      "loss": 0.7193,
+      "step": 4328
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.163673388256973e-05,
+      "loss": 0.8035,
+      "step": 4329
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.163235968263901e-05,
+      "loss": 0.7589,
+      "step": 4330
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.162798444355507e-05,
+      "loss": 0.7948,
+      "step": 4331
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.162360816542714e-05,
+      "loss": 20.4228,
+      "step": 4332
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.161923084836444e-05,
+      "loss": 0.7776,
+      "step": 4333
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.161485249247624e-05,
+      "loss": 0.7696,
+      "step": 4334
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.161047309787182e-05,
+      "loss": 21.4384,
+      "step": 4335
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.160609266466048e-05,
+      "loss": 0.7853,
+      "step": 4336
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.160171119295158e-05,
+      "loss": 21.8279,
+      "step": 4337
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.159732868285446e-05,
+      "loss": 24.2821,
+      "step": 4338
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.159294513447853e-05,
+      "loss": 19.8141,
+      "step": 4339
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.158856054793319e-05,
+      "loss": 22.2027,
+      "step": 4340
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.158417492332787e-05,
+      "loss": 0.7913,
+      "step": 4341
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.157978826077207e-05,
+      "loss": 0.7162,
+      "step": 4342
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.157540056037524e-05,
+      "loss": 20.9179,
+      "step": 4343
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.157101182224694e-05,
+      "loss": 0.7114,
+      "step": 4344
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.156662204649667e-05,
+      "loss": 26.1807,
+      "step": 4345
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.156223123323405e-05,
+      "loss": 21.8348,
+      "step": 4346
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.155783938256862e-05,
+      "loss": 21.7467,
+      "step": 4347
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.155344649461004e-05,
+      "loss": 0.8054,
+      "step": 4348
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.154905256946796e-05,
+      "loss": 21.9621,
+      "step": 4349
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.154465760725202e-05,
+      "loss": 0.7778,
+      "step": 4350
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.154026160807196e-05,
+      "loss": 0.7781,
+      "step": 4351
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.153586457203747e-05,
+      "loss": 21.0675,
+      "step": 4352
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.153146649925832e-05,
+      "loss": 0.8093,
+      "step": 4353
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.15270673898443e-05,
+      "loss": 23.4166,
+      "step": 4354
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.152266724390518e-05,
+      "loss": 0.7441,
+      "step": 4355
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.15182660615508e-05,
+      "loss": 24.6428,
+      "step": 4356
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.151386384289103e-05,
+      "loss": 0.7656,
+      "step": 4357
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.150946058803574e-05,
+      "loss": 0.7643,
+      "step": 4358
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.150505629709483e-05,
+      "loss": 20.6916,
+      "step": 4359
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.150065097017824e-05,
+      "loss": 22.7555,
+      "step": 4360
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.149624460739594e-05,
+      "loss": 0.7117,
+      "step": 4361
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.149183720885787e-05,
+      "loss": 22.1088,
+      "step": 4362
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.148742877467409e-05,
+      "loss": 21.4232,
+      "step": 4363
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.148301930495461e-05,
+      "loss": 0.8135,
+      "step": 4364
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.14786087998095e-05,
+      "loss": 0.7245,
+      "step": 4365
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.147419725934886e-05,
+      "loss": 0.7098,
+      "step": 4366
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.146978468368275e-05,
+      "loss": 19.7336,
+      "step": 4367
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.146537107292137e-05,
+      "loss": 22.56,
+      "step": 4368
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.146095642717485e-05,
+      "loss": 22.9945,
+      "step": 4369
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.145654074655338e-05,
+      "loss": 0.7964,
+      "step": 4370
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.14521240311672e-05,
+      "loss": 0.7987,
+      "step": 4371
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.144770628112655e-05,
+      "loss": 23.2434,
+      "step": 4372
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.144328749654167e-05,
+      "loss": 22.2855,
+      "step": 4373
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.143886767752289e-05,
+      "loss": 0.8238,
+      "step": 4374
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.14344468241805e-05,
+      "loss": 0.7309,
+      "step": 4375
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.143002493662485e-05,
+      "loss": 0.7856,
+      "step": 4376
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.142560201496633e-05,
+      "loss": 21.8835,
+      "step": 4377
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.142117805931532e-05,
+      "loss": 0.7596,
+      "step": 4378
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.141675306978224e-05,
+      "loss": 0.8127,
+      "step": 4379
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.141232704647756e-05,
+      "loss": 21.2065,
+      "step": 4380
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.140789998951174e-05,
+      "loss": 0.8235,
+      "step": 4381
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.140347189899527e-05,
+      "loss": 0.7803,
+      "step": 4382
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.13990427750387e-05,
+      "loss": 0.7615,
+      "step": 4383
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.139461261775256e-05,
+      "loss": 21.1705,
+      "step": 4384
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.139018142724744e-05,
+      "loss": 24.0864,
+      "step": 4385
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.138574920363394e-05,
+      "loss": 24.8074,
+      "step": 4386
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.13813159470227e-05,
+      "loss": 22.0158,
+      "step": 4387
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.137688165752436e-05,
+      "loss": 22.551,
+      "step": 4388
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.137244633524961e-05,
+      "loss": 24.1307,
+      "step": 4389
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.136800998030914e-05,
+      "loss": 21.7563,
+      "step": 4390
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.136357259281371e-05,
+      "loss": 22.6953,
+      "step": 4391
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.135913417287405e-05,
+      "loss": 20.9422,
+      "step": 4392
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.135469472060096e-05,
+      "loss": 0.7675,
+      "step": 4393
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.135025423610526e-05,
+      "loss": 24.6384,
+      "step": 4394
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.134581271949775e-05,
+      "loss": 0.79,
+      "step": 4395
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.134137017088935e-05,
+      "loss": 23.5862,
+      "step": 4396
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.133692659039088e-05,
+      "loss": 0.759,
+      "step": 4397
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.13324819781133e-05,
+      "loss": 21.072,
+      "step": 4398
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.132803633416752e-05,
+      "loss": 0.7401,
+      "step": 4399
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.132358965866453e-05,
+      "loss": 0.7863,
+      "step": 4400
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.131914195171529e-05,
+      "loss": 20.8119,
+      "step": 4401
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.131469321343084e-05,
+      "loss": 19.6962,
+      "step": 4402
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.131024344392223e-05,
+      "loss": 0.8297,
+      "step": 4403
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.130579264330048e-05,
+      "loss": 0.8025,
+      "step": 4404
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.130134081167674e-05,
+      "loss": 0.7678,
+      "step": 4405
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.129688794916209e-05,
+      "loss": 19.6048,
+      "step": 4406
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.12924340558677e-05,
+      "loss": 0.7623,
+      "step": 4407
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.128797913190473e-05,
+      "loss": 0.8044,
+      "step": 4408
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.128352317738435e-05,
+      "loss": 19.5587,
+      "step": 4409
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.127906619241782e-05,
+      "loss": 0.787,
+      "step": 4410
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.127460817711637e-05,
+      "loss": 0.7281,
+      "step": 4411
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.127014913159127e-05,
+      "loss": 22.2427,
+      "step": 4412
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.126568905595383e-05,
+      "loss": 23.0438,
+      "step": 4413
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.126122795031538e-05,
+      "loss": 20.8679,
+      "step": 4414
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.125676581478724e-05,
+      "loss": 0.7651,
+      "step": 4415
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.125230264948081e-05,
+      "loss": 0.7696,
+      "step": 4416
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.12478384545075e-05,
+      "loss": 23.0402,
+      "step": 4417
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.124337322997871e-05,
+      "loss": 0.8062,
+      "step": 4418
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.123890697600592e-05,
+      "loss": 21.8359,
+      "step": 4419
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.123443969270058e-05,
+      "loss": 22.0593,
+      "step": 4420
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.122997138017423e-05,
+      "loss": 21.2739,
+      "step": 4421
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.122550203853837e-05,
+      "loss": 23.0416,
+      "step": 4422
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.122103166790457e-05,
+      "loss": 0.8211,
+      "step": 4423
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.121656026838438e-05,
+      "loss": 0.7766,
+      "step": 4424
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.121208784008947e-05,
+      "loss": 19.429,
+      "step": 4425
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.120761438313143e-05,
+      "loss": 20.8906,
+      "step": 4426
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.120313989762192e-05,
+      "loss": 0.7852,
+      "step": 4427
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.119866438367263e-05,
+      "loss": 0.7766,
+      "step": 4428
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.119418784139526e-05,
+      "loss": 21.3243,
+      "step": 4429
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.118971027090158e-05,
+      "loss": 0.8025,
+      "step": 4430
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.11852316723033e-05,
+      "loss": 21.3092,
+      "step": 4431
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.118075204571226e-05,
+      "loss": 0.6994,
+      "step": 4432
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.117627139124023e-05,
+      "loss": 20.5228,
+      "step": 4433
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.117178970899907e-05,
+      "loss": 0.7924,
+      "step": 4434
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.116730699910064e-05,
+      "loss": 20.8245,
+      "step": 4435
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.116282326165681e-05,
+      "loss": 22.0047,
+      "step": 4436
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.115833849677953e-05,
+      "loss": 0.7533,
+      "step": 4437
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.115385270458072e-05,
+      "loss": 0.7039,
+      "step": 4438
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.114936588517236e-05,
+      "loss": 0.8397,
+      "step": 4439
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.11448780386664e-05,
+      "loss": 20.4651,
+      "step": 4440
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.114038916517492e-05,
+      "loss": 0.7647,
+      "step": 4441
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.113589926480992e-05,
+      "loss": 20.549,
+      "step": 4442
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.11314083376835e-05,
+      "loss": 23.3499,
+      "step": 4443
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.112691638390772e-05,
+      "loss": 23.2389,
+      "step": 4444
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.112242340359471e-05,
+      "loss": 19.2431,
+      "step": 4445
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.111792939685664e-05,
+      "loss": 22.0246,
+      "step": 4446
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.111343436380565e-05,
+      "loss": 22.351,
+      "step": 4447
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.110893830455396e-05,
+      "loss": 0.8321,
+      "step": 4448
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.110444121921378e-05,
+      "loss": 0.7832,
+      "step": 4449
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.109994310789735e-05,
+      "loss": 0.7306,
+      "step": 4450
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.109544397071697e-05,
+      "loss": 0.7675,
+      "step": 4451
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.109094380778492e-05,
+      "loss": 0.797,
+      "step": 4452
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.108644261921353e-05,
+      "loss": 24.6121,
+      "step": 4453
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.108194040511515e-05,
+      "loss": 24.4983,
+      "step": 4454
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.107743716560216e-05,
+      "loss": 0.8047,
+      "step": 4455
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.107293290078697e-05,
+      "loss": 22.2692,
+      "step": 4456
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.106842761078198e-05,
+      "loss": 22.0587,
+      "step": 4457
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.106392129569967e-05,
+      "loss": 0.7455,
+      "step": 4458
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.10594139556525e-05,
+      "loss": 23.116,
+      "step": 4459
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.105490559075298e-05,
+      "loss": 22.1278,
+      "step": 4460
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.105039620111367e-05,
+      "loss": 21.2493,
+      "step": 4461
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.104588578684706e-05,
+      "loss": 0.7553,
+      "step": 4462
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.10413743480658e-05,
+      "loss": 0.7714,
+      "step": 4463
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.103686188488243e-05,
+      "loss": 20.3449,
+      "step": 4464
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.103234839740965e-05,
+      "loss": 0.7627,
+      "step": 4465
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.102783388576007e-05,
+      "loss": 22.5152,
+      "step": 4466
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.10233183500464e-05,
+      "loss": 22.9424,
+      "step": 4467
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.101880179038132e-05,
+      "loss": 0.8242,
+      "step": 4468
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.101428420687759e-05,
+      "loss": 0.7387,
+      "step": 4469
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.100976559964796e-05,
+      "loss": 0.7652,
+      "step": 4470
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.100524596880522e-05,
+      "loss": 21.933,
+      "step": 4471
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.100072531446218e-05,
+      "loss": 0.7664,
+      "step": 4472
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.099620363673167e-05,
+      "loss": 0.8108,
+      "step": 4473
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.099168093572657e-05,
+      "loss": 0.7696,
+      "step": 4474
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.098715721155973e-05,
+      "loss": 19.4772,
+      "step": 4475
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.098263246434411e-05,
+      "loss": 21.8658,
+      "step": 4476
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.09781066941926e-05,
+      "loss": 0.7846,
+      "step": 4477
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.097357990121822e-05,
+      "loss": 0.8074,
+      "step": 4478
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.09690520855339e-05,
+      "loss": 23.7039,
+      "step": 4479
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.096452324725271e-05,
+      "loss": 0.7864,
+      "step": 4480
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.095999338648766e-05,
+      "loss": 0.7567,
+      "step": 4481
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.095546250335183e-05,
+      "loss": 23.6049,
+      "step": 4482
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.095093059795829e-05,
+      "loss": 24.3686,
+      "step": 4483
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.094639767042017e-05,
+      "loss": 23.8005,
+      "step": 4484
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.094186372085063e-05,
+      "loss": 0.7753,
+      "step": 4485
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.09373287493628e-05,
+      "loss": 24.6134,
+      "step": 4486
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.093279275606991e-05,
+      "loss": 0.7268,
+      "step": 4487
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.092825574108518e-05,
+      "loss": 0.7563,
+      "step": 4488
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.092371770452184e-05,
+      "loss": 0.7699,
+      "step": 4489
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.091917864649313e-05,
+      "loss": 23.7933,
+      "step": 4490
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.091463856711243e-05,
+      "loss": 22.0147,
+      "step": 4491
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.091009746649297e-05,
+      "loss": 0.8298,
+      "step": 4492
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.090555534474815e-05,
+      "loss": 0.7502,
+      "step": 4493
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.090101220199132e-05,
+      "loss": 23.5202,
+      "step": 4494
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.089646803833589e-05,
+      "loss": 0.7388,
+      "step": 4495
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.089192285389528e-05,
+      "loss": 0.7927,
+      "step": 4496
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.088737664878294e-05,
+      "loss": 18.1974,
+      "step": 4497
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.088282942311235e-05,
+      "loss": 0.8154,
+      "step": 4498
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.087828117699699e-05,
+      "loss": 0.7466,
+      "step": 4499
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.087373191055039e-05,
+      "loss": 0.7402,
+      "step": 4500
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.086918162388614e-05,
+      "loss": 21.4714,
+      "step": 4501
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.086463031711776e-05,
+      "loss": 21.8888,
+      "step": 4502
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.086007799035889e-05,
+      "loss": 20.424,
+      "step": 4503
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.085552464372313e-05,
+      "loss": 0.8082,
+      "step": 4504
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.085097027732417e-05,
+      "loss": 24.9436,
+      "step": 4505
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.084641489127565e-05,
+      "loss": 22.974,
+      "step": 4506
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.084185848569126e-05,
+      "loss": 0.7571,
+      "step": 4507
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.08373010606848e-05,
+      "loss": 21.4839,
+      "step": 4508
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.083274261636996e-05,
+      "loss": 22.2413,
+      "step": 4509
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.082818315286055e-05,
+      "loss": 21.5275,
+      "step": 4510
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.082362267027036e-05,
+      "loss": 0.7499,
+      "step": 4511
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.081906116871324e-05,
+      "loss": 0.8284,
+      "step": 4512
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.081449864830302e-05,
+      "loss": 19.2867,
+      "step": 4513
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.08099351091536e-05,
+      "loss": 0.7722,
+      "step": 4514
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.080537055137888e-05,
+      "loss": 0.8036,
+      "step": 4515
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.080080497509278e-05,
+      "loss": 0.7361,
+      "step": 4516
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.079623838040929e-05,
+      "loss": 0.7507,
+      "step": 4517
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.079167076744237e-05,
+      "loss": 21.5841,
+      "step": 4518
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.078710213630603e-05,
+      "loss": 22.9687,
+      "step": 4519
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.07825324871143e-05,
+      "loss": 21.6229,
+      "step": 4520
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.077796181998126e-05,
+      "loss": 0.7769,
+      "step": 4521
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.077339013502099e-05,
+      "loss": 25.5959,
+      "step": 4522
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.076881743234757e-05,
+      "loss": 20.9566,
+      "step": 4523
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.076424371207518e-05,
+      "loss": 21.5318,
+      "step": 4524
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.075966897431795e-05,
+      "loss": 19.6459,
+      "step": 4525
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.075509321919007e-05,
+      "loss": 0.7892,
+      "step": 4526
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.075051644680577e-05,
+      "loss": 0.7711,
+      "step": 4527
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.074593865727926e-05,
+      "loss": 0.7346,
+      "step": 4528
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.074135985072482e-05,
+      "loss": 20.3591,
+      "step": 4529
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.073678002725675e-05,
+      "loss": 23.985,
+      "step": 4530
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.073219918698934e-05,
+      "loss": 0.7618,
+      "step": 4531
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.072761733003694e-05,
+      "loss": 21.1594,
+      "step": 4532
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.072303445651391e-05,
+      "loss": 0.7826,
+      "step": 4533
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.071845056653464e-05,
+      "loss": 20.9857,
+      "step": 4534
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.071386566021355e-05,
+      "loss": 20.8175,
+      "step": 4535
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.070927973766508e-05,
+      "loss": 0.8125,
+      "step": 4536
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.070469279900369e-05,
+      "loss": 22.2891,
+      "step": 4537
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.070010484434387e-05,
+      "loss": 19.978,
+      "step": 4538
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.069551587380014e-05,
+      "loss": 0.7561,
+      "step": 4539
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.069092588748704e-05,
+      "loss": 24.2381,
+      "step": 4540
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.068633488551914e-05,
+      "loss": 20.6805,
+      "step": 4541
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.068174286801104e-05,
+      "loss": 0.8323,
+      "step": 4542
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.067714983507734e-05,
+      "loss": 22.3137,
+      "step": 4543
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.06725557868327e-05,
+      "loss": 0.7304,
+      "step": 4544
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.066796072339178e-05,
+      "loss": 0.719,
+      "step": 4545
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.066336464486927e-05,
+      "loss": 19.8958,
+      "step": 4546
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.065876755137989e-05,
+      "loss": 0.7614,
+      "step": 4547
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.065416944303839e-05,
+      "loss": 0.7732,
+      "step": 4548
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.064957031995956e-05,
+      "loss": 0.8573,
+      "step": 4549
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.064497018225815e-05,
+      "loss": 21.9444,
+      "step": 4550
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.0640369030049e-05,
+      "loss": 21.1525,
+      "step": 4551
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.063576686344697e-05,
+      "loss": 0.7922,
+      "step": 4552
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.063116368256692e-05,
+      "loss": 0.7608,
+      "step": 4553
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.062655948752374e-05,
+      "loss": 21.9653,
+      "step": 4554
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.062195427843237e-05,
+      "loss": 0.8209,
+      "step": 4555
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.061734805540774e-05,
+      "loss": 0.7642,
+      "step": 4556
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.061274081856484e-05,
+      "loss": 20.0831,
+      "step": 4557
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.060813256801864e-05,
+      "loss": 20.5089,
+      "step": 4558
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.060352330388418e-05,
+      "loss": 20.3907,
+      "step": 4559
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.059891302627651e-05,
+      "loss": 22.9318,
+      "step": 4560
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.05943017353107e-05,
+      "loss": 21.3587,
+      "step": 4561
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.058968943110186e-05,
+      "loss": 21.7269,
+      "step": 4562
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.058507611376509e-05,
+      "loss": 22.931,
+      "step": 4563
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.058046178341556e-05,
+      "loss": 23.7195,
+      "step": 4564
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.057584644016844e-05,
+      "loss": 0.7567,
+      "step": 4565
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.057123008413893e-05,
+      "loss": 0.7694,
+      "step": 4566
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.056661271544224e-05,
+      "loss": 23.0457,
+      "step": 4567
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.056199433419365e-05,
+      "loss": 20.0907,
+      "step": 4568
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.055737494050842e-05,
+      "loss": 0.7542,
+      "step": 4569
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.055275453450184e-05,
+      "loss": 0.7201,
+      "step": 4570
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.054813311628925e-05,
+      "loss": 20.0777,
+      "step": 4571
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.0543510685986e-05,
+      "loss": 0.7732,
+      "step": 4572
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.053888724370746e-05,
+      "loss": 0.7885,
+      "step": 4573
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.053426278956905e-05,
+      "loss": 19.6245,
+      "step": 4574
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.052963732368618e-05,
+      "loss": 22.3513,
+      "step": 4575
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.05250108461743e-05,
+      "loss": 0.8005,
+      "step": 4576
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.052038335714892e-05,
+      "loss": 22.1921,
+      "step": 4577
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.05157548567255e-05,
+      "loss": 22.0332,
+      "step": 4578
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.05111253450196e-05,
+      "loss": 21.7462,
+      "step": 4579
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.050649482214675e-05,
+      "loss": 19.1604,
+      "step": 4580
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.050186328822254e-05,
+      "loss": 0.7935,
+      "step": 4581
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.049723074336258e-05,
+      "loss": 0.7342,
+      "step": 4582
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.04925971876825e-05,
+      "loss": 0.7781,
+      "step": 4583
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.048796262129794e-05,
+      "loss": 20.4409,
+      "step": 4584
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.048332704432459e-05,
+      "loss": 20.2758,
+      "step": 4585
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.047869045687815e-05,
+      "loss": 0.758,
+      "step": 4586
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.047405285907435e-05,
+      "loss": 0.7223,
+      "step": 4587
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.046941425102898e-05,
+      "loss": 0.8023,
+      "step": 4588
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.046477463285776e-05,
+      "loss": 20.8249,
+      "step": 4589
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.046013400467652e-05,
+      "loss": 0.7851,
+      "step": 4590
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.045549236660112e-05,
+      "loss": 0.7676,
+      "step": 4591
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.045084971874738e-05,
+      "loss": 21.1998,
+      "step": 4592
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.04462060612312e-05,
+      "loss": 0.7136,
+      "step": 4593
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.044156139416847e-05,
+      "loss": 19.6936,
+      "step": 4594
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.043691571767514e-05,
+      "loss": 0.7584,
+      "step": 4595
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.043226903186717e-05,
+      "loss": 0.7484,
+      "step": 4596
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.042762133686052e-05,
+      "loss": 22.9612,
+      "step": 4597
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.042297263277122e-05,
+      "loss": 21.7302,
+      "step": 4598
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.041832291971529e-05,
+      "loss": 25.3494,
+      "step": 4599
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.04136721978088e-05,
+      "loss": 22.5535,
+      "step": 4600
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.040902046716782e-05,
+      "loss": 0.7684,
+      "step": 4601
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.040436772790845e-05,
+      "loss": 23.1696,
+      "step": 4602
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.039971398014684e-05,
+      "loss": 0.8343,
+      "step": 4603
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.039505922399916e-05,
+      "loss": 0.7816,
+      "step": 4604
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.039040345958156e-05,
+      "loss": 0.7638,
+      "step": 4605
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.038574668701027e-05,
+      "loss": 20.8308,
+      "step": 4606
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.038108890640153e-05,
+      "loss": 21.579,
+      "step": 4607
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.037643011787158e-05,
+      "loss": 20.6066,
+      "step": 4608
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.037177032153673e-05,
+      "loss": 0.7173,
+      "step": 4609
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.036710951751326e-05,
+      "loss": 0.752,
+      "step": 4610
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.036244770591753e-05,
+      "loss": 18.9663,
+      "step": 4611
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.035778488686588e-05,
+      "loss": 0.7794,
+      "step": 4612
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.035312106047471e-05,
+      "loss": 23.9866,
+      "step": 4613
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.034845622686043e-05,
+      "loss": 21.8075,
+      "step": 4614
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.034379038613947e-05,
+      "loss": 0.7285,
+      "step": 4615
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.033912353842828e-05,
+      "loss": 0.7543,
+      "step": 4616
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.033445568384335e-05,
+      "loss": 24.6365,
+      "step": 4617
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.032978682250122e-05,
+      "loss": 0.7272,
+      "step": 4618
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.032511695451839e-05,
+      "loss": 23.2813,
+      "step": 4619
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.032044608001142e-05,
+      "loss": 23.1124,
+      "step": 4620
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.031577419909694e-05,
+      "loss": 22.2526,
+      "step": 4621
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.031110131189152e-05,
+      "loss": 0.7526,
+      "step": 4622
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.030642741851178e-05,
+      "loss": 0.7274,
+      "step": 4623
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.030175251907444e-05,
+      "loss": 0.7744,
+      "step": 4624
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.029707661369615e-05,
+      "loss": 22.4316,
+      "step": 4625
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.029239970249362e-05,
+      "loss": 21.918,
+      "step": 4626
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.028772178558358e-05,
+      "loss": 0.8071,
+      "step": 4627
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.028304286308282e-05,
+      "loss": 0.7799,
+      "step": 4628
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.027836293510812e-05,
+      "loss": 0.7554,
+      "step": 4629
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.027368200177626e-05,
+      "loss": 0.8116,
+      "step": 4630
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.026900006320412e-05,
+      "loss": 0.694,
+      "step": 4631
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.026431711950855e-05,
+      "loss": 26.5413,
+      "step": 4632
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.025963317080641e-05,
+      "loss": 0.7556,
+      "step": 4633
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.025494821721465e-05,
+      "loss": 0.7713,
+      "step": 4634
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.025026225885019e-05,
+      "loss": 0.7811,
+      "step": 4635
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.024557529583e-05,
+      "loss": 21.3002,
+      "step": 4636
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.024088732827104e-05,
+      "loss": 20.1094,
+      "step": 4637
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.023619835629036e-05,
+      "loss": 21.5152,
+      "step": 4638
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.023150838000499e-05,
+      "loss": 22.1518,
+      "step": 4639
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.022681739953196e-05,
+      "loss": 20.2095,
+      "step": 4640
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.02221254149884e-05,
+      "loss": 21.716,
+      "step": 4641
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.021743242649139e-05,
+      "loss": 22.996,
+      "step": 4642
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.02127384341581e-05,
+      "loss": 0.7983,
+      "step": 4643
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.020804343810567e-05,
+      "loss": 21.3963,
+      "step": 4644
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.02033474384513e-05,
+      "loss": 0.7755,
+      "step": 4645
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.01986504353122e-05,
+      "loss": 18.4819,
+      "step": 4646
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.019395242880559e-05,
+      "loss": 20.1537,
+      "step": 4647
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.018925341904877e-05,
+      "loss": 22.5154,
+      "step": 4648
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.0184553406159e-05,
+      "loss": 0.7675,
+      "step": 4649
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.01798523902536e-05,
+      "loss": 0.6995,
+      "step": 4650
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.017515037144992e-05,
+      "loss": 0.7869,
+      "step": 4651
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.017044734986531e-05,
+      "loss": 22.9313,
+      "step": 4652
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.016574332561715e-05,
+      "loss": 23.6013,
+      "step": 4653
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.016103829882287e-05,
+      "loss": 0.751,
+      "step": 4654
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.01563322695999e-05,
+      "loss": 0.7397,
+      "step": 4655
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.015162523806573e-05,
+      "loss": 20.6912,
+      "step": 4656
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.014691720433782e-05,
+      "loss": 22.2108,
+      "step": 4657
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.014220816853367e-05,
+      "loss": 0.7802,
+      "step": 4658
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.013749813077084e-05,
+      "loss": 20.4174,
+      "step": 4659
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.01327870911669e-05,
+      "loss": 0.7296,
+      "step": 4660
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.012807504983945e-05,
+      "loss": 22.5085,
+      "step": 4661
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.012336200690605e-05,
+      "loss": 21.9078,
+      "step": 4662
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.011864796248439e-05,
+      "loss": 0.7871,
+      "step": 4663
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.011393291669212e-05,
+      "loss": 0.7705,
+      "step": 4664
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.010921686964693e-05,
+      "loss": 22.6313,
+      "step": 4665
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.010449982146651e-05,
+      "loss": 0.7742,
+      "step": 4666
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.009978177226863e-05,
+      "loss": 20.3072,
+      "step": 4667
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.009506272217103e-05,
+      "loss": 22.266,
+      "step": 4668
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.009034267129153e-05,
+      "loss": 0.7637,
+      "step": 4669
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.00856216197479e-05,
+      "loss": 23.3846,
+      "step": 4670
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.008089956765801e-05,
+      "loss": 20.5177,
+      "step": 4671
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.007617651513973e-05,
+      "loss": 0.7452,
+      "step": 4672
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.007145246231093e-05,
+      "loss": 0.7697,
+      "step": 4673
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.006672740928952e-05,
+      "loss": 20.4165,
+      "step": 4674
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.006200135619343e-05,
+      "loss": 21.829,
+      "step": 4675
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.005727430314066e-05,
+      "loss": 21.5938,
+      "step": 4676
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.005254625024917e-05,
+      "loss": 21.8804,
+      "step": 4677
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.004781719763698e-05,
+      "loss": 24.1719,
+      "step": 4678
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.004308714542212e-05,
+      "loss": 22.2554,
+      "step": 4679
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.003835609372266e-05,
+      "loss": 0.7639,
+      "step": 4680
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.003362404265671e-05,
+      "loss": 19.4886,
+      "step": 4681
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.002889099234235e-05,
+      "loss": 22.0879,
+      "step": 4682
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.002415694289774e-05,
+      "loss": 20.4277,
+      "step": 4683
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.001942189444103e-05,
+      "loss": 0.7447,
+      "step": 4684
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.00146858470904e-05,
+      "loss": 0.756,
+      "step": 4685
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.000994880096409e-05,
+      "loss": 21.1576,
+      "step": 4686
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.000521075618032e-05,
+      "loss": 0.76,
+      "step": 4687
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.000047171285737e-05,
+      "loss": 21.5496,
+      "step": 4688
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.999573167111348e-05,
+      "loss": 20.742,
+      "step": 4689
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.999099063106702e-05,
+      "loss": 0.7594,
+      "step": 4690
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.998624859283629e-05,
+      "loss": 20.876,
+      "step": 4691
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.998150555653967e-05,
+      "loss": 23.6851,
+      "step": 4692
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.997676152229555e-05,
+      "loss": 21.0011,
+      "step": 4693
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.997201649022231e-05,
+      "loss": 0.7701,
+      "step": 4694
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.996727046043843e-05,
+      "loss": 0.8071,
+      "step": 4695
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.996252343306235e-05,
+      "loss": 23.5058,
+      "step": 4696
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.995777540821257e-05,
+      "loss": 21.5185,
+      "step": 4697
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.995302638600758e-05,
+      "loss": 0.8143,
+      "step": 4698
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.994827636656593e-05,
+      "loss": 24.8334,
+      "step": 4699
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.994352535000619e-05,
+      "loss": 19.575,
+      "step": 4700
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.993877333644693e-05,
+      "loss": 20.1142,
+      "step": 4701
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.993402032600676e-05,
+      "loss": 0.7529,
+      "step": 4702
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.992926631880434e-05,
+      "loss": 22.4321,
+      "step": 4703
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.99245113149583e-05,
+      "loss": 21.6447,
+      "step": 4704
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.991975531458736e-05,
+      "loss": 0.7352,
+      "step": 4705
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.99149983178102e-05,
+      "loss": 0.7296,
+      "step": 4706
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.991024032474558e-05,
+      "loss": 23.4203,
+      "step": 4707
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.990548133551223e-05,
+      "loss": 20.8726,
+      "step": 4708
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.990072135022897e-05,
+      "loss": 20.6472,
+      "step": 4709
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.989596036901457e-05,
+      "loss": 21.4844,
+      "step": 4710
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.989119839198792e-05,
+      "loss": 23.28,
+      "step": 4711
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.988643541926785e-05,
+      "loss": 19.9796,
+      "step": 4712
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.988167145097322e-05,
+      "loss": 20.9981,
+      "step": 4713
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.987690648722298e-05,
+      "loss": 18.4563,
+      "step": 4714
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.987214052813604e-05,
+      "loss": 19.5256,
+      "step": 4715
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.986737357383139e-05,
+      "loss": 20.3393,
+      "step": 4716
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.986260562442797e-05,
+      "loss": 0.7516,
+      "step": 4717
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.985783668004482e-05,
+      "loss": 21.0796,
+      "step": 4718
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.985306674080096e-05,
+      "loss": 21.7058,
+      "step": 4719
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.984829580681547e-05,
+      "loss": 21.2953,
+      "step": 4720
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.98435238782074e-05,
+      "loss": 0.7525,
+      "step": 4721
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.983875095509588e-05,
+      "loss": 21.8065,
+      "step": 4722
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.983397703760004e-05,
+      "loss": 0.8108,
+      "step": 4723
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.982920212583905e-05,
+      "loss": 0.7659,
+      "step": 4724
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.982442621993207e-05,
+      "loss": 0.811,
+      "step": 4725
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.981964931999832e-05,
+      "loss": 21.5955,
+      "step": 4726
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.981487142615703e-05,
+      "loss": 0.8014,
+      "step": 4727
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.981009253852745e-05,
+      "loss": 22.0302,
+      "step": 4728
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.98053126572289e-05,
+      "loss": 0.7407,
+      "step": 4729
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.980053178238064e-05,
+      "loss": 0.7933,
+      "step": 4730
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.979574991410201e-05,
+      "loss": 0.7701,
+      "step": 4731
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.979096705251237e-05,
+      "loss": 22.8865,
+      "step": 4732
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.978618319773114e-05,
+      "loss": 20.1712,
+      "step": 4733
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.978139834987766e-05,
+      "loss": 22.4862,
+      "step": 4734
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.97766125090714e-05,
+      "loss": 0.7558,
+      "step": 4735
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.977182567543183e-05,
+      "loss": 0.7759,
+      "step": 4736
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.976703784907839e-05,
+      "loss": 22.0891,
+      "step": 4737
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.976224903013062e-05,
+      "loss": 0.7331,
+      "step": 4738
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.975745921870802e-05,
+      "loss": 20.4026,
+      "step": 4739
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.975266841493018e-05,
+      "loss": 22.518,
+      "step": 4740
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.974787661891664e-05,
+      "loss": 0.7056,
+      "step": 4741
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.974308383078702e-05,
+      "loss": 21.916,
+      "step": 4742
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.973829005066097e-05,
+      "loss": 19.7084,
+      "step": 4743
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.973349527865811e-05,
+      "loss": 0.7603,
+      "step": 4744
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.972869951489815e-05,
+      "loss": 0.7579,
+      "step": 4745
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.972390275950077e-05,
+      "loss": 0.7062,
+      "step": 4746
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.971910501258573e-05,
+      "loss": 0.7106,
+      "step": 4747
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.971430627427274e-05,
+      "loss": 0.708,
+      "step": 4748
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.97095065446816e-05,
+      "loss": 19.3882,
+      "step": 4749
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.97047058239321e-05,
+      "loss": 0.7201,
+      "step": 4750
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.969990411214409e-05,
+      "loss": 0.7757,
+      "step": 4751
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.96951014094374e-05,
+      "loss": 0.7325,
+      "step": 4752
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.969029771593192e-05,
+      "loss": 21.8894,
+      "step": 4753
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.968549303174754e-05,
+      "loss": 0.7537,
+      "step": 4754
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.968068735700421e-05,
+      "loss": 20.5077,
+      "step": 4755
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.967588069182185e-05,
+      "loss": 0.718,
+      "step": 4756
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.967107303632046e-05,
+      "loss": 0.7649,
+      "step": 4757
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.966626439062e-05,
+      "loss": 0.7349,
+      "step": 4758
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.966145475484056e-05,
+      "loss": 0.7705,
+      "step": 4759
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.965664412910213e-05,
+      "loss": 23.0844,
+      "step": 4760
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.965183251352482e-05,
+      "loss": 22.7358,
+      "step": 4761
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.96470199082287e-05,
+      "loss": 22.602,
+      "step": 4762
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.964220631333394e-05,
+      "loss": 0.7318,
+      "step": 4763
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.963739172896063e-05,
+      "loss": 0.8191,
+      "step": 4764
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.963257615522898e-05,
+      "loss": 25.6077,
+      "step": 4765
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.962775959225918e-05,
+      "loss": 0.8117,
+      "step": 4766
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.962294204017146e-05,
+      "loss": 0.7632,
+      "step": 4767
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.961812349908605e-05,
+      "loss": 23.1002,
+      "step": 4768
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.961330396912324e-05,
+      "loss": 0.7565,
+      "step": 4769
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.960848345040331e-05,
+      "loss": 0.6867,
+      "step": 4770
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.96036619430466e-05,
+      "loss": 0.7524,
+      "step": 4771
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.959883944717344e-05,
+      "loss": 21.4733,
+      "step": 4772
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.95940159629042e-05,
+      "loss": 22.5601,
+      "step": 4773
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.95891914903593e-05,
+      "loss": 0.7616,
+      "step": 4774
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.958436602965911e-05,
+      "loss": 0.8139,
+      "step": 4775
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.957953958092414e-05,
+      "loss": 21.7099,
+      "step": 4776
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.957471214427479e-05,
+      "loss": 23.428,
+      "step": 4777
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.956988371983162e-05,
+      "loss": 22.3828,
+      "step": 4778
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.956505430771509e-05,
+      "loss": 0.7505,
+      "step": 4779
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.95602239080458e-05,
+      "loss": 18.5071,
+      "step": 4780
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.955539252094426e-05,
+      "loss": 0.7081,
+      "step": 4781
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.955056014653111e-05,
+      "loss": 0.8057,
+      "step": 4782
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.954572678492694e-05,
+      "loss": 23.1884,
+      "step": 4783
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.95408924362524e-05,
+      "loss": 0.7196,
+      "step": 4784
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.953605710062816e-05,
+      "loss": 0.786,
+      "step": 4785
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.953122077817491e-05,
+      "loss": 0.7416,
+      "step": 4786
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.952638346901335e-05,
+      "loss": 19.814,
+      "step": 4787
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.952154517326424e-05,
+      "loss": 21.5785,
+      "step": 4788
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.951670589104835e-05,
+      "loss": 21.5266,
+      "step": 4789
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.951186562248644e-05,
+      "loss": 0.7568,
+      "step": 4790
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.950702436769935e-05,
+      "loss": 21.701,
+      "step": 4791
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.950218212680792e-05,
+      "loss": 0.7654,
+      "step": 4792
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.949733889993299e-05,
+      "loss": 0.7765,
+      "step": 4793
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.949249468719547e-05,
+      "loss": 0.7687,
+      "step": 4794
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.948764948871626e-05,
+      "loss": 0.8455,
+      "step": 4795
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.948280330461632e-05,
+      "loss": 22.4191,
+      "step": 4796
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.947795613501658e-05,
+      "loss": 22.968,
+      "step": 4797
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.947310798003804e-05,
+      "loss": 20.9899,
+      "step": 4798
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.946825883980172e-05,
+      "loss": 0.7778,
+      "step": 4799
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.946340871442864e-05,
+      "loss": 22.2009,
+      "step": 4800
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.945855760403987e-05,
+      "loss": 22.8431,
+      "step": 4801
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.945370550875649e-05,
+      "loss": 0.7237,
+      "step": 4802
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.944885242869962e-05,
+      "loss": 22.3769,
+      "step": 4803
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.944399836399037e-05,
+      "loss": 0.776,
+      "step": 4804
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.943914331474994e-05,
+      "loss": 0.7942,
+      "step": 4805
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.943428728109946e-05,
+      "loss": 0.7807,
+      "step": 4806
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.942943026316018e-05,
+      "loss": 0.765,
+      "step": 4807
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.942457226105331e-05,
+      "loss": 23.1143,
+      "step": 4808
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.941971327490013e-05,
+      "loss": 21.4969,
+      "step": 4809
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.94148533048219e-05,
+      "loss": 22.8823,
+      "step": 4810
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.940999235093993e-05,
+      "loss": 21.6487,
+      "step": 4811
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.940513041337555e-05,
+      "loss": 0.7649,
+      "step": 4812
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.940026749225013e-05,
+      "loss": 25.9432,
+      "step": 4813
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.939540358768501e-05,
+      "loss": 24.1019,
+      "step": 4814
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.939053869980165e-05,
+      "loss": 0.7723,
+      "step": 4815
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.938567282872143e-05,
+      "loss": 22.7478,
+      "step": 4816
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.938080597456584e-05,
+      "loss": 0.7528,
+      "step": 4817
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.937593813745633e-05,
+      "loss": 0.7264,
+      "step": 4818
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.937106931751441e-05,
+      "loss": 0.7503,
+      "step": 4819
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.936619951486162e-05,
+      "loss": 24.3006,
+      "step": 4820
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.936132872961948e-05,
+      "loss": 24.6773,
+      "step": 4821
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.935645696190961e-05,
+      "loss": 24.5182,
+      "step": 4822
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.93515842118536e-05,
+      "loss": 22.155,
+      "step": 4823
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.934671047957304e-05,
+      "loss": 20.929,
+      "step": 4824
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.93418357651896e-05,
+      "loss": 19.9464,
+      "step": 4825
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.933696006882498e-05,
+      "loss": 0.7963,
+      "step": 4826
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.933208339060083e-05,
+      "loss": 23.1068,
+      "step": 4827
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.932720573063891e-05,
+      "loss": 0.7483,
+      "step": 4828
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.932232708906094e-05,
+      "loss": 0.7351,
+      "step": 4829
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.931744746598874e-05,
+      "loss": 0.74,
+      "step": 4830
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.931256686154403e-05,
+      "loss": 24.6952,
+      "step": 4831
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.930768527584871e-05,
+      "loss": 20.5631,
+      "step": 4832
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.930280270902458e-05,
+      "loss": 19.2688,
+      "step": 4833
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.929791916119352e-05,
+      "loss": 23.8852,
+      "step": 4834
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.929303463247743e-05,
+      "loss": 0.7973,
+      "step": 4835
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.928814912299821e-05,
+      "loss": 0.7756,
+      "step": 4836
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.928326263287783e-05,
+      "loss": 0.7934,
+      "step": 4837
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.927837516223824e-05,
+      "loss": 0.7875,
+      "step": 4838
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.927348671120144e-05,
+      "loss": 0.7598,
+      "step": 4839
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.926859727988943e-05,
+      "loss": 0.8207,
+      "step": 4840
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.92637068684243e-05,
+      "loss": 0.8458,
+      "step": 4841
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.925881547692805e-05,
+      "loss": 22.1167,
+      "step": 4842
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.92539231055228e-05,
+      "loss": 0.7782,
+      "step": 4843
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.92490297543307e-05,
+      "loss": 0.7792,
+      "step": 4844
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.924413542347381e-05,
+      "loss": 24.0928,
+      "step": 4845
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.923924011307438e-05,
+      "loss": 0.7411,
+      "step": 4846
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.923434382325453e-05,
+      "loss": 0.827,
+      "step": 4847
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.922944655413652e-05,
+      "loss": 19.2096,
+      "step": 4848
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.922454830584255e-05,
+      "loss": 21.4802,
+      "step": 4849
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.92196490784949e-05,
+      "loss": 21.6431,
+      "step": 4850
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.921474887221586e-05,
+      "loss": 19.1286,
+      "step": 4851
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.920984768712773e-05,
+      "loss": 22.4672,
+      "step": 4852
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.920494552335286e-05,
+      "loss": 0.7782,
+      "step": 4853
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.920004238101359e-05,
+      "loss": 0.7689,
+      "step": 4854
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.919513826023231e-05,
+      "loss": 21.9373,
+      "step": 4855
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.919023316113143e-05,
+      "loss": 0.7759,
+      "step": 4856
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.918532708383339e-05,
+      "loss": 21.4223,
+      "step": 4857
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.918042002846063e-05,
+      "loss": 0.7413,
+      "step": 4858
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.917551199513565e-05,
+      "loss": 0.7605,
+      "step": 4859
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.917060298398092e-05,
+      "loss": 0.7571,
+      "step": 4860
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.916569299511902e-05,
+      "loss": 0.7886,
+      "step": 4861
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.916078202867248e-05,
+      "loss": 24.619,
+      "step": 4862
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.915587008476388e-05,
+      "loss": 0.7534,
+      "step": 4863
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.91509571635158e-05,
+      "loss": 0.7564,
+      "step": 4864
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.91460432650509e-05,
+      "loss": 19.4287,
+      "step": 4865
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.914112838949183e-05,
+      "loss": 0.7586,
+      "step": 4866
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.913621253696126e-05,
+      "loss": 20.3793,
+      "step": 4867
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.913129570758187e-05,
+      "loss": 19.4266,
+      "step": 4868
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.912637790147643e-05,
+      "loss": 20.8489,
+      "step": 4869
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.912145911876766e-05,
+      "loss": 19.3501,
+      "step": 4870
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.911653935957833e-05,
+      "loss": 0.7721,
+      "step": 4871
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.911161862403124e-05,
+      "loss": 22.0339,
+      "step": 4872
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.910669691224922e-05,
+      "loss": 22.3041,
+      "step": 4873
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.910177422435513e-05,
+      "loss": 0.77,
+      "step": 4874
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.909685056047182e-05,
+      "loss": 0.7276,
+      "step": 4875
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.909192592072219e-05,
+      "loss": 0.7721,
+      "step": 4876
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.908700030522916e-05,
+      "loss": 19.4244,
+      "step": 4877
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.908207371411569e-05,
+      "loss": 23.138,
+      "step": 4878
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.907714614750473e-05,
+      "loss": 0.7873,
+      "step": 4879
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.907221760551927e-05,
+      "loss": 0.754,
+      "step": 4880
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.906728808828235e-05,
+      "loss": 22.7423,
+      "step": 4881
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.906235759591698e-05,
+      "loss": 23.1592,
+      "step": 4882
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.905742612854628e-05,
+      "loss": 0.7248,
+      "step": 4883
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.905249368629328e-05,
+      "loss": 18.6043,
+      "step": 4884
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.904756026928113e-05,
+      "loss": 21.0275,
+      "step": 4885
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.904262587763294e-05,
+      "loss": 0.707,
+      "step": 4886
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.903769051147192e-05,
+      "loss": 20.8577,
+      "step": 4887
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.903275417092123e-05,
+      "loss": 0.7375,
+      "step": 4888
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.902781685610407e-05,
+      "loss": 22.4618,
+      "step": 4889
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.902287856714368e-05,
+      "loss": 22.0814,
+      "step": 4890
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.901793930416336e-05,
+      "loss": 0.8097,
+      "step": 4891
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.901299906728635e-05,
+      "loss": 0.7575,
+      "step": 4892
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.900805785663597e-05,
+      "loss": 0.8114,
+      "step": 4893
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.900311567233556e-05,
+      "loss": 21.1582,
+      "step": 4894
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.89981725145085e-05,
+      "loss": 21.8179,
+      "step": 4895
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.899322838327813e-05,
+      "loss": 23.1507,
+      "step": 4896
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.898828327876788e-05,
+      "loss": 0.8451,
+      "step": 4897
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.898333720110117e-05,
+      "loss": 20.2757,
+      "step": 4898
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.897839015040148e-05,
+      "loss": 20.2792,
+      "step": 4899
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.897344212679227e-05,
+      "loss": 19.2974,
+      "step": 4900
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.896849313039702e-05,
+      "loss": 0.7795,
+      "step": 4901
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.89635431613393e-05,
+      "loss": 20.72,
+      "step": 4902
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.895859221974266e-05,
+      "loss": 21.2526,
+      "step": 4903
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.895364030573065e-05,
+      "loss": 22.7612,
+      "step": 4904
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.894868741942689e-05,
+      "loss": 18.8452,
+      "step": 4905
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.894373356095497e-05,
+      "loss": 0.8086,
+      "step": 4906
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.89387787304386e-05,
+      "loss": 22.7335,
+      "step": 4907
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.893382292800142e-05,
+      "loss": 22.5208,
+      "step": 4908
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.892886615376712e-05,
+      "loss": 20.0169,
+      "step": 4909
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.892390840785942e-05,
+      "loss": 20.2359,
+      "step": 4910
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.891894969040212e-05,
+      "loss": 18.245,
+      "step": 4911
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.89139900015189e-05,
+      "loss": 0.7869,
+      "step": 4912
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.890902934133363e-05,
+      "loss": 21.3917,
+      "step": 4913
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.89040677099701e-05,
+      "loss": 0.8078,
+      "step": 4914
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.889910510755216e-05,
+      "loss": 22.8535,
+      "step": 4915
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.889414153420367e-05,
+      "loss": 21.4691,
+      "step": 4916
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.888917699004853e-05,
+      "loss": 22.2472,
+      "step": 4917
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.888421147521066e-05,
+      "loss": 21.9117,
+      "step": 4918
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.887924498981399e-05,
+      "loss": 0.7324,
+      "step": 4919
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.887427753398248e-05,
+      "loss": 20.1569,
+      "step": 4920
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.886930910784013e-05,
+      "loss": 0.7212,
+      "step": 4921
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.886433971151094e-05,
+      "loss": 0.7358,
+      "step": 4922
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.885936934511895e-05,
+      "loss": 22.2629,
+      "step": 4923
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.885439800878825e-05,
+      "loss": 22.9013,
+      "step": 4924
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.884942570264291e-05,
+      "loss": 0.7958,
+      "step": 4925
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.884445242680703e-05,
+      "loss": 0.8366,
+      "step": 4926
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.883947818140473e-05,
+      "loss": 0.7416,
+      "step": 4927
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.883450296656019e-05,
+      "loss": 19.7518,
+      "step": 4928
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.88295267823976e-05,
+      "loss": 21.8655,
+      "step": 4929
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.882454962904115e-05,
+      "loss": 0.8062,
+      "step": 4930
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.881957150661508e-05,
+      "loss": 21.6145,
+      "step": 4931
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.881459241524363e-05,
+      "loss": 0.7803,
+      "step": 4932
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.880961235505112e-05,
+      "loss": 23.1489,
+      "step": 4933
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.880463132616178e-05,
+      "loss": 0.7502,
+      "step": 4934
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.879964932870002e-05,
+      "loss": 23.5082,
+      "step": 4935
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.879466636279015e-05,
+      "loss": 0.7458,
+      "step": 4936
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.878968242855653e-05,
+      "loss": 21.3788,
+      "step": 4937
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.87846975261236e-05,
+      "loss": 20.714,
+      "step": 4938
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.877971165561574e-05,
+      "loss": 0.7854,
+      "step": 4939
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.877472481715744e-05,
+      "loss": 21.3994,
+      "step": 4940
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.876973701087314e-05,
+      "loss": 0.7481,
+      "step": 4941
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.876474823688737e-05,
+      "loss": 21.1806,
+      "step": 4942
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.875975849532464e-05,
+      "loss": 0.7431,
+      "step": 4943
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.875476778630946e-05,
+      "loss": 20.6397,
+      "step": 4944
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.874977610996644e-05,
+      "loss": 18.8181,
+      "step": 4945
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.874478346642016e-05,
+      "loss": 20.2905,
+      "step": 4946
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.873978985579525e-05,
+      "loss": 0.7684,
+      "step": 4947
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.873479527821632e-05,
+      "loss": 0.7508,
+      "step": 4948
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.872979973380806e-05,
+      "loss": 23.1733,
+      "step": 4949
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.872480322269517e-05,
+      "loss": 22.4979,
+      "step": 4950
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.871980574500234e-05,
+      "loss": 0.7885,
+      "step": 4951
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.871480730085432e-05,
+      "loss": 21.9084,
+      "step": 4952
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.870980789037586e-05,
+      "loss": 0.7911,
+      "step": 4953
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.870480751369176e-05,
+      "loss": 0.7998,
+      "step": 4954
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.869980617092683e-05,
+      "loss": 0.7354,
+      "step": 4955
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.869480386220591e-05,
+      "loss": 21.1137,
+      "step": 4956
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.868980058765383e-05,
+      "loss": 0.8091,
+      "step": 4957
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.868479634739553e-05,
+      "loss": 23.2908,
+      "step": 4958
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.867979114155584e-05,
+      "loss": 0.7695,
+      "step": 4959
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.867478497025976e-05,
+      "loss": 22.7511,
+      "step": 4960
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.86697778336322e-05,
+      "loss": 0.7787,
+      "step": 4961
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.866476973179816e-05,
+      "loss": 19.9908,
+      "step": 4962
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.865976066488264e-05,
+      "loss": 19.6505,
+      "step": 4963
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.865475063301067e-05,
+      "loss": 0.7736,
+      "step": 4964
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.864973963630729e-05,
+      "loss": 0.7422,
+      "step": 4965
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.864472767489759e-05,
+      "loss": 24.2429,
+      "step": 4966
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.863971474890667e-05,
+      "loss": 0.7631,
+      "step": 4967
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.863470085845963e-05,
+      "loss": 22.6659,
+      "step": 4968
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.862968600368164e-05,
+      "loss": 0.7583,
+      "step": 4969
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.862467018469788e-05,
+      "loss": 21.0904,
+      "step": 4970
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.861965340163351e-05,
+      "loss": 0.7599,
+      "step": 4971
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.86146356546138e-05,
+      "loss": 19.887,
+      "step": 4972
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.860961694376396e-05,
+      "loss": 0.8177,
+      "step": 4973
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.860459726920925e-05,
+      "loss": 22.5949,
+      "step": 4974
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.859957663107498e-05,
+      "loss": 21.8928,
+      "step": 4975
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.859455502948648e-05,
+      "loss": 22.198,
+      "step": 4976
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.858953246456907e-05,
+      "loss": 0.7312,
+      "step": 4977
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.858450893644811e-05,
+      "loss": 20.0188,
+      "step": 4978
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.857948444524899e-05,
+      "loss": 21.2192,
+      "step": 4979
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.857445899109715e-05,
+      "loss": 0.8527,
+      "step": 4980
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.8569432574118e-05,
+      "loss": 22.1054,
+      "step": 4981
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.8564405194437e-05,
+      "loss": 21.2857,
+      "step": 4982
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.855937685217965e-05,
+      "loss": 20.7102,
+      "step": 4983
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.855434754747144e-05,
+      "loss": 0.7625,
+      "step": 4984
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.854931728043792e-05,
+      "loss": 21.0354,
+      "step": 4985
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.854428605120463e-05,
+      "loss": 18.883,
+      "step": 4986
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.853925385989716e-05,
+      "loss": 0.738,
+      "step": 4987
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.853422070664113e-05,
+      "loss": 19.4503,
+      "step": 4988
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.852918659156213e-05,
+      "loss": 23.909,
+      "step": 4989
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.852415151478583e-05,
+      "loss": 0.7749,
+      "step": 4990
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.851911547643794e-05,
+      "loss": 21.5629,
+      "step": 4991
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.851407847664413e-05,
+      "loss": 0.7735,
+      "step": 4992
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.85090405155301e-05,
+      "loss": 0.765,
+      "step": 4993
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.850400159322165e-05,
+      "loss": 0.799,
+      "step": 4994
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.849896170984452e-05,
+      "loss": 0.8103,
+      "step": 4995
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.849392086552452e-05,
+      "loss": 0.7479,
+      "step": 4996
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.848887906038747e-05,
+      "loss": 22.6052,
+      "step": 4997
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.84838362945592e-05,
+      "loss": 20.4339,
+      "step": 4998
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.847879256816559e-05,
+      "loss": 21.5431,
+      "step": 4999
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.847374788133254e-05,
+      "loss": 0.7351,
+      "step": 5000
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.846870223418596e-05,
+      "loss": 18.6651,
+      "step": 5001
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.846365562685177e-05,
+      "loss": 19.5045,
+      "step": 5002
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.845860805945596e-05,
+      "loss": 0.7535,
+      "step": 5003
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.845355953212452e-05,
+      "loss": 21.7944,
+      "step": 5004
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.844851004498345e-05,
+      "loss": 0.781,
+      "step": 5005
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.844345959815879e-05,
+      "loss": 0.809,
+      "step": 5006
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.84384081917766e-05,
+      "loss": 21.8648,
+      "step": 5007
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.843335582596297e-05,
+      "loss": 23.0398,
+      "step": 5008
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.8428302500844e-05,
+      "loss": 18.8526,
+      "step": 5009
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.842324821654582e-05,
+      "loss": 19.3421,
+      "step": 5010
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.841819297319459e-05,
+      "loss": 20.341,
+      "step": 5011
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.84131367709165e-05,
+      "loss": 21.2722,
+      "step": 5012
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.840807960983775e-05,
+      "loss": 21.267,
+      "step": 5013
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.840302149008455e-05,
+      "loss": 0.7256,
+      "step": 5014
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.839796241178318e-05,
+      "loss": 19.9793,
+      "step": 5015
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.839290237505988e-05,
+      "loss": 19.9944,
+      "step": 5016
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.838784138004098e-05,
+      "loss": 0.7502,
+      "step": 5017
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.838277942685281e-05,
+      "loss": 20.7216,
+      "step": 5018
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.837771651562168e-05,
+      "loss": 23.0094,
+      "step": 5019
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.837265264647399e-05,
+      "loss": 21.4393,
+      "step": 5020
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.836758781953612e-05,
+      "loss": 21.2232,
+      "step": 5021
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.836252203493452e-05,
+      "loss": 21.5172,
+      "step": 5022
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.835745529279559e-05,
+      "loss": 21.3573,
+      "step": 5023
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.835238759324583e-05,
+      "loss": 22.5759,
+      "step": 5024
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.83473189364117e-05,
+      "loss": 0.7572,
+      "step": 5025
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.834224932241975e-05,
+      "loss": 21.0083,
+      "step": 5026
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.83371787513965e-05,
+      "loss": 21.2539,
+      "step": 5027
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.833210722346851e-05,
+      "loss": 0.803,
+      "step": 5028
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.832703473876239e-05,
+      "loss": 19.9926,
+      "step": 5029
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.832196129740471e-05,
+      "loss": 23.0857,
+      "step": 5030
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.831688689952214e-05,
+      "loss": 0.8077,
+      "step": 5031
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.83118115452413e-05,
+      "loss": 20.6209,
+      "step": 5032
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.830673523468893e-05,
+      "loss": 0.8097,
+      "step": 5033
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.830165796799168e-05,
+      "loss": 0.7979,
+      "step": 5034
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.82965797452763e-05,
+      "loss": 0.7536,
+      "step": 5035
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.829150056666955e-05,
+      "loss": 19.8606,
+      "step": 5036
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.828642043229822e-05,
+      "loss": 0.7489,
+      "step": 5037
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.828133934228908e-05,
+      "loss": 21.5654,
+      "step": 5038
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.827625729676897e-05,
+      "loss": 0.8225,
+      "step": 5039
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.827117429586474e-05,
+      "loss": 0.7555,
+      "step": 5040
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.826609033970325e-05,
+      "loss": 19.8836,
+      "step": 5041
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.82610054284114e-05,
+      "loss": 23.248,
+      "step": 5042
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.825591956211614e-05,
+      "loss": 0.7579,
+      "step": 5043
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.825083274094439e-05,
+      "loss": 21.6933,
+      "step": 5044
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.824574496502312e-05,
+      "loss": 0.8103,
+      "step": 5045
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.824065623447932e-05,
+      "loss": 0.7179,
+      "step": 5046
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.823556654944001e-05,
+      "loss": 20.2165,
+      "step": 5047
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.823047591003222e-05,
+      "loss": 21.4799,
+      "step": 5048
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.822538431638304e-05,
+      "loss": 21.7526,
+      "step": 5049
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.822029176861954e-05,
+      "loss": 23.2528,
+      "step": 5050
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.821519826686882e-05,
+      "loss": 22.4728,
+      "step": 5051
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.821010381125802e-05,
+      "loss": 0.7894,
+      "step": 5052
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.82050084019143e-05,
+      "loss": 24.4334,
+      "step": 5053
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.819991203896487e-05,
+      "loss": 22.5262,
+      "step": 5054
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.81948147225369e-05,
+      "loss": 0.7789,
+      "step": 5055
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.818971645275763e-05,
+      "loss": 0.7898,
+      "step": 5056
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.818461722975434e-05,
+      "loss": 21.5765,
+      "step": 5057
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.817951705365424e-05,
+      "loss": 22.1707,
+      "step": 5058
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.817441592458472e-05,
+      "loss": 21.0316,
+      "step": 5059
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.816931384267304e-05,
+      "loss": 22.8581,
+      "step": 5060
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.816421080804657e-05,
+      "loss": 21.0347,
+      "step": 5061
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.81591068208327e-05,
+      "loss": 0.8166,
+      "step": 5062
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.81540018811588e-05,
+      "loss": 0.7812,
+      "step": 5063
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.814889598915231e-05,
+      "loss": 0.7328,
+      "step": 5064
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.814378914494066e-05,
+      "loss": 22.2312,
+      "step": 5065
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.813868134865132e-05,
+      "loss": 0.7657,
+      "step": 5066
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.813357260041178e-05,
+      "loss": 0.7906,
+      "step": 5067
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.812846290034957e-05,
+      "loss": 0.755,
+      "step": 5068
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.812335224859223e-05,
+      "loss": 0.7974,
+      "step": 5069
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.81182406452673e-05,
+      "loss": 21.0543,
+      "step": 5070
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.811312809050239e-05,
+      "loss": 0.7927,
+      "step": 5071
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.810801458442508e-05,
+      "loss": 23.8395,
+      "step": 5072
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.810290012716305e-05,
+      "loss": 0.7488,
+      "step": 5073
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.809778471884391e-05,
+      "loss": 21.7277,
+      "step": 5074
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.809266835959538e-05,
+      "loss": 20.6829,
+      "step": 5075
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.808755104954515e-05,
+      "loss": 0.7688,
+      "step": 5076
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.808243278882094e-05,
+      "loss": 19.4353,
+      "step": 5077
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.807731357755051e-05,
+      "loss": 20.6026,
+      "step": 5078
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.807219341586165e-05,
+      "loss": 0.7796,
+      "step": 5079
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.806707230388214e-05,
+      "loss": 0.7459,
+      "step": 5080
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.806195024173982e-05,
+      "loss": 0.8099,
+      "step": 5081
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.805682722956254e-05,
+      "loss": 21.3789,
+      "step": 5082
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.805170326747816e-05,
+      "loss": 0.7888,
+      "step": 5083
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.804657835561456e-05,
+      "loss": 19.9474,
+      "step": 5084
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.804145249409969e-05,
+      "loss": 18.8897,
+      "step": 5085
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.803632568306147e-05,
+      "loss": 0.7216,
+      "step": 5086
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.803119792262788e-05,
+      "loss": 23.5725,
+      "step": 5087
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.802606921292692e-05,
+      "loss": 0.7192,
+      "step": 5088
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.802093955408658e-05,
+      "loss": 0.7682,
+      "step": 5089
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.801580894623491e-05,
+      "loss": 0.7252,
+      "step": 5090
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.801067738949997e-05,
+      "loss": 22.5654,
+      "step": 5091
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.800554488400983e-05,
+      "loss": 0.7554,
+      "step": 5092
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.800041142989262e-05,
+      "loss": 0.7909,
+      "step": 5093
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.799527702727647e-05,
+      "loss": 0.7145,
+      "step": 5094
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.799014167628953e-05,
+      "loss": 0.7591,
+      "step": 5095
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.798500537705997e-05,
+      "loss": 0.8007,
+      "step": 5096
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.797986812971601e-05,
+      "loss": 22.6314,
+      "step": 5097
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.797472993438587e-05,
+      "loss": 22.7847,
+      "step": 5098
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.79695907911978e-05,
+      "loss": 22.2943,
+      "step": 5099
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.796445070028005e-05,
+      "loss": 19.9608,
+      "step": 5100
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.795930966176098e-05,
+      "loss": 0.7719,
+      "step": 5101
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.795416767576885e-05,
+      "loss": 19.6431,
+      "step": 5102
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.794902474243204e-05,
+      "loss": 0.7606,
+      "step": 5103
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.79438808618789e-05,
+      "loss": 22.7387,
+      "step": 5104
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.793873603423783e-05,
+      "loss": 0.8002,
+      "step": 5105
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.793359025963726e-05,
+      "loss": 21.319,
+      "step": 5106
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.79284435382056e-05,
+      "loss": 0.7703,
+      "step": 5107
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.792329587007133e-05,
+      "loss": 18.5518,
+      "step": 5108
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.791814725536295e-05,
+      "loss": 21.4029,
+      "step": 5109
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.791299769420894e-05,
+      "loss": 19.2801,
+      "step": 5110
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.790784718673786e-05,
+      "loss": 0.771,
+      "step": 5111
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.790269573307826e-05,
+      "loss": 21.2037,
+      "step": 5112
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.789754333335871e-05,
+      "loss": 20.0228,
+      "step": 5113
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.789238998770784e-05,
+      "loss": 0.7897,
+      "step": 5114
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.788723569625426e-05,
+      "loss": 0.799,
+      "step": 5115
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.788208045912661e-05,
+      "loss": 21.5822,
+      "step": 5116
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.78769242764536e-05,
+      "loss": 0.7659,
+      "step": 5117
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.78717671483639e-05,
+      "loss": 0.7955,
+      "step": 5118
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.786660907498625e-05,
+      "loss": 0.7574,
+      "step": 5119
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.78614500564494e-05,
+      "loss": 0.7952,
+      "step": 5120
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.78562900928821e-05,
+      "loss": 0.7273,
+      "step": 5121
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.785112918441316e-05,
+      "loss": 17.8019,
+      "step": 5122
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.784596733117139e-05,
+      "loss": 20.7759,
+      "step": 5123
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.784080453328565e-05,
+      "loss": 0.8178,
+      "step": 5124
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.783564079088477e-05,
+      "loss": 20.1902,
+      "step": 5125
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.783047610409765e-05,
+      "loss": 0.7599,
+      "step": 5126
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.782531047305323e-05,
+      "loss": 21.3084,
+      "step": 5127
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.782014389788041e-05,
+      "loss": 18.482,
+      "step": 5128
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.781497637870817e-05,
+      "loss": 23.6543,
+      "step": 5129
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.780980791566549e-05,
+      "loss": 20.2908,
+      "step": 5130
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.780463850888135e-05,
+      "loss": 18.6173,
+      "step": 5131
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.779946815848483e-05,
+      "loss": 0.7497,
+      "step": 5132
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.779429686460494e-05,
+      "loss": 21.8987,
+      "step": 5133
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.778912462737077e-05,
+      "loss": 0.7349,
+      "step": 5134
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.778395144691142e-05,
+      "loss": 21.4036,
+      "step": 5135
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.7778777323356e-05,
+      "loss": 21.6327,
+      "step": 5136
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.777360225683369e-05,
+      "loss": 0.7549,
+      "step": 5137
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.776842624747364e-05,
+      "loss": 0.6952,
+      "step": 5138
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.776324929540505e-05,
+      "loss": 0.802,
+      "step": 5139
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.775807140075714e-05,
+      "loss": 23.4144,
+      "step": 5140
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.775289256365913e-05,
+      "loss": 19.0821,
+      "step": 5141
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.774771278424032e-05,
+      "loss": 0.7803,
+      "step": 5142
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.774253206262997e-05,
+      "loss": 21.7883,
+      "step": 5143
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.773735039895741e-05,
+      "loss": 0.7883,
+      "step": 5144
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.773216779335195e-05,
+      "loss": 23.5272,
+      "step": 5145
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.772698424594297e-05,
+      "loss": 19.4143,
+      "step": 5146
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.772179975685985e-05,
+      "loss": 0.7429,
+      "step": 5147
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.771661432623201e-05,
+      "loss": 22.0884,
+      "step": 5148
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.771142795418885e-05,
+      "loss": 20.6871,
+      "step": 5149
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.770624064085985e-05,
+      "loss": 21.0941,
+      "step": 5150
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.770105238637447e-05,
+      "loss": 18.1122,
+      "step": 5151
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.76958631908622e-05,
+      "loss": 20.0301,
+      "step": 5152
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.76906730544526e-05,
+      "loss": 22.0639,
+      "step": 5153
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.768548197727517e-05,
+      "loss": 22.9781,
+      "step": 5154
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.768028995945953e-05,
+      "loss": 20.9875,
+      "step": 5155
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.767509700113524e-05,
+      "loss": 21.1978,
+      "step": 5156
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.766990310243193e-05,
+      "loss": 0.7916,
+      "step": 5157
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.766470826347922e-05,
+      "loss": 20.866,
+      "step": 5158
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.765951248440681e-05,
+      "loss": 23.1416,
+      "step": 5159
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.765431576534436e-05,
+      "loss": 21.8338,
+      "step": 5160
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.76491181064216e-05,
+      "loss": 19.4845,
+      "step": 5161
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.764391950776824e-05,
+      "loss": 21.5477,
+      "step": 5162
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.763871996951408e-05,
+      "loss": 18.7274,
+      "step": 5163
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.763351949178884e-05,
+      "loss": 23.149,
+      "step": 5164
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.762831807472239e-05,
+      "loss": 20.1094,
+      "step": 5165
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.762311571844451e-05,
+      "loss": 0.7484,
+      "step": 5166
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.761791242308508e-05,
+      "loss": 18.8788,
+      "step": 5167
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.761270818877397e-05,
+      "loss": 21.3152,
+      "step": 5168
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.760750301564107e-05,
+      "loss": 0.776,
+      "step": 5169
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.760229690381632e-05,
+      "loss": 20.2646,
+      "step": 5170
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.759708985342964e-05,
+      "loss": 21.9219,
+      "step": 5171
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.759188186461101e-05,
+      "loss": 22.5633,
+      "step": 5172
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.758667293749044e-05,
+      "loss": 0.7845,
+      "step": 5173
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.758146307219792e-05,
+      "loss": 18.9051,
+      "step": 5174
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.75762522688635e-05,
+      "loss": 0.7701,
+      "step": 5175
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.757104052761724e-05,
+      "loss": 20.6353,
+      "step": 5176
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.756582784858925e-05,
+      "loss": 22.2758,
+      "step": 5177
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.756061423190959e-05,
+      "loss": 20.8053,
+      "step": 5178
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.755539967770845e-05,
+      "loss": 21.5809,
+      "step": 5179
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.755018418611596e-05,
+      "loss": 19.2584,
+      "step": 5180
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.754496775726228e-05,
+      "loss": 0.7901,
+      "step": 5181
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.753975039127764e-05,
+      "loss": 0.7677,
+      "step": 5182
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.753453208829226e-05,
+      "loss": 0.7259,
+      "step": 5183
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.752931284843639e-05,
+      "loss": 19.7218,
+      "step": 5184
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.752409267184029e-05,
+      "loss": 0.7512,
+      "step": 5185
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.751887155863428e-05,
+      "loss": 0.8057,
+      "step": 5186
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.751364950894867e-05,
+      "loss": 0.7635,
+      "step": 5187
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.75084265229138e-05,
+      "loss": 0.7337,
+      "step": 5188
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.750320260066004e-05,
+      "loss": 24.1843,
+      "step": 5189
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.749797774231778e-05,
+      "loss": 22.1324,
+      "step": 5190
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.749275194801742e-05,
+      "loss": 21.4375,
+      "step": 5191
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.748752521788943e-05,
+      "loss": 19.5622,
+      "step": 5192
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.748229755206425e-05,
+      "loss": 0.7353,
+      "step": 5193
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.747706895067236e-05,
+      "loss": 0.8043,
+      "step": 5194
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.747183941384426e-05,
+      "loss": 20.6429,
+      "step": 5195
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.746660894171051e-05,
+      "loss": 0.742,
+      "step": 5196
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.746137753440164e-05,
+      "loss": 20.9049,
+      "step": 5197
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.745614519204823e-05,
+      "loss": 21.5948,
+      "step": 5198
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.74509119147809e-05,
+      "loss": 18.9223,
+      "step": 5199
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.744567770273024e-05,
+      "loss": 20.6202,
+      "step": 5200
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.744044255602693e-05,
+      "loss": 0.7687,
+      "step": 5201
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.743520647480162e-05,
+      "loss": 22.0836,
+      "step": 5202
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.742996945918499e-05,
+      "loss": 0.7673,
+      "step": 5203
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.74247315093078e-05,
+      "loss": 0.7282,
+      "step": 5204
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.741949262530074e-05,
+      "loss": 0.7922,
+      "step": 5205
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.741425280729463e-05,
+      "loss": 0.7727,
+      "step": 5206
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.740901205542021e-05,
+      "loss": 21.7231,
+      "step": 5207
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.74037703698083e-05,
+      "loss": 22.2743,
+      "step": 5208
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.739852775058976e-05,
+      "loss": 21.9854,
+      "step": 5209
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.739328419789541e-05,
+      "loss": 19.8385,
+      "step": 5210
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.738803971185616e-05,
+      "loss": 0.7885,
+      "step": 5211
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.738279429260289e-05,
+      "loss": 23.1561,
+      "step": 5212
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.737754794026655e-05,
+      "loss": 0.787,
+      "step": 5213
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.737230065497809e-05,
+      "loss": 19.9586,
+      "step": 5214
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.736705243686845e-05,
+      "loss": 0.7535,
+      "step": 5215
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.736180328606866e-05,
+      "loss": 0.7342,
+      "step": 5216
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.735655320270974e-05,
+      "loss": 0.7965,
+      "step": 5217
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.735130218692271e-05,
+      "loss": 0.7749,
+      "step": 5218
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.734605023883864e-05,
+      "loss": 21.948,
+      "step": 5219
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.734079735858865e-05,
+      "loss": 0.7697,
+      "step": 5220
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.733554354630382e-05,
+      "loss": 20.1388,
+      "step": 5221
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.733028880211528e-05,
+      "loss": 0.719,
+      "step": 5222
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.732503312615424e-05,
+      "loss": 0.756,
+      "step": 5223
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.731977651855183e-05,
+      "loss": 22.1476,
+      "step": 5224
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.73145189794393e-05,
+      "loss": 22.8508,
+      "step": 5225
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.730926050894783e-05,
+      "loss": 20.8749,
+      "step": 5226
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.730400110720872e-05,
+      "loss": 21.8913,
+      "step": 5227
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.72987407743532e-05,
+      "loss": 23.463,
+      "step": 5228
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.729347951051261e-05,
+      "loss": 0.7327,
+      "step": 5229
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.728821731581826e-05,
+      "loss": 0.8198,
+      "step": 5230
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.728295419040149e-05,
+      "loss": 22.7113,
+      "step": 5231
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.727769013439367e-05,
+      "loss": 0.7052,
+      "step": 5232
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.727242514792618e-05,
+      "loss": 21.6528,
+      "step": 5233
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.726715923113047e-05,
+      "loss": 0.7425,
+      "step": 5234
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.726189238413793e-05,
+      "loss": 0.7761,
+      "step": 5235
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.725662460708007e-05,
+      "loss": 19.719,
+      "step": 5236
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.725135590008835e-05,
+      "loss": 0.7354,
+      "step": 5237
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.724608626329426e-05,
+      "loss": 0.7938,
+      "step": 5238
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.724081569682938e-05,
+      "loss": 0.8017,
+      "step": 5239
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.723554420082522e-05,
+      "loss": 21.0807,
+      "step": 5240
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.723027177541339e-05,
+      "loss": 21.4134,
+      "step": 5241
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.722499842072547e-05,
+      "loss": 0.766,
+      "step": 5242
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.721972413689307e-05,
+      "loss": 22.711,
+      "step": 5243
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.721444892404789e-05,
+      "loss": 0.7304,
+      "step": 5244
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.720917278232154e-05,
+      "loss": 0.7391,
+      "step": 5245
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.720389571184576e-05,
+      "loss": 0.7528,
+      "step": 5246
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.719861771275223e-05,
+      "loss": 0.7861,
+      "step": 5247
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.719333878517273e-05,
+      "loss": 0.7841,
+      "step": 5248
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.718805892923898e-05,
+      "loss": 25.1452,
+      "step": 5249
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.71827781450828e-05,
+      "loss": 0.7874,
+      "step": 5250
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.7177496432836e-05,
+      "loss": 20.9491,
+      "step": 5251
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.717221379263037e-05,
+      "loss": 0.7774,
+      "step": 5252
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.716693022459782e-05,
+      "loss": 20.7759,
+      "step": 5253
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.71616457288702e-05,
+      "loss": 20.0662,
+      "step": 5254
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.715636030557942e-05,
+      "loss": 0.7162,
+      "step": 5255
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.715107395485738e-05,
+      "loss": 0.7714,
+      "step": 5256
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.714578667683608e-05,
+      "loss": 0.7799,
+      "step": 5257
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.714049847164745e-05,
+      "loss": 21.2767,
+      "step": 5258
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.71352093394235e-05,
+      "loss": 22.2103,
+      "step": 5259
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.712991928029625e-05,
+      "loss": 22.5527,
+      "step": 5260
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.712462829439774e-05,
+      "loss": 0.7734,
+      "step": 5261
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.711933638186e-05,
+      "loss": 0.8179,
+      "step": 5262
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.711404354281518e-05,
+      "loss": 0.8053,
+      "step": 5263
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.710874977739536e-05,
+      "loss": 19.6778,
+      "step": 5264
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.710345508573267e-05,
+      "loss": 21.4845,
+      "step": 5265
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.709815946795925e-05,
+      "loss": 20.5009,
+      "step": 5266
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.709286292420732e-05,
+      "loss": 20.0047,
+      "step": 5267
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.708756545460905e-05,
+      "loss": 0.7619,
+      "step": 5268
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.708226705929669e-05,
+      "loss": 0.7911,
+      "step": 5269
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.707696773840247e-05,
+      "loss": 22.8148,
+      "step": 5270
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.707166749205866e-05,
+      "loss": 0.8135,
+      "step": 5271
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.706636632039757e-05,
+      "loss": 0.754,
+      "step": 5272
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.706106422355154e-05,
+      "loss": 0.7985,
+      "step": 5273
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.705576120165284e-05,
+      "loss": 20.0805,
+      "step": 5274
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.705045725483392e-05,
+      "loss": 0.7797,
+      "step": 5275
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.704515238322711e-05,
+      "loss": 20.8973,
+      "step": 5276
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.703984658696483e-05,
+      "loss": 0.8044,
+      "step": 5277
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.703453986617953e-05,
+      "loss": 21.171,
+      "step": 5278
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.702923222100365e-05,
+      "loss": 21.7994,
+      "step": 5279
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.70239236515697e-05,
+      "loss": 23.7462,
+      "step": 5280
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.701861415801013e-05,
+      "loss": 0.7451,
+      "step": 5281
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.701330374045752e-05,
+      "loss": 22.6988,
+      "step": 5282
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.70079923990444e-05,
+      "loss": 0.7838,
+      "step": 5283
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.700268013390333e-05,
+      "loss": 0.7391,
+      "step": 5284
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.699736694516693e-05,
+      "loss": 0.7062,
+      "step": 5285
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.699205283296779e-05,
+      "loss": 20.1907,
+      "step": 5286
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.698673779743856e-05,
+      "loss": 18.7945,
+      "step": 5287
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.698142183871192e-05,
+      "loss": 0.7406,
+      "step": 5288
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.697610495692054e-05,
+      "loss": 0.763,
+      "step": 5289
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.697078715219715e-05,
+      "loss": 19.4058,
+      "step": 5290
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.696546842467446e-05,
+      "loss": 0.722,
+      "step": 5291
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.696014877448524e-05,
+      "loss": 21.0279,
+      "step": 5292
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.695482820176227e-05,
+      "loss": 0.7863,
+      "step": 5293
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.694950670663836e-05,
+      "loss": 0.7624,
+      "step": 5294
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.694418428924628e-05,
+      "loss": 21.7366,
+      "step": 5295
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.693886094971897e-05,
+      "loss": 0.8138,
+      "step": 5296
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.693353668818924e-05,
+      "loss": 0.7839,
+      "step": 5297
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.692821150479002e-05,
+      "loss": 23.145,
+      "step": 5298
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.692288539965418e-05,
+      "loss": 0.8165,
+      "step": 5299
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.691755837291473e-05,
+      "loss": 20.134,
+      "step": 5300
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.691223042470456e-05,
+      "loss": 19.2744,
+      "step": 5301
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.69069015551567e-05,
+      "loss": 22.2636,
+      "step": 5302
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.690157176440415e-05,
+      "loss": 20.808,
+      "step": 5303
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.689624105257994e-05,
+      "loss": 21.2238,
+      "step": 5304
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.689090941981712e-05,
+      "loss": 19.7155,
+      "step": 5305
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.68855768662488e-05,
+      "loss": 18.9082,
+      "step": 5306
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.688024339200804e-05,
+      "loss": 17.4066,
+      "step": 5307
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.6874908997228e-05,
+      "loss": 0.7928,
+      "step": 5308
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.686957368204181e-05,
+      "loss": 0.7357,
+      "step": 5309
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.686423744658263e-05,
+      "loss": 18.8519,
+      "step": 5310
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.685890029098366e-05,
+      "loss": 0.7427,
+      "step": 5311
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.685356221537815e-05,
+      "loss": 18.4416,
+      "step": 5312
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.684822321989929e-05,
+      "loss": 0.7798,
+      "step": 5313
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.684288330468038e-05,
+      "loss": 0.8092,
+      "step": 5314
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.683754246985468e-05,
+      "loss": 0.778,
+      "step": 5315
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.683220071555552e-05,
+      "loss": 18.7345,
+      "step": 5316
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.68268580419162e-05,
+      "loss": 0.7612,
+      "step": 5317
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.682151444907012e-05,
+      "loss": 20.7078,
+      "step": 5318
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.68161699371506e-05,
+      "loss": 19.555,
+      "step": 5319
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.68108245062911e-05,
+      "loss": 20.5329,
+      "step": 5320
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.6805478156625e-05,
+      "loss": 0.7428,
+      "step": 5321
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.680013088828577e-05,
+      "loss": 0.7377,
+      "step": 5322
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.679478270140687e-05,
+      "loss": 21.7722,
+      "step": 5323
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.678943359612179e-05,
+      "loss": 0.7345,
+      "step": 5324
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.678408357256405e-05,
+      "loss": 19.333,
+      "step": 5325
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.677873263086718e-05,
+      "loss": 0.7898,
+      "step": 5326
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.677338077116475e-05,
+      "loss": 0.8098,
+      "step": 5327
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.676802799359034e-05,
+      "loss": 20.4058,
+      "step": 5328
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.676267429827755e-05,
+      "loss": 0.7775,
+      "step": 5329
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.675731968536002e-05,
+      "loss": 20.3329,
+      "step": 5330
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.67519641549714e-05,
+      "loss": 19.6099,
+      "step": 5331
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.674660770724536e-05,
+      "loss": 0.7413,
+      "step": 5332
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.67412503423156e-05,
+      "loss": 0.7301,
+      "step": 5333
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.673589206031585e-05,
+      "loss": 0.7398,
+      "step": 5334
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.673053286137985e-05,
+      "loss": 18.1568,
+      "step": 5335
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.672517274564134e-05,
+      "loss": 22.0864,
+      "step": 5336
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.671981171323415e-05,
+      "loss": 0.7485,
+      "step": 5337
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.671444976429208e-05,
+      "loss": 0.8015,
+      "step": 5338
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.670908689894895e-05,
+      "loss": 0.7773,
+      "step": 5339
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.670372311733863e-05,
+      "loss": 19.2134,
+      "step": 5340
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.669835841959499e-05,
+      "loss": 0.7398,
+      "step": 5341
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.669299280585196e-05,
+      "loss": 0.8061,
+      "step": 5342
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.668762627624345e-05,
+      "loss": 0.799,
+      "step": 5343
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.668225883090341e-05,
+      "loss": 0.7531,
+      "step": 5344
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.667689046996583e-05,
+      "loss": 0.7284,
+      "step": 5345
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.667152119356466e-05,
+      "loss": 0.7365,
+      "step": 5346
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.666615100183398e-05,
+      "loss": 0.8111,
+      "step": 5347
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.666077989490777e-05,
+      "loss": 23.1573,
+      "step": 5348
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.665540787292012e-05,
+      "loss": 20.5227,
+      "step": 5349
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.665003493600514e-05,
+      "loss": 0.7708,
+      "step": 5350
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.664466108429692e-05,
+      "loss": 22.1384,
+      "step": 5351
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.663928631792958e-05,
+      "loss": 20.166,
+      "step": 5352
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.66339106370373e-05,
+      "loss": 22.5417,
+      "step": 5353
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.662853404175424e-05,
+      "loss": 0.728,
+      "step": 5354
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.66231565322146e-05,
+      "loss": 0.7149,
+      "step": 5355
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.661777810855262e-05,
+      "loss": 0.7912,
+      "step": 5356
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.661239877090252e-05,
+      "loss": 21.0749,
+      "step": 5357
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.660701851939861e-05,
+      "loss": 0.7265,
+      "step": 5358
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.660163735417514e-05,
+      "loss": 21.8336,
+      "step": 5359
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.659625527536646e-05,
+      "loss": 21.4694,
+      "step": 5360
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.659087228310689e-05,
+      "loss": 0.8641,
+      "step": 5361
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.65854883775308e-05,
+      "loss": 21.2897,
+      "step": 5362
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.658010355877255e-05,
+      "loss": 18.9757,
+      "step": 5363
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.657471782696657e-05,
+      "loss": 0.7992,
+      "step": 5364
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.656933118224728e-05,
+      "loss": 19.3956,
+      "step": 5365
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.656394362474913e-05,
+      "loss": 0.7768,
+      "step": 5366
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.65585551546066e-05,
+      "loss": 0.8359,
+      "step": 5367
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.65531657719542e-05,
+      "loss": 19.0816,
+      "step": 5368
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.654777547692643e-05,
+      "loss": 0.7751,
+      "step": 5369
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.654238426965783e-05,
+      "loss": 0.7333,
+      "step": 5370
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.653699215028296e-05,
+      "loss": 0.7425,
+      "step": 5371
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.653159911893647e-05,
+      "loss": 19.8927,
+      "step": 5372
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.65262051757529e-05,
+      "loss": 0.7851,
+      "step": 5373
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.65208103208669e-05,
+      "loss": 0.7571,
+      "step": 5374
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.651541455441314e-05,
+      "loss": 0.8011,
+      "step": 5375
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.651001787652629e-05,
+      "loss": 0.8234,
+      "step": 5376
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.650462028734107e-05,
+      "loss": 0.7569,
+      "step": 5377
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.649922178699219e-05,
+      "loss": 20.129,
+      "step": 5378
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.64938223756144e-05,
+      "loss": 21.8865,
+      "step": 5379
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.648842205334247e-05,
+      "loss": 0.8451,
+      "step": 5380
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.648302082031119e-05,
+      "loss": 0.7921,
+      "step": 5381
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.647761867665538e-05,
+      "loss": 21.0171,
+      "step": 5382
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.647221562250989e-05,
+      "loss": 0.7529,
+      "step": 5383
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.646681165800956e-05,
+      "loss": 0.7749,
+      "step": 5384
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.646140678328929e-05,
+      "loss": 0.7791,
+      "step": 5385
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.645600099848397e-05,
+      "loss": 0.7672,
+      "step": 5386
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.645059430372854e-05,
+      "loss": 20.1393,
+      "step": 5387
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.644518669915798e-05,
+      "loss": 0.7615,
+      "step": 5388
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.643977818490722e-05,
+      "loss": 20.6756,
+      "step": 5389
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.643436876111127e-05,
+      "loss": 24.2089,
+      "step": 5390
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.642895842790518e-05,
+      "loss": 0.7187,
+      "step": 5391
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.642354718542395e-05,
+      "loss": 22.0977,
+      "step": 5392
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.641813503380266e-05,
+      "loss": 20.7633,
+      "step": 5393
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.641272197317642e-05,
+      "loss": 21.1428,
+      "step": 5394
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.640730800368031e-05,
+      "loss": 0.7664,
+      "step": 5395
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.64018931254495e-05,
+      "loss": 19.7291,
+      "step": 5396
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.63964773386191e-05,
+      "loss": 18.1395,
+      "step": 5397
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.639106064332434e-05,
+      "loss": 22.9028,
+      "step": 5398
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.638564303970039e-05,
+      "loss": 0.7558,
+      "step": 5399
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.638022452788248e-05,
+      "loss": 22.0465,
+      "step": 5400
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.637480510800585e-05,
+      "loss": 19.9988,
+      "step": 5401
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.63693847802058e-05,
+      "loss": 19.4935,
+      "step": 5402
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.636396354461757e-05,
+      "loss": 22.5272,
+      "step": 5403
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.635854140137653e-05,
+      "loss": 19.4318,
+      "step": 5404
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.6353118350618e-05,
+      "loss": 0.7731,
+      "step": 5405
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.634769439247732e-05,
+      "loss": 21.0816,
+      "step": 5406
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.63422695270899e-05,
+      "loss": 0.7202,
+      "step": 5407
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.633684375459112e-05,
+      "loss": 0.8027,
+      "step": 5408
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.633141707511644e-05,
+      "loss": 0.7663,
+      "step": 5409
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.632598948880129e-05,
+      "loss": 20.4124,
+      "step": 5410
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.632056099578115e-05,
+      "loss": 23.202,
+      "step": 5411
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.631513159619151e-05,
+      "loss": 0.7756,
+      "step": 5412
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.63097012901679e-05,
+      "loss": 22.5657,
+      "step": 5413
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.630427007784582e-05,
+      "loss": 0.7598,
+      "step": 5414
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.629883795936091e-05,
+      "loss": 25.1297,
+      "step": 5415
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.62934049348487e-05,
+      "loss": 0.7617,
+      "step": 5416
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.628797100444483e-05,
+      "loss": 0.8028,
+      "step": 5417
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.62825361682849e-05,
+      "loss": 0.7566,
+      "step": 5418
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.627710042650458e-05,
+      "loss": 22.5901,
+      "step": 5419
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.627166377923958e-05,
+      "loss": 20.7177,
+      "step": 5420
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.626622622662554e-05,
+      "loss": 0.7561,
+      "step": 5421
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.626078776879822e-05,
+      "loss": 0.7548,
+      "step": 5422
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.625534840589338e-05,
+      "loss": 18.729,
+      "step": 5423
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.624990813804674e-05,
+      "loss": 19.919,
+      "step": 5424
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.624446696539412e-05,
+      "loss": 20.2627,
+      "step": 5425
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.623902488807134e-05,
+      "loss": 0.7461,
+      "step": 5426
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.623358190621423e-05,
+      "loss": 20.1289,
+      "step": 5427
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.622813801995861e-05,
+      "loss": 23.072,
+      "step": 5428
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.62226932294404e-05,
+      "loss": 0.7752,
+      "step": 5429
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.621724753479552e-05,
+      "loss": 20.0556,
+      "step": 5430
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.621180093615986e-05,
+      "loss": 0.7434,
+      "step": 5431
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.620635343366938e-05,
+      "loss": 19.984,
+      "step": 5432
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.620090502746006e-05,
+      "loss": 22.2271,
+      "step": 5433
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.619545571766788e-05,
+      "loss": 0.7396,
+      "step": 5434
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.619000550442886e-05,
+      "loss": 0.768,
+      "step": 5435
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.618455438787903e-05,
+      "loss": 0.7482,
+      "step": 5436
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.617910236815447e-05,
+      "loss": 0.7557,
+      "step": 5437
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.617364944539125e-05,
+      "loss": 21.0497,
+      "step": 5438
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.616819561972549e-05,
+      "loss": 21.5909,
+      "step": 5439
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.61627408912933e-05,
+      "loss": 19.5972,
+      "step": 5440
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.615728526023084e-05,
+      "loss": 21.1336,
+      "step": 5441
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.615182872667429e-05,
+      "loss": 0.7527,
+      "step": 5442
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.614637129075983e-05,
+      "loss": 0.7416,
+      "step": 5443
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.61409129526237e-05,
+      "loss": 20.5203,
+      "step": 5444
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.613545371240212e-05,
+      "loss": 0.7996,
+      "step": 5445
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.612999357023137e-05,
+      "loss": 0.7111,
+      "step": 5446
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.612453252624773e-05,
+      "loss": 23.3299,
+      "step": 5447
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.611907058058752e-05,
+      "loss": 20.7797,
+      "step": 5448
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.611360773338706e-05,
+      "loss": 20.267,
+      "step": 5449
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.610814398478268e-05,
+      "loss": 0.7493,
+      "step": 5450
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.61026793349108e-05,
+      "loss": 0.763,
+      "step": 5451
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.60972137839078e-05,
+      "loss": 0.7866,
+      "step": 5452
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.609174733191011e-05,
+      "loss": 21.6387,
+      "step": 5453
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.608627997905416e-05,
+      "loss": 21.2323,
+      "step": 5454
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.608081172547642e-05,
+      "loss": 21.4638,
+      "step": 5455
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.607534257131338e-05,
+      "loss": 20.8878,
+      "step": 5456
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.606987251670155e-05,
+      "loss": 0.7546,
+      "step": 5457
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.606440156177746e-05,
+      "loss": 0.8091,
+      "step": 5458
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.605892970667769e-05,
+      "loss": 21.9426,
+      "step": 5459
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.605345695153879e-05,
+      "loss": 0.7918,
+      "step": 5460
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.604798329649736e-05,
+      "loss": 0.7895,
+      "step": 5461
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.604250874169007e-05,
+      "loss": 20.8224,
+      "step": 5462
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.603703328725352e-05,
+      "loss": 20.6792,
+      "step": 5463
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.603155693332437e-05,
+      "loss": 0.7617,
+      "step": 5464
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.602607968003935e-05,
+      "loss": 0.7694,
+      "step": 5465
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.602060152753515e-05,
+      "loss": 20.7048,
+      "step": 5466
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.601512247594851e-05,
+      "loss": 19.3747,
+      "step": 5467
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.600964252541621e-05,
+      "loss": 21.1066,
+      "step": 5468
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.600416167607499e-05,
+      "loss": 0.7769,
+      "step": 5469
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.599867992806167e-05,
+      "loss": 0.7886,
+      "step": 5470
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.59931972815131e-05,
+      "loss": 0.7216,
+      "step": 5471
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.59877137365661e-05,
+      "loss": 23.0482,
+      "step": 5472
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.598222929335753e-05,
+      "loss": 17.7328,
+      "step": 5473
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.597674395202431e-05,
+      "loss": 21.2649,
+      "step": 5474
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.597125771270336e-05,
+      "loss": 20.684,
+      "step": 5475
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.596577057553159e-05,
+      "loss": 0.7871,
+      "step": 5476
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.596028254064596e-05,
+      "loss": 22.8092,
+      "step": 5477
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.595479360818349e-05,
+      "loss": 0.7866,
+      "step": 5478
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.594930377828116e-05,
+      "loss": 0.8009,
+      "step": 5479
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.5943813051076e-05,
+      "loss": 0.7701,
+      "step": 5480
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.593832142670507e-05,
+      "loss": 0.7837,
+      "step": 5481
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.59328289053054e-05,
+      "loss": 0.7897,
+      "step": 5482
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.592733548701414e-05,
+      "loss": 0.784,
+      "step": 5483
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.592184117196838e-05,
+      "loss": 19.5336,
+      "step": 5484
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.591634596030524e-05,
+      "loss": 0.7162,
+      "step": 5485
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.591084985216193e-05,
+      "loss": 22.3129,
+      "step": 5486
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.59053528476756e-05,
+      "loss": 19.6419,
+      "step": 5487
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.589985494698347e-05,
+      "loss": 0.7487,
+      "step": 5488
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.589435615022277e-05,
+      "loss": 21.0709,
+      "step": 5489
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.588885645753072e-05,
+      "loss": 20.6344,
+      "step": 5490
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.588335586904463e-05,
+      "loss": 20.893,
+      "step": 5491
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.58778543849018e-05,
+      "loss": 0.7539,
+      "step": 5492
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.587235200523951e-05,
+      "loss": 0.7933,
+      "step": 5493
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.586684873019513e-05,
+      "loss": 21.3243,
+      "step": 5494
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.586134455990602e-05,
+      "loss": 20.4593,
+      "step": 5495
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.585583949450956e-05,
+      "loss": 0.7915,
+      "step": 5496
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.585033353414316e-05,
+      "loss": 0.7705,
+      "step": 5497
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.584482667894424e-05,
+      "loss": 22.3147,
+      "step": 5498
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.583931892905027e-05,
+      "loss": 20.8709,
+      "step": 5499
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.583381028459871e-05,
+      "loss": 25.0456,
+      "step": 5500
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.582830074572707e-05,
+      "loss": 21.693,
+      "step": 5501
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.582279031257288e-05,
+      "loss": 0.7721,
+      "step": 5502
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.581727898527364e-05,
+      "loss": 0.7668,
+      "step": 5503
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.581176676396694e-05,
+      "loss": 0.7361,
+      "step": 5504
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.580625364879038e-05,
+      "loss": 0.6988,
+      "step": 5505
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.580073963988154e-05,
+      "loss": 22.0601,
+      "step": 5506
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.579522473737807e-05,
+      "loss": 19.4575,
+      "step": 5507
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.578970894141762e-05,
+      "loss": 21.0756,
+      "step": 5508
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.578419225213785e-05,
+      "loss": 20.3276,
+      "step": 5509
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.577867466967648e-05,
+      "loss": 0.7744,
+      "step": 5510
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.577315619417121e-05,
+      "loss": 18.3309,
+      "step": 5511
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.576763682575982e-05,
+      "loss": 22.2996,
+      "step": 5512
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.576211656458002e-05,
+      "loss": 0.7538,
+      "step": 5513
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.575659541076964e-05,
+      "loss": 18.9104,
+      "step": 5514
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.575107336446647e-05,
+      "loss": 21.37,
+      "step": 5515
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.574555042580834e-05,
+      "loss": 20.8512,
+      "step": 5516
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.574002659493313e-05,
+      "loss": 19.4389,
+      "step": 5517
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.573450187197867e-05,
+      "loss": 19.3703,
+      "step": 5518
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.57289762570829e-05,
+      "loss": 0.7809,
+      "step": 5519
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.572344975038371e-05,
+      "loss": 0.8042,
+      "step": 5520
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.571792235201908e-05,
+      "loss": 0.818,
+      "step": 5521
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.571239406212693e-05,
+      "loss": 19.8402,
+      "step": 5522
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.570686488084528e-05,
+      "loss": 20.9188,
+      "step": 5523
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.570133480831212e-05,
+      "loss": 21.2993,
+      "step": 5524
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.569580384466551e-05,
+      "loss": 0.727,
+      "step": 5525
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.569027199004349e-05,
+      "loss": 0.7584,
+      "step": 5526
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.568473924458411e-05,
+      "loss": 0.7415,
+      "step": 5527
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.56792056084255e-05,
+      "loss": 0.7992,
+      "step": 5528
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.567367108170577e-05,
+      "loss": 21.2421,
+      "step": 5529
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.566813566456306e-05,
+      "loss": 0.7572,
+      "step": 5530
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.566259935713554e-05,
+      "loss": 19.0747,
+      "step": 5531
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.565706215956142e-05,
+      "loss": 0.7216,
+      "step": 5532
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.565152407197886e-05,
+      "loss": 21.7188,
+      "step": 5533
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.564598509452613e-05,
+      "loss": 0.7422,
+      "step": 5534
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.564044522734147e-05,
+      "loss": 0.7699,
+      "step": 5535
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.563490447056316e-05,
+      "loss": 21.3754,
+      "step": 5536
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.562936282432949e-05,
+      "loss": 0.7874,
+      "step": 5537
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.56238202887788e-05,
+      "loss": 0.8042,
+      "step": 5538
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.561827686404942e-05,
+      "loss": 0.823,
+      "step": 5539
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.561273255027972e-05,
+      "loss": 22.0234,
+      "step": 5540
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.560718734760807e-05,
+      "loss": 0.7449,
+      "step": 5541
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.560164125617291e-05,
+      "loss": 20.4439,
+      "step": 5542
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.559609427611264e-05,
+      "loss": 20.3832,
+      "step": 5543
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.559054640756574e-05,
+      "loss": 0.7864,
+      "step": 5544
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.558499765067068e-05,
+      "loss": 0.7627,
+      "step": 5545
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.557944800556594e-05,
+      "loss": 21.1685,
+      "step": 5546
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.557389747239007e-05,
+      "loss": 20.6524,
+      "step": 5547
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.556834605128159e-05,
+      "loss": 22.9844,
+      "step": 5548
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.556279374237905e-05,
+      "loss": 0.7224,
+      "step": 5549
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.555724054582108e-05,
+      "loss": 20.5488,
+      "step": 5550
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.555168646174627e-05,
+      "loss": 0.7407,
+      "step": 5551
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.554613149029325e-05,
+      "loss": 18.7749,
+      "step": 5552
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.554057563160065e-05,
+      "loss": 0.8406,
+      "step": 5553
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.553501888580718e-05,
+      "loss": 0.7784,
+      "step": 5554
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.552946125305153e-05,
+      "loss": 21.072,
+      "step": 5555
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.552390273347242e-05,
+      "loss": 0.7576,
+      "step": 5556
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.551834332720856e-05,
+      "loss": 0.7126,
+      "step": 5557
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.551278303439877e-05,
+      "loss": 0.7945,
+      "step": 5558
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.550722185518179e-05,
+      "loss": 0.8203,
+      "step": 5559
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.550165978969645e-05,
+      "loss": 22.1451,
+      "step": 5560
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.549609683808158e-05,
+      "loss": 20.5468,
+      "step": 5561
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.549053300047603e-05,
+      "loss": 0.7429,
+      "step": 5562
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.548496827701866e-05,
+      "loss": 21.2354,
+      "step": 5563
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.547940266784838e-05,
+      "loss": 0.7874,
+      "step": 5564
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.547383617310412e-05,
+      "loss": 21.5493,
+      "step": 5565
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.54682687929248e-05,
+      "loss": 18.6182,
+      "step": 5566
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.54627005274494e-05,
+      "loss": 20.101,
+      "step": 5567
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.545713137681689e-05,
+      "loss": 0.7334,
+      "step": 5568
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.545156134116629e-05,
+      "loss": 18.6319,
+      "step": 5569
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.544599042063661e-05,
+      "loss": 0.7539,
+      "step": 5570
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.544041861536693e-05,
+      "loss": 0.7584,
+      "step": 5571
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.543484592549631e-05,
+      "loss": 19.6106,
+      "step": 5572
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.542927235116385e-05,
+      "loss": 20.9771,
+      "step": 5573
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.542369789250866e-05,
+      "loss": 19.1985,
+      "step": 5574
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.541812254966988e-05,
+      "loss": 0.7951,
+      "step": 5575
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.541254632278665e-05,
+      "loss": 0.7158,
+      "step": 5576
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.540696921199822e-05,
+      "loss": 18.28,
+      "step": 5577
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.540139121744373e-05,
+      "loss": 0.7869,
+      "step": 5578
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.539581233926243e-05,
+      "loss": 0.7834,
+      "step": 5579
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.539023257759357e-05,
+      "loss": 0.7843,
+      "step": 5580
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.538465193257644e-05,
+      "loss": 20.6033,
+      "step": 5581
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.537907040435032e-05,
+      "loss": 17.7805,
+      "step": 5582
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.537348799305452e-05,
+      "loss": 0.8203,
+      "step": 5583
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.536790469882836e-05,
+      "loss": 21.7267,
+      "step": 5584
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.536232052181124e-05,
+      "loss": 0.7809,
+      "step": 5585
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.535673546214254e-05,
+      "loss": 20.0833,
+      "step": 5586
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.535114951996164e-05,
+      "loss": 19.1831,
+      "step": 5587
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.534556269540797e-05,
+      "loss": 0.7172,
+      "step": 5588
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.533997498862098e-05,
+      "loss": 0.7653,
+      "step": 5589
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.533438639974015e-05,
+      "loss": 18.7041,
+      "step": 5590
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.532879692890498e-05,
+      "loss": 0.8032,
+      "step": 5591
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.532320657625497e-05,
+      "loss": 0.7455,
+      "step": 5592
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.531761534192963e-05,
+      "loss": 18.8735,
+      "step": 5593
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.531202322606857e-05,
+      "loss": 21.6835,
+      "step": 5594
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.530643022881136e-05,
+      "loss": 0.8074,
+      "step": 5595
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.530083635029757e-05,
+      "loss": 23.6299,
+      "step": 5596
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.529524159066684e-05,
+      "loss": 21.7852,
+      "step": 5597
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.528964595005884e-05,
+      "loss": 0.7885,
+      "step": 5598
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.528404942861321e-05,
+      "loss": 20.9984,
+      "step": 5599
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.527845202646965e-05,
+      "loss": 0.801,
+      "step": 5600
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.527285374376787e-05,
+      "loss": 18.0008,
+      "step": 5601
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.526725458064761e-05,
+      "loss": 20.0037,
+      "step": 5602
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.526165453724863e-05,
+      "loss": 19.984,
+      "step": 5603
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.525605361371068e-05,
+      "loss": 19.8994,
+      "step": 5604
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.52504518101736e-05,
+      "loss": 0.7494,
+      "step": 5605
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.524484912677719e-05,
+      "loss": 20.4857,
+      "step": 5606
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.52392455636613e-05,
+      "loss": 18.7918,
+      "step": 5607
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.52336411209658e-05,
+      "loss": 19.7776,
+      "step": 5608
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.522803579883057e-05,
+      "loss": 20.1425,
+      "step": 5609
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.52224295973955e-05,
+      "loss": 21.8152,
+      "step": 5610
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.521682251680057e-05,
+      "loss": 16.8422,
+      "step": 5611
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.52112145571857e-05,
+      "loss": 21.9991,
+      "step": 5612
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.520560571869087e-05,
+      "loss": 0.7854,
+      "step": 5613
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.51999960014561e-05,
+      "loss": 0.7788,
+      "step": 5614
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.519438540562137e-05,
+      "loss": 20.1794,
+      "step": 5615
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.518877393132677e-05,
+      "loss": 0.7266,
+      "step": 5616
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.518316157871231e-05,
+      "loss": 20.9971,
+      "step": 5617
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.517754834791813e-05,
+      "loss": 19.7652,
+      "step": 5618
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.517193423908429e-05,
+      "loss": 0.8151,
+      "step": 5619
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.516631925235096e-05,
+      "loss": 0.7951,
+      "step": 5620
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.516070338785826e-05,
+      "loss": 18.6797,
+      "step": 5621
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.515508664574637e-05,
+      "loss": 0.7842,
+      "step": 5622
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.51494690261555e-05,
+      "loss": 19.4249,
+      "step": 5623
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.514385052922586e-05,
+      "loss": 0.7616,
+      "step": 5624
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.513823115509766e-05,
+      "loss": 20.4321,
+      "step": 5625
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.513261090391122e-05,
+      "loss": 0.7349,
+      "step": 5626
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.512698977580679e-05,
+      "loss": 21.8621,
+      "step": 5627
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.512136777092465e-05,
+      "loss": 20.2072,
+      "step": 5628
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.511574488940516e-05,
+      "loss": 20.6457,
+      "step": 5629
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.511012113138868e-05,
+      "loss": 0.8235,
+      "step": 5630
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.510449649701554e-05,
+      "loss": 0.7973,
+      "step": 5631
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.509887098642615e-05,
+      "loss": 19.9311,
+      "step": 5632
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.509324459976093e-05,
+      "loss": 17.7875,
+      "step": 5633
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.508761733716033e-05,
+      "loss": 20.1928,
+      "step": 5634
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.508198919876475e-05,
+      "loss": 0.7446,
+      "step": 5635
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.507636018471475e-05,
+      "loss": 0.7495,
+      "step": 5636
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.507073029515078e-05,
+      "loss": 20.0875,
+      "step": 5637
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.506509953021335e-05,
+      "loss": 0.7664,
+      "step": 5638
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.505946789004305e-05,
+      "loss": 0.8072,
+      "step": 5639
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.505383537478043e-05,
+      "loss": 0.7693,
+      "step": 5640
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.504820198456607e-05,
+      "loss": 0.8024,
+      "step": 5641
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.504256771954057e-05,
+      "loss": 0.8132,
+      "step": 5642
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.503693257984461e-05,
+      "loss": 0.7877,
+      "step": 5643
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.503129656561879e-05,
+      "loss": 21.1349,
+      "step": 5644
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.502565967700383e-05,
+      "loss": 19.5162,
+      "step": 5645
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.502002191414038e-05,
+      "loss": 0.7435,
+      "step": 5646
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.50143832771692e-05,
+      "loss": 20.7342,
+      "step": 5647
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.500874376623103e-05,
+      "loss": 0.7994,
+      "step": 5648
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.50031033814666e-05,
+      "loss": 0.8158,
+      "step": 5649
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.499746212301672e-05,
+      "loss": 0.6932,
+      "step": 5650
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.499181999102221e-05,
+      "loss": 21.0069,
+      "step": 5651
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.498617698562387e-05,
+      "loss": 23.1752,
+      "step": 5652
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.498053310696256e-05,
+      "loss": 0.7527,
+      "step": 5653
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.497488835517918e-05,
+      "loss": 19.7128,
+      "step": 5654
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.496924273041458e-05,
+      "loss": 0.7912,
+      "step": 5655
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.49635962328097e-05,
+      "loss": 20.9508,
+      "step": 5656
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.495794886250547e-05,
+      "loss": 0.7487,
+      "step": 5657
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.495230061964288e-05,
+      "loss": 21.3266,
+      "step": 5658
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.494665150436288e-05,
+      "loss": 0.7688,
+      "step": 5659
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.494100151680645e-05,
+      "loss": 23.8623,
+      "step": 5660
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.493535065711466e-05,
+      "loss": 0.779,
+      "step": 5661
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.492969892542855e-05,
+      "loss": 19.9767,
+      "step": 5662
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.492404632188917e-05,
+      "loss": 0.7707,
+      "step": 5663
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.491839284663763e-05,
+      "loss": 21.4047,
+      "step": 5664
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.491273849981501e-05,
+      "loss": 0.7606,
+      "step": 5665
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.490708328156248e-05,
+      "loss": 0.773,
+      "step": 5666
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.490142719202117e-05,
+      "loss": 0.6963,
+      "step": 5667
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.489577023133228e-05,
+      "loss": 0.7395,
+      "step": 5668
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.4890112399637e-05,
+      "loss": 0.7506,
+      "step": 5669
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.488445369707654e-05,
+      "loss": 22.4177,
+      "step": 5670
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.487879412379217e-05,
+      "loss": 20.4276,
+      "step": 5671
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.48731336799251e-05,
+      "loss": 20.3896,
+      "step": 5672
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.486747236561667e-05,
+      "loss": 0.7519,
+      "step": 5673
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.486181018100818e-05,
+      "loss": 22.3496,
+      "step": 5674
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.485614712624092e-05,
+      "loss": 21.5489,
+      "step": 5675
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.48504832014563e-05,
+      "loss": 20.7346,
+      "step": 5676
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.484481840679564e-05,
+      "loss": 20.5022,
+      "step": 5677
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.483915274240037e-05,
+      "loss": 0.7677,
+      "step": 5678
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.483348620841189e-05,
+      "loss": 20.11,
+      "step": 5679
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.482781880497163e-05,
+      "loss": 0.7334,
+      "step": 5680
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.482215053222107e-05,
+      "loss": 0.7877,
+      "step": 5681
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.481648139030167e-05,
+      "loss": 22.0706,
+      "step": 5682
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.481081137935496e-05,
+      "loss": 0.7513,
+      "step": 5683
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.480514049952244e-05,
+      "loss": 0.7252,
+      "step": 5684
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.479946875094566e-05,
+      "loss": 0.7753,
+      "step": 5685
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.479379613376621e-05,
+      "loss": 22.3053,
+      "step": 5686
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.478812264812565e-05,
+      "loss": 0.7302,
+      "step": 5687
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.478244829416561e-05,
+      "loss": 0.7767,
+      "step": 5688
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.477677307202774e-05,
+      "loss": 0.8041,
+      "step": 5689
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.477109698185365e-05,
+      "loss": 21.7984,
+      "step": 5690
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.476542002378505e-05,
+      "loss": 22.2414,
+      "step": 5691
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.475974219796362e-05,
+      "loss": 19.363,
+      "step": 5692
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.475406350453108e-05,
+      "loss": 21.606,
+      "step": 5693
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.47483839436292e-05,
+      "loss": 21.6704,
+      "step": 5694
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.474270351539971e-05,
+      "loss": 21.6671,
+      "step": 5695
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.473702221998442e-05,
+      "loss": 0.7503,
+      "step": 5696
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.47313400575251e-05,
+      "loss": 0.742,
+      "step": 5697
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.472565702816362e-05,
+      "loss": 0.7437,
+      "step": 5698
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.471997313204181e-05,
+      "loss": 21.6092,
+      "step": 5699
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.471428836930156e-05,
+      "loss": 20.6878,
+      "step": 5700
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.470860274008472e-05,
+      "loss": 18.4967,
+      "step": 5701
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.470291624453324e-05,
+      "loss": 20.3544,
+      "step": 5702
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.469722888278904e-05,
+      "loss": 21.189,
+      "step": 5703
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.46915406549941e-05,
+      "loss": 21.9434,
+      "step": 5704
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.468585156129036e-05,
+      "loss": 21.6154,
+      "step": 5705
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.468016160181987e-05,
+      "loss": 20.8339,
+      "step": 5706
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.46744707767246e-05,
+      "loss": 23.3118,
+      "step": 5707
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.466877908614665e-05,
+      "loss": 21.6999,
+      "step": 5708
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.466308653022805e-05,
+      "loss": 0.7578,
+      "step": 5709
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.46573931091109e-05,
+      "loss": 18.1168,
+      "step": 5710
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.465169882293729e-05,
+      "loss": 19.2496,
+      "step": 5711
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.464600367184937e-05,
+      "loss": 0.799,
+      "step": 5712
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.46403076559893e-05,
+      "loss": 23.3361,
+      "step": 5713
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.463461077549923e-05,
+      "loss": 0.7615,
+      "step": 5714
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.462891303052136e-05,
+      "loss": 0.7548,
+      "step": 5715
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.462321442119794e-05,
+      "loss": 21.9625,
+      "step": 5716
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.461751494767117e-05,
+      "loss": 22.3111,
+      "step": 5717
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.461181461008331e-05,
+      "loss": 0.7639,
+      "step": 5718
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.460611340857665e-05,
+      "loss": 21.6854,
+      "step": 5719
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.460041134329352e-05,
+      "loss": 19.7881,
+      "step": 5720
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.459470841437622e-05,
+      "loss": 0.7725,
+      "step": 5721
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.458900462196709e-05,
+      "loss": 23.444,
+      "step": 5722
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.45832999662085e-05,
+      "loss": 20.4378,
+      "step": 5723
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.457759444724285e-05,
+      "loss": 0.7724,
+      "step": 5724
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.457188806521257e-05,
+      "loss": 22.4147,
+      "step": 5725
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.456618082026004e-05,
+      "loss": 18.0397,
+      "step": 5726
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.456047271252776e-05,
+      "loss": 21.5711,
+      "step": 5727
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.455476374215818e-05,
+      "loss": 0.7724,
+      "step": 5728
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.454905390929382e-05,
+      "loss": 21.2873,
+      "step": 5729
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.454334321407719e-05,
+      "loss": 21.6719,
+      "step": 5730
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.45376316566508e-05,
+      "loss": 0.7442,
+      "step": 5731
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.453191923715727e-05,
+      "loss": 19.3745,
+      "step": 5732
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.452620595573913e-05,
+      "loss": 20.1243,
+      "step": 5733
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.4520491812539e-05,
+      "loss": 0.7846,
+      "step": 5734
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.451477680769953e-05,
+      "loss": 20.4599,
+      "step": 5735
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.450906094136335e-05,
+      "loss": 20.903,
+      "step": 5736
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.450334421367312e-05,
+      "loss": 0.7412,
+      "step": 5737
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.449762662477154e-05,
+      "loss": 0.7409,
+      "step": 5738
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.449190817480133e-05,
+      "loss": 0.7694,
+      "step": 5739
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.448618886390522e-05,
+      "loss": 0.7443,
+      "step": 5740
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.448046869222594e-05,
+      "loss": 0.7395,
+      "step": 5741
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.447474765990631e-05,
+      "loss": 0.7844,
+      "step": 5742
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.44690257670891e-05,
+      "loss": 0.7612,
+      "step": 5743
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.446330301391713e-05,
+      "loss": 0.7569,
+      "step": 5744
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.445757940053326e-05,
+      "loss": 0.7527,
+      "step": 5745
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.445185492708034e-05,
+      "loss": 0.7563,
+      "step": 5746
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.444612959370125e-05,
+      "loss": 0.8048,
+      "step": 5747
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.44404034005389e-05,
+      "loss": 0.7378,
+      "step": 5748
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.443467634773621e-05,
+      "loss": 19.6578,
+      "step": 5749
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.442894843543614e-05,
+      "loss": 20.4386,
+      "step": 5750
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.442321966378166e-05,
+      "loss": 0.7889,
+      "step": 5751
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.441749003291576e-05,
+      "loss": 24.3365,
+      "step": 5752
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.441175954298144e-05,
+      "loss": 0.76,
+      "step": 5753
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.440602819412175e-05,
+      "loss": 22.5438,
+      "step": 5754
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.440029598647974e-05,
+      "loss": 0.7819,
+      "step": 5755
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.43945629201985e-05,
+      "loss": 0.725,
+      "step": 5756
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.43888289954211e-05,
+      "loss": 0.8069,
+      "step": 5757
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.438309421229069e-05,
+      "loss": 22.0813,
+      "step": 5758
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.437735857095038e-05,
+      "loss": 21.1816,
+      "step": 5759
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.437162207154336e-05,
+      "loss": 0.7634,
+      "step": 5760
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.43658847142128e-05,
+      "loss": 0.777,
+      "step": 5761
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.436014649910192e-05,
+      "loss": 0.7463,
+      "step": 5762
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.435440742635394e-05,
+      "loss": 0.7558,
+      "step": 5763
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.43486674961121e-05,
+      "loss": 23.0139,
+      "step": 5764
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.434292670851968e-05,
+      "loss": 19.3832,
+      "step": 5765
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.433718506371995e-05,
+      "loss": 20.9227,
+      "step": 5766
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.433144256185625e-05,
+      "loss": 20.6767,
+      "step": 5767
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.432569920307192e-05,
+      "loss": 0.6994,
+      "step": 5768
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.431995498751029e-05,
+      "loss": 21.2999,
+      "step": 5769
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.431420991531475e-05,
+      "loss": 20.9715,
+      "step": 5770
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.430846398662868e-05,
+      "loss": 21.522,
+      "step": 5771
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.430271720159555e-05,
+      "loss": 20.3874,
+      "step": 5772
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.429696956035872e-05,
+      "loss": 0.7917,
+      "step": 5773
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.429122106306173e-05,
+      "loss": 0.8254,
+      "step": 5774
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.428547170984805e-05,
+      "loss": 20.6581,
+      "step": 5775
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.427972150086111e-05,
+      "loss": 19.56,
+      "step": 5776
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.427397043624454e-05,
+      "loss": 0.7385,
+      "step": 5777
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.426821851614184e-05,
+      "loss": 19.9597,
+      "step": 5778
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.426246574069657e-05,
+      "loss": 0.8073,
+      "step": 5779
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.425671211005232e-05,
+      "loss": 0.7611,
+      "step": 5780
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.425095762435273e-05,
+      "loss": 0.8021,
+      "step": 5781
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.424520228374142e-05,
+      "loss": 22.3197,
+      "step": 5782
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.423944608836204e-05,
+      "loss": 21.2409,
+      "step": 5783
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.423368903835826e-05,
+      "loss": 0.77,
+      "step": 5784
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.42279311338738e-05,
+      "loss": 0.7189,
+      "step": 5785
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.422217237505235e-05,
+      "loss": 0.8134,
+      "step": 5786
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.421641276203767e-05,
+      "loss": 23.7295,
+      "step": 5787
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.42106522949735e-05,
+      "loss": 0.7953,
+      "step": 5788
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.420489097400366e-05,
+      "loss": 20.9296,
+      "step": 5789
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.41991287992719e-05,
+      "loss": 0.7551,
+      "step": 5790
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.419336577092211e-05,
+      "loss": 0.7253,
+      "step": 5791
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.418760188909809e-05,
+      "loss": 0.7568,
+      "step": 5792
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.418183715394373e-05,
+      "loss": 21.597,
+      "step": 5793
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.417607156560288e-05,
+      "loss": 19.3218,
+      "step": 5794
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.417030512421951e-05,
+      "loss": 0.7231,
+      "step": 5795
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.416453782993752e-05,
+      "loss": 20.0863,
+      "step": 5796
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.415876968290085e-05,
+      "loss": 0.7798,
+      "step": 5797
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.41530006832535e-05,
+      "loss": 0.7946,
+      "step": 5798
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.414723083113945e-05,
+      "loss": 0.7765,
+      "step": 5799
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.414146012670272e-05,
+      "loss": 0.7606,
+      "step": 5800
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.413568857008735e-05,
+      "loss": 19.1407,
+      "step": 5801
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.41299161614374e-05,
+      "loss": 0.7553,
+      "step": 5802
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.412414290089695e-05,
+      "loss": 0.7706,
+      "step": 5803
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.41183687886101e-05,
+      "loss": 18.326,
+      "step": 5804
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.411259382472099e-05,
+      "loss": 0.7551,
+      "step": 5805
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.410681800937374e-05,
+      "loss": 0.7759,
+      "step": 5806
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.410104134271251e-05,
+      "loss": 0.8001,
+      "step": 5807
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.409526382488151e-05,
+      "loss": 0.7909,
+      "step": 5808
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.408948545602495e-05,
+      "loss": 19.1531,
+      "step": 5809
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.408370623628705e-05,
+      "loss": 0.7171,
+      "step": 5810
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.407792616581205e-05,
+      "loss": 20.742,
+      "step": 5811
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.407214524474423e-05,
+      "loss": 23.1955,
+      "step": 5812
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.406636347322789e-05,
+      "loss": 0.7751,
+      "step": 5813
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.406058085140734e-05,
+      "loss": 0.8231,
+      "step": 5814
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.405479737942692e-05,
+      "loss": 0.6869,
+      "step": 5815
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.404901305743099e-05,
+      "loss": 0.7913,
+      "step": 5816
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.404322788556389e-05,
+      "loss": 0.7641,
+      "step": 5817
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.403744186397006e-05,
+      "loss": 0.7902,
+      "step": 5818
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.403165499279391e-05,
+      "loss": 0.78,
+      "step": 5819
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.402586727217989e-05,
+      "loss": 19.6944,
+      "step": 5820
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.402007870227245e-05,
+      "loss": 0.8024,
+      "step": 5821
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.401428928321607e-05,
+      "loss": 20.5152,
+      "step": 5822
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.400849901515529e-05,
+      "loss": 0.7459,
+      "step": 5823
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.400270789823458e-05,
+      "loss": 21.1494,
+      "step": 5824
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.399691593259852e-05,
+      "loss": 0.7725,
+      "step": 5825
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.399112311839168e-05,
+      "loss": 0.7704,
+      "step": 5826
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.398532945575866e-05,
+      "loss": 0.7753,
+      "step": 5827
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.397953494484403e-05,
+      "loss": 19.1883,
+      "step": 5828
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.397373958579246e-05,
+      "loss": 18.5547,
+      "step": 5829
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.39679433787486e-05,
+      "loss": 0.7912,
+      "step": 5830
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.396214632385713e-05,
+      "loss": 17.6449,
+      "step": 5831
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.39563484212627e-05,
+      "loss": 0.7751,
+      "step": 5832
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.395054967111008e-05,
+      "loss": 20.1385,
+      "step": 5833
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.394475007354399e-05,
+      "loss": 0.7384,
+      "step": 5834
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.393894962870919e-05,
+      "loss": 20.8059,
+      "step": 5835
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.393314833675043e-05,
+      "loss": 0.781,
+      "step": 5836
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.392734619781258e-05,
+      "loss": 0.7184,
+      "step": 5837
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.39215432120404e-05,
+      "loss": 0.737,
+      "step": 5838
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.391573937957875e-05,
+      "loss": 20.7324,
+      "step": 5839
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.390993470057251e-05,
+      "loss": 0.7566,
+      "step": 5840
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.390412917516656e-05,
+      "loss": 0.804,
+      "step": 5841
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.38983228035058e-05,
+      "loss": 22.1824,
+      "step": 5842
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.389251558573516e-05,
+      "loss": 20.9396,
+      "step": 5843
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.388670752199961e-05,
+      "loss": 22.1616,
+      "step": 5844
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.388089861244408e-05,
+      "loss": 0.7435,
+      "step": 5845
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.38750888572136e-05,
+      "loss": 17.7525,
+      "step": 5846
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.386927825645316e-05,
+      "loss": 22.9512,
+      "step": 5847
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.386346681030781e-05,
+      "loss": 18.8058,
+      "step": 5848
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.385765451892259e-05,
+      "loss": 18.4196,
+      "step": 5849
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.385184138244258e-05,
+      "loss": 18.9678,
+      "step": 5850
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.384602740101288e-05,
+      "loss": 18.2717,
+      "step": 5851
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.38402125747786e-05,
+      "loss": 22.0322,
+      "step": 5852
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.38343969038849e-05,
+      "loss": 0.7966,
+      "step": 5853
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.382858038847694e-05,
+      "loss": 19.8941,
+      "step": 5854
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.382276302869987e-05,
+      "loss": 20.5531,
+      "step": 5855
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.381694482469892e-05,
+      "loss": 20.3763,
+      "step": 5856
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.381112577661929e-05,
+      "loss": 0.7891,
+      "step": 5857
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.380530588460626e-05,
+      "loss": 0.7851,
+      "step": 5858
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.379948514880507e-05,
+      "loss": 24.7107,
+      "step": 5859
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.3793663569361e-05,
+      "loss": 0.7878,
+      "step": 5860
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.378784114641939e-05,
+      "loss": 0.704,
+      "step": 5861
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.378201788012555e-05,
+      "loss": 22.2358,
+      "step": 5862
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.377619377062482e-05,
+      "loss": 0.7783,
+      "step": 5863
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.377036881806259e-05,
+      "loss": 0.739,
+      "step": 5864
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.376454302258423e-05,
+      "loss": 21.6577,
+      "step": 5865
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.375871638433519e-05,
+      "loss": 0.7832,
+      "step": 5866
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.375288890346087e-05,
+      "loss": 17.5127,
+      "step": 5867
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.374706058010672e-05,
+      "loss": 19.4039,
+      "step": 5868
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.374123141441825e-05,
+      "loss": 23.0122,
+      "step": 5869
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.373540140654095e-05,
+      "loss": 19.1073,
+      "step": 5870
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.37295705566203e-05,
+      "loss": 19.7405,
+      "step": 5871
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.372373886480189e-05,
+      "loss": 0.7622,
+      "step": 5872
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.371790633123124e-05,
+      "loss": 21.5211,
+      "step": 5873
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.371207295605396e-05,
+      "loss": 22.5408,
+      "step": 5874
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.370623873941563e-05,
+      "loss": 20.7389,
+      "step": 5875
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.37004036814619e-05,
+      "loss": 18.7195,
+      "step": 5876
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.36945677823384e-05,
+      "loss": 22.5005,
+      "step": 5877
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.368873104219079e-05,
+      "loss": 22.3388,
+      "step": 5878
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.368289346116476e-05,
+      "loss": 18.2507,
+      "step": 5879
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.367705503940602e-05,
+      "loss": 22.1591,
+      "step": 5880
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.367121577706029e-05,
+      "loss": 17.9134,
+      "step": 5881
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.366537567427333e-05,
+      "loss": 0.7898,
+      "step": 5882
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.36595347311909e-05,
+      "loss": 20.8562,
+      "step": 5883
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.365369294795881e-05,
+      "loss": 19.9052,
+      "step": 5884
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.364785032472286e-05,
+      "loss": 19.3945,
+      "step": 5885
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.364200686162887e-05,
+      "loss": 0.7654,
+      "step": 5886
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.363616255882272e-05,
+      "loss": 21.0654,
+      "step": 5887
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.363031741645027e-05,
+      "loss": 18.1091,
+      "step": 5888
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.362447143465741e-05,
+      "loss": 0.7687,
+      "step": 5889
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.361862461359007e-05,
+      "loss": 18.1065,
+      "step": 5890
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.361277695339418e-05,
+      "loss": 0.7864,
+      "step": 5891
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.360692845421572e-05,
+      "loss": 0.838,
+      "step": 5892
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.360107911620063e-05,
+      "loss": 0.841,
+      "step": 5893
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.359522893949494e-05,
+      "loss": 17.4602,
+      "step": 5894
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.358937792424468e-05,
+      "loss": 0.7522,
+      "step": 5895
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.358352607059585e-05,
+      "loss": 20.9188,
+      "step": 5896
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.357767337869456e-05,
+      "loss": 22.2365,
+      "step": 5897
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.357181984868687e-05,
+      "loss": 18.4762,
+      "step": 5898
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.356596548071891e-05,
+      "loss": 0.7738,
+      "step": 5899
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.356011027493677e-05,
+      "loss": 0.7803,
+      "step": 5900
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.35542542314866e-05,
+      "loss": 0.7978,
+      "step": 5901
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.354839735051461e-05,
+      "loss": 20.1781,
+      "step": 5902
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.354253963216693e-05,
+      "loss": 19.8257,
+      "step": 5903
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.353668107658984e-05,
+      "loss": 0.6707,
+      "step": 5904
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.353082168392951e-05,
+      "loss": 0.7517,
+      "step": 5905
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.352496145433223e-05,
+      "loss": 0.785,
+      "step": 5906
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.351910038794424e-05,
+      "loss": 0.7504,
+      "step": 5907
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.351323848491186e-05,
+      "loss": 0.7425,
+      "step": 5908
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.35073757453814e-05,
+      "loss": 21.9327,
+      "step": 5909
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.350151216949918e-05,
+      "loss": 21.3066,
+      "step": 5910
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.349564775741156e-05,
+      "loss": 18.6175,
+      "step": 5911
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.348978250926494e-05,
+      "loss": 0.7877,
+      "step": 5912
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.34839164252057e-05,
+      "loss": 0.7817,
+      "step": 5913
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.347804950538025e-05,
+      "loss": 18.8877,
+      "step": 5914
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.347218174993505e-05,
+      "loss": 0.7646,
+      "step": 5915
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.346631315901656e-05,
+      "loss": 0.767,
+      "step": 5916
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.346044373277125e-05,
+      "loss": 0.7413,
+      "step": 5917
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.34545734713456e-05,
+      "loss": 21.7159,
+      "step": 5918
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.344870237488618e-05,
+      "loss": 18.0966,
+      "step": 5919
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.344283044353951e-05,
+      "loss": 20.4938,
+      "step": 5920
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.343695767745214e-05,
+      "loss": 19.6319,
+      "step": 5921
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.343108407677069e-05,
+      "loss": 0.7269,
+      "step": 5922
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.342520964164173e-05,
+      "loss": 0.8001,
+      "step": 5923
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.341933437221192e-05,
+      "loss": 0.771,
+      "step": 5924
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.341345826862789e-05,
+      "loss": 22.4381,
+      "step": 5925
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.340758133103631e-05,
+      "loss": 23.7533,
+      "step": 5926
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.340170355958388e-05,
+      "loss": 0.7751,
+      "step": 5927
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.339582495441729e-05,
+      "loss": 0.8135,
+      "step": 5928
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.338994551568328e-05,
+      "loss": 0.7416,
+      "step": 5929
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.338406524352862e-05,
+      "loss": 0.7527,
+      "step": 5930
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.337818413810007e-05,
+      "loss": 0.7284,
+      "step": 5931
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.337230219954441e-05,
+      "loss": 0.7723,
+      "step": 5932
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.336641942800848e-05,
+      "loss": 20.3779,
+      "step": 5933
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.33605358236391e-05,
+      "loss": 19.4022,
+      "step": 5934
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.335465138658313e-05,
+      "loss": 21.0398,
+      "step": 5935
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.334876611698742e-05,
+      "loss": 19.5948,
+      "step": 5936
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.334288001499893e-05,
+      "loss": 21.8902,
+      "step": 5937
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.333699308076453e-05,
+      "loss": 20.5094,
+      "step": 5938
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.333110531443116e-05,
+      "loss": 0.7667,
+      "step": 5939
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.33252167161458e-05,
+      "loss": 21.7921,
+      "step": 5940
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.331932728605543e-05,
+      "loss": 0.754,
+      "step": 5941
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.331343702430702e-05,
+      "loss": 0.7479,
+      "step": 5942
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.330754593104764e-05,
+      "loss": 22.3214,
+      "step": 5943
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.330165400642429e-05,
+      "loss": 17.5771,
+      "step": 5944
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.329576125058406e-05,
+      "loss": 21.341,
+      "step": 5945
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.328986766367403e-05,
+      "loss": 20.4591,
+      "step": 5946
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.32839732458413e-05,
+      "loss": 20.5644,
+      "step": 5947
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.327807799723299e-05,
+      "loss": 0.7941,
+      "step": 5948
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.327218191799626e-05,
+      "loss": 18.1872,
+      "step": 5949
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.326628500827826e-05,
+      "loss": 0.7314,
+      "step": 5950
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.326038726822622e-05,
+      "loss": 19.5243,
+      "step": 5951
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.32544886979873e-05,
+      "loss": 19.4049,
+      "step": 5952
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.324858929770876e-05,
+      "loss": 0.7285,
+      "step": 5953
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.324268906753782e-05,
+      "loss": 19.7348,
+      "step": 5954
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.323678800762179e-05,
+      "loss": 0.832,
+      "step": 5955
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.323088611810793e-05,
+      "loss": 20.6151,
+      "step": 5956
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.322498339914357e-05,
+      "loss": 19.0424,
+      "step": 5957
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.321907985087605e-05,
+      "loss": 20.4233,
+      "step": 5958
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.321317547345268e-05,
+      "loss": 17.0241,
+      "step": 5959
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.320727026702089e-05,
+      "loss": 0.7509,
+      "step": 5960
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.320136423172804e-05,
+      "loss": 0.761,
+      "step": 5961
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.319545736772158e-05,
+      "loss": 0.783,
+      "step": 5962
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.318954967514891e-05,
+      "loss": 21.5622,
+      "step": 5963
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.31836411541575e-05,
+      "loss": 18.3888,
+      "step": 5964
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.317773180489483e-05,
+      "loss": 0.7375,
+      "step": 5965
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.31718216275084e-05,
+      "loss": 0.8148,
+      "step": 5966
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.316591062214574e-05,
+      "loss": 20.9607,
+      "step": 5967
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.315999878895437e-05,
+      "loss": 0.7126,
+      "step": 5968
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.315408612808185e-05,
+      "loss": 21.0421,
+      "step": 5969
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.314817263967579e-05,
+      "loss": 0.7874,
+      "step": 5970
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.314225832388374e-05,
+      "loss": 19.111,
+      "step": 5971
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.313634318085338e-05,
+      "loss": 0.7316,
+      "step": 5972
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.313042721073232e-05,
+      "loss": 19.5163,
+      "step": 5973
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.312451041366824e-05,
+      "loss": 0.7805,
+      "step": 5974
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.31185927898088e-05,
+      "loss": 0.739,
+      "step": 5975
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.311267433930174e-05,
+      "loss": 0.7614,
+      "step": 5976
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.310675506229475e-05,
+      "loss": 19.1236,
+      "step": 5977
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.310083495893561e-05,
+      "loss": 20.0823,
+      "step": 5978
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.309491402937207e-05,
+      "loss": 19.4759,
+      "step": 5979
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.308899227375191e-05,
+      "loss": 0.8138,
+      "step": 5980
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.308306969222296e-05,
+      "loss": 19.6308,
+      "step": 5981
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.307714628493304e-05,
+      "loss": 20.8018,
+      "step": 5982
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.307122205202998e-05,
+      "loss": 0.7216,
+      "step": 5983
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.306529699366167e-05,
+      "loss": 21.235,
+      "step": 5984
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.3059371109976e-05,
+      "loss": 0.7404,
+      "step": 5985
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.305344440112089e-05,
+      "loss": 0.732,
+      "step": 5986
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.304751686724423e-05,
+      "loss": 20.3602,
+      "step": 5987
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.304158850849403e-05,
+      "loss": 0.775,
+      "step": 5988
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.303565932501823e-05,
+      "loss": 20.6293,
+      "step": 5989
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.302972931696482e-05,
+      "loss": 0.761,
+      "step": 5990
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.302379848448184e-05,
+      "loss": 17.8791,
+      "step": 5991
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.30178668277173e-05,
+      "loss": 20.4416,
+      "step": 5992
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.301193434681927e-05,
+      "loss": 0.8287,
+      "step": 5993
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.300600104193582e-05,
+      "loss": 19.7379,
+      "step": 5994
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.300006691321505e-05,
+      "loss": 0.7697,
+      "step": 5995
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.299413196080507e-05,
+      "loss": 18.2232,
+      "step": 5996
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.298819618485403e-05,
+      "loss": 20.4028,
+      "step": 5997
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.298225958551007e-05,
+      "loss": 22.1073,
+      "step": 5998
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.297632216292138e-05,
+      "loss": 22.6576,
+      "step": 5999
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.297038391723618e-05,
+      "loss": 20.7098,
+      "step": 6000
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.296444484860264e-05,
+      "loss": 0.748,
+      "step": 6001
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.295850495716904e-05,
+      "loss": 0.7726,
+      "step": 6002
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.295256424308364e-05,
+      "loss": 22.8016,
+      "step": 6003
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.294662270649471e-05,
+      "loss": 21.7801,
+      "step": 6004
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.294068034755056e-05,
+      "loss": 20.1981,
+      "step": 6005
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.29347371663995e-05,
+      "loss": 18.1436,
+      "step": 6006
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.292879316318987e-05,
+      "loss": 0.814,
+      "step": 6007
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.292284833807006e-05,
+      "loss": 0.7841,
+      "step": 6008
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.291690269118843e-05,
+      "loss": 19.7064,
+      "step": 6009
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.29109562226934e-05,
+      "loss": 23.6416,
+      "step": 6010
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.290500893273338e-05,
+      "loss": 19.3072,
+      "step": 6011
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.289906082145682e-05,
+      "loss": 20.0528,
+      "step": 6012
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.289311188901221e-05,
+      "loss": 18.999,
+      "step": 6013
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.2887162135548e-05,
+      "loss": 0.6961,
+      "step": 6014
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.288121156121272e-05,
+      "loss": 17.7043,
+      "step": 6015
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.287526016615488e-05,
+      "loss": 21.1567,
+      "step": 6016
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.286930795052306e-05,
+      "loss": 20.2627,
+      "step": 6017
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.28633549144658e-05,
+      "loss": 0.6969,
+      "step": 6018
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.285740105813169e-05,
+      "loss": 0.7923,
+      "step": 6019
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.285144638166935e-05,
+      "loss": 21.1652,
+      "step": 6020
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.284549088522742e-05,
+      "loss": 0.7479,
+      "step": 6021
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.283953456895452e-05,
+      "loss": 0.7437,
+      "step": 6022
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.283357743299935e-05,
+      "loss": 21.5457,
+      "step": 6023
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.282761947751059e-05,
+      "loss": 22.2076,
+      "step": 6024
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.282166070263695e-05,
+      "loss": 0.8082,
+      "step": 6025
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.281570110852715e-05,
+      "loss": 19.6861,
+      "step": 6026
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.280974069532998e-05,
+      "loss": 20.3076,
+      "step": 6027
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.280377946319417e-05,
+      "loss": 0.7834,
+      "step": 6028
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.279781741226856e-05,
+      "loss": 23.0373,
+      "step": 6029
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.279185454270191e-05,
+      "loss": 19.7728,
+      "step": 6030
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.278589085464312e-05,
+      "loss": 18.8282,
+      "step": 6031
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.277992634824098e-05,
+      "loss": 19.6556,
+      "step": 6032
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.27739610236444e-05,
+      "loss": 0.7227,
+      "step": 6033
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.276799488100228e-05,
+      "loss": 20.914,
+      "step": 6034
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.27620279204635e-05,
+      "loss": 21.9636,
+      "step": 6035
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.275606014217705e-05,
+      "loss": 21.5661,
+      "step": 6036
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.275009154629185e-05,
+      "loss": 0.7852,
+      "step": 6037
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.274412213295688e-05,
+      "loss": 19.7188,
+      "step": 6038
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.273815190232115e-05,
+      "loss": 22.4187,
+      "step": 6039
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.273218085453367e-05,
+      "loss": 0.7061,
+      "step": 6040
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.27262089897435e-05,
+      "loss": 0.8034,
+      "step": 6041
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.272023630809967e-05,
+      "loss": 20.2234,
+      "step": 6042
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.271426280975125e-05,
+      "loss": 20.1617,
+      "step": 6043
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.270828849484739e-05,
+      "loss": 22.041,
+      "step": 6044
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.270231336353718e-05,
+      "loss": 0.7231,
+      "step": 6045
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.269633741596974e-05,
+      "loss": 0.8234,
+      "step": 6046
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.269036065229427e-05,
+      "loss": 0.8038,
+      "step": 6047
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.268438307265993e-05,
+      "loss": 0.7346,
+      "step": 6048
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.267840467721593e-05,
+      "loss": 0.771,
+      "step": 6049
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.267242546611146e-05,
+      "loss": 0.8175,
+      "step": 6050
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.266644543949582e-05,
+      "loss": 0.7338,
+      "step": 6051
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.266046459751824e-05,
+      "loss": 22.675,
+      "step": 6052
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.2654482940328e-05,
+      "loss": 20.9969,
+      "step": 6053
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.264850046807441e-05,
+      "loss": 0.7317,
+      "step": 6054
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.26425171809068e-05,
+      "loss": 0.7117,
+      "step": 6055
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.26365330789745e-05,
+      "loss": 0.7613,
+      "step": 6056
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.263054816242688e-05,
+      "loss": 18.2983,
+      "step": 6057
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.262456243141335e-05,
+      "loss": 23.0299,
+      "step": 6058
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.261857588608326e-05,
+      "loss": 0.7077,
+      "step": 6059
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.261258852658608e-05,
+      "loss": 24.495,
+      "step": 6060
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.260660035307125e-05,
+      "loss": 0.707,
+      "step": 6061
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.260061136568823e-05,
+      "loss": 0.7741,
+      "step": 6062
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.25946215645865e-05,
+      "loss": 0.7993,
+      "step": 6063
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.258863094991558e-05,
+      "loss": 0.7708,
+      "step": 6064
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.258263952182499e-05,
+      "loss": 0.776,
+      "step": 6065
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.257664728046428e-05,
+      "loss": 21.2197,
+      "step": 6066
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.2570654225983e-05,
+      "loss": 0.7868,
+      "step": 6067
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.256466035853076e-05,
+      "loss": 22.7494,
+      "step": 6068
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.255866567825719e-05,
+      "loss": 19.5829,
+      "step": 6069
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.255267018531185e-05,
+      "loss": 0.7402,
+      "step": 6070
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.254667387984446e-05,
+      "loss": 0.7688,
+      "step": 6071
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.254067676200464e-05,
+      "loss": 19.1034,
+      "step": 6072
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.253467883194211e-05,
+      "loss": 20.5758,
+      "step": 6073
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.252868008980655e-05,
+      "loss": 18.937,
+      "step": 6074
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.252268053574772e-05,
+      "loss": 0.794,
+      "step": 6075
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.251668016991537e-05,
+      "loss": 0.7549,
+      "step": 6076
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.251067899245923e-05,
+      "loss": 22.4723,
+      "step": 6077
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.250467700352915e-05,
+      "loss": 0.749,
+      "step": 6078
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.249867420327488e-05,
+      "loss": 20.9704,
+      "step": 6079
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.249267059184632e-05,
+      "loss": 0.7468,
+      "step": 6080
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.248666616939324e-05,
+      "loss": 18.7596,
+      "step": 6081
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.248066093606559e-05,
+      "loss": 0.7539,
+      "step": 6082
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.24746548920132e-05,
+      "loss": 20.7837,
+      "step": 6083
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.246864803738602e-05,
+      "loss": 19.1871,
+      "step": 6084
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.246264037233396e-05,
+      "loss": 0.7965,
+      "step": 6085
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.245663189700701e-05,
+      "loss": 0.8164,
+      "step": 6086
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.24506226115551e-05,
+      "loss": 21.9935,
+      "step": 6087
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.244461251612824e-05,
+      "loss": 0.7743,
+      "step": 6088
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.243860161087646e-05,
+      "loss": 0.76,
+      "step": 6089
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.243258989594975e-05,
+      "loss": 17.7949,
+      "step": 6090
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.24265773714982e-05,
+      "loss": 20.0338,
+      "step": 6091
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.242056403767189e-05,
+      "loss": 0.7802,
+      "step": 6092
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.241454989462089e-05,
+      "loss": 0.7442,
+      "step": 6093
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.240853494249532e-05,
+      "loss": 20.2785,
+      "step": 6094
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.240251918144531e-05,
+      "loss": 0.7632,
+      "step": 6095
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.239650261162104e-05,
+      "loss": 20.9068,
+      "step": 6096
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.239048523317266e-05,
+      "loss": 19.9377,
+      "step": 6097
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.238446704625036e-05,
+      "loss": 19.9713,
+      "step": 6098
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.237844805100438e-05,
+      "loss": 21.9236,
+      "step": 6099
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.237242824758495e-05,
+      "loss": 17.9527,
+      "step": 6100
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.236640763614232e-05,
+      "loss": 0.819,
+      "step": 6101
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.236038621682675e-05,
+      "loss": 20.8428,
+      "step": 6102
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.235436398978858e-05,
+      "loss": 19.9293,
+      "step": 6103
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.234834095517808e-05,
+      "loss": 19.4147,
+      "step": 6104
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.234231711314559e-05,
+      "loss": 0.7438,
+      "step": 6105
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.23362924638415e-05,
+      "loss": 0.7508,
+      "step": 6106
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.233026700741616e-05,
+      "loss": 22.5096,
+      "step": 6107
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.232424074401997e-05,
+      "loss": 0.7925,
+      "step": 6108
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.231821367380335e-05,
+      "loss": 17.28,
+      "step": 6109
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.231218579691673e-05,
+      "loss": 0.7141,
+      "step": 6110
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.230615711351059e-05,
+      "loss": 21.4971,
+      "step": 6111
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.230012762373537e-05,
+      "loss": 0.7432,
+      "step": 6112
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.229409732774162e-05,
+      "loss": 0.7957,
+      "step": 6113
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.228806622567982e-05,
+      "loss": 19.6631,
+      "step": 6114
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.228203431770049e-05,
+      "loss": 20.9035,
+      "step": 6115
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.227600160395423e-05,
+      "loss": 21.7256,
+      "step": 6116
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.226996808459162e-05,
+      "loss": 17.931,
+      "step": 6117
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.226393375976322e-05,
+      "loss": 19.76,
+      "step": 6118
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.225789862961966e-05,
+      "loss": 0.7706,
+      "step": 6119
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.22518626943116e-05,
+      "loss": 0.754,
+      "step": 6120
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.224582595398966e-05,
+      "loss": 20.0768,
+      "step": 6121
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.223978840880455e-05,
+      "loss": 18.3881,
+      "step": 6122
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.223375005890696e-05,
+      "loss": 0.7819,
+      "step": 6123
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.222771090444762e-05,
+      "loss": 20.9957,
+      "step": 6124
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.222167094557725e-05,
+      "loss": 0.8353,
+      "step": 6125
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.221563018244663e-05,
+      "loss": 0.7987,
+      "step": 6126
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.220958861520651e-05,
+      "loss": 0.7405,
+      "step": 6127
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.220354624400771e-05,
+      "loss": 0.759,
+      "step": 6128
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.219750306900102e-05,
+      "loss": 0.7453,
+      "step": 6129
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.219145909033732e-05,
+      "loss": 0.7503,
+      "step": 6130
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.218541430816746e-05,
+      "loss": 20.4003,
+      "step": 6131
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.21793687226423e-05,
+      "loss": 20.5687,
+      "step": 6132
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.217332233391275e-05,
+      "loss": 18.7897,
+      "step": 6133
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.216727514212974e-05,
+      "loss": 0.7825,
+      "step": 6134
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.216122714744418e-05,
+      "loss": 0.8033,
+      "step": 6135
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.215517835000706e-05,
+      "loss": 0.7654,
+      "step": 6136
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.214912874996932e-05,
+      "loss": 19.9628,
+      "step": 6137
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.214307834748202e-05,
+      "loss": 19.5546,
+      "step": 6138
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.213702714269612e-05,
+      "loss": 0.7547,
+      "step": 6139
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.21309751357627e-05,
+      "loss": 0.7835,
+      "step": 6140
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.21249223268328e-05,
+      "loss": 0.7123,
+      "step": 6141
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.211886871605749e-05,
+      "loss": 0.8006,
+      "step": 6142
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.21128143035879e-05,
+      "loss": 0.779,
+      "step": 6143
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.210675908957512e-05,
+      "loss": 0.8152,
+      "step": 6144
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.210070307417032e-05,
+      "loss": 20.1129,
+      "step": 6145
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.209464625752464e-05,
+      "loss": 0.7307,
+      "step": 6146
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.208858863978926e-05,
+      "loss": 0.8384,
+      "step": 6147
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.208253022111538e-05,
+      "loss": 21.1396,
+      "step": 6148
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.207647100165421e-05,
+      "loss": 19.8405,
+      "step": 6149
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.2070410981557e-05,
+      "loss": 18.7258,
+      "step": 6150
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.206435016097501e-05,
+      "loss": 20.2304,
+      "step": 6151
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.205828854005953e-05,
+      "loss": 0.7644,
+      "step": 6152
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.205222611896185e-05,
+      "loss": 18.8059,
+      "step": 6153
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.204616289783326e-05,
+      "loss": 22.0911,
+      "step": 6154
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.204009887682515e-05,
+      "loss": 0.7621,
+      "step": 6155
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.203403405608885e-05,
+      "loss": 21.2195,
+      "step": 6156
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.202796843577576e-05,
+      "loss": 0.7813,
+      "step": 6157
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.202190201603725e-05,
+      "loss": 0.7323,
+      "step": 6158
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.201583479702475e-05,
+      "loss": 21.2704,
+      "step": 6159
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.200976677888973e-05,
+      "loss": 0.7692,
+      "step": 6160
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.200369796178359e-05,
+      "loss": 17.9594,
+      "step": 6161
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.199762834585784e-05,
+      "loss": 0.7546,
+      "step": 6162
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.199155793126401e-05,
+      "loss": 19.606,
+      "step": 6163
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.198548671815357e-05,
+      "loss": 19.0009,
+      "step": 6164
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.197941470667809e-05,
+      "loss": 0.7207,
+      "step": 6165
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.197334189698911e-05,
+      "loss": 0.7407,
+      "step": 6166
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.19672682892382e-05,
+      "loss": 21.6349,
+      "step": 6167
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.1961193883577e-05,
+      "loss": 0.7258,
+      "step": 6168
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.195511868015709e-05,
+      "loss": 21.518,
+      "step": 6169
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.194904267913013e-05,
+      "loss": 20.4163,
+      "step": 6170
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.194296588064776e-05,
+      "loss": 18.3307,
+      "step": 6171
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.193688828486166e-05,
+      "loss": 17.5277,
+      "step": 6172
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.193080989192357e-05,
+      "loss": 18.6282,
+      "step": 6173
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.192473070198514e-05,
+      "loss": 18.0636,
+      "step": 6174
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.191865071519815e-05,
+      "loss": 0.7837,
+      "step": 6175
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.191256993171437e-05,
+      "loss": 0.7831,
+      "step": 6176
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.190648835168553e-05,
+      "loss": 19.7261,
+      "step": 6177
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.190040597526347e-05,
+      "loss": 18.7484,
+      "step": 6178
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.189432280259998e-05,
+      "loss": 21.6376,
+      "step": 6179
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.188823883384693e-05,
+      "loss": 19.477,
+      "step": 6180
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.188215406915614e-05,
+      "loss": 19.7269,
+      "step": 6181
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.187606850867952e-05,
+      "loss": 20.5708,
+      "step": 6182
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.186998215256896e-05,
+      "loss": 20.4374,
+      "step": 6183
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.186389500097634e-05,
+      "loss": 0.7354,
+      "step": 6184
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.185780705405363e-05,
+      "loss": 0.7578,
+      "step": 6185
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.185171831195282e-05,
+      "loss": 20.7525,
+      "step": 6186
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.184562877482581e-05,
+      "loss": 19.3118,
+      "step": 6187
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.183953844282464e-05,
+      "loss": 18.7748,
+      "step": 6188
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.183344731610132e-05,
+      "loss": 19.2725,
+      "step": 6189
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.18273553948079e-05,
+      "loss": 18.8032,
+      "step": 6190
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.182126267909641e-05,
+      "loss": 18.4481,
+      "step": 6191
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.181516916911894e-05,
+      "loss": 19.8642,
+      "step": 6192
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.180907486502759e-05,
+      "loss": 0.7954,
+      "step": 6193
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.180297976697446e-05,
+      "loss": 0.7558,
+      "step": 6194
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.179688387511169e-05,
+      "loss": 0.7524,
+      "step": 6195
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.179078718959145e-05,
+      "loss": 0.7926,
+      "step": 6196
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.17846897105659e-05,
+      "loss": 20.5172,
+      "step": 6197
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.177859143818723e-05,
+      "loss": 0.686,
+      "step": 6198
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.177249237260767e-05,
+      "loss": 19.9962,
+      "step": 6199
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.176639251397945e-05,
+      "loss": 0.7184,
+      "step": 6200
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.176029186245482e-05,
+      "loss": 0.7626,
+      "step": 6201
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.175419041818604e-05,
+      "loss": 19.316,
+      "step": 6202
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.174808818132542e-05,
+      "loss": 20.6726,
+      "step": 6203
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.174198515202527e-05,
+      "loss": 18.3796,
+      "step": 6204
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.173588133043793e-05,
+      "loss": 0.8147,
+      "step": 6205
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.172977671671573e-05,
+      "loss": 22.1038,
+      "step": 6206
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.172367131101108e-05,
+      "loss": 19.2625,
+      "step": 6207
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.171756511347635e-05,
+      "loss": 21.3743,
+      "step": 6208
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.171145812426392e-05,
+      "loss": 19.0623,
+      "step": 6209
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.170535034352629e-05,
+      "loss": 0.7412,
+      "step": 6210
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.169924177141585e-05,
+      "loss": 19.3392,
+      "step": 6211
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.169313240808511e-05,
+      "loss": 0.8209,
+      "step": 6212
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.168702225368656e-05,
+      "loss": 0.7913,
+      "step": 6213
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.16809113083727e-05,
+      "loss": 17.9623,
+      "step": 6214
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.167479957229605e-05,
+      "loss": 19.5901,
+      "step": 6215
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.166868704560916e-05,
+      "loss": 19.8551,
+      "step": 6216
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.166257372846462e-05,
+      "loss": 0.7648,
+      "step": 6217
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.165645962101501e-05,
+      "loss": 0.7204,
+      "step": 6218
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.165034472341295e-05,
+      "loss": 19.9892,
+      "step": 6219
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.164422903581103e-05,
+      "loss": 0.7498,
+      "step": 6220
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.163811255836196e-05,
+      "loss": 0.75,
+      "step": 6221
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.163199529121837e-05,
+      "loss": 20.2815,
+      "step": 6222
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.162587723453294e-05,
+      "loss": 19.1,
+      "step": 6223
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.16197583884584e-05,
+      "loss": 20.0474,
+      "step": 6224
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.161363875314747e-05,
+      "loss": 0.7949,
+      "step": 6225
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.16075183287529e-05,
+      "loss": 0.751,
+      "step": 6226
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.160139711542746e-05,
+      "loss": 20.5165,
+      "step": 6227
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.159527511332393e-05,
+      "loss": 0.7548,
+      "step": 6228
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.158915232259512e-05,
+      "loss": 0.7622,
+      "step": 6229
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.158302874339385e-05,
+      "loss": 23.5576,
+      "step": 6230
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.157690437587297e-05,
+      "loss": 19.4587,
+      "step": 6231
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.157077922018537e-05,
+      "loss": 0.7978,
+      "step": 6232
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.156465327648389e-05,
+      "loss": 18.8842,
+      "step": 6233
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.155852654492145e-05,
+      "loss": 0.7693,
+      "step": 6234
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.155239902565099e-05,
+      "loss": 21.8697,
+      "step": 6235
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.154627071882546e-05,
+      "loss": 17.4916,
+      "step": 6236
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.154014162459779e-05,
+      "loss": 0.7853,
+      "step": 6237
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.153401174312098e-05,
+      "loss": 0.7273,
+      "step": 6238
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.152788107454806e-05,
+      "loss": 0.8262,
+      "step": 6239
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.152174961903202e-05,
+      "loss": 20.5003,
+      "step": 6240
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.151561737672591e-05,
+      "loss": 19.5548,
+      "step": 6241
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.15094843477828e-05,
+      "loss": 0.7032,
+      "step": 6242
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.150335053235575e-05,
+      "loss": 0.7126,
+      "step": 6243
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.149721593059789e-05,
+      "loss": 17.0249,
+      "step": 6244
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.149108054266233e-05,
+      "loss": 0.7151,
+      "step": 6245
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.148494436870221e-05,
+      "loss": 0.7803,
+      "step": 6246
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.147880740887069e-05,
+      "loss": 0.7843,
+      "step": 6247
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.147266966332094e-05,
+      "loss": 20.2413,
+      "step": 6248
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.146653113220616e-05,
+      "loss": 0.7813,
+      "step": 6249
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.14603918156796e-05,
+      "loss": 18.812,
+      "step": 6250
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.145425171389448e-05,
+      "loss": 19.2402,
+      "step": 6251
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.144811082700404e-05,
+      "loss": 19.9521,
+      "step": 6252
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.144196915516158e-05,
+      "loss": 23.5727,
+      "step": 6253
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.143582669852039e-05,
+      "loss": 20.0766,
+      "step": 6254
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.142968345723378e-05,
+      "loss": 0.7624,
+      "step": 6255
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.142353943145511e-05,
+      "loss": 18.4545,
+      "step": 6256
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.14173946213377e-05,
+      "loss": 0.7432,
+      "step": 6257
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.141124902703497e-05,
+      "loss": 0.7565,
+      "step": 6258
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.140510264870027e-05,
+      "loss": 0.7846,
+      "step": 6259
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.139895548648706e-05,
+      "loss": 0.7227,
+      "step": 6260
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.139280754054872e-05,
+      "loss": 0.7716,
+      "step": 6261
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.138665881103877e-05,
+      "loss": 19.8542,
+      "step": 6262
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.138050929811063e-05,
+      "loss": 0.7475,
+      "step": 6263
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.137435900191782e-05,
+      "loss": 0.7694,
+      "step": 6264
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.136820792261384e-05,
+      "loss": 19.7405,
+      "step": 6265
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.136205606035223e-05,
+      "loss": 0.8179,
+      "step": 6266
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.135590341528655e-05,
+      "loss": 0.7247,
+      "step": 6267
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.134974998757035e-05,
+      "loss": 20.8198,
+      "step": 6268
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.134359577735724e-05,
+      "loss": 19.4967,
+      "step": 6269
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.133744078480081e-05,
+      "loss": 21.4145,
+      "step": 6270
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.133128501005471e-05,
+      "loss": 0.7717,
+      "step": 6271
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.13251284532726e-05,
+      "loss": 0.7693,
+      "step": 6272
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.13189711146081e-05,
+      "loss": 21.624,
+      "step": 6273
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.131281299421494e-05,
+      "loss": 0.7796,
+      "step": 6274
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.130665409224682e-05,
+      "loss": 19.301,
+      "step": 6275
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.130049440885748e-05,
+      "loss": 0.7011,
+      "step": 6276
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.129433394420063e-05,
+      "loss": 0.8218,
+      "step": 6277
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.128817269843007e-05,
+      "loss": 19.0376,
+      "step": 6278
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.128201067169958e-05,
+      "loss": 21.996,
+      "step": 6279
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.127584786416296e-05,
+      "loss": 0.7533,
+      "step": 6280
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.126968427597403e-05,
+      "loss": 23.0188,
+      "step": 6281
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.126351990728664e-05,
+      "loss": 19.958,
+      "step": 6282
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.125735475825467e-05,
+      "loss": 21.4985,
+      "step": 6283
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.125118882903197e-05,
+      "loss": 0.7527,
+      "step": 6284
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.124502211977246e-05,
+      "loss": 0.7257,
+      "step": 6285
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.123885463063008e-05,
+      "loss": 18.3498,
+      "step": 6286
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.123268636175874e-05,
+      "loss": 0.721,
+      "step": 6287
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.122651731331244e-05,
+      "loss": 0.7407,
+      "step": 6288
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.12203474854451e-05,
+      "loss": 0.7211,
+      "step": 6289
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.121417687831078e-05,
+      "loss": 0.809,
+      "step": 6290
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.120800549206348e-05,
+      "loss": 21.7165,
+      "step": 6291
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.120183332685725e-05,
+      "loss": 19.5009,
+      "step": 6292
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.119566038284612e-05,
+      "loss": 0.7095,
+      "step": 6293
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.118948666018418e-05,
+      "loss": 19.8342,
+      "step": 6294
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.118331215902553e-05,
+      "loss": 0.7581,
+      "step": 6295
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.11771368795243e-05,
+      "loss": 0.7736,
+      "step": 6296
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.117096082183461e-05,
+      "loss": 0.7535,
+      "step": 6297
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.116478398611062e-05,
+      "loss": 22.3152,
+      "step": 6298
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.11586063725065e-05,
+      "loss": 20.1549,
+      "step": 6299
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.115242798117646e-05,
+      "loss": 20.0339,
+      "step": 6300
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.114624881227471e-05,
+      "loss": 20.0665,
+      "step": 6301
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.114006886595547e-05,
+      "loss": 0.7664,
+      "step": 6302
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.1133888142373e-05,
+      "loss": 20.6178,
+      "step": 6303
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.112770664168157e-05,
+      "loss": 19.1105,
+      "step": 6304
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.112152436403547e-05,
+      "loss": 18.3499,
+      "step": 6305
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.111534130958902e-05,
+      "loss": 23.4211,
+      "step": 6306
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.110915747849656e-05,
+      "loss": 0.7526,
+      "step": 6307
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.110297287091242e-05,
+      "loss": 19.722,
+      "step": 6308
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.109678748699096e-05,
+      "loss": 0.7987,
+      "step": 6309
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.10906013268866e-05,
+      "loss": 0.7183,
+      "step": 6310
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.108441439075374e-05,
+      "loss": 19.818,
+      "step": 6311
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.107822667874678e-05,
+      "loss": 0.7886,
+      "step": 6312
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.107203819102018e-05,
+      "loss": 20.3165,
+      "step": 6313
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.106584892772844e-05,
+      "loss": 20.1253,
+      "step": 6314
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.1059658889026e-05,
+      "loss": 0.7412,
+      "step": 6315
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.105346807506738e-05,
+      "loss": 19.3926,
+      "step": 6316
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.104727648600712e-05,
+      "loss": 0.7419,
+      "step": 6317
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.104108412199973e-05,
+      "loss": 20.4664,
+      "step": 6318
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.10348909831998e-05,
+      "loss": 19.2346,
+      "step": 6319
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.102869706976192e-05,
+      "loss": 17.1388,
+      "step": 6320
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.102250238184067e-05,
+      "loss": 0.7398,
+      "step": 6321
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.101630691959067e-05,
+      "loss": 0.7833,
+      "step": 6322
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.101011068316657e-05,
+      "loss": 18.774,
+      "step": 6323
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.100391367272303e-05,
+      "loss": 0.7681,
+      "step": 6324
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.099771588841472e-05,
+      "loss": 0.7231,
+      "step": 6325
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.099151733039633e-05,
+      "loss": 0.8051,
+      "step": 6326
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.098531799882263e-05,
+      "loss": 22.1765,
+      "step": 6327
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.097911789384829e-05,
+      "loss": 20.4081,
+      "step": 6328
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.09729170156281e-05,
+      "loss": 0.7512,
+      "step": 6329
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.096671536431683e-05,
+      "loss": 19.7317,
+      "step": 6330
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.096051294006928e-05,
+      "loss": 21.4364,
+      "step": 6331
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.095430974304023e-05,
+      "loss": 0.8145,
+      "step": 6332
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.094810577338457e-05,
+      "loss": 0.735,
+      "step": 6333
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.09419010312571e-05,
+      "loss": 0.7181,
+      "step": 6334
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.093569551681275e-05,
+      "loss": 0.7035,
+      "step": 6335
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.092948923020635e-05,
+      "loss": 0.7717,
+      "step": 6336
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.092328217159284e-05,
+      "loss": 0.7476,
+      "step": 6337
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.091707434112716e-05,
+      "loss": 18.9501,
+      "step": 6338
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.091086573896422e-05,
+      "loss": 0.7848,
+      "step": 6339
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.090465636525903e-05,
+      "loss": 19.3732,
+      "step": 6340
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.089844622016655e-05,
+      "loss": 21.0696,
+      "step": 6341
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.089223530384182e-05,
+      "loss": 20.1961,
+      "step": 6342
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.088602361643982e-05,
+      "loss": 0.7816,
+      "step": 6343
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.087981115811562e-05,
+      "loss": 20.8717,
+      "step": 6344
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.087359792902431e-05,
+      "loss": 19.8955,
+      "step": 6345
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.086738392932092e-05,
+      "loss": 0.7901,
+      "step": 6346
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.08611691591606e-05,
+      "loss": 0.8233,
+      "step": 6347
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.085495361869842e-05,
+      "loss": 0.8218,
+      "step": 6348
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.084873730808958e-05,
+      "loss": 21.8033,
+      "step": 6349
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.084252022748919e-05,
+      "loss": 0.7229,
+      "step": 6350
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.083630237705247e-05,
+      "loss": 20.0343,
+      "step": 6351
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.08300837569346e-05,
+      "loss": 22.2055,
+      "step": 6352
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.082386436729079e-05,
+      "loss": 19.9879,
+      "step": 6353
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.08176442082763e-05,
+      "loss": 20.6575,
+      "step": 6354
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.081142328004637e-05,
+      "loss": 0.756,
+      "step": 6355
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.080520158275629e-05,
+      "loss": 20.0317,
+      "step": 6356
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.079897911656131e-05,
+      "loss": 0.7075,
+      "step": 6357
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.079275588161681e-05,
+      "loss": 0.7146,
+      "step": 6358
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.078653187807809e-05,
+      "loss": 0.7315,
+      "step": 6359
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.07803071061005e-05,
+      "loss": 18.9751,
+      "step": 6360
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.07740815658394e-05,
+      "loss": 0.7714,
+      "step": 6361
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.076785525745024e-05,
+      "loss": 18.5047,
+      "step": 6362
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.076162818108835e-05,
+      "loss": 20.1452,
+      "step": 6363
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.075540033690921e-05,
+      "loss": 19.6772,
+      "step": 6364
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.074917172506824e-05,
+      "loss": 20.6122,
+      "step": 6365
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.074294234572095e-05,
+      "loss": 0.739,
+      "step": 6366
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.073671219902277e-05,
+      "loss": 0.7722,
+      "step": 6367
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.073048128512926e-05,
+      "loss": 0.6921,
+      "step": 6368
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.072424960419591e-05,
+      "loss": 18.6092,
+      "step": 6369
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.071801715637828e-05,
+      "loss": 18.9709,
+      "step": 6370
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.071178394183192e-05,
+      "loss": 22.4491,
+      "step": 6371
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.070554996071242e-05,
+      "loss": 0.7327,
+      "step": 6372
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.06993152131754e-05,
+      "loss": 21.7908,
+      "step": 6373
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.069307969937643e-05,
+      "loss": 0.7615,
+      "step": 6374
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.06868434194712e-05,
+      "loss": 21.3985,
+      "step": 6375
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.068060637361536e-05,
+      "loss": 19.6578,
+      "step": 6376
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.067436856196458e-05,
+      "loss": 20.9122,
+      "step": 6377
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.066812998467455e-05,
+      "loss": 18.2254,
+      "step": 6378
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.066189064190099e-05,
+      "loss": 20.7007,
+      "step": 6379
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.065565053379965e-05,
+      "loss": 19.6454,
+      "step": 6380
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.064940966052625e-05,
+      "loss": 0.7838,
+      "step": 6381
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.06431680222366e-05,
+      "loss": 22.2218,
+      "step": 6382
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.063692561908648e-05,
+      "loss": 19.4721,
+      "step": 6383
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.063068245123171e-05,
+      "loss": 21.4607,
+      "step": 6384
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.06244385188281e-05,
+      "loss": 0.7862,
+      "step": 6385
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.061819382203151e-05,
+      "loss": 0.7394,
+      "step": 6386
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.061194836099781e-05,
+      "loss": 21.1538,
+      "step": 6387
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.06057021358829e-05,
+      "loss": 21.7623,
+      "step": 6388
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.059945514684268e-05,
+      "loss": 17.4084,
+      "step": 6389
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.059320739403306e-05,
+      "loss": 19.1379,
+      "step": 6390
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.058695887761001e-05,
+      "loss": 20.3787,
+      "step": 6391
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.058070959772947e-05,
+      "loss": 20.3603,
+      "step": 6392
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.057445955454744e-05,
+      "loss": 20.6489,
+      "step": 6393
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.056820874821993e-05,
+      "loss": 0.695,
+      "step": 6394
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.056195717890294e-05,
+      "loss": 21.6386,
+      "step": 6395
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.055570484675251e-05,
+      "loss": 17.742,
+      "step": 6396
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.054945175192472e-05,
+      "loss": 20.9569,
+      "step": 6397
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.054319789457565e-05,
+      "loss": 21.3552,
+      "step": 6398
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.053694327486137e-05,
+      "loss": 18.6846,
+      "step": 6399
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.053068789293802e-05,
+      "loss": 0.7395,
+      "step": 6400
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.052443174896174e-05,
+      "loss": 19.3815,
+      "step": 6401
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.051817484308866e-05,
+      "loss": 0.7934,
+      "step": 6402
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.051191717547496e-05,
+      "loss": 20.3979,
+      "step": 6403
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.050565874627686e-05,
+      "loss": 0.7679,
+      "step": 6404
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.049939955565053e-05,
+      "loss": 0.7158,
+      "step": 6405
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.049313960375224e-05,
+      "loss": 22.4685,
+      "step": 6406
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.048687889073822e-05,
+      "loss": 19.4952,
+      "step": 6407
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.048061741676476e-05,
+      "loss": 0.7702,
+      "step": 6408
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.04743551819881e-05,
+      "loss": 0.7639,
+      "step": 6409
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.046809218656459e-05,
+      "loss": 0.7718,
+      "step": 6410
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.046182843065054e-05,
+      "loss": 0.7286,
+      "step": 6411
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.04555639144023e-05,
+      "loss": 22.8436,
+      "step": 6412
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.044929863797622e-05,
+      "loss": 0.7736,
+      "step": 6413
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.04430326015287e-05,
+      "loss": 0.776,
+      "step": 6414
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.043676580521614e-05,
+      "loss": 19.9992,
+      "step": 6415
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.043049824919495e-05,
+      "loss": 0.7624,
+      "step": 6416
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.042422993362157e-05,
+      "loss": 21.1812,
+      "step": 6417
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.041796085865249e-05,
+      "loss": 0.7857,
+      "step": 6418
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.041169102444414e-05,
+      "loss": 20.5535,
+      "step": 6419
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.040542043115304e-05,
+      "loss": 21.4636,
+      "step": 6420
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.03991490789357e-05,
+      "loss": 0.7711,
+      "step": 6421
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.039287696794867e-05,
+      "loss": 20.6917,
+      "step": 6422
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.038660409834848e-05,
+      "loss": 19.0132,
+      "step": 6423
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.038033047029172e-05,
+      "loss": 19.1007,
+      "step": 6424
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.037405608393498e-05,
+      "loss": 21.1081,
+      "step": 6425
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.036778093943486e-05,
+      "loss": 19.4477,
+      "step": 6426
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.036150503694798e-05,
+      "loss": 18.3038,
+      "step": 6427
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.035522837663101e-05,
+      "loss": 0.6757,
+      "step": 6428
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.034895095864062e-05,
+      "loss": 0.8209,
+      "step": 6429
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.034267278313348e-05,
+      "loss": 0.7785,
+      "step": 6430
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.03363938502663e-05,
+      "loss": 0.7876,
+      "step": 6431
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.033011416019581e-05,
+      "loss": 0.7867,
+      "step": 6432
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.032383371307873e-05,
+      "loss": 20.2977,
+      "step": 6433
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.031755250907183e-05,
+      "loss": 22.0845,
+      "step": 6434
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.03112705483319e-05,
+      "loss": 0.7489,
+      "step": 6435
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.030498783101574e-05,
+      "loss": 18.9908,
+      "step": 6436
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.029870435728018e-05,
+      "loss": 19.4525,
+      "step": 6437
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.029242012728203e-05,
+      "loss": 19.4579,
+      "step": 6438
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.028613514117815e-05,
+      "loss": 20.9375,
+      "step": 6439
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.027984939912543e-05,
+      "loss": 19.2893,
+      "step": 6440
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.027356290128076e-05,
+      "loss": 18.7318,
+      "step": 6441
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.026727564780103e-05,
+      "loss": 19.1191,
+      "step": 6442
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.026098763884319e-05,
+      "loss": 0.7129,
+      "step": 6443
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.02546988745642e-05,
+      "loss": 0.7816,
+      "step": 6444
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.024840935512103e-05,
+      "loss": 20.6663,
+      "step": 6445
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.024211908067065e-05,
+      "loss": 23.7801,
+      "step": 6446
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.023582805137006e-05,
+      "loss": 0.8037,
+      "step": 6447
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.022953626737631e-05,
+      "loss": 0.787,
+      "step": 6448
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.022324372884643e-05,
+      "loss": 0.7819,
+      "step": 6449
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.021695043593746e-05,
+      "loss": 0.7315,
+      "step": 6450
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.021065638880654e-05,
+      "loss": 20.9104,
+      "step": 6451
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.020436158761073e-05,
+      "loss": 0.8059,
+      "step": 6452
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.019806603250715e-05,
+      "loss": 0.7692,
+      "step": 6453
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.019176972365297e-05,
+      "loss": 0.7831,
+      "step": 6454
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.018547266120528e-05,
+      "loss": 0.7764,
+      "step": 6455
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.017917484532133e-05,
+      "loss": 21.3789,
+      "step": 6456
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.017287627615829e-05,
+      "loss": 21.0224,
+      "step": 6457
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.016657695387335e-05,
+      "loss": 20.026,
+      "step": 6458
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.016027687862378e-05,
+      "loss": 22.3735,
+      "step": 6459
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.01539760505668e-05,
+      "loss": 21.2195,
+      "step": 6460
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.014767446985971e-05,
+      "loss": 18.5293,
+      "step": 6461
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.014137213665976e-05,
+      "loss": 0.8228,
+      "step": 6462
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.013506905112429e-05,
+      "loss": 19.0158,
+      "step": 6463
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.012876521341061e-05,
+      "loss": 19.5103,
+      "step": 6464
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.012246062367607e-05,
+      "loss": 19.1382,
+      "step": 6465
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.011615528207805e-05,
+      "loss": 19.7612,
+      "step": 6466
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.010984918877391e-05,
+      "loss": 19.4749,
+      "step": 6467
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.010354234392105e-05,
+      "loss": 18.8013,
+      "step": 6468
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.009723474767692e-05,
+      "loss": 0.7202,
+      "step": 6469
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.009092640019892e-05,
+      "loss": 19.323,
+      "step": 6470
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.008461730164454e-05,
+      "loss": 17.9481,
+      "step": 6471
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.007830745217121e-05,
+      "loss": 19.2027,
+      "step": 6472
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.007199685193649e-05,
+      "loss": 0.709,
+      "step": 6473
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.006568550109786e-05,
+      "loss": 19.0187,
+      "step": 6474
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.005937339981286e-05,
+      "loss": 0.7934,
+      "step": 6475
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.005306054823902e-05,
+      "loss": 0.8303,
+      "step": 6476
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.004674694653393e-05,
+      "loss": 21.3004,
+      "step": 6477
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.004043259485519e-05,
+      "loss": 19.9925,
+      "step": 6478
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.003411749336038e-05,
+      "loss": 19.1089,
+      "step": 6479
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.002780164220714e-05,
+      "loss": 0.7752,
+      "step": 6480
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.002148504155312e-05,
+      "loss": 0.7625,
+      "step": 6481
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.001516769155597e-05,
+      "loss": 22.1286,
+      "step": 6482
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.000884959237339e-05,
+      "loss": 0.7795,
+      "step": 6483
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.000253074416307e-05,
+      "loss": 18.4848,
+      "step": 6484
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.999621114708273e-05,
+      "loss": 0.7761,
+      "step": 6485
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.99898908012901e-05,
+      "loss": 19.2184,
+      "step": 6486
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.998356970694297e-05,
+      "loss": 0.8006,
+      "step": 6487
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.997724786419908e-05,
+      "loss": 20.1497,
+      "step": 6488
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.997092527321624e-05,
+      "loss": 23.8647,
+      "step": 6489
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.996460193415227e-05,
+      "loss": 0.7556,
+      "step": 6490
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.995827784716499e-05,
+      "loss": 0.7816,
+      "step": 6491
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.995195301241225e-05,
+      "loss": 22.9108,
+      "step": 6492
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.994562743005192e-05,
+      "loss": 18.1416,
+      "step": 6493
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.99393011002419e-05,
+      "loss": 0.7693,
+      "step": 6494
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.993297402314008e-05,
+      "loss": 0.7001,
+      "step": 6495
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.992664619890441e-05,
+      "loss": 19.6984,
+      "step": 6496
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.99203176276928e-05,
+      "loss": 23.0849,
+      "step": 6497
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.991398830966324e-05,
+      "loss": 0.8073,
+      "step": 6498
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.99076582449737e-05,
+      "loss": 0.8192,
+      "step": 6499
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.990132743378217e-05,
+      "loss": 0.7307,
+      "step": 6500
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.989499587624672e-05,
+      "loss": 22.2266,
+      "step": 6501
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.988866357252532e-05,
+      "loss": 20.441,
+      "step": 6502
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.988233052277604e-05,
+      "loss": 0.7363,
+      "step": 6503
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.987599672715698e-05,
+      "loss": 0.7353,
+      "step": 6504
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.986966218582621e-05,
+      "loss": 0.71,
+      "step": 6505
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.986332689894187e-05,
+      "loss": 22.4949,
+      "step": 6506
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.985699086666207e-05,
+      "loss": 19.1744,
+      "step": 6507
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.985065408914494e-05,
+      "loss": 0.7585,
+      "step": 6508
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.984431656654868e-05,
+      "loss": 0.7653,
+      "step": 6509
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.983797829903147e-05,
+      "loss": 19.6546,
+      "step": 6510
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.983163928675151e-05,
+      "loss": 21.3685,
+      "step": 6511
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.9825299529867e-05,
+      "loss": 0.7136,
+      "step": 6512
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.981895902853622e-05,
+      "loss": 20.869,
+      "step": 6513
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.981261778291742e-05,
+      "loss": 19.9122,
+      "step": 6514
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.980627579316886e-05,
+      "loss": 0.7808,
+      "step": 6515
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.979993305944886e-05,
+      "loss": 0.7259,
+      "step": 6516
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.979358958191573e-05,
+      "loss": 0.7913,
+      "step": 6517
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.978724536072779e-05,
+      "loss": 0.7405,
+      "step": 6518
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.978090039604341e-05,
+      "loss": 0.852,
+      "step": 6519
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.977455468802095e-05,
+      "loss": 0.764,
+      "step": 6520
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.976820823681883e-05,
+      "loss": 20.9,
+      "step": 6521
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.976186104259542e-05,
+      "loss": 20.0039,
+      "step": 6522
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.975551310550916e-05,
+      "loss": 0.7706,
+      "step": 6523
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.97491644257185e-05,
+      "loss": 22.1235,
+      "step": 6524
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.97428150033819e-05,
+      "loss": 21.0548,
+      "step": 6525
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.973646483865784e-05,
+      "loss": 18.5606,
+      "step": 6526
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.973011393170485e-05,
+      "loss": 17.4953,
+      "step": 6527
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.972376228268142e-05,
+      "loss": 23.39,
+      "step": 6528
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.971740989174607e-05,
+      "loss": 19.687,
+      "step": 6529
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.971105675905741e-05,
+      "loss": 0.7447,
+      "step": 6530
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.970470288477398e-05,
+      "loss": 19.8573,
+      "step": 6531
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.96983482690544e-05,
+      "loss": 0.7421,
+      "step": 6532
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.969199291205724e-05,
+      "loss": 0.7881,
+      "step": 6533
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.968563681394117e-05,
+      "loss": 22.0742,
+      "step": 6534
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.967927997486485e-05,
+      "loss": 0.7813,
+      "step": 6535
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.96729223949869e-05,
+      "loss": 19.6341,
+      "step": 6536
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.966656407446602e-05,
+      "loss": 0.8051,
+      "step": 6537
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.966020501346094e-05,
+      "loss": 0.7479,
+      "step": 6538
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.965384521213039e-05,
+      "loss": 0.7621,
+      "step": 6539
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.964748467063307e-05,
+      "loss": 17.5131,
+      "step": 6540
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.964112338912776e-05,
+      "loss": 0.7908,
+      "step": 6541
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.963476136777327e-05,
+      "loss": 0.7851,
+      "step": 6542
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.962839860672834e-05,
+      "loss": 18.6855,
+      "step": 6543
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.962203510615182e-05,
+      "loss": 19.6343,
+      "step": 6544
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.961567086620256e-05,
+      "loss": 0.7075,
+      "step": 6545
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.960930588703939e-05,
+      "loss": 20.6684,
+      "step": 6546
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.960294016882117e-05,
+      "loss": 18.5916,
+      "step": 6547
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.95965737117068e-05,
+      "loss": 20.3561,
+      "step": 6548
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.95902065158552e-05,
+      "loss": 18.2622,
+      "step": 6549
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.95838385814253e-05,
+      "loss": 20.2484,
+      "step": 6550
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.957746990857601e-05,
+      "loss": 20.8077,
+      "step": 6551
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.957110049746635e-05,
+      "loss": 0.7737,
+      "step": 6552
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.956473034825525e-05,
+      "loss": 0.801,
+      "step": 6553
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.955835946110173e-05,
+      "loss": 17.2633,
+      "step": 6554
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.95519878361648e-05,
+      "loss": 20.1206,
+      "step": 6555
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.954561547360353e-05,
+      "loss": 21.7226,
+      "step": 6556
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.953924237357692e-05,
+      "loss": 19.9823,
+      "step": 6557
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.953286853624412e-05,
+      "loss": 0.7739,
+      "step": 6558
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.952649396176414e-05,
+      "loss": 0.7288,
+      "step": 6559
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.952011865029614e-05,
+      "loss": 21.004,
+      "step": 6560
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.951374260199924e-05,
+      "loss": 19.4214,
+      "step": 6561
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.950736581703257e-05,
+      "loss": 0.8262,
+      "step": 6562
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.950098829555535e-05,
+      "loss": 20.2608,
+      "step": 6563
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.94946100377267e-05,
+      "loss": 0.8162,
+      "step": 6564
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.948823104370584e-05,
+      "loss": 0.7895,
+      "step": 6565
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.948185131365203e-05,
+      "loss": 19.2434,
+      "step": 6566
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.947547084772444e-05,
+      "loss": 0.8445,
+      "step": 6567
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.94690896460824e-05,
+      "loss": 20.3282,
+      "step": 6568
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.946270770888513e-05,
+      "loss": 0.8518,
+      "step": 6569
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.945632503629197e-05,
+      "loss": 0.7619,
+      "step": 6570
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.944994162846219e-05,
+      "loss": 0.7107,
+      "step": 6571
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.944355748555516e-05,
+      "loss": 20.1354,
+      "step": 6572
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.943717260773022e-05,
+      "loss": 0.739,
+      "step": 6573
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.943078699514672e-05,
+      "loss": 19.0719,
+      "step": 6574
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.942440064796404e-05,
+      "loss": 22.788,
+      "step": 6575
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.941801356634162e-05,
+      "loss": 18.6218,
+      "step": 6576
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.941162575043887e-05,
+      "loss": 20.6231,
+      "step": 6577
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.940523720041521e-05,
+      "loss": 0.7471,
+      "step": 6578
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.939884791643012e-05,
+      "loss": 0.757,
+      "step": 6579
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.939245789864308e-05,
+      "loss": 0.723,
+      "step": 6580
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.938606714721357e-05,
+      "loss": 0.7643,
+      "step": 6581
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.93796756623011e-05,
+      "loss": 0.7679,
+      "step": 6582
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.937328344406524e-05,
+      "loss": 0.8311,
+      "step": 6583
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.936689049266553e-05,
+      "loss": 21.852,
+      "step": 6584
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.936049680826151e-05,
+      "loss": 19.0583,
+      "step": 6585
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.935410239101278e-05,
+      "loss": 0.7712,
+      "step": 6586
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.934770724107897e-05,
+      "loss": 0.7804,
+      "step": 6587
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.934131135861968e-05,
+      "loss": 20.0006,
+      "step": 6588
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.933491474379456e-05,
+      "loss": 19.054,
+      "step": 6589
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.932851739676326e-05,
+      "loss": 0.7731,
+      "step": 6590
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.932211931768548e-05,
+      "loss": 0.7305,
+      "step": 6591
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.931572050672089e-05,
+      "loss": 0.7031,
+      "step": 6592
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.930932096402924e-05,
+      "loss": 18.9905,
+      "step": 6593
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.930292068977024e-05,
+      "loss": 0.7186,
+      "step": 6594
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.929651968410366e-05,
+      "loss": 0.8071,
+      "step": 6595
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.929011794718924e-05,
+      "loss": 21.3286,
+      "step": 6596
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.92837154791868e-05,
+      "loss": 19.6689,
+      "step": 6597
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.927731228025612e-05,
+      "loss": 0.7991,
+      "step": 6598
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.927090835055706e-05,
+      "loss": 18.1425,
+      "step": 6599
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.926450369024943e-05,
+      "loss": 0.8052,
+      "step": 6600
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.925809829949312e-05,
+      "loss": 0.7714,
+      "step": 6601
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.925169217844799e-05,
+      "loss": 18.2203,
+      "step": 6602
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.924528532727394e-05,
+      "loss": 19.3873,
+      "step": 6603
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.923887774613091e-05,
+      "loss": 21.1432,
+      "step": 6604
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.923246943517879e-05,
+      "loss": 0.7724,
+      "step": 6605
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.922606039457757e-05,
+      "loss": 0.7603,
+      "step": 6606
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.921965062448719e-05,
+      "loss": 0.7366,
+      "step": 6607
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.921324012506767e-05,
+      "loss": 23.1526,
+      "step": 6608
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.920682889647903e-05,
+      "loss": 0.76,
+      "step": 6609
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.920041693888125e-05,
+      "loss": 0.7644,
+      "step": 6610
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.91940042524344e-05,
+      "loss": 0.7714,
+      "step": 6611
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.918759083729855e-05,
+      "loss": 0.7907,
+      "step": 6612
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.918117669363374e-05,
+      "loss": 22.1684,
+      "step": 6613
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.91747618216001e-05,
+      "loss": 0.7773,
+      "step": 6614
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.916834622135777e-05,
+      "loss": 20.0043,
+      "step": 6615
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.916192989306684e-05,
+      "loss": 0.7161,
+      "step": 6616
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.91555128368875e-05,
+      "loss": 18.9615,
+      "step": 6617
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.914909505297988e-05,
+      "loss": 0.7746,
+      "step": 6618
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.91426765415042e-05,
+      "loss": 0.7431,
+      "step": 6619
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.913625730262065e-05,
+      "loss": 0.7906,
+      "step": 6620
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.912983733648949e-05,
+      "loss": 20.3712,
+      "step": 6621
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.912341664327092e-05,
+      "loss": 0.7686,
+      "step": 6622
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.911699522312522e-05,
+      "loss": 0.7854,
+      "step": 6623
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.911057307621267e-05,
+      "loss": 0.7095,
+      "step": 6624
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.910415020269358e-05,
+      "loss": 20.7564,
+      "step": 6625
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.909772660272825e-05,
+      "loss": 0.7294,
+      "step": 6626
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.909130227647701e-05,
+      "loss": 19.7089,
+      "step": 6627
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.908487722410023e-05,
+      "loss": 18.2943,
+      "step": 6628
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.907845144575829e-05,
+      "loss": 19.618,
+      "step": 6629
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.907202494161154e-05,
+      "loss": 0.7733,
+      "step": 6630
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.906559771182043e-05,
+      "loss": 20.1343,
+      "step": 6631
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.905916975654534e-05,
+      "loss": 0.7972,
+      "step": 6632
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.905274107594676e-05,
+      "loss": 21.2019,
+      "step": 6633
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.904631167018512e-05,
+      "loss": 21.9224,
+      "step": 6634
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.90398815394209e-05,
+      "loss": 0.7459,
+      "step": 6635
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.903345068381462e-05,
+      "loss": 20.4012,
+      "step": 6636
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.902701910352677e-05,
+      "loss": 18.0379,
+      "step": 6637
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.90205867987179e-05,
+      "loss": 18.7594,
+      "step": 6638
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.901415376954857e-05,
+      "loss": 18.7245,
+      "step": 6639
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.900772001617933e-05,
+      "loss": 0.7929,
+      "step": 6640
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.900128553877077e-05,
+      "loss": 0.7602,
+      "step": 6641
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.89948503374835e-05,
+      "loss": 0.7748,
+      "step": 6642
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.898841441247815e-05,
+      "loss": 0.7099,
+      "step": 6643
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.898197776391535e-05,
+      "loss": 0.7933,
+      "step": 6644
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.897554039195577e-05,
+      "loss": 0.7728,
+      "step": 6645
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.896910229676009e-05,
+      "loss": 21.5691,
+      "step": 6646
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.8962663478489e-05,
+      "loss": 0.7523,
+      "step": 6647
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.895622393730322e-05,
+      "loss": 21.5967,
+      "step": 6648
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.894978367336347e-05,
+      "loss": 0.7566,
+      "step": 6649
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.894334268683052e-05,
+      "loss": 0.7532,
+      "step": 6650
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.893690097786513e-05,
+      "loss": 0.7647,
+      "step": 6651
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.893045854662805e-05,
+      "loss": 0.7737,
+      "step": 6652
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.892401539328018e-05,
+      "loss": 0.7566,
+      "step": 6653
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.891757151798224e-05,
+      "loss": 0.7565,
+      "step": 6654
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.89111269208951e-05,
+      "loss": 21.7644,
+      "step": 6655
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.890468160217967e-05,
+      "loss": 0.7368,
+      "step": 6656
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.889823556199676e-05,
+      "loss": 20.1919,
+      "step": 6657
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.889178880050732e-05,
+      "loss": 19.7876,
+      "step": 6658
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.888534131787221e-05,
+      "loss": 0.6959,
+      "step": 6659
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.887889311425238e-05,
+      "loss": 20.9579,
+      "step": 6660
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.88724441898088e-05,
+      "loss": 19.908,
+      "step": 6661
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.886599454470241e-05,
+      "loss": 0.7406,
+      "step": 6662
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.885954417909421e-05,
+      "loss": 0.768,
+      "step": 6663
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.88530930931452e-05,
+      "loss": 20.8203,
+      "step": 6664
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.884664128701639e-05,
+      "loss": 0.7638,
+      "step": 6665
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.884018876086884e-05,
+      "loss": 18.3986,
+      "step": 6666
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.88337355148636e-05,
+      "loss": 0.7432,
+      "step": 6667
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.882728154916171e-05,
+      "loss": 21.4576,
+      "step": 6668
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.882082686392432e-05,
+      "loss": 18.4792,
+      "step": 6669
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.881437145931248e-05,
+      "loss": 0.7661,
+      "step": 6670
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.880791533548737e-05,
+      "loss": 0.7637,
+      "step": 6671
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.880145849261011e-05,
+      "loss": 19.7511,
+      "step": 6672
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.879500093084188e-05,
+      "loss": 19.3162,
+      "step": 6673
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.878854265034383e-05,
+      "loss": 0.8109,
+      "step": 6674
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.87820836512772e-05,
+      "loss": 0.769,
+      "step": 6675
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.877562393380317e-05,
+      "loss": 21.2933,
+      "step": 6676
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.876916349808304e-05,
+      "loss": 20.3865,
+      "step": 6677
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.876270234427798e-05,
+      "loss": 23.0378,
+      "step": 6678
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.875624047254932e-05,
+      "loss": 0.8142,
+      "step": 6679
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.874977788305833e-05,
+      "loss": 21.7845,
+      "step": 6680
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.874331457596631e-05,
+      "loss": 20.1807,
+      "step": 6681
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.87368505514346e-05,
+      "loss": 19.5583,
+      "step": 6682
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.873038580962452e-05,
+      "loss": 19.2294,
+      "step": 6683
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.87239203506975e-05,
+      "loss": 0.7987,
+      "step": 6684
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.871745417481482e-05,
+      "loss": 19.4035,
+      "step": 6685
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.871098728213794e-05,
+      "loss": 21.3191,
+      "step": 6686
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.870451967282829e-05,
+      "loss": 0.7786,
+      "step": 6687
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.869805134704724e-05,
+      "loss": 0.761,
+      "step": 6688
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.869158230495628e-05,
+      "loss": 22.2497,
+      "step": 6689
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.868511254671688e-05,
+      "loss": 20.7727,
+      "step": 6690
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.867864207249053e-05,
+      "loss": 19.8457,
+      "step": 6691
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.867217088243873e-05,
+      "loss": 0.8019,
+      "step": 6692
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.866569897672298e-05,
+      "loss": 19.6659,
+      "step": 6693
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.865922635550485e-05,
+      "loss": 0.7448,
+      "step": 6694
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.865275301894588e-05,
+      "loss": 18.1437,
+      "step": 6695
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.864627896720766e-05,
+      "loss": 0.749,
+      "step": 6696
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.863980420045176e-05,
+      "loss": 0.7572,
+      "step": 6697
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.863332871883982e-05,
+      "loss": 17.5496,
+      "step": 6698
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.862685252253346e-05,
+      "loss": 0.7823,
+      "step": 6699
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.862037561169432e-05,
+      "loss": 0.788,
+      "step": 6700
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.861389798648407e-05,
+      "loss": 21.4482,
+      "step": 6701
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.86074196470644e-05,
+      "loss": 17.9185,
+      "step": 6702
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.8600940593597e-05,
+      "loss": 19.0399,
+      "step": 6703
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.859446082624359e-05,
+      "loss": 21.3307,
+      "step": 6704
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.85879803451659e-05,
+      "loss": 21.6672,
+      "step": 6705
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.858149915052571e-05,
+      "loss": 0.8331,
+      "step": 6706
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.857501724248476e-05,
+      "loss": 18.598,
+      "step": 6707
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.856853462120486e-05,
+      "loss": 19.545,
+      "step": 6708
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.856205128684781e-05,
+      "loss": 0.7761,
+      "step": 6709
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.855556723957544e-05,
+      "loss": 0.7733,
+      "step": 6710
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.854908247954961e-05,
+      "loss": 20.7176,
+      "step": 6711
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.854259700693214e-05,
+      "loss": 19.638,
+      "step": 6712
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.853611082188495e-05,
+      "loss": 21.3749,
+      "step": 6713
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.85296239245699e-05,
+      "loss": 0.79,
+      "step": 6714
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.852313631514894e-05,
+      "loss": 0.7353,
+      "step": 6715
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.851664799378395e-05,
+      "loss": 19.531,
+      "step": 6716
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.851015896063695e-05,
+      "loss": 0.7396,
+      "step": 6717
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.850366921586987e-05,
+      "loss": 0.7997,
+      "step": 6718
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.849717875964469e-05,
+      "loss": 0.7855,
+      "step": 6719
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.849068759212342e-05,
+      "loss": 0.7337,
+      "step": 6720
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.848419571346807e-05,
+      "loss": 0.7575,
+      "step": 6721
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.847770312384071e-05,
+      "loss": 21.0246,
+      "step": 6722
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.847120982340336e-05,
+      "loss": 0.8028,
+      "step": 6723
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.846471581231814e-05,
+      "loss": 0.7664,
+      "step": 6724
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.845822109074711e-05,
+      "loss": 0.6935,
+      "step": 6725
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.845172565885237e-05,
+      "loss": 0.7389,
+      "step": 6726
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.844522951679606e-05,
+      "loss": 21.1663,
+      "step": 6727
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.843873266474032e-05,
+      "loss": 21.2741,
+      "step": 6728
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.843223510284734e-05,
+      "loss": 0.6787,
+      "step": 6729
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.842573683127927e-05,
+      "loss": 0.7454,
+      "step": 6730
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.841923785019831e-05,
+      "loss": 18.9981,
+      "step": 6731
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.84127381597667e-05,
+      "loss": 0.7668,
+      "step": 6732
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.840623776014665e-05,
+      "loss": 0.7877,
+      "step": 6733
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.839973665150043e-05,
+      "loss": 18.591,
+      "step": 6734
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.839323483399029e-05,
+      "loss": 18.8258,
+      "step": 6735
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.838673230777851e-05,
+      "loss": 19.9619,
+      "step": 6736
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.838022907302742e-05,
+      "loss": 20.3667,
+      "step": 6737
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.837372512989934e-05,
+      "loss": 22.1205,
+      "step": 6738
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.83672204785566e-05,
+      "loss": 19.4191,
+      "step": 6739
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.836071511916155e-05,
+      "loss": 17.9019,
+      "step": 6740
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.835420905187659e-05,
+      "loss": 19.8795,
+      "step": 6741
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.83477022768641e-05,
+      "loss": 17.9677,
+      "step": 6742
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.834119479428648e-05,
+      "loss": 0.8214,
+      "step": 6743
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.833468660430617e-05,
+      "loss": 18.8642,
+      "step": 6744
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.83281777070856e-05,
+      "loss": 20.1295,
+      "step": 6745
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.832166810278724e-05,
+      "loss": 18.8273,
+      "step": 6746
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.831515779157362e-05,
+      "loss": 0.7374,
+      "step": 6747
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.830864677360715e-05,
+      "loss": 0.7324,
+      "step": 6748
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.830213504905042e-05,
+      "loss": 15.9824,
+      "step": 6749
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.829562261806593e-05,
+      "loss": 19.9531,
+      "step": 6750
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.828910948081624e-05,
+      "loss": 0.7399,
+      "step": 6751
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.82825956374639e-05,
+      "loss": 0.7836,
+      "step": 6752
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.827608108817154e-05,
+      "loss": 0.7588,
+      "step": 6753
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.826956583310173e-05,
+      "loss": 0.8316,
+      "step": 6754
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.82630498724171e-05,
+      "loss": 19.1836,
+      "step": 6755
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.825653320628031e-05,
+      "loss": 19.0395,
+      "step": 6756
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.825001583485398e-05,
+      "loss": 0.7723,
+      "step": 6757
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.82434977583008e-05,
+      "loss": 0.762,
+      "step": 6758
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.823697897678348e-05,
+      "loss": 21.3563,
+      "step": 6759
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.823045949046471e-05,
+      "loss": 22.3805,
+      "step": 6760
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.822393929950723e-05,
+      "loss": 19.571,
+      "step": 6761
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.821741840407377e-05,
+      "loss": 17.9477,
+      "step": 6762
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.821089680432713e-05,
+      "loss": 20.9439,
+      "step": 6763
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.820437450043004e-05,
+      "loss": 0.7436,
+      "step": 6764
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.819785149254532e-05,
+      "loss": 18.0078,
+      "step": 6765
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.81913277808358e-05,
+      "loss": 19.802,
+      "step": 6766
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.81848033654643e-05,
+      "loss": 20.4041,
+      "step": 6767
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.817827824659368e-05,
+      "loss": 0.8117,
+      "step": 6768
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.817175242438679e-05,
+      "loss": 0.7783,
+      "step": 6769
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.816522589900654e-05,
+      "loss": 20.3985,
+      "step": 6770
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.815869867061581e-05,
+      "loss": 0.8208,
+      "step": 6771
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.815217073937753e-05,
+      "loss": 0.7831,
+      "step": 6772
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.814564210545465e-05,
+      "loss": 20.074,
+      "step": 6773
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.81391127690101e-05,
+      "loss": 21.0421,
+      "step": 6774
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.813258273020688e-05,
+      "loss": 0.7123,
+      "step": 6775
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.812605198920797e-05,
+      "loss": 19.3503,
+      "step": 6776
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.81195205461764e-05,
+      "loss": 0.7805,
+      "step": 6777
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.811298840127515e-05,
+      "loss": 19.9009,
+      "step": 6778
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.810645555466729e-05,
+      "loss": 0.75,
+      "step": 6779
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.809992200651591e-05,
+      "loss": 0.746,
+      "step": 6780
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.809338775698403e-05,
+      "loss": 0.7402,
+      "step": 6781
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.808685280623478e-05,
+      "loss": 0.7255,
+      "step": 6782
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.808031715443128e-05,
+      "loss": 0.7273,
+      "step": 6783
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.807378080173666e-05,
+      "loss": 0.8319,
+      "step": 6784
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.806724374831405e-05,
+      "loss": 19.5178,
+      "step": 6785
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.806070599432662e-05,
+      "loss": 19.4512,
+      "step": 6786
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.805416753993758e-05,
+      "loss": 0.7224,
+      "step": 6787
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.80476283853101e-05,
+      "loss": 0.8123,
+      "step": 6788
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.804108853060743e-05,
+      "loss": 18.069,
+      "step": 6789
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.803454797599277e-05,
+      "loss": 18.4537,
+      "step": 6790
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.80280067216294e-05,
+      "loss": 19.1383,
+      "step": 6791
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.802146476768058e-05,
+      "loss": 20.4663,
+      "step": 6792
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.801492211430958e-05,
+      "loss": 21.8316,
+      "step": 6793
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.800837876167975e-05,
+      "loss": 18.0363,
+      "step": 6794
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.80018347099544e-05,
+      "loss": 0.7044,
+      "step": 6795
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.799528995929684e-05,
+      "loss": 0.792,
+      "step": 6796
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.798874450987046e-05,
+      "loss": 18.5593,
+      "step": 6797
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.798219836183862e-05,
+      "loss": 0.792,
+      "step": 6798
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.797565151536473e-05,
+      "loss": 20.8739,
+      "step": 6799
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.796910397061217e-05,
+      "loss": 20.3024,
+      "step": 6800
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.796255572774442e-05,
+      "loss": 0.7144,
+      "step": 6801
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.795600678692487e-05,
+      "loss": 17.8951,
+      "step": 6802
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.794945714831701e-05,
+      "loss": 0.8208,
+      "step": 6803
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.794290681208433e-05,
+      "loss": 20.1448,
+      "step": 6804
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.793635577839032e-05,
+      "loss": 0.8095,
+      "step": 6805
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.792980404739848e-05,
+      "loss": 21.4525,
+      "step": 6806
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.792325161927235e-05,
+      "loss": 0.8061,
+      "step": 6807
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.79166984941755e-05,
+      "loss": 20.3696,
+      "step": 6808
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.791014467227147e-05,
+      "loss": 0.7559,
+      "step": 6809
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.790359015372384e-05,
+      "loss": 20.1697,
+      "step": 6810
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.789703493869625e-05,
+      "loss": 0.7224,
+      "step": 6811
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.789047902735229e-05,
+      "loss": 19.7982,
+      "step": 6812
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.788392241985561e-05,
+      "loss": 17.6742,
+      "step": 6813
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.787736511636986e-05,
+      "loss": 18.4028,
+      "step": 6814
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.787080711705872e-05,
+      "loss": 17.9444,
+      "step": 6815
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.786424842208587e-05,
+      "loss": 18.7785,
+      "step": 6816
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.785768903161499e-05,
+      "loss": 0.7536,
+      "step": 6817
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.785112894580986e-05,
+      "loss": 0.7626,
+      "step": 6818
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.784456816483416e-05,
+      "loss": 20.6554,
+      "step": 6819
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.783800668885171e-05,
+      "loss": 0.8099,
+      "step": 6820
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.783144451802624e-05,
+      "loss": 0.6836,
+      "step": 6821
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.782488165252157e-05,
+      "loss": 19.4753,
+      "step": 6822
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.781831809250151e-05,
+      "loss": 0.7698,
+      "step": 6823
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.781175383812986e-05,
+      "loss": 20.4867,
+      "step": 6824
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.780518888957051e-05,
+      "loss": 0.7846,
+      "step": 6825
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.779862324698727e-05,
+      "loss": 0.8179,
+      "step": 6826
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.779205691054406e-05,
+      "loss": 20.2236,
+      "step": 6827
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.778548988040474e-05,
+      "loss": 19.0912,
+      "step": 6828
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.777892215673328e-05,
+      "loss": 0.7546,
+      "step": 6829
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.777235373969356e-05,
+      "loss": 0.8035,
+      "step": 6830
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.776578462944956e-05,
+      "loss": 0.8111,
+      "step": 6831
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.775921482616524e-05,
+      "loss": 0.7476,
+      "step": 6832
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.775264433000456e-05,
+      "loss": 0.7268,
+      "step": 6833
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.774607314113156e-05,
+      "loss": 20.8261,
+      "step": 6834
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.773950125971024e-05,
+      "loss": 0.7454,
+      "step": 6835
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.773292868590462e-05,
+      "loss": 20.2549,
+      "step": 6836
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.772635541987877e-05,
+      "loss": 0.7262,
+      "step": 6837
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.771978146179675e-05,
+      "loss": 18.4033,
+      "step": 6838
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.771320681182268e-05,
+      "loss": 0.7512,
+      "step": 6839
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.770663147012062e-05,
+      "loss": 19.5649,
+      "step": 6840
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.770005543685474e-05,
+      "loss": 0.7742,
+      "step": 6841
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.769347871218913e-05,
+      "loss": 0.8492,
+      "step": 6842
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.768690129628797e-05,
+      "loss": 0.7814,
+      "step": 6843
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.768032318931543e-05,
+      "loss": 0.7375,
+      "step": 6844
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.76737443914357e-05,
+      "loss": 17.1385,
+      "step": 6845
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.7667164902813e-05,
+      "loss": 17.9122,
+      "step": 6846
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.766058472361154e-05,
+      "loss": 19.8667,
+      "step": 6847
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.765400385399557e-05,
+      "loss": 0.7785,
+      "step": 6848
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.764742229412936e-05,
+      "loss": 20.2761,
+      "step": 6849
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.764084004417716e-05,
+      "loss": 23.0967,
+      "step": 6850
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.76342571043033e-05,
+      "loss": 0.7547,
+      "step": 6851
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.762767347467206e-05,
+      "loss": 19.3694,
+      "step": 6852
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.762108915544779e-05,
+      "loss": 0.8221,
+      "step": 6853
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.761450414679482e-05,
+      "loss": 0.7907,
+      "step": 6854
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.760791844887752e-05,
+      "loss": 0.7047,
+      "step": 6855
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.760133206186029e-05,
+      "loss": 0.7484,
+      "step": 6856
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.759474498590748e-05,
+      "loss": 18.2393,
+      "step": 6857
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.758815722118356e-05,
+      "loss": 0.7268,
+      "step": 6858
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.758156876785291e-05,
+      "loss": 20.0495,
+      "step": 6859
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.757497962608003e-05,
+      "loss": 19.9955,
+      "step": 6860
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.756838979602936e-05,
+      "loss": 21.2385,
+      "step": 6861
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.756179927786538e-05,
+      "loss": 0.7492,
+      "step": 6862
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.75552080717526e-05,
+      "loss": 20.8713,
+      "step": 6863
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.754861617785553e-05,
+      "loss": 18.8617,
+      "step": 6864
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.75420235963387e-05,
+      "loss": 25.2453,
+      "step": 6865
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.75354303273667e-05,
+      "loss": 22.4191,
+      "step": 6866
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.752883637110405e-05,
+      "loss": 0.7736,
+      "step": 6867
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.752224172771537e-05,
+      "loss": 17.7091,
+      "step": 6868
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.751564639736525e-05,
+      "loss": 20.6287,
+      "step": 6869
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.750905038021831e-05,
+      "loss": 0.6928,
+      "step": 6870
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.75024536764392e-05,
+      "loss": 18.6464,
+      "step": 6871
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.749585628619256e-05,
+      "loss": 17.2703,
+      "step": 6872
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.748925820964306e-05,
+      "loss": 21.7691,
+      "step": 6873
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.74826594469554e-05,
+      "loss": 20.1631,
+      "step": 6874
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.74760599982943e-05,
+      "loss": 19.7792,
+      "step": 6875
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.746945986382446e-05,
+      "loss": 0.6913,
+      "step": 6876
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.746285904371063e-05,
+      "loss": 0.772,
+      "step": 6877
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.745625753811757e-05,
+      "loss": 21.4486,
+      "step": 6878
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.744965534721005e-05,
+      "loss": 19.2698,
+      "step": 6879
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.744305247115287e-05,
+      "loss": 18.8871,
+      "step": 6880
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.743644891011082e-05,
+      "loss": 20.2898,
+      "step": 6881
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.742984466424875e-05,
+      "loss": 0.7973,
+      "step": 6882
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.742323973373149e-05,
+      "loss": 0.7958,
+      "step": 6883
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.741663411872393e-05,
+      "loss": 0.8078,
+      "step": 6884
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.741002781939088e-05,
+      "loss": 20.7833,
+      "step": 6885
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.740342083589732e-05,
+      "loss": 21.4694,
+      "step": 6886
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.739681316840809e-05,
+      "loss": 0.8343,
+      "step": 6887
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.739020481708815e-05,
+      "loss": 0.7628,
+      "step": 6888
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.738359578210246e-05,
+      "loss": 0.7545,
+      "step": 6889
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.737698606361595e-05,
+      "loss": 19.7273,
+      "step": 6890
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.737037566179362e-05,
+      "loss": 0.7975,
+      "step": 6891
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.736376457680047e-05,
+      "loss": 20.3802,
+      "step": 6892
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.73571528088015e-05,
+      "loss": 0.7653,
+      "step": 6893
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.735054035796177e-05,
+      "loss": 0.7406,
+      "step": 6894
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.734392722444628e-05,
+      "loss": 20.6049,
+      "step": 6895
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.733731340842014e-05,
+      "loss": 20.7574,
+      "step": 6896
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.73306989100484e-05,
+      "loss": 17.7277,
+      "step": 6897
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.73240837294962e-05,
+      "loss": 0.8231,
+      "step": 6898
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.731746786692861e-05,
+      "loss": 18.9545,
+      "step": 6899
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.73108513225108e-05,
+      "loss": 18.7658,
+      "step": 6900
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.730423409640788e-05,
+      "loss": 18.0811,
+      "step": 6901
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.729761618878506e-05,
+      "loss": 0.7104,
+      "step": 6902
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.729099759980752e-05,
+      "loss": 0.7552,
+      "step": 6903
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.728437832964043e-05,
+      "loss": 19.57,
+      "step": 6904
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.727775837844904e-05,
+      "loss": 0.7002,
+      "step": 6905
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.727113774639856e-05,
+      "loss": 20.6703,
+      "step": 6906
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.726451643365426e-05,
+      "loss": 0.7392,
+      "step": 6907
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.72578944403814e-05,
+      "loss": 17.1708,
+      "step": 6908
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.725127176674527e-05,
+      "loss": 18.8588,
+      "step": 6909
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.724464841291118e-05,
+      "loss": 0.7612,
+      "step": 6910
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.723802437904443e-05,
+      "loss": 20.3323,
+      "step": 6911
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.723139966531042e-05,
+      "loss": 20.1713,
+      "step": 6912
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.72247742718744e-05,
+      "loss": 21.2071,
+      "step": 6913
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.721814819890183e-05,
+      "loss": 0.8327,
+      "step": 6914
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.721152144655806e-05,
+      "loss": 18.5976,
+      "step": 6915
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.72048940150085e-05,
+      "loss": 19.2959,
+      "step": 6916
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.719826590441856e-05,
+      "loss": 0.7516,
+      "step": 6917
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.719163711495372e-05,
+      "loss": 21.4143,
+      "step": 6918
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.718500764677938e-05,
+      "loss": 0.793,
+      "step": 6919
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.717837750006106e-05,
+      "loss": 16.8685,
+      "step": 6920
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.717174667496423e-05,
+      "loss": 19.9708,
+      "step": 6921
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.71651151716544e-05,
+      "loss": 20.4765,
+      "step": 6922
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.715848299029709e-05,
+      "loss": 19.3913,
+      "step": 6923
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.715185013105783e-05,
+      "loss": 20.7224,
+      "step": 6924
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.714521659410223e-05,
+      "loss": 19.8212,
+      "step": 6925
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.71385823795958e-05,
+      "loss": 0.746,
+      "step": 6926
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.713194748770414e-05,
+      "loss": 18.9838,
+      "step": 6927
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.712531191859289e-05,
+      "loss": 18.3118,
+      "step": 6928
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.711867567242768e-05,
+      "loss": 19.1108,
+      "step": 6929
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.711203874937411e-05,
+      "loss": 19.047,
+      "step": 6930
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.710540114959786e-05,
+      "loss": 21.2559,
+      "step": 6931
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.709876287326463e-05,
+      "loss": 0.7601,
+      "step": 6932
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.709212392054009e-05,
+      "loss": 19.8227,
+      "step": 6933
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.708548429158993e-05,
+      "loss": 20.2969,
+      "step": 6934
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.707884398657989e-05,
+      "loss": 22.0759,
+      "step": 6935
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.707220300567576e-05,
+      "loss": 0.798,
+      "step": 6936
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.706556134904324e-05,
+      "loss": 0.7646,
+      "step": 6937
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.705891901684812e-05,
+      "loss": 18.8979,
+      "step": 6938
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.705227600925622e-05,
+      "loss": 17.4684,
+      "step": 6939
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.704563232643333e-05,
+      "loss": 18.23,
+      "step": 6940
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.703898796854527e-05,
+      "loss": 19.9634,
+      "step": 6941
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.70323429357579e-05,
+      "loss": 0.7903,
+      "step": 6942
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.702569722823709e-05,
+      "loss": 0.7499,
+      "step": 6943
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.70190508461487e-05,
+      "loss": 18.2632,
+      "step": 6944
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.701240378965862e-05,
+      "loss": 0.7529,
+      "step": 6945
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.700575605893281e-05,
+      "loss": 19.6176,
+      "step": 6946
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.699910765413712e-05,
+      "loss": 20.1607,
+      "step": 6947
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.699245857543755e-05,
+      "loss": 16.589,
+      "step": 6948
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.698580882300007e-05,
+      "loss": 18.623,
+      "step": 6949
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.697915839699062e-05,
+      "loss": 19.3054,
+      "step": 6950
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.697250729757524e-05,
+      "loss": 20.2099,
+      "step": 6951
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.696585552491989e-05,
+      "loss": 0.7016,
+      "step": 6952
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.695920307919065e-05,
+      "loss": 19.9318,
+      "step": 6953
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.695254996055352e-05,
+      "loss": 0.7511,
+      "step": 6954
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.694589616917461e-05,
+      "loss": 0.8069,
+      "step": 6955
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.693924170521998e-05,
+      "loss": 18.0704,
+      "step": 6956
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.693258656885572e-05,
+      "loss": 19.2382,
+      "step": 6957
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.692593076024794e-05,
+      "loss": 0.7841,
+      "step": 6958
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.69192742795628e-05,
+      "loss": 18.8865,
+      "step": 6959
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.691261712696643e-05,
+      "loss": 23.2074,
+      "step": 6960
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.690595930262496e-05,
+      "loss": 0.7597,
+      "step": 6961
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.689930080670462e-05,
+      "loss": 19.9963,
+      "step": 6962
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.68926416393716e-05,
+      "loss": 20.0544,
+      "step": 6963
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.68859818007921e-05,
+      "loss": 19.2912,
+      "step": 6964
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.687932129113235e-05,
+      "loss": 0.7407,
+      "step": 6965
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.68726601105586e-05,
+      "loss": 0.7707,
+      "step": 6966
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.686599825923713e-05,
+      "loss": 0.7968,
+      "step": 6967
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.685933573733419e-05,
+      "loss": 0.7855,
+      "step": 6968
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.685267254501611e-05,
+      "loss": 18.7689,
+      "step": 6969
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.68460086824492e-05,
+      "loss": 0.7529,
+      "step": 6970
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.683934414979976e-05,
+      "loss": 0.809,
+      "step": 6971
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.683267894723418e-05,
+      "loss": 17.9837,
+      "step": 6972
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.68260130749188e-05,
+      "loss": 19.4947,
+      "step": 6973
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.681934653302004e-05,
+      "loss": 18.5554,
+      "step": 6974
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.681267932170422e-05,
+      "loss": 0.7379,
+      "step": 6975
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.680601144113781e-05,
+      "loss": 20.0304,
+      "step": 6976
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.679934289148725e-05,
+      "loss": 19.4401,
+      "step": 6977
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.679267367291897e-05,
+      "loss": 20.5845,
+      "step": 6978
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.678600378559943e-05,
+      "loss": 0.7678,
+      "step": 6979
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.677933322969513e-05,
+      "loss": 18.546,
+      "step": 6980
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.677266200537255e-05,
+      "loss": 19.4138,
+      "step": 6981
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.676599011279821e-05,
+      "loss": 0.759,
+      "step": 6982
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.675931755213865e-05,
+      "loss": 0.8009,
+      "step": 6983
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.675264432356042e-05,
+      "loss": 20.0241,
+      "step": 6984
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.674597042723007e-05,
+      "loss": 18.0081,
+      "step": 6985
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.673929586331419e-05,
+      "loss": 0.7915,
+      "step": 6986
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.673262063197939e-05,
+      "loss": 19.9777,
+      "step": 6987
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.672594473339227e-05,
+      "loss": 17.4912,
+      "step": 6988
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.671926816771947e-05,
+      "loss": 18.3778,
+      "step": 6989
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.671259093512763e-05,
+      "loss": 0.7687,
+      "step": 6990
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.670591303578343e-05,
+      "loss": 22.9993,
+      "step": 6991
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.669923446985353e-05,
+      "loss": 0.7499,
+      "step": 6992
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.669255523750465e-05,
+      "loss": 21.6312,
+      "step": 6993
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.66858753389035e-05,
+      "loss": 19.7018,
+      "step": 6994
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.66791947742168e-05,
+      "loss": 0.7693,
+      "step": 6995
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.667251354361131e-05,
+      "loss": 0.7058,
+      "step": 6996
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.666583164725379e-05,
+      "loss": 19.4508,
+      "step": 6997
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.665914908531103e-05,
+      "loss": 19.0252,
+      "step": 6998
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.66524658579498e-05,
+      "loss": 18.0665,
+      "step": 6999
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.664578196533696e-05,
+      "loss": 18.1489,
+      "step": 7000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.663909740763932e-05,
+      "loss": 19.4256,
+      "step": 7001
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.663241218502372e-05,
+      "loss": 17.9761,
+      "step": 7002
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.662572629765703e-05,
+      "loss": 0.7656,
+      "step": 7003
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.661903974570612e-05,
+      "loss": 0.7573,
+      "step": 7004
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.661235252933792e-05,
+      "loss": 0.764,
+      "step": 7005
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.660566464871931e-05,
+      "loss": 20.9391,
+      "step": 7006
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.659897610401725e-05,
+      "loss": 19.5791,
+      "step": 7007
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.659228689539866e-05,
+      "loss": 0.7328,
+      "step": 7008
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.658559702303051e-05,
+      "loss": 20.9246,
+      "step": 7009
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.65789064870798e-05,
+      "loss": 20.3398,
+      "step": 7010
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.657221528771352e-05,
+      "loss": 20.5438,
+      "step": 7011
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.656552342509868e-05,
+      "loss": 18.8372,
+      "step": 7012
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.65588308994023e-05,
+      "loss": 0.7096,
+      "step": 7013
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.655213771079144e-05,
+      "loss": 0.7474,
+      "step": 7014
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.654544385943317e-05,
+      "loss": 20.8569,
+      "step": 7015
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.653874934549454e-05,
+      "loss": 18.5936,
+      "step": 7016
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.653205416914267e-05,
+      "loss": 20.2108,
+      "step": 7017
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.652535833054468e-05,
+      "loss": 18.6133,
+      "step": 7018
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.651866182986769e-05,
+      "loss": 0.7502,
+      "step": 7019
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.651196466727884e-05,
+      "loss": 0.7292,
+      "step": 7020
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.65052668429453e-05,
+      "loss": 20.4344,
+      "step": 7021
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.649856835703426e-05,
+      "loss": 19.5492,
+      "step": 7022
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.649186920971288e-05,
+      "loss": 0.7635,
+      "step": 7023
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.64851694011484e-05,
+      "loss": 0.7743,
+      "step": 7024
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.647846893150807e-05,
+      "loss": 0.8136,
+      "step": 7025
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.647176780095908e-05,
+      "loss": 0.734,
+      "step": 7026
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.646506600966874e-05,
+      "loss": 0.8218,
+      "step": 7027
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.64583635578043e-05,
+      "loss": 22.4018,
+      "step": 7028
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.645166044553306e-05,
+      "loss": 0.7247,
+      "step": 7029
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.644495667302236e-05,
+      "loss": 17.6447,
+      "step": 7030
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.643825224043948e-05,
+      "loss": 20.5468,
+      "step": 7031
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.643154714795177e-05,
+      "loss": 18.6164,
+      "step": 7032
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.642484139572665e-05,
+      "loss": 18.9395,
+      "step": 7033
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.641813498393143e-05,
+      "loss": 0.7648,
+      "step": 7034
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.641142791273355e-05,
+      "loss": 0.753,
+      "step": 7035
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.640472018230039e-05,
+      "loss": 18.8162,
+      "step": 7036
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.639801179279936e-05,
+      "loss": 0.7141,
+      "step": 7037
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.639130274439795e-05,
+      "loss": 20.8101,
+      "step": 7038
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.63845930372636e-05,
+      "loss": 0.7753,
+      "step": 7039
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.637788267156379e-05,
+      "loss": 21.9756,
+      "step": 7040
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.637117164746598e-05,
+      "loss": 0.7372,
+      "step": 7041
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.636445996513772e-05,
+      "loss": 20.7429,
+      "step": 7042
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.635774762474651e-05,
+      "loss": 19.7012,
+      "step": 7043
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.635103462645992e-05,
+      "loss": 18.4853,
+      "step": 7044
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.634432097044547e-05,
+      "loss": 0.7315,
+      "step": 7045
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.633760665687077e-05,
+      "loss": 0.7485,
+      "step": 7046
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.633089168590337e-05,
+      "loss": 21.5483,
+      "step": 7047
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.632417605771091e-05,
+      "loss": 19.7751,
+      "step": 7048
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.631745977246102e-05,
+      "loss": 0.7588,
+      "step": 7049
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.631074283032134e-05,
+      "loss": 0.8011,
+      "step": 7050
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.630402523145948e-05,
+      "loss": 0.7519,
+      "step": 7051
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.629730697604314e-05,
+      "loss": 0.7687,
+      "step": 7052
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.629058806424003e-05,
+      "loss": 0.7297,
+      "step": 7053
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.628386849621784e-05,
+      "loss": 17.8348,
+      "step": 7054
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.627714827214428e-05,
+      "loss": 19.4523,
+      "step": 7055
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.627042739218711e-05,
+      "loss": 20.0136,
+      "step": 7056
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.626370585651406e-05,
+      "loss": 0.753,
+      "step": 7057
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.625698366529292e-05,
+      "loss": 0.8302,
+      "step": 7058
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.625026081869148e-05,
+      "loss": 18.888,
+      "step": 7059
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.624353731687755e-05,
+      "loss": 0.8003,
+      "step": 7060
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.623681316001891e-05,
+      "loss": 0.8003,
+      "step": 7061
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.623008834828342e-05,
+      "loss": 21.7922,
+      "step": 7062
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.622336288183895e-05,
+      "loss": 0.7673,
+      "step": 7063
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.621663676085337e-05,
+      "loss": 19.9735,
+      "step": 7064
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.620990998549454e-05,
+      "loss": 0.7929,
+      "step": 7065
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.620318255593035e-05,
+      "loss": 21.2949,
+      "step": 7066
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.619645447232875e-05,
+      "loss": 19.3385,
+      "step": 7067
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.618972573485768e-05,
+      "loss": 0.7218,
+      "step": 7068
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.618299634368507e-05,
+      "loss": 18.4937,
+      "step": 7069
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.617626629897888e-05,
+      "loss": 0.7816,
+      "step": 7070
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.616953560090711e-05,
+      "loss": 20.6053,
+      "step": 7071
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.616280424963776e-05,
+      "loss": 20.7535,
+      "step": 7072
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.615607224533886e-05,
+      "loss": 17.7308,
+      "step": 7073
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.61493395881784e-05,
+      "loss": 18.2256,
+      "step": 7074
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.614260627832446e-05,
+      "loss": 21.8279,
+      "step": 7075
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.61358723159451e-05,
+      "loss": 18.5439,
+      "step": 7076
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.61291377012084e-05,
+      "loss": 17.8966,
+      "step": 7077
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.612240243428247e-05,
+      "loss": 17.5598,
+      "step": 7078
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.61156665153354e-05,
+      "loss": 16.7668,
+      "step": 7079
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.610892994453531e-05,
+      "loss": 0.7659,
+      "step": 7080
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.610219272205039e-05,
+      "loss": 18.8189,
+      "step": 7081
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.609545484804879e-05,
+      "loss": 0.7828,
+      "step": 7082
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.608871632269865e-05,
+      "loss": 0.7719,
+      "step": 7083
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.608197714616821e-05,
+      "loss": 21.1795,
+      "step": 7084
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.607523731862567e-05,
+      "loss": 0.74,
+      "step": 7085
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.606849684023923e-05,
+      "loss": 0.7777,
+      "step": 7086
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.606175571117718e-05,
+      "loss": 0.7665,
+      "step": 7087
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.605501393160776e-05,
+      "loss": 0.7699,
+      "step": 7088
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.604827150169922e-05,
+      "loss": 18.4576,
+      "step": 7089
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.60415284216199e-05,
+      "loss": 0.7097,
+      "step": 7090
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.603478469153806e-05,
+      "loss": 0.7919,
+      "step": 7091
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.602804031162207e-05,
+      "loss": 18.3579,
+      "step": 7092
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.602129528204023e-05,
+      "loss": 18.7489,
+      "step": 7093
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.601454960296093e-05,
+      "loss": 18.5308,
+      "step": 7094
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.600780327455254e-05,
+      "loss": 20.0501,
+      "step": 7095
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.600105629698344e-05,
+      "loss": 19.5596,
+      "step": 7096
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.599430867042203e-05,
+      "loss": 21.0934,
+      "step": 7097
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.598756039503673e-05,
+      "loss": 17.4068,
+      "step": 7098
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.5980811470996e-05,
+      "loss": 19.0945,
+      "step": 7099
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.59740618984683e-05,
+      "loss": 0.7868,
+      "step": 7100
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.596731167762207e-05,
+      "loss": 0.7333,
+      "step": 7101
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.596056080862581e-05,
+      "loss": 18.5642,
+      "step": 7102
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.595380929164802e-05,
+      "loss": 17.5082,
+      "step": 7103
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.594705712685722e-05,
+      "loss": 21.5402,
+      "step": 7104
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.594030431442198e-05,
+      "loss": 0.7306,
+      "step": 7105
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.593355085451079e-05,
+      "loss": 18.6378,
+      "step": 7106
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.592679674729227e-05,
+      "loss": 19.5014,
+      "step": 7107
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.592004199293496e-05,
+      "loss": 19.518,
+      "step": 7108
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.59132865916075e-05,
+      "loss": 0.787,
+      "step": 7109
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.59065305434785e-05,
+      "loss": 17.9402,
+      "step": 7110
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.589977384871656e-05,
+      "loss": 0.7304,
+      "step": 7111
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.589301650749036e-05,
+      "loss": 21.7738,
+      "step": 7112
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.588625851996856e-05,
+      "loss": 19.1644,
+      "step": 7113
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.587949988631982e-05,
+      "loss": 0.7705,
+      "step": 7114
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.587274060671287e-05,
+      "loss": 21.3652,
+      "step": 7115
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.58659806813164e-05,
+      "loss": 18.3769,
+      "step": 7116
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.585922011029916e-05,
+      "loss": 17.2724,
+      "step": 7117
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.585245889382986e-05,
+      "loss": 20.0195,
+      "step": 7118
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.584569703207731e-05,
+      "loss": 18.0476,
+      "step": 7119
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.583893452521024e-05,
+      "loss": 19.76,
+      "step": 7120
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.583217137339746e-05,
+      "loss": 20.0089,
+      "step": 7121
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.582540757680781e-05,
+      "loss": 18.4922,
+      "step": 7122
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.581864313561007e-05,
+      "loss": 20.2216,
+      "step": 7123
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.581187804997311e-05,
+      "loss": 19.97,
+      "step": 7124
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.580511232006577e-05,
+      "loss": 0.7227,
+      "step": 7125
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.579834594605695e-05,
+      "loss": 0.7454,
+      "step": 7126
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.579157892811551e-05,
+      "loss": 0.709,
+      "step": 7127
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.578481126641038e-05,
+      "loss": 21.2049,
+      "step": 7128
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.577804296111048e-05,
+      "loss": 0.7338,
+      "step": 7129
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.577127401238471e-05,
+      "loss": 20.1855,
+      "step": 7130
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.576450442040208e-05,
+      "loss": 0.7936,
+      "step": 7131
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.575773418533154e-05,
+      "loss": 17.9194,
+      "step": 7132
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.575096330734207e-05,
+      "loss": 0.7879,
+      "step": 7133
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.574419178660268e-05,
+      "loss": 0.7758,
+      "step": 7134
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.573741962328238e-05,
+      "loss": 20.3841,
+      "step": 7135
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.57306468175502e-05,
+      "loss": 19.5988,
+      "step": 7136
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.572387336957521e-05,
+      "loss": 0.7683,
+      "step": 7137
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.571709927952647e-05,
+      "loss": 21.9057,
+      "step": 7138
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.571032454757305e-05,
+      "loss": 0.7329,
+      "step": 7139
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.570354917388407e-05,
+      "loss": 18.9005,
+      "step": 7140
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.569677315862863e-05,
+      "loss": 0.7502,
+      "step": 7141
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.568999650197585e-05,
+      "loss": 0.7916,
+      "step": 7142
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.568321920409494e-05,
+      "loss": 21.7182,
+      "step": 7143
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.567644126515496e-05,
+      "loss": 0.7529,
+      "step": 7144
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.566966268532516e-05,
+      "loss": 20.6566,
+      "step": 7145
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.566288346477472e-05,
+      "loss": 0.7406,
+      "step": 7146
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.565610360367285e-05,
+      "loss": 0.7518,
+      "step": 7147
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.564932310218878e-05,
+      "loss": 0.7557,
+      "step": 7148
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.564254196049175e-05,
+      "loss": 21.0456,
+      "step": 7149
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.563576017875102e-05,
+      "loss": 0.773,
+      "step": 7150
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.562897775713584e-05,
+      "loss": 0.7827,
+      "step": 7151
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.562219469581556e-05,
+      "loss": 23.0313,
+      "step": 7152
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.561541099495942e-05,
+      "loss": 21.6816,
+      "step": 7153
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.560862665473678e-05,
+      "loss": 19.8684,
+      "step": 7154
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.560184167531696e-05,
+      "loss": 0.7231,
+      "step": 7155
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.559505605686933e-05,
+      "loss": 17.6703,
+      "step": 7156
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.558826979956329e-05,
+      "loss": 0.7871,
+      "step": 7157
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.558148290356817e-05,
+      "loss": 18.5579,
+      "step": 7158
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.55746953690534e-05,
+      "loss": 19.4314,
+      "step": 7159
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.556790719618839e-05,
+      "loss": 21.5884,
+      "step": 7160
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.556111838514257e-05,
+      "loss": 0.7849,
+      "step": 7161
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.55543289360854e-05,
+      "loss": 23.1158,
+      "step": 7162
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.554753884918635e-05,
+      "loss": 17.296,
+      "step": 7163
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.554074812461491e-05,
+      "loss": 0.68,
+      "step": 7164
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.553395676254054e-05,
+      "loss": 0.8028,
+      "step": 7165
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.55271647631328e-05,
+      "loss": 0.7794,
+      "step": 7166
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.55203721265612e-05,
+      "loss": 17.8659,
+      "step": 7167
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.551357885299527e-05,
+      "loss": 19.1602,
+      "step": 7168
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.550678494260459e-05,
+      "loss": 21.1898,
+      "step": 7169
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.549999039555873e-05,
+      "loss": 20.2756,
+      "step": 7170
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.54931952120273e-05,
+      "loss": 19.6995,
+      "step": 7171
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.54863993921799e-05,
+      "loss": 20.0855,
+      "step": 7172
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.547960293618614e-05,
+      "loss": 21.6704,
+      "step": 7173
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.547280584421568e-05,
+      "loss": 0.7596,
+      "step": 7174
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.546600811643816e-05,
+      "loss": 0.7465,
+      "step": 7175
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.545920975302327e-05,
+      "loss": 19.4855,
+      "step": 7176
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.545241075414068e-05,
+      "loss": 19.7716,
+      "step": 7177
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.544561111996012e-05,
+      "loss": 19.1164,
+      "step": 7178
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.543881085065129e-05,
+      "loss": 19.5863,
+      "step": 7179
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.543200994638393e-05,
+      "loss": 20.0241,
+      "step": 7180
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.54252084073278e-05,
+      "loss": 18.6927,
+      "step": 7181
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.541840623365266e-05,
+      "loss": 21.9807,
+      "step": 7182
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.541160342552831e-05,
+      "loss": 17.8736,
+      "step": 7183
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.54047999831245e-05,
+      "loss": 17.5229,
+      "step": 7184
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.539799590661114e-05,
+      "loss": 0.735,
+      "step": 7185
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.539119119615796e-05,
+      "loss": 0.6916,
+      "step": 7186
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.538438585193486e-05,
+      "loss": 0.7773,
+      "step": 7187
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.53775798741117e-05,
+      "loss": 20.1509,
+      "step": 7188
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.537077326285833e-05,
+      "loss": 19.0516,
+      "step": 7189
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.536396601834469e-05,
+      "loss": 0.7642,
+      "step": 7190
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.535715814074067e-05,
+      "loss": 0.7297,
+      "step": 7191
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.535034963021617e-05,
+      "loss": 20.2636,
+      "step": 7192
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.534354048694117e-05,
+      "loss": 0.7637,
+      "step": 7193
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.53367307110856e-05,
+      "loss": 18.042,
+      "step": 7194
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.532992030281946e-05,
+      "loss": 17.6956,
+      "step": 7195
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.532310926231272e-05,
+      "loss": 22.5696,
+      "step": 7196
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.53162975897354e-05,
+      "loss": 19.2075,
+      "step": 7197
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.53094852852575e-05,
+      "loss": 0.7601,
+      "step": 7198
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.530267234904908e-05,
+      "loss": 23.2421,
+      "step": 7199
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.529585878128015e-05,
+      "loss": 18.9749,
+      "step": 7200
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.528904458212085e-05,
+      "loss": 0.7466,
+      "step": 7201
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.52822297517412e-05,
+      "loss": 18.4266,
+      "step": 7202
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.527541429031133e-05,
+      "loss": 0.8213,
+      "step": 7203
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.526859819800135e-05,
+      "loss": 0.7491,
+      "step": 7204
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.526178147498139e-05,
+      "loss": 0.7838,
+      "step": 7205
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.52549641214216e-05,
+      "loss": 21.212,
+      "step": 7206
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.524814613749212e-05,
+      "loss": 19.5185,
+      "step": 7207
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.524132752336315e-05,
+      "loss": 19.5681,
+      "step": 7208
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.523450827920489e-05,
+      "loss": 18.2792,
+      "step": 7209
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.522768840518754e-05,
+      "loss": 0.7151,
+      "step": 7210
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.522086790148133e-05,
+      "loss": 0.8167,
+      "step": 7211
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.521404676825649e-05,
+      "loss": 20.5685,
+      "step": 7212
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.520722500568332e-05,
+      "loss": 0.8135,
+      "step": 7213
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.520040261393202e-05,
+      "loss": 0.745,
+      "step": 7214
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.519357959317292e-05,
+      "loss": 0.7984,
+      "step": 7215
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.518675594357633e-05,
+      "loss": 0.7655,
+      "step": 7216
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.517993166531256e-05,
+      "loss": 20.588,
+      "step": 7217
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.517310675855194e-05,
+      "loss": 0.8131,
+      "step": 7218
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.516628122346484e-05,
+      "loss": 0.8113,
+      "step": 7219
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.51594550602216e-05,
+      "loss": 21.7801,
+      "step": 7220
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.515262826899263e-05,
+      "loss": 21.5571,
+      "step": 7221
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.514580084994831e-05,
+      "loss": 17.2797,
+      "step": 7222
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.513897280325906e-05,
+      "loss": 0.7325,
+      "step": 7223
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.513214412909531e-05,
+      "loss": 0.6901,
+      "step": 7224
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.51253148276275e-05,
+      "loss": 0.7309,
+      "step": 7225
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.511848489902609e-05,
+      "loss": 0.793,
+      "step": 7226
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.511165434346158e-05,
+      "loss": 0.8035,
+      "step": 7227
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.510482316110443e-05,
+      "loss": 0.7499,
+      "step": 7228
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.509799135212516e-05,
+      "loss": 19.5517,
+      "step": 7229
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.50911589166943e-05,
+      "loss": 0.7119,
+      "step": 7230
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.508432585498239e-05,
+      "loss": 0.7972,
+      "step": 7231
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.507749216715997e-05,
+      "loss": 19.1089,
+      "step": 7232
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.507065785339764e-05,
+      "loss": 18.4525,
+      "step": 7233
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.506382291386594e-05,
+      "loss": 18.0104,
+      "step": 7234
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.505698734873549e-05,
+      "loss": 18.4473,
+      "step": 7235
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.505015115817694e-05,
+      "loss": 17.9295,
+      "step": 7236
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.504331434236089e-05,
+      "loss": 0.76,
+      "step": 7237
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.503647690145799e-05,
+      "loss": 19.3272,
+      "step": 7238
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.502963883563892e-05,
+      "loss": 15.9632,
+      "step": 7239
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.502280014507433e-05,
+      "loss": 20.1877,
+      "step": 7240
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.501596082993496e-05,
+      "loss": 20.2191,
+      "step": 7241
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.500912089039147e-05,
+      "loss": 0.7862,
+      "step": 7242
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.500228032661463e-05,
+      "loss": 17.8902,
+      "step": 7243
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.499543913877517e-05,
+      "loss": 16.6952,
+      "step": 7244
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.498859732704382e-05,
+      "loss": 21.5681,
+      "step": 7245
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.49817548915914e-05,
+      "loss": 0.7657,
+      "step": 7246
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.497491183258867e-05,
+      "loss": 0.7622,
+      "step": 7247
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.496806815020642e-05,
+      "loss": 18.8527,
+      "step": 7248
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.49612238446155e-05,
+      "loss": 0.7736,
+      "step": 7249
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.495437891598672e-05,
+      "loss": 21.6288,
+      "step": 7250
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.494753336449096e-05,
+      "loss": 0.762,
+      "step": 7251
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.494068719029906e-05,
+      "loss": 19.0804,
+      "step": 7252
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.493384039358192e-05,
+      "loss": 18.5354,
+      "step": 7253
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.492699297451043e-05,
+      "loss": 18.9669,
+      "step": 7254
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.492014493325551e-05,
+      "loss": 0.7712,
+      "step": 7255
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.491329626998808e-05,
+      "loss": 18.1833,
+      "step": 7256
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.490644698487909e-05,
+      "loss": 0.7471,
+      "step": 7257
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.489959707809948e-05,
+      "loss": 0.758,
+      "step": 7258
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.489274654982024e-05,
+      "loss": 0.7799,
+      "step": 7259
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.488589540021238e-05,
+      "loss": 19.7701,
+      "step": 7260
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.487904362944688e-05,
+      "loss": 0.7521,
+      "step": 7261
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.487219123769477e-05,
+      "loss": 0.7891,
+      "step": 7262
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.486533822512709e-05,
+      "loss": 17.8198,
+      "step": 7263
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.48584845919149e-05,
+      "loss": 19.283,
+      "step": 7264
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.485163033822925e-05,
+      "loss": 18.8807,
+      "step": 7265
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.484477546424122e-05,
+      "loss": 0.7596,
+      "step": 7266
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.483791997012195e-05,
+      "loss": 0.7704,
+      "step": 7267
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.48310638560425e-05,
+      "loss": 19.8994,
+      "step": 7268
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.482420712217403e-05,
+      "loss": 19.2053,
+      "step": 7269
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.48173497686877e-05,
+      "loss": 20.5417,
+      "step": 7270
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.481049179575463e-05,
+      "loss": 19.8668,
+      "step": 7271
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.480363320354604e-05,
+      "loss": 17.0611,
+      "step": 7272
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.479677399223308e-05,
+      "loss": 0.7142,
+      "step": 7273
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.4789914161987e-05,
+      "loss": 20.2591,
+      "step": 7274
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.4783053712979e-05,
+      "loss": 0.6762,
+      "step": 7275
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.477619264538031e-05,
+      "loss": 19.6155,
+      "step": 7276
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.47693309593622e-05,
+      "loss": 0.7609,
+      "step": 7277
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.476246865509593e-05,
+      "loss": 20.39,
+      "step": 7278
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.47556057327528e-05,
+      "loss": 0.7822,
+      "step": 7279
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.474874219250408e-05,
+      "loss": 17.7844,
+      "step": 7280
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.474187803452113e-05,
+      "loss": 19.5172,
+      "step": 7281
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.473501325897523e-05,
+      "loss": 0.7944,
+      "step": 7282
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.472814786603775e-05,
+      "loss": 0.7947,
+      "step": 7283
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.472128185588007e-05,
+      "loss": 17.8932,
+      "step": 7284
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.471441522867355e-05,
+      "loss": 0.78,
+      "step": 7285
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.470754798458956e-05,
+      "loss": 0.7624,
+      "step": 7286
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.470068012379956e-05,
+      "loss": 0.8188,
+      "step": 7287
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.469381164647494e-05,
+      "loss": 17.7493,
+      "step": 7288
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.468694255278713e-05,
+      "loss": 0.7399,
+      "step": 7289
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.46800728429076e-05,
+      "loss": 0.7575,
+      "step": 7290
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.467320251700781e-05,
+      "loss": 0.7744,
+      "step": 7291
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.466633157525927e-05,
+      "loss": 0.7985,
+      "step": 7292
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.465946001783345e-05,
+      "loss": 19.1533,
+      "step": 7293
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.465258784490188e-05,
+      "loss": 0.7373,
+      "step": 7294
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.46457150566361e-05,
+      "loss": 22.7553,
+      "step": 7295
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.463884165320762e-05,
+      "loss": 0.7591,
+      "step": 7296
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.463196763478804e-05,
+      "loss": 0.7799,
+      "step": 7297
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.462509300154892e-05,
+      "loss": 0.7568,
+      "step": 7298
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.461821775366186e-05,
+      "loss": 19.0915,
+      "step": 7299
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.461134189129846e-05,
+      "loss": 0.7469,
+      "step": 7300
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.460446541463034e-05,
+      "loss": 19.0708,
+      "step": 7301
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.459758832382917e-05,
+      "loss": 19.3265,
+      "step": 7302
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.459071061906656e-05,
+      "loss": 19.9918,
+      "step": 7303
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.45838323005142e-05,
+      "loss": 0.7102,
+      "step": 7304
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.457695336834378e-05,
+      "loss": 0.7594,
+      "step": 7305
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.457007382272698e-05,
+      "loss": 0.7979,
+      "step": 7306
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.456319366383554e-05,
+      "loss": 0.82,
+      "step": 7307
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.455631289184116e-05,
+      "loss": 0.7803,
+      "step": 7308
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.454943150691563e-05,
+      "loss": 0.7647,
+      "step": 7309
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.454254950923066e-05,
+      "loss": 19.4655,
+      "step": 7310
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.453566689895806e-05,
+      "loss": 19.3423,
+      "step": 7311
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.452878367626962e-05,
+      "loss": 17.4276,
+      "step": 7312
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.452189984133713e-05,
+      "loss": 20.9811,
+      "step": 7313
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.451501539433242e-05,
+      "loss": 0.782,
+      "step": 7314
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.450813033542732e-05,
+      "loss": 0.8048,
+      "step": 7315
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.45012446647937e-05,
+      "loss": 19.1589,
+      "step": 7316
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.449435838260343e-05,
+      "loss": 18.2185,
+      "step": 7317
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.448747148902836e-05,
+      "loss": 0.6791,
+      "step": 7318
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.448058398424043e-05,
+      "loss": 19.2136,
+      "step": 7319
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.447369586841152e-05,
+      "loss": 19.7078,
+      "step": 7320
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.446680714171359e-05,
+      "loss": 0.7728,
+      "step": 7321
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.445991780431855e-05,
+      "loss": 20.1223,
+      "step": 7322
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.445302785639838e-05,
+      "loss": 19.0575,
+      "step": 7323
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.444613729812506e-05,
+      "loss": 21.8307,
+      "step": 7324
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.443924612967054e-05,
+      "loss": 19.1874,
+      "step": 7325
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.443235435120691e-05,
+      "loss": 0.8001,
+      "step": 7326
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.44254619629061e-05,
+      "loss": 19.7427,
+      "step": 7327
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.441856896494019e-05,
+      "loss": 20.3586,
+      "step": 7328
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.441167535748122e-05,
+      "loss": 18.6957,
+      "step": 7329
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.440478114070126e-05,
+      "loss": 0.75,
+      "step": 7330
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.43978863147724e-05,
+      "loss": 0.6628,
+      "step": 7331
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.439099087986672e-05,
+      "loss": 0.7512,
+      "step": 7332
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.438409483615634e-05,
+      "loss": 20.0202,
+      "step": 7333
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.437719818381338e-05,
+      "loss": 18.2367,
+      "step": 7334
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.437030092301e-05,
+      "loss": 19.6809,
+      "step": 7335
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.436340305391833e-05,
+      "loss": 0.7832,
+      "step": 7336
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.435650457671058e-05,
+      "loss": 18.9848,
+      "step": 7337
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.434960549155889e-05,
+      "loss": 20.0105,
+      "step": 7338
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.434270579863548e-05,
+      "loss": 18.6023,
+      "step": 7339
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.433580549811261e-05,
+      "loss": 19.6947,
+      "step": 7340
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.432890459016244e-05,
+      "loss": 20.2525,
+      "step": 7341
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.432200307495728e-05,
+      "loss": 19.8285,
+      "step": 7342
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.431510095266936e-05,
+      "loss": 18.8724,
+      "step": 7343
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.430819822347097e-05,
+      "loss": 19.7646,
+      "step": 7344
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.430129488753441e-05,
+      "loss": 17.4922,
+      "step": 7345
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.429439094503197e-05,
+      "loss": 19.8696,
+      "step": 7346
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.4287486396136e-05,
+      "loss": 0.7761,
+      "step": 7347
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.42805812410188e-05,
+      "loss": 0.7699,
+      "step": 7348
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.427367547985278e-05,
+      "loss": 0.7775,
+      "step": 7349
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.426676911281025e-05,
+      "loss": 0.8582,
+      "step": 7350
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.425986214006364e-05,
+      "loss": 0.7622,
+      "step": 7351
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.425295456178532e-05,
+      "loss": 20.5315,
+      "step": 7352
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.424604637814772e-05,
+      "loss": 0.7254,
+      "step": 7353
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.423913758932327e-05,
+      "loss": 21.9531,
+      "step": 7354
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.42322281954844e-05,
+      "loss": 0.7697,
+      "step": 7355
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.422531819680358e-05,
+      "loss": 0.7321,
+      "step": 7356
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.42184075934533e-05,
+      "loss": 0.7582,
+      "step": 7357
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.421149638560604e-05,
+      "loss": 18.6236,
+      "step": 7358
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.42045845734343e-05,
+      "loss": 18.0507,
+      "step": 7359
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.419767215711059e-05,
+      "loss": 0.73,
+      "step": 7360
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.419075913680746e-05,
+      "loss": 0.7784,
+      "step": 7361
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.418384551269747e-05,
+      "loss": 18.4142,
+      "step": 7362
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.417693128495317e-05,
+      "loss": 18.9817,
+      "step": 7363
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.417001645374715e-05,
+      "loss": 0.7296,
+      "step": 7364
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.416310101925199e-05,
+      "loss": 19.7234,
+      "step": 7365
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.41561849816403e-05,
+      "loss": 0.7644,
+      "step": 7366
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.414926834108473e-05,
+      "loss": 0.7683,
+      "step": 7367
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.414235109775792e-05,
+      "loss": 0.7617,
+      "step": 7368
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.41354332518325e-05,
+      "loss": 0.8528,
+      "step": 7369
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.412851480348115e-05,
+      "loss": 0.7564,
+      "step": 7370
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.412159575287659e-05,
+      "loss": 0.7826,
+      "step": 7371
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.411467610019146e-05,
+      "loss": 20.578,
+      "step": 7372
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.410775584559853e-05,
+      "loss": 0.8143,
+      "step": 7373
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.41008349892705e-05,
+      "loss": 20.7471,
+      "step": 7374
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.409391353138012e-05,
+      "loss": 0.7426,
+      "step": 7375
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.408699147210016e-05,
+      "loss": 18.0315,
+      "step": 7376
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.408006881160341e-05,
+      "loss": 20.8973,
+      "step": 7377
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.407314555006262e-05,
+      "loss": 0.771,
+      "step": 7378
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.406622168765064e-05,
+      "loss": 0.7738,
+      "step": 7379
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.405929722454026e-05,
+      "loss": 0.7514,
+      "step": 7380
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.405237216090432e-05,
+      "loss": 22.3494,
+      "step": 7381
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.404544649691571e-05,
+      "loss": 18.6264,
+      "step": 7382
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.403852023274723e-05,
+      "loss": 21.0519,
+      "step": 7383
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.40315933685718e-05,
+      "loss": 18.6886,
+      "step": 7384
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.402466590456233e-05,
+      "loss": 0.753,
+      "step": 7385
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.40177378408917e-05,
+      "loss": 0.7908,
+      "step": 7386
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.401080917773287e-05,
+      "loss": 0.8034,
+      "step": 7387
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.400387991525875e-05,
+      "loss": 19.729,
+      "step": 7388
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.39969500536423e-05,
+      "loss": 18.6093,
+      "step": 7389
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.39900195930565e-05,
+      "loss": 0.7261,
+      "step": 7390
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.398308853367434e-05,
+      "loss": 19.6633,
+      "step": 7391
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.397615687566881e-05,
+      "loss": 19.7069,
+      "step": 7392
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.396922461921291e-05,
+      "loss": 0.755,
+      "step": 7393
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.396229176447972e-05,
+      "loss": 21.4682,
+      "step": 7394
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.395535831164224e-05,
+      "loss": 20.0838,
+      "step": 7395
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.394842426087356e-05,
+      "loss": 18.7651,
+      "step": 7396
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.394148961234673e-05,
+      "loss": 0.797,
+      "step": 7397
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.393455436623486e-05,
+      "loss": 0.7405,
+      "step": 7398
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.392761852271103e-05,
+      "loss": 0.7548,
+      "step": 7399
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.39206820819484e-05,
+      "loss": 20.0356,
+      "step": 7400
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.391374504412009e-05,
+      "loss": 0.6927,
+      "step": 7401
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.390680740939923e-05,
+      "loss": 23.3492,
+      "step": 7402
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.3899869177959e-05,
+      "loss": 0.7265,
+      "step": 7403
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.389293034997256e-05,
+      "loss": 0.7255,
+      "step": 7404
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.388599092561315e-05,
+      "loss": 19.9975,
+      "step": 7405
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.387905090505395e-05,
+      "loss": 0.7463,
+      "step": 7406
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.387211028846818e-05,
+      "loss": 0.7565,
+      "step": 7407
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.386516907602907e-05,
+      "loss": 0.7532,
+      "step": 7408
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.385822726790992e-05,
+      "loss": 18.1859,
+      "step": 7409
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.385128486428397e-05,
+      "loss": 0.8345,
+      "step": 7410
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.384434186532447e-05,
+      "loss": 0.7659,
+      "step": 7411
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.383739827120477e-05,
+      "loss": 0.7848,
+      "step": 7412
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.383045408209818e-05,
+      "loss": 0.7587,
+      "step": 7413
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.3823509298178e-05,
+      "loss": 0.6947,
+      "step": 7414
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.38165639196176e-05,
+      "loss": 21.2762,
+      "step": 7415
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.380961794659031e-05,
+      "loss": 20.0701,
+      "step": 7416
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.380267137926953e-05,
+      "loss": 18.7018,
+      "step": 7417
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.379572421782863e-05,
+      "loss": 0.7539,
+      "step": 7418
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.378877646244103e-05,
+      "loss": 0.7415,
+      "step": 7419
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.378182811328016e-05,
+      "loss": 0.7715,
+      "step": 7420
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.377487917051939e-05,
+      "loss": 0.8029,
+      "step": 7421
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.376792963433222e-05,
+      "loss": 18.0413,
+      "step": 7422
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.37609795048921e-05,
+      "loss": 0.723,
+      "step": 7423
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.375402878237252e-05,
+      "loss": 21.8572,
+      "step": 7424
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.374707746694693e-05,
+      "loss": 0.7185,
+      "step": 7425
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.374012555878888e-05,
+      "loss": 18.3302,
+      "step": 7426
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.373317305807188e-05,
+      "loss": 17.8602,
+      "step": 7427
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.372621996496945e-05,
+      "loss": 0.751,
+      "step": 7428
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.371926627965516e-05,
+      "loss": 19.2652,
+      "step": 7429
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.371231200230258e-05,
+      "loss": 0.751,
+      "step": 7430
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.370535713308526e-05,
+      "loss": 0.8154,
+      "step": 7431
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.369840167217681e-05,
+      "loss": 0.7877,
+      "step": 7432
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.369144561975085e-05,
+      "loss": 0.7493,
+      "step": 7433
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.368448897598101e-05,
+      "loss": 0.7229,
+      "step": 7434
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.367753174104091e-05,
+      "loss": 0.7793,
+      "step": 7435
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.36705739151042e-05,
+      "loss": 0.7556,
+      "step": 7436
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.366361549834457e-05,
+      "loss": 20.8853,
+      "step": 7437
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.36566564909357e-05,
+      "loss": 20.9735,
+      "step": 7438
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.364969689305127e-05,
+      "loss": 19.921,
+      "step": 7439
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.364273670486501e-05,
+      "loss": 20.5418,
+      "step": 7440
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.363577592655064e-05,
+      "loss": 0.7357,
+      "step": 7441
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.362881455828192e-05,
+      "loss": 18.7738,
+      "step": 7442
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.362185260023258e-05,
+      "loss": 0.7211,
+      "step": 7443
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.361489005257642e-05,
+      "loss": 0.7644,
+      "step": 7444
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.36079269154872e-05,
+      "loss": 20.1195,
+      "step": 7445
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.360096318913872e-05,
+      "loss": 0.7438,
+      "step": 7446
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.359399887370484e-05,
+      "loss": 0.7774,
+      "step": 7447
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.358703396935935e-05,
+      "loss": 21.0321,
+      "step": 7448
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.358006847627609e-05,
+      "loss": 0.7484,
+      "step": 7449
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.357310239462894e-05,
+      "loss": 0.7267,
+      "step": 7450
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.356613572459178e-05,
+      "loss": 19.3176,
+      "step": 7451
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.35591684663385e-05,
+      "loss": 17.4125,
+      "step": 7452
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.3552200620043e-05,
+      "loss": 19.1923,
+      "step": 7453
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.354523218587918e-05,
+      "loss": 20.9093,
+      "step": 7454
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.3538263164021e-05,
+      "loss": 19.5592,
+      "step": 7455
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.353129355464238e-05,
+      "loss": 0.8174,
+      "step": 7456
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.352432335791732e-05,
+      "loss": 0.7951,
+      "step": 7457
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.351735257401976e-05,
+      "loss": 20.6399,
+      "step": 7458
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.35103812031237e-05,
+      "loss": 0.7297,
+      "step": 7459
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.350340924540318e-05,
+      "loss": 20.1734,
+      "step": 7460
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.34964367010322e-05,
+      "loss": 0.744,
+      "step": 7461
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.348946357018479e-05,
+      "loss": 0.732,
+      "step": 7462
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.3482489853035e-05,
+      "loss": 21.0614,
+      "step": 7463
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.347551554975691e-05,
+      "loss": 0.7695,
+      "step": 7464
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.346854066052459e-05,
+      "loss": 19.8613,
+      "step": 7465
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.346156518551213e-05,
+      "loss": 0.7846,
+      "step": 7466
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.345458912489365e-05,
+      "loss": 0.7732,
+      "step": 7467
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.344761247884328e-05,
+      "loss": 0.7853,
+      "step": 7468
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.344063524753513e-05,
+      "loss": 0.745,
+      "step": 7469
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.343365743114338e-05,
+      "loss": 19.5197,
+      "step": 7470
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.34266790298422e-05,
+      "loss": 0.736,
+      "step": 7471
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.341970004380574e-05,
+      "loss": 0.7587,
+      "step": 7472
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.341272047320823e-05,
+      "loss": 0.7221,
+      "step": 7473
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.340574031822387e-05,
+      "loss": 18.5294,
+      "step": 7474
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.33987595790269e-05,
+      "loss": 0.7807,
+      "step": 7475
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.339177825579155e-05,
+      "loss": 17.2436,
+      "step": 7476
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.338479634869204e-05,
+      "loss": 0.7523,
+      "step": 7477
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.337781385790271e-05,
+      "loss": 0.7507,
+      "step": 7478
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.337083078359778e-05,
+      "loss": 18.0959,
+      "step": 7479
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.33638471259516e-05,
+      "loss": 19.1929,
+      "step": 7480
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.335686288513846e-05,
+      "loss": 20.3024,
+      "step": 7481
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.334987806133269e-05,
+      "loss": 0.767,
+      "step": 7482
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.334289265470862e-05,
+      "loss": 21.5402,
+      "step": 7483
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.333590666544062e-05,
+      "loss": 19.1384,
+      "step": 7484
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.332892009370306e-05,
+      "loss": 20.4601,
+      "step": 7485
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.332193293967034e-05,
+      "loss": 0.7703,
+      "step": 7486
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.331494520351684e-05,
+      "loss": 17.9455,
+      "step": 7487
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.330795688541698e-05,
+      "loss": 19.9999,
+      "step": 7488
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.33009679855452e-05,
+      "loss": 21.5061,
+      "step": 7489
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.329397850407596e-05,
+      "loss": 0.7468,
+      "step": 7490
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.328698844118366e-05,
+      "loss": 0.7776,
+      "step": 7491
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.327999779704283e-05,
+      "loss": 0.7438,
+      "step": 7492
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.327300657182794e-05,
+      "loss": 19.7175,
+      "step": 7493
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.326601476571349e-05,
+      "loss": 18.1042,
+      "step": 7494
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.3259022378874e-05,
+      "loss": 0.8028,
+      "step": 7495
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.3252029411484e-05,
+      "loss": 18.799,
+      "step": 7496
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.324503586371802e-05,
+      "loss": 0.792,
+      "step": 7497
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.323804173575066e-05,
+      "loss": 18.4566,
+      "step": 7498
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.323104702775648e-05,
+      "loss": 16.9055,
+      "step": 7499
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.322405173991005e-05,
+      "loss": 0.8054,
+      "step": 7500
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.321705587238597e-05,
+      "loss": 20.4938,
+      "step": 7501
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.32100594253589e-05,
+      "loss": 0.7668,
+      "step": 7502
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.320306239900343e-05,
+      "loss": 21.874,
+      "step": 7503
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.319606479349424e-05,
+      "loss": 22.084,
+      "step": 7504
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.318906660900596e-05,
+      "loss": 0.7471,
+      "step": 7505
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.318206784571329e-05,
+      "loss": 0.7336,
+      "step": 7506
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.31750685037909e-05,
+      "loss": 0.7754,
+      "step": 7507
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.316806858341351e-05,
+      "loss": 0.764,
+      "step": 7508
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.316106808475588e-05,
+      "loss": 0.776,
+      "step": 7509
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.315406700799267e-05,
+      "loss": 22.0629,
+      "step": 7510
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.314706535329866e-05,
+      "loss": 19.3135,
+      "step": 7511
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.31400631208486e-05,
+      "loss": 18.2852,
+      "step": 7512
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.31330603108173e-05,
+      "loss": 0.7711,
+      "step": 7513
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.312605692337953e-05,
+      "loss": 21.0306,
+      "step": 7514
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.31190529587101e-05,
+      "loss": 0.7118,
+      "step": 7515
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.311204841698384e-05,
+      "loss": 21.0845,
+      "step": 7516
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.310504329837556e-05,
+      "loss": 0.7212,
+      "step": 7517
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.309803760306012e-05,
+      "loss": 18.9383,
+      "step": 7518
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.309103133121239e-05,
+      "loss": 18.3532,
+      "step": 7519
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.308402448300725e-05,
+      "loss": 0.7751,
+      "step": 7520
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.307701705861958e-05,
+      "loss": 0.8184,
+      "step": 7521
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.30700090582243e-05,
+      "loss": 0.7483,
+      "step": 7522
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.306300048199635e-05,
+      "loss": 20.3316,
+      "step": 7523
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.305599133011061e-05,
+      "loss": 21.0941,
+      "step": 7524
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.304898160274208e-05,
+      "loss": 19.1247,
+      "step": 7525
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.304197130006568e-05,
+      "loss": 0.7962,
+      "step": 7526
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.303496042225643e-05,
+      "loss": 19.4648,
+      "step": 7527
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.302794896948932e-05,
+      "loss": 0.8049,
+      "step": 7528
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.302093694193935e-05,
+      "loss": 0.7477,
+      "step": 7529
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.301392433978153e-05,
+      "loss": 0.7728,
+      "step": 7530
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.300691116319089e-05,
+      "loss": 0.7885,
+      "step": 7531
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.29998974123425e-05,
+      "loss": 0.7256,
+      "step": 7532
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.299288308741143e-05,
+      "loss": 0.7569,
+      "step": 7533
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.298586818857273e-05,
+      "loss": 18.5372,
+      "step": 7534
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.297885271600152e-05,
+      "loss": 0.7717,
+      "step": 7535
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.29718366698729e-05,
+      "loss": 0.8237,
+      "step": 7536
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.296482005036201e-05,
+      "loss": 19.6811,
+      "step": 7537
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.295780285764396e-05,
+      "loss": 18.1973,
+      "step": 7538
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.295078509189388e-05,
+      "loss": 18.6082,
+      "step": 7539
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.294376675328698e-05,
+      "loss": 0.7544,
+      "step": 7540
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.293674784199843e-05,
+      "loss": 19.9314,
+      "step": 7541
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.292972835820341e-05,
+      "loss": 19.8815,
+      "step": 7542
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.292270830207715e-05,
+      "loss": 0.7356,
+      "step": 7543
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.291568767379484e-05,
+      "loss": 0.746,
+      "step": 7544
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.290866647353175e-05,
+      "loss": 0.7437,
+      "step": 7545
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.29016447014631e-05,
+      "loss": 19.5373,
+      "step": 7546
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.289462235776417e-05,
+      "loss": 0.7138,
+      "step": 7547
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.288759944261024e-05,
+      "loss": 19.7671,
+      "step": 7548
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.288057595617658e-05,
+      "loss": 0.8014,
+      "step": 7549
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.287355189863856e-05,
+      "loss": 0.733,
+      "step": 7550
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.286652727017144e-05,
+      "loss": 0.7346,
+      "step": 7551
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.28595020709506e-05,
+      "loss": 19.3463,
+      "step": 7552
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.285247630115132e-05,
+      "loss": 20.4666,
+      "step": 7553
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.284544996094906e-05,
+      "loss": 21.1916,
+      "step": 7554
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.283842305051913e-05,
+      "loss": 0.7613,
+      "step": 7555
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.283139557003697e-05,
+      "loss": 19.5556,
+      "step": 7556
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.282436751967793e-05,
+      "loss": 17.6043,
+      "step": 7557
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.28173388996175e-05,
+      "loss": 20.0963,
+      "step": 7558
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.281030971003106e-05,
+      "loss": 19.7088,
+      "step": 7559
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.280327995109409e-05,
+      "loss": 22.1156,
+      "step": 7560
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.279624962298203e-05,
+      "loss": 19.783,
+      "step": 7561
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.27892187258704e-05,
+      "loss": 17.722,
+      "step": 7562
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.278218725993463e-05,
+      "loss": 19.9414,
+      "step": 7563
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.277515522535028e-05,
+      "loss": 0.805,
+      "step": 7564
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.276812262229287e-05,
+      "loss": 19.0281,
+      "step": 7565
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.276108945093791e-05,
+      "loss": 18.1647,
+      "step": 7566
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.275405571146097e-05,
+      "loss": 0.7623,
+      "step": 7567
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.27470214040376e-05,
+      "loss": 19.3871,
+      "step": 7568
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.273998652884337e-05,
+      "loss": 0.7951,
+      "step": 7569
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.273295108605389e-05,
+      "loss": 20.776,
+      "step": 7570
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.272591507584478e-05,
+      "loss": 0.7345,
+      "step": 7571
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.27188784983916e-05,
+      "loss": 21.2908,
+      "step": 7572
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.271184135387005e-05,
+      "loss": 0.7881,
+      "step": 7573
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.270480364245576e-05,
+      "loss": 0.77,
+      "step": 7574
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.269776536432439e-05,
+      "loss": 0.7729,
+      "step": 7575
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.269072651965159e-05,
+      "loss": 19.2763,
+      "step": 7576
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.268368710861308e-05,
+      "loss": 0.806,
+      "step": 7577
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.267664713138458e-05,
+      "loss": 0.733,
+      "step": 7578
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.26696065881418e-05,
+      "loss": 20.6702,
+      "step": 7579
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.266256547906043e-05,
+      "loss": 0.7582,
+      "step": 7580
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.265552380431625e-05,
+      "loss": 20.2144,
+      "step": 7581
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.264848156408504e-05,
+      "loss": 20.3261,
+      "step": 7582
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.264143875854254e-05,
+      "loss": 0.787,
+      "step": 7583
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.263439538786459e-05,
+      "loss": 0.7621,
+      "step": 7584
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.262735145222696e-05,
+      "loss": 20.7525,
+      "step": 7585
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.262030695180545e-05,
+      "loss": 19.8513,
+      "step": 7586
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.26132618867759e-05,
+      "loss": 0.7478,
+      "step": 7587
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.260621625731421e-05,
+      "loss": 0.7914,
+      "step": 7588
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.25991700635962e-05,
+      "loss": 0.7732,
+      "step": 7589
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.259212330579771e-05,
+      "loss": 19.6297,
+      "step": 7590
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.258507598409468e-05,
+      "loss": 0.7416,
+      "step": 7591
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.2578028098663e-05,
+      "loss": 18.2256,
+      "step": 7592
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.257097964967859e-05,
+      "loss": 18.0513,
+      "step": 7593
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.256393063731736e-05,
+      "loss": 0.7368,
+      "step": 7594
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.255688106175527e-05,
+      "loss": 20.8819,
+      "step": 7595
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.254983092316828e-05,
+      "loss": 0.7914,
+      "step": 7596
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.254278022173237e-05,
+      "loss": 17.0996,
+      "step": 7597
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.253572895762351e-05,
+      "loss": 18.7186,
+      "step": 7598
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.252867713101771e-05,
+      "loss": 0.7329,
+      "step": 7599
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.252162474209099e-05,
+      "loss": 0.74,
+      "step": 7600
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.251457179101938e-05,
+      "loss": 0.7892,
+      "step": 7601
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.25075182779789e-05,
+      "loss": 19.9612,
+      "step": 7602
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.250046420314565e-05,
+      "loss": 17.0575,
+      "step": 7603
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.249340956669565e-05,
+      "loss": 18.3086,
+      "step": 7604
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.248635436880501e-05,
+      "loss": 0.7417,
+      "step": 7605
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.247929860964985e-05,
+      "loss": 20.922,
+      "step": 7606
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.247224228940625e-05,
+      "loss": 0.7152,
+      "step": 7607
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.246518540825036e-05,
+      "loss": 18.4533,
+      "step": 7608
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.245812796635831e-05,
+      "loss": 0.7676,
+      "step": 7609
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.245106996390627e-05,
+      "loss": 0.777,
+      "step": 7610
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.244401140107037e-05,
+      "loss": 0.771,
+      "step": 7611
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.243695227802684e-05,
+      "loss": 18.0625,
+      "step": 7612
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.242989259495186e-05,
+      "loss": 18.6189,
+      "step": 7613
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.242283235202164e-05,
+      "loss": 18.3102,
+      "step": 7614
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.241577154941238e-05,
+      "loss": 20.2708,
+      "step": 7615
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.240871018730037e-05,
+      "loss": 18.3808,
+      "step": 7616
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.240164826586184e-05,
+      "loss": 0.7667,
+      "step": 7617
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.239458578527303e-05,
+      "loss": 19.3202,
+      "step": 7618
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.238752274571025e-05,
+      "loss": 19.5599,
+      "step": 7619
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.238045914734981e-05,
+      "loss": 0.8208,
+      "step": 7620
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.237339499036799e-05,
+      "loss": 0.7721,
+      "step": 7621
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.236633027494112e-05,
+      "loss": 0.7553,
+      "step": 7622
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.235926500124555e-05,
+      "loss": 0.7232,
+      "step": 7623
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.23521991694576e-05,
+      "loss": 0.78,
+      "step": 7624
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.234513277975366e-05,
+      "loss": 19.8658,
+      "step": 7625
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.233806583231011e-05,
+      "loss": 0.7508,
+      "step": 7626
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.233099832730335e-05,
+      "loss": 18.3943,
+      "step": 7627
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.232393026490976e-05,
+      "loss": 20.6784,
+      "step": 7628
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.231686164530577e-05,
+      "loss": 0.738,
+      "step": 7629
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.230979246866782e-05,
+      "loss": 0.7644,
+      "step": 7630
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.230272273517235e-05,
+      "loss": 18.7996,
+      "step": 7631
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.229565244499584e-05,
+      "loss": 18.3536,
+      "step": 7632
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.228858159831474e-05,
+      "loss": 0.7573,
+      "step": 7633
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.228151019530558e-05,
+      "loss": 0.7225,
+      "step": 7634
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.227443823614482e-05,
+      "loss": 18.7872,
+      "step": 7635
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.2267365721009e-05,
+      "loss": 0.7844,
+      "step": 7636
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.226029265007464e-05,
+      "loss": 0.8004,
+      "step": 7637
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.225321902351829e-05,
+      "loss": 0.7148,
+      "step": 7638
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.224614484151652e-05,
+      "loss": 21.2403,
+      "step": 7639
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.223907010424589e-05,
+      "loss": 0.7552,
+      "step": 7640
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.223199481188301e-05,
+      "loss": 18.5597,
+      "step": 7641
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.222491896460444e-05,
+      "loss": 0.7661,
+      "step": 7642
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.221784256258683e-05,
+      "loss": 18.7679,
+      "step": 7643
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.221076560600679e-05,
+      "loss": 18.2021,
+      "step": 7644
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.220368809504099e-05,
+      "loss": 18.2856,
+      "step": 7645
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.219661002986604e-05,
+      "loss": 0.7786,
+      "step": 7646
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.218953141065865e-05,
+      "loss": 0.792,
+      "step": 7647
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.218245223759548e-05,
+      "loss": 0.7287,
+      "step": 7648
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.217537251085324e-05,
+      "loss": 20.8334,
+      "step": 7649
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.216829223060864e-05,
+      "loss": 18.8152,
+      "step": 7650
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.216121139703841e-05,
+      "loss": 18.3704,
+      "step": 7651
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.215413001031927e-05,
+      "loss": 20.3579,
+      "step": 7652
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.2147048070628e-05,
+      "loss": 0.7597,
+      "step": 7653
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.213996557814135e-05,
+      "loss": 0.7647,
+      "step": 7654
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.213288253303609e-05,
+      "loss": 18.1245,
+      "step": 7655
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.212579893548905e-05,
+      "loss": 19.5325,
+      "step": 7656
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.2118714785677e-05,
+      "loss": 17.4662,
+      "step": 7657
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.21116300837768e-05,
+      "loss": 19.6607,
+      "step": 7658
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.210454482996525e-05,
+      "loss": 20.5055,
+      "step": 7659
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.20974590244192e-05,
+      "loss": 0.7742,
+      "step": 7660
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.209037266731555e-05,
+      "loss": 0.819,
+      "step": 7661
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.208328575883114e-05,
+      "loss": 0.7678,
+      "step": 7662
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.207619829914287e-05,
+      "loss": 21.208,
+      "step": 7663
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.206911028842766e-05,
+      "loss": 19.3825,
+      "step": 7664
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.206202172686243e-05,
+      "loss": 0.7944,
+      "step": 7665
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.205493261462408e-05,
+      "loss": 20.0702,
+      "step": 7666
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.204784295188959e-05,
+      "loss": 0.796,
+      "step": 7667
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.20407527388359e-05,
+      "loss": 20.0644,
+      "step": 7668
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.203366197564e-05,
+      "loss": 22.0243,
+      "step": 7669
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.202657066247884e-05,
+      "loss": 0.7517,
+      "step": 7670
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.201947879952946e-05,
+      "loss": 18.7285,
+      "step": 7671
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.201238638696888e-05,
+      "loss": 0.7598,
+      "step": 7672
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.20052934249741e-05,
+      "loss": 0.7623,
+      "step": 7673
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.199819991372216e-05,
+      "loss": 0.8152,
+      "step": 7674
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.199110585339012e-05,
+      "loss": 0.7736,
+      "step": 7675
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.198401124415508e-05,
+      "loss": 0.798,
+      "step": 7676
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.197691608619408e-05,
+      "loss": 18.614,
+      "step": 7677
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.196982037968424e-05,
+      "loss": 0.7654,
+      "step": 7678
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.19627241248027e-05,
+      "loss": 0.7127,
+      "step": 7679
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.19556273217265e-05,
+      "loss": 17.4982,
+      "step": 7680
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.194852997063286e-05,
+      "loss": 0.7516,
+      "step": 7681
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.194143207169889e-05,
+      "loss": 19.6599,
+      "step": 7682
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.193433362510178e-05,
+      "loss": 0.8034,
+      "step": 7683
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.192723463101867e-05,
+      "loss": 0.7267,
+      "step": 7684
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.192013508962681e-05,
+      "loss": 17.8641,
+      "step": 7685
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.191303500110333e-05,
+      "loss": 18.9444,
+      "step": 7686
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.190593436562553e-05,
+      "loss": 20.21,
+      "step": 7687
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.189883318337059e-05,
+      "loss": 0.7643,
+      "step": 7688
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.189173145451578e-05,
+      "loss": 18.7882,
+      "step": 7689
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.188462917923835e-05,
+      "loss": 17.0947,
+      "step": 7690
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.187752635771557e-05,
+      "loss": 19.2902,
+      "step": 7691
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.187042299012477e-05,
+      "loss": 18.5247,
+      "step": 7692
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.186331907664319e-05,
+      "loss": 19.0258,
+      "step": 7693
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.185621461744816e-05,
+      "loss": 0.7707,
+      "step": 7694
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.184910961271705e-05,
+      "loss": 0.7961,
+      "step": 7695
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.184200406262717e-05,
+      "loss": 0.7826,
+      "step": 7696
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.183489796735587e-05,
+      "loss": 20.0559,
+      "step": 7697
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.182779132708054e-05,
+      "loss": 19.6317,
+      "step": 7698
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.182068414197856e-05,
+      "loss": 19.2871,
+      "step": 7699
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.181357641222731e-05,
+      "loss": 0.7605,
+      "step": 7700
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.180646813800423e-05,
+      "loss": 20.1781,
+      "step": 7701
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.17993593194867e-05,
+      "loss": 17.2221,
+      "step": 7702
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.17922499568522e-05,
+      "loss": 19.3802,
+      "step": 7703
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.178514005027817e-05,
+      "loss": 0.752,
+      "step": 7704
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.177802959994205e-05,
+      "loss": 18.3918,
+      "step": 7705
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.177091860602137e-05,
+      "loss": 0.7273,
+      "step": 7706
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.176380706869357e-05,
+      "loss": 16.9257,
+      "step": 7707
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.175669498813617e-05,
+      "loss": 0.7404,
+      "step": 7708
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.174958236452669e-05,
+      "loss": 19.2924,
+      "step": 7709
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.174246919804269e-05,
+      "loss": 0.7403,
+      "step": 7710
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.17353554888617e-05,
+      "loss": 17.0351,
+      "step": 7711
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.172824123716124e-05,
+      "loss": 21.81,
+      "step": 7712
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.172112644311894e-05,
+      "loss": 19.875,
+      "step": 7713
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.171401110691237e-05,
+      "loss": 0.763,
+      "step": 7714
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.170689522871911e-05,
+      "loss": 0.7871,
+      "step": 7715
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.16997788087168e-05,
+      "loss": 0.7325,
+      "step": 7716
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.169266184708306e-05,
+      "loss": 0.8112,
+      "step": 7717
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.168554434399551e-05,
+      "loss": 18.7421,
+      "step": 7718
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.167842629963183e-05,
+      "loss": 0.7965,
+      "step": 7719
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.16713077141697e-05,
+      "loss": 19.1087,
+      "step": 7720
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.166418858778675e-05,
+      "loss": 0.7731,
+      "step": 7721
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.165706892066072e-05,
+      "loss": 18.298,
+      "step": 7722
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.16499487129693e-05,
+      "loss": 19.2937,
+      "step": 7723
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.164282796489021e-05,
+      "loss": 18.6005,
+      "step": 7724
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.163570667660122e-05,
+      "loss": 19.0307,
+      "step": 7725
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.162858484828002e-05,
+      "loss": 19.0557,
+      "step": 7726
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.162146248010444e-05,
+      "loss": 18.6193,
+      "step": 7727
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.161433957225219e-05,
+      "loss": 21.0313,
+      "step": 7728
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.160721612490109e-05,
+      "loss": 0.7341,
+      "step": 7729
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.160009213822896e-05,
+      "loss": 0.7553,
+      "step": 7730
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.159296761241358e-05,
+      "loss": 0.74,
+      "step": 7731
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.15858425476328e-05,
+      "loss": 20.7007,
+      "step": 7732
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.157871694406446e-05,
+      "loss": 0.7822,
+      "step": 7733
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.157159080188644e-05,
+      "loss": 18.6097,
+      "step": 7734
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.156446412127656e-05,
+      "loss": 21.4699,
+      "step": 7735
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.155733690241274e-05,
+      "loss": 17.5565,
+      "step": 7736
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.155020914547288e-05,
+      "loss": 17.952,
+      "step": 7737
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.154308085063486e-05,
+      "loss": 19.7473,
+      "step": 7738
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.153595201807665e-05,
+      "loss": 19.1575,
+      "step": 7739
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.152882264797613e-05,
+      "loss": 20.4711,
+      "step": 7740
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.15216927405113e-05,
+      "loss": 0.8031,
+      "step": 7741
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.15145622958601e-05,
+      "loss": 19.9961,
+      "step": 7742
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.150743131420052e-05,
+      "loss": 18.5888,
+      "step": 7743
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.150029979571054e-05,
+      "loss": 18.8493,
+      "step": 7744
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.149316774056815e-05,
+      "loss": 19.057,
+      "step": 7745
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.14860351489514e-05,
+      "loss": 19.987,
+      "step": 7746
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.14789020210383e-05,
+      "loss": 17.1347,
+      "step": 7747
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.147176835700693e-05,
+      "loss": 0.7528,
+      "step": 7748
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.146463415703531e-05,
+      "loss": 0.6986,
+      "step": 7749
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.145749942130149e-05,
+      "loss": 21.0982,
+      "step": 7750
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.14503641499836e-05,
+      "loss": 18.8507,
+      "step": 7751
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.144322834325973e-05,
+      "loss": 0.7587,
+      "step": 7752
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.143609200130799e-05,
+      "loss": 18.8491,
+      "step": 7753
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.142895512430649e-05,
+      "loss": 21.6821,
+      "step": 7754
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.142181771243338e-05,
+      "loss": 20.3615,
+      "step": 7755
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.14146797658668e-05,
+      "loss": 0.7098,
+      "step": 7756
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.140754128478493e-05,
+      "loss": 0.7671,
+      "step": 7757
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.140040226936596e-05,
+      "loss": 0.8019,
+      "step": 7758
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.139326271978804e-05,
+      "loss": 0.7411,
+      "step": 7759
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.138612263622939e-05,
+      "loss": 19.4467,
+      "step": 7760
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.137898201886826e-05,
+      "loss": 0.7408,
+      "step": 7761
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.137184086788284e-05,
+      "loss": 0.7755,
+      "step": 7762
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.136469918345139e-05,
+      "loss": 0.7589,
+      "step": 7763
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.135755696575217e-05,
+      "loss": 0.7095,
+      "step": 7764
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.135041421496346e-05,
+      "loss": 0.7493,
+      "step": 7765
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.134327093126353e-05,
+      "loss": 20.7373,
+      "step": 7766
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.133612711483072e-05,
+      "loss": 0.7566,
+      "step": 7767
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.132898276584327e-05,
+      "loss": 20.7915,
+      "step": 7768
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.132183788447954e-05,
+      "loss": 0.8003,
+      "step": 7769
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.131469247091786e-05,
+      "loss": 17.3584,
+      "step": 7770
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.130754652533661e-05,
+      "loss": 19.8513,
+      "step": 7771
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.130040004791415e-05,
+      "loss": 0.7606,
+      "step": 7772
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.129325303882879e-05,
+      "loss": 0.7494,
+      "step": 7773
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.128610549825901e-05,
+      "loss": 0.7676,
+      "step": 7774
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.127895742638316e-05,
+      "loss": 19.0241,
+      "step": 7775
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.12718088233797e-05,
+      "loss": 19.2898,
+      "step": 7776
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.126465968942699e-05,
+      "loss": 19.7294,
+      "step": 7777
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.125751002470354e-05,
+      "loss": 18.8106,
+      "step": 7778
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.12503598293878e-05,
+      "loss": 0.8189,
+      "step": 7779
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.124320910365821e-05,
+      "loss": 17.0541,
+      "step": 7780
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.123605784769328e-05,
+      "loss": 0.7763,
+      "step": 7781
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.12289060616715e-05,
+      "loss": 0.7196,
+      "step": 7782
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.122175374577137e-05,
+      "loss": 0.7077,
+      "step": 7783
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.121460090017142e-05,
+      "loss": 0.6871,
+      "step": 7784
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.120744752505019e-05,
+      "loss": 20.3477,
+      "step": 7785
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.120029362058623e-05,
+      "loss": 20.2404,
+      "step": 7786
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.11931391869581e-05,
+      "loss": 21.109,
+      "step": 7787
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.118598422434438e-05,
+      "loss": 0.744,
+      "step": 7788
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.117882873292364e-05,
+      "loss": 0.7667,
+      "step": 7789
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.117167271287453e-05,
+      "loss": 0.7899,
+      "step": 7790
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.11645161643756e-05,
+      "loss": 0.7589,
+      "step": 7791
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.115735908760553e-05,
+      "loss": 0.7449,
+      "step": 7792
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.115020148274295e-05,
+      "loss": 0.8053,
+      "step": 7793
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.11430433499665e-05,
+      "loss": 19.723,
+      "step": 7794
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.113588468945486e-05,
+      "loss": 20.4654,
+      "step": 7795
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.112872550138674e-05,
+      "loss": 0.7831,
+      "step": 7796
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.112156578594078e-05,
+      "loss": 19.539,
+      "step": 7797
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.111440554329572e-05,
+      "loss": 19.8747,
+      "step": 7798
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.110724477363029e-05,
+      "loss": 0.8551,
+      "step": 7799
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.11000834771232e-05,
+      "loss": 0.7647,
+      "step": 7800
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.10929216539532e-05,
+      "loss": 0.6882,
+      "step": 7801
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.108575930429906e-05,
+      "loss": 0.7293,
+      "step": 7802
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.107859642833958e-05,
+      "loss": 20.5633,
+      "step": 7803
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.10714330262535e-05,
+      "loss": 0.7465,
+      "step": 7804
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.106426909821964e-05,
+      "loss": 19.2289,
+      "step": 7805
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.105710464441681e-05,
+      "loss": 0.7618,
+      "step": 7806
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.104993966502384e-05,
+      "loss": 0.7798,
+      "step": 7807
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.104277416021959e-05,
+      "loss": 20.7202,
+      "step": 7808
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.103560813018288e-05,
+      "loss": 19.7416,
+      "step": 7809
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.102844157509258e-05,
+      "loss": 0.7773,
+      "step": 7810
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.102127449512759e-05,
+      "loss": 20.0638,
+      "step": 7811
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.101410689046677e-05,
+      "loss": 0.7246,
+      "step": 7812
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.100693876128906e-05,
+      "loss": 18.3981,
+      "step": 7813
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.099977010777337e-05,
+      "loss": 20.3443,
+      "step": 7814
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.09926009300986e-05,
+      "loss": 0.7447,
+      "step": 7815
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.098543122844373e-05,
+      "loss": 0.7929,
+      "step": 7816
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.09782610029877e-05,
+      "loss": 19.7712,
+      "step": 7817
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.097109025390948e-05,
+      "loss": 19.0617,
+      "step": 7818
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.096391898138808e-05,
+      "loss": 0.7647,
+      "step": 7819
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.095674718560247e-05,
+      "loss": 21.5336,
+      "step": 7820
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.094957486673165e-05,
+      "loss": 21.4463,
+      "step": 7821
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.094240202495465e-05,
+      "loss": 0.7991,
+      "step": 7822
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.093522866045054e-05,
+      "loss": 0.7518,
+      "step": 7823
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.092805477339835e-05,
+      "loss": 0.8133,
+      "step": 7824
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.09208803639771e-05,
+      "loss": 18.9819,
+      "step": 7825
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.091370543236591e-05,
+      "loss": 0.7755,
+      "step": 7826
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.090652997874388e-05,
+      "loss": 20.3146,
+      "step": 7827
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.089935400329006e-05,
+      "loss": 0.7754,
+      "step": 7828
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.08921775061836e-05,
+      "loss": 0.7615,
+      "step": 7829
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.088500048760362e-05,
+      "loss": 0.7627,
+      "step": 7830
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.087782294772927e-05,
+      "loss": 0.7717,
+      "step": 7831
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.087064488673969e-05,
+      "loss": 0.7579,
+      "step": 7832
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.086346630481405e-05,
+      "loss": 19.0141,
+      "step": 7833
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.085628720213152e-05,
+      "loss": 19.2648,
+      "step": 7834
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.084910757887129e-05,
+      "loss": 20.7473,
+      "step": 7835
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.084192743521258e-05,
+      "loss": 19.2834,
+      "step": 7836
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.083474677133461e-05,
+      "loss": 0.7721,
+      "step": 7837
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.08275655874166e-05,
+      "loss": 0.7619,
+      "step": 7838
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.08203838836378e-05,
+      "loss": 19.8967,
+      "step": 7839
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.081320166017745e-05,
+      "loss": 0.7276,
+      "step": 7840
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.080601891721485e-05,
+      "loss": 20.2789,
+      "step": 7841
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.079883565492927e-05,
+      "loss": 0.7749,
+      "step": 7842
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.079165187349998e-05,
+      "loss": 0.7524,
+      "step": 7843
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.078446757310634e-05,
+      "loss": 18.9892,
+      "step": 7844
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.077728275392763e-05,
+      "loss": 18.9363,
+      "step": 7845
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.077009741614319e-05,
+      "loss": 0.7371,
+      "step": 7846
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.07629115599324e-05,
+      "loss": 0.6942,
+      "step": 7847
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.075572518547457e-05,
+      "loss": 20.0456,
+      "step": 7848
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.074853829294912e-05,
+      "loss": 0.7683,
+      "step": 7849
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.07413508825354e-05,
+      "loss": 0.7351,
+      "step": 7850
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.073416295441284e-05,
+      "loss": 22.1538,
+      "step": 7851
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.072697450876085e-05,
+      "loss": 20.5504,
+      "step": 7852
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.071978554575881e-05,
+      "loss": 18.5807,
+      "step": 7853
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.07125960655862e-05,
+      "loss": 0.8066,
+      "step": 7854
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.070540606842248e-05,
+      "loss": 20.0771,
+      "step": 7855
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.06982155544471e-05,
+      "loss": 19.816,
+      "step": 7856
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.069102452383951e-05,
+      "loss": 0.7482,
+      "step": 7857
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.068383297677926e-05,
+      "loss": 17.7489,
+      "step": 7858
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.067664091344577e-05,
+      "loss": 20.0318,
+      "step": 7859
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.066944833401863e-05,
+      "loss": 0.8337,
+      "step": 7860
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.066225523867733e-05,
+      "loss": 18.9877,
+      "step": 7861
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.065506162760143e-05,
+      "loss": 17.0593,
+      "step": 7862
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.064786750097046e-05,
+      "loss": 0.7601,
+      "step": 7863
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.064067285896402e-05,
+      "loss": 0.7611,
+      "step": 7864
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.063347770176168e-05,
+      "loss": 0.8113,
+      "step": 7865
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.0626282029543e-05,
+      "loss": 0.7107,
+      "step": 7866
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.061908584248761e-05,
+      "loss": 18.4467,
+      "step": 7867
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.061188914077515e-05,
+      "loss": 17.7865,
+      "step": 7868
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.060469192458523e-05,
+      "loss": 18.7252,
+      "step": 7869
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.059749419409748e-05,
+      "loss": 18.1261,
+      "step": 7870
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.059029594949158e-05,
+      "loss": 0.7429,
+      "step": 7871
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.05830971909472e-05,
+      "loss": 18.9366,
+      "step": 7872
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.0575897918644e-05,
+      "loss": 0.7592,
+      "step": 7873
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.056869813276169e-05,
+      "loss": 19.5238,
+      "step": 7874
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.056149783348001e-05,
+      "loss": 0.7422,
+      "step": 7875
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.055429702097862e-05,
+      "loss": 0.7933,
+      "step": 7876
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.054709569543727e-05,
+      "loss": 0.7888,
+      "step": 7877
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.053989385703574e-05,
+      "loss": 20.9274,
+      "step": 7878
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.053269150595376e-05,
+      "loss": 0.756,
+      "step": 7879
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.052548864237113e-05,
+      "loss": 0.8092,
+      "step": 7880
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.05182852664676e-05,
+      "loss": 0.7038,
+      "step": 7881
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.051108137842299e-05,
+      "loss": 20.574,
+      "step": 7882
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.050387697841709e-05,
+      "loss": 0.7921,
+      "step": 7883
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.049667206662976e-05,
+      "loss": 18.3804,
+      "step": 7884
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.048946664324081e-05,
+      "loss": 19.8856,
+      "step": 7885
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.04822607084301e-05,
+      "loss": 0.7348,
+      "step": 7886
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.047505426237747e-05,
+      "loss": 0.7782,
+      "step": 7887
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.046784730526282e-05,
+      "loss": 19.5613,
+      "step": 7888
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.046063983726604e-05,
+      "loss": 20.0436,
+      "step": 7889
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.045343185856701e-05,
+      "loss": 0.7543,
+      "step": 7890
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.044622336934563e-05,
+      "loss": 0.7976,
+      "step": 7891
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.043901436978186e-05,
+      "loss": 0.7615,
+      "step": 7892
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.043180486005563e-05,
+      "loss": 0.7374,
+      "step": 7893
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.042459484034689e-05,
+      "loss": 21.234,
+      "step": 7894
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.041738431083558e-05,
+      "loss": 20.4366,
+      "step": 7895
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.041017327170171e-05,
+      "loss": 17.7105,
+      "step": 7896
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.040296172312526e-05,
+      "loss": 20.1787,
+      "step": 7897
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.039574966528622e-05,
+      "loss": 18.5911,
+      "step": 7898
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.038853709836461e-05,
+      "loss": 0.8284,
+      "step": 7899
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.038132402254046e-05,
+      "loss": 0.7734,
+      "step": 7900
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.03741104379938e-05,
+      "loss": 19.3208,
+      "step": 7901
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.03668963449047e-05,
+      "loss": 0.7635,
+      "step": 7902
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.035968174345323e-05,
+      "loss": 17.6705,
+      "step": 7903
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.035246663381944e-05,
+      "loss": 19.3064,
+      "step": 7904
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.034525101618343e-05,
+      "loss": 17.3586,
+      "step": 7905
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.033803489072531e-05,
+      "loss": 0.8006,
+      "step": 7906
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.03308182576252e-05,
+      "loss": 18.2637,
+      "step": 7907
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.032360111706323e-05,
+      "loss": 0.7127,
+      "step": 7908
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.031638346921954e-05,
+      "loss": 0.6694,
+      "step": 7909
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.030916531427427e-05,
+      "loss": 0.8132,
+      "step": 7910
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.030194665240759e-05,
+      "loss": 18.9998,
+      "step": 7911
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.029472748379971e-05,
+      "loss": 0.771,
+      "step": 7912
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.028750780863078e-05,
+      "loss": 0.7809,
+      "step": 7913
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.0280287627081e-05,
+      "loss": 0.7132,
+      "step": 7914
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.027306693933065e-05,
+      "loss": 0.7497,
+      "step": 7915
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.02658457455599e-05,
+      "loss": 0.7746,
+      "step": 7916
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.025862404594903e-05,
+      "loss": 0.7681,
+      "step": 7917
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.025140184067826e-05,
+      "loss": 18.3218,
+      "step": 7918
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.024417912992788e-05,
+      "loss": 0.7383,
+      "step": 7919
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.023695591387816e-05,
+      "loss": 0.7281,
+      "step": 7920
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.022973219270941e-05,
+      "loss": 0.7575,
+      "step": 7921
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.022250796660193e-05,
+      "loss": 18.493,
+      "step": 7922
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.021528323573603e-05,
+      "loss": 0.7574,
+      "step": 7923
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.020805800029204e-05,
+      "loss": 0.7512,
+      "step": 7924
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.02008322604503e-05,
+      "loss": 17.8728,
+      "step": 7925
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.01936060163912e-05,
+      "loss": 17.3462,
+      "step": 7926
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.018637926829508e-05,
+      "loss": 0.7444,
+      "step": 7927
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.01791520163423e-05,
+      "loss": 0.739,
+      "step": 7928
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.017192426071328e-05,
+      "loss": 0.7605,
+      "step": 7929
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.016469600158844e-05,
+      "loss": 17.4354,
+      "step": 7930
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.015746723914818e-05,
+      "loss": 19.4524,
+      "step": 7931
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.015023797357295e-05,
+      "loss": 19.7074,
+      "step": 7932
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.014300820504314e-05,
+      "loss": 18.1474,
+      "step": 7933
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.013577793373928e-05,
+      "loss": 21.4613,
+      "step": 7934
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.012854715984178e-05,
+      "loss": 18.7339,
+      "step": 7935
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.012131588353117e-05,
+      "loss": 0.807,
+      "step": 7936
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.011408410498791e-05,
+      "loss": 19.2113,
+      "step": 7937
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.010685182439253e-05,
+      "loss": 0.7499,
+      "step": 7938
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.009961904192553e-05,
+      "loss": 18.4329,
+      "step": 7939
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.009238575776744e-05,
+      "loss": 20.8346,
+      "step": 7940
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.008515197209882e-05,
+      "loss": 0.7329,
+      "step": 7941
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.007791768510022e-05,
+      "loss": 17.8785,
+      "step": 7942
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.007068289695221e-05,
+      "loss": 21.4461,
+      "step": 7943
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.006344760783537e-05,
+      "loss": 20.7625,
+      "step": 7944
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.005621181793032e-05,
+      "loss": 18.1404,
+      "step": 7945
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.004897552741761e-05,
+      "loss": 0.8004,
+      "step": 7946
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.00417387364779e-05,
+      "loss": 0.7609,
+      "step": 7947
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.003450144529184e-05,
+      "loss": 16.6582,
+      "step": 7948
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.002726365404002e-05,
+      "loss": 18.1141,
+      "step": 7949
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.002002536290315e-05,
+      "loss": 0.7605,
+      "step": 7950
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.001278657206187e-05,
+      "loss": 0.7496,
+      "step": 7951
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 7.000554728169685e-05,
+      "loss": 0.7513,
+      "step": 7952
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.999830749198881e-05,
+      "loss": 0.7098,
+      "step": 7953
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.999106720311845e-05,
+      "loss": 19.6643,
+      "step": 7954
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.998382641526651e-05,
+      "loss": 22.8888,
+      "step": 7955
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.997658512861367e-05,
+      "loss": 17.1605,
+      "step": 7956
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.99693433433407e-05,
+      "loss": 18.2348,
+      "step": 7957
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.996210105962839e-05,
+      "loss": 0.8243,
+      "step": 7958
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.995485827765747e-05,
+      "loss": 22.0535,
+      "step": 7959
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.994761499760872e-05,
+      "loss": 0.7184,
+      "step": 7960
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.994037121966294e-05,
+      "loss": 0.7609,
+      "step": 7961
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.993312694400096e-05,
+      "loss": 0.7253,
+      "step": 7962
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.992588217080358e-05,
+      "loss": 0.78,
+      "step": 7963
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.991863690025162e-05,
+      "loss": 20.0952,
+      "step": 7964
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.991139113252593e-05,
+      "loss": 0.7434,
+      "step": 7965
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.990414486780738e-05,
+      "loss": 19.5114,
+      "step": 7966
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.989689810627683e-05,
+      "loss": 18.6614,
+      "step": 7967
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.988965084811516e-05,
+      "loss": 19.6341,
+      "step": 7968
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.988240309350327e-05,
+      "loss": 18.3084,
+      "step": 7969
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.987515484262205e-05,
+      "loss": 0.7759,
+      "step": 7970
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.986790609565242e-05,
+      "loss": 0.7742,
+      "step": 7971
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.986065685277534e-05,
+      "loss": 17.578,
+      "step": 7972
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.985340711417172e-05,
+      "loss": 18.1943,
+      "step": 7973
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.98461568800225e-05,
+      "loss": 0.7644,
+      "step": 7974
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.98389061505087e-05,
+      "loss": 0.7547,
+      "step": 7975
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.983165492581127e-05,
+      "loss": 17.6755,
+      "step": 7976
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.98244032061112e-05,
+      "loss": 20.8098,
+      "step": 7977
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.981715099158951e-05,
+      "loss": 21.8946,
+      "step": 7978
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.98098982824272e-05,
+      "loss": 18.509,
+      "step": 7979
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.98026450788053e-05,
+      "loss": 0.7488,
+      "step": 7980
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.979539138090485e-05,
+      "loss": 0.7678,
+      "step": 7981
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.978813718890693e-05,
+      "loss": 17.3598,
+      "step": 7982
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.978088250299258e-05,
+      "loss": 0.7326,
+      "step": 7983
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.977362732334289e-05,
+      "loss": 20.7412,
+      "step": 7984
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.976637165013892e-05,
+      "loss": 0.7478,
+      "step": 7985
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.975911548356185e-05,
+      "loss": 18.5954,
+      "step": 7986
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.975185882379271e-05,
+      "loss": 19.6666,
+      "step": 7987
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.974460167101267e-05,
+      "loss": 0.7806,
+      "step": 7988
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.973734402540288e-05,
+      "loss": 0.7536,
+      "step": 7989
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.973008588714445e-05,
+      "loss": 18.3182,
+      "step": 7990
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.97228272564186e-05,
+      "loss": 0.744,
+      "step": 7991
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.971556813340646e-05,
+      "loss": 0.716,
+      "step": 7992
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.970830851828923e-05,
+      "loss": 18.9901,
+      "step": 7993
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.970104841124814e-05,
+      "loss": 18.1248,
+      "step": 7994
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.969378781246436e-05,
+      "loss": 21.1367,
+      "step": 7995
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.968652672211916e-05,
+      "loss": 19.0504,
+      "step": 7996
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.967926514039375e-05,
+      "loss": 19.6156,
+      "step": 7997
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.96720030674694e-05,
+      "loss": 20.8323,
+      "step": 7998
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.966474050352734e-05,
+      "loss": 21.0045,
+      "step": 7999
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.965747744874887e-05,
+      "loss": 18.7745,
+      "step": 8000
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.965021390331529e-05,
+      "loss": 18.0351,
+      "step": 8001
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.964294986740788e-05,
+      "loss": 19.1863,
+      "step": 8002
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.963568534120796e-05,
+      "loss": 0.7348,
+      "step": 8003
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.962842032489684e-05,
+      "loss": 0.7451,
+      "step": 8004
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.962115481865586e-05,
+      "loss": 0.758,
+      "step": 8005
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.96138888226664e-05,
+      "loss": 0.7105,
+      "step": 8006
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.960662233710978e-05,
+      "loss": 0.7244,
+      "step": 8007
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.95993553621674e-05,
+      "loss": 0.7227,
+      "step": 8008
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.95920878980206e-05,
+      "loss": 19.0689,
+      "step": 8009
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.958481994485087e-05,
+      "loss": 0.747,
+      "step": 8010
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.957755150283953e-05,
+      "loss": 0.7498,
+      "step": 8011
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.957028257216803e-05,
+      "loss": 0.7831,
+      "step": 8012
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.95630131530178e-05,
+      "loss": 0.7762,
+      "step": 8013
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.955574324557032e-05,
+      "loss": 21.2845,
+      "step": 8014
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.954847285000703e-05,
+      "loss": 19.145,
+      "step": 8015
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.954120196650937e-05,
+      "loss": 0.729,
+      "step": 8016
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.953393059525885e-05,
+      "loss": 22.0053,
+      "step": 8017
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.952665873643698e-05,
+      "loss": 0.7867,
+      "step": 8018
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.951938639022522e-05,
+      "loss": 0.7801,
+      "step": 8019
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.951211355680515e-05,
+      "loss": 0.716,
+      "step": 8020
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.950484023635826e-05,
+      "loss": 22.1116,
+      "step": 8021
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.949756642906609e-05,
+      "loss": 0.7737,
+      "step": 8022
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.949029213511021e-05,
+      "loss": 0.818,
+      "step": 8023
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.948301735467221e-05,
+      "loss": 20.1806,
+      "step": 8024
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.947574208793363e-05,
+      "loss": 16.8256,
+      "step": 8025
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.946846633507608e-05,
+      "loss": 18.5276,
+      "step": 8026
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.94611900962812e-05,
+      "loss": 18.9958,
+      "step": 8027
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.945391337173053e-05,
+      "loss": 19.8135,
+      "step": 8028
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.944663616160576e-05,
+      "loss": 15.663,
+      "step": 8029
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.94393584660885e-05,
+      "loss": 0.6965,
+      "step": 8030
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.943208028536043e-05,
+      "loss": 0.7232,
+      "step": 8031
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.942480161960318e-05,
+      "loss": 21.6036,
+      "step": 8032
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.941752246899845e-05,
+      "loss": 0.7519,
+      "step": 8033
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.941024283372794e-05,
+      "loss": 0.7633,
+      "step": 8034
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.940296271397333e-05,
+      "loss": 19.7111,
+      "step": 8035
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.939568210991633e-05,
+      "loss": 16.5483,
+      "step": 8036
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.938840102173868e-05,
+      "loss": 16.2922,
+      "step": 8037
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.938111944962213e-05,
+      "loss": 21.4694,
+      "step": 8038
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.937383739374839e-05,
+      "loss": 0.7514,
+      "step": 8039
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.936655485429927e-05,
+      "loss": 17.031,
+      "step": 8040
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.93592718314565e-05,
+      "loss": 20.01,
+      "step": 8041
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.935198832540189e-05,
+      "loss": 0.7545,
+      "step": 8042
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.934470433631722e-05,
+      "loss": 19.2013,
+      "step": 8043
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.933741986438434e-05,
+      "loss": 0.7199,
+      "step": 8044
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.933013490978503e-05,
+      "loss": 0.7374,
+      "step": 8045
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.932284947270114e-05,
+      "loss": 18.0868,
+      "step": 8046
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.93155635533145e-05,
+      "loss": 0.7435,
+      "step": 8047
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.9308277151807e-05,
+      "loss": 20.4116,
+      "step": 8048
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.93009902683605e-05,
+      "loss": 23.2436,
+      "step": 8049
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.929370290315686e-05,
+      "loss": 0.7585,
+      "step": 8050
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.928641505637798e-05,
+      "loss": 20.6093,
+      "step": 8051
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.927912672820581e-05,
+      "loss": 0.7331,
+      "step": 8052
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.927183791882223e-05,
+      "loss": 0.7296,
+      "step": 8053
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.926454862840915e-05,
+      "loss": 19.5625,
+      "step": 8054
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.925725885714856e-05,
+      "loss": 0.8119,
+      "step": 8055
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.924996860522237e-05,
+      "loss": 0.6893,
+      "step": 8056
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.924267787281258e-05,
+      "loss": 0.7523,
+      "step": 8057
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.923538666010116e-05,
+      "loss": 18.3714,
+      "step": 8058
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.922809496727011e-05,
+      "loss": 19.3612,
+      "step": 8059
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.92208027945014e-05,
+      "loss": 17.8743,
+      "step": 8060
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.921351014197706e-05,
+      "loss": 0.7819,
+      "step": 8061
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.920621700987913e-05,
+      "loss": 17.5938,
+      "step": 8062
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.919892339838965e-05,
+      "loss": 20.2702,
+      "step": 8063
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.919162930769063e-05,
+      "loss": 16.2383,
+      "step": 8064
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.918433473796419e-05,
+      "loss": 0.7573,
+      "step": 8065
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.917703968939235e-05,
+      "loss": 18.6645,
+      "step": 8066
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.916974416215723e-05,
+      "loss": 0.7321,
+      "step": 8067
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.916244815644091e-05,
+      "loss": 19.7099,
+      "step": 8068
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.915515167242553e-05,
+      "loss": 18.8919,
+      "step": 8069
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.914785471029318e-05,
+      "loss": 0.7936,
+      "step": 8070
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.9140557270226e-05,
+      "loss": 18.7235,
+      "step": 8071
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.913325935240615e-05,
+      "loss": 0.7491,
+      "step": 8072
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.912596095701576e-05,
+      "loss": 17.7645,
+      "step": 8073
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.911866208423703e-05,
+      "loss": 0.7215,
+      "step": 8074
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.911136273425212e-05,
+      "loss": 0.7747,
+      "step": 8075
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.910406290724324e-05,
+      "loss": 18.3462,
+      "step": 8076
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.90967626033926e-05,
+      "loss": 18.4597,
+      "step": 8077
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.908946182288239e-05,
+      "loss": 0.8091,
+      "step": 8078
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.908216056589485e-05,
+      "loss": 0.722,
+      "step": 8079
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.907485883261221e-05,
+      "loss": 0.7819,
+      "step": 8080
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.906755662321678e-05,
+      "loss": 19.6969,
+      "step": 8081
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.906025393789077e-05,
+      "loss": 0.7717,
+      "step": 8082
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.905295077681646e-05,
+      "loss": 15.9364,
+      "step": 8083
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.904564714017614e-05,
+      "loss": 0.7651,
+      "step": 8084
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.903834302815213e-05,
+      "loss": 0.7297,
+      "step": 8085
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.903103844092675e-05,
+      "loss": 18.6463,
+      "step": 8086
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.902373337868228e-05,
+      "loss": 0.7865,
+      "step": 8087
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.901642784160107e-05,
+      "loss": 0.7607,
+      "step": 8088
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.90091218298655e-05,
+      "loss": 0.7334,
+      "step": 8089
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.90018153436579e-05,
+      "loss": 17.7068,
+      "step": 8090
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.899450838316067e-05,
+      "loss": 20.1919,
+      "step": 8091
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.898720094855616e-05,
+      "loss": 17.0985,
+      "step": 8092
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.897989304002678e-05,
+      "loss": 21.1316,
+      "step": 8093
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.897258465775492e-05,
+      "loss": 20.1677,
+      "step": 8094
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.896527580192304e-05,
+      "loss": 18.6441,
+      "step": 8095
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.895796647271353e-05,
+      "loss": 18.5021,
+      "step": 8096
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.895065667030885e-05,
+      "loss": 16.9182,
+      "step": 8097
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.894334639489146e-05,
+      "loss": 19.6725,
+      "step": 8098
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.89360356466438e-05,
+      "loss": 0.7313,
+      "step": 8099
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.892872442574841e-05,
+      "loss": 0.7426,
+      "step": 8100
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.89214127323877e-05,
+      "loss": 0.7098,
+      "step": 8101
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.89141005667442e-05,
+      "loss": 0.7453,
+      "step": 8102
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.890678792900045e-05,
+      "loss": 21.6983,
+      "step": 8103
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.889947481933895e-05,
+      "loss": 0.8009,
+      "step": 8104
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.889216123794225e-05,
+      "loss": 17.9604,
+      "step": 8105
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.888484718499288e-05,
+      "loss": 17.5327,
+      "step": 8106
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.887753266067343e-05,
+      "loss": 19.6819,
+      "step": 8107
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.887021766516643e-05,
+      "loss": 19.4271,
+      "step": 8108
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.886290219865451e-05,
+      "loss": 19.036,
+      "step": 8109
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.885558626132024e-05,
+      "loss": 18.4094,
+      "step": 8110
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.884826985334623e-05,
+      "loss": 16.8032,
+      "step": 8111
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.884095297491508e-05,
+      "loss": 18.357,
+      "step": 8112
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.883363562620946e-05,
+      "loss": 18.7233,
+      "step": 8113
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.882631780741203e-05,
+      "loss": 0.7197,
+      "step": 8114
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.881899951870536e-05,
+      "loss": 0.7765,
+      "step": 8115
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.881168076027217e-05,
+      "loss": 17.0087,
+      "step": 8116
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.880436153229515e-05,
+      "loss": 21.3186,
+      "step": 8117
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.879704183495695e-05,
+      "loss": 0.7395,
+      "step": 8118
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.878972166844032e-05,
+      "loss": 19.2172,
+      "step": 8119
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.878240103292793e-05,
+      "loss": 19.4175,
+      "step": 8120
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.877507992860253e-05,
+      "loss": 0.7763,
+      "step": 8121
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.876775835564683e-05,
+      "loss": 18.5955,
+      "step": 8122
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.876043631424363e-05,
+      "loss": 21.1819,
+      "step": 8123
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.875311380457564e-05,
+      "loss": 18.609,
+      "step": 8124
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.874579082682564e-05,
+      "loss": 19.0876,
+      "step": 8125
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.873846738117642e-05,
+      "loss": 0.7503,
+      "step": 8126
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.873114346781078e-05,
+      "loss": 0.7595,
+      "step": 8127
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.872381908691155e-05,
+      "loss": 21.4432,
+      "step": 8128
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.871649423866148e-05,
+      "loss": 0.7696,
+      "step": 8129
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.870916892324344e-05,
+      "loss": 0.7861,
+      "step": 8130
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.87018431408403e-05,
+      "loss": 0.7414,
+      "step": 8131
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.869451689163486e-05,
+      "loss": 0.7305,
+      "step": 8132
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.868719017581004e-05,
+      "loss": 0.771,
+      "step": 8133
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.867986299354865e-05,
+      "loss": 21.6538,
+      "step": 8134
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.867253534503364e-05,
+      "loss": 18.5195,
+      "step": 8135
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.866520723044787e-05,
+      "loss": 0.7866,
+      "step": 8136
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.865787864997429e-05,
+      "loss": 0.7185,
+      "step": 8137
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.865054960379577e-05,
+      "loss": 0.7757,
+      "step": 8138
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.864322009209528e-05,
+      "loss": 0.7672,
+      "step": 8139
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.863589011505576e-05,
+      "loss": 20.8127,
+      "step": 8140
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.862855967286017e-05,
+      "loss": 18.8931,
+      "step": 8141
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.86212287656915e-05,
+      "loss": 0.7526,
+      "step": 8142
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.861389739373268e-05,
+      "loss": 17.8002,
+      "step": 8143
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.860656555716672e-05,
+      "loss": 19.8841,
+      "step": 8144
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.859923325617666e-05,
+      "loss": 18.8304,
+      "step": 8145
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.859190049094548e-05,
+      "loss": 19.8732,
+      "step": 8146
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.858456726165623e-05,
+      "loss": 0.8212,
+      "step": 8147
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.857723356849194e-05,
+      "loss": 0.7384,
+      "step": 8148
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.856989941163564e-05,
+      "loss": 0.7924,
+      "step": 8149
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.856256479127042e-05,
+      "loss": 0.7767,
+      "step": 8150
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.855522970757934e-05,
+      "loss": 0.8352,
+      "step": 8151
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.854789416074549e-05,
+      "loss": 20.3993,
+      "step": 8152
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.854055815095198e-05,
+      "loss": 0.7272,
+      "step": 8153
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.853322167838188e-05,
+      "loss": 18.4568,
+      "step": 8154
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.852588474321836e-05,
+      "loss": 18.4077,
+      "step": 8155
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.851854734564453e-05,
+      "loss": 19.9751,
+      "step": 8156
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.85112094858435e-05,
+      "loss": 20.3227,
+      "step": 8157
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.850387116399847e-05,
+      "loss": 20.0631,
+      "step": 8158
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.84965323802926e-05,
+      "loss": 0.7182,
+      "step": 8159
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.848919313490904e-05,
+      "loss": 0.7315,
+      "step": 8160
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.848185342803103e-05,
+      "loss": 18.2687,
+      "step": 8161
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.847451325984173e-05,
+      "loss": 0.747,
+      "step": 8162
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.846717263052433e-05,
+      "loss": 0.7206,
+      "step": 8163
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.845983154026209e-05,
+      "loss": 17.4497,
+      "step": 8164
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.845248998923827e-05,
+      "loss": 0.7212,
+      "step": 8165
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.844514797763607e-05,
+      "loss": 0.7571,
+      "step": 8166
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.843780550563876e-05,
+      "loss": 0.7444,
+      "step": 8167
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.843046257342961e-05,
+      "loss": 19.1791,
+      "step": 8168
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.842311918119193e-05,
+      "loss": 18.6061,
+      "step": 8169
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.841577532910897e-05,
+      "loss": 19.5565,
+      "step": 8170
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.840843101736405e-05,
+      "loss": 0.7366,
+      "step": 8171
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.84010862461405e-05,
+      "loss": 0.7545,
+      "step": 8172
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.839374101562162e-05,
+      "loss": 0.7975,
+      "step": 8173
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.838639532599078e-05,
+      "loss": 16.9521,
+      "step": 8174
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.837904917743132e-05,
+      "loss": 0.7633,
+      "step": 8175
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.837170257012659e-05,
+      "loss": 18.2688,
+      "step": 8176
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.836435550425995e-05,
+      "loss": 0.7439,
+      "step": 8177
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.835700798001482e-05,
+      "loss": 19.4099,
+      "step": 8178
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.834965999757458e-05,
+      "loss": 0.7601,
+      "step": 8179
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.834231155712263e-05,
+      "loss": 18.4458,
+      "step": 8180
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.833496265884241e-05,
+      "loss": 18.5881,
+      "step": 8181
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.832761330291732e-05,
+      "loss": 0.6934,
+      "step": 8182
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.832026348953082e-05,
+      "loss": 19.1602,
+      "step": 8183
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.831291321886636e-05,
+      "loss": 0.7578,
+      "step": 8184
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.830556249110742e-05,
+      "loss": 0.7821,
+      "step": 8185
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.829821130643745e-05,
+      "loss": 0.7814,
+      "step": 8186
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.829085966503994e-05,
+      "loss": 19.1537,
+      "step": 8187
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.82835075670984e-05,
+      "loss": 0.7961,
+      "step": 8188
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.827615501279634e-05,
+      "loss": 0.8013,
+      "step": 8189
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.826880200231729e-05,
+      "loss": 19.231,
+      "step": 8190
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.826144853584475e-05,
+      "loss": 18.3787,
+      "step": 8191
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.825409461356228e-05,
+      "loss": 18.231,
+      "step": 8192
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.824674023565346e-05,
+      "loss": 19.5264,
+      "step": 8193
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.823938540230184e-05,
+      "loss": 0.7468,
+      "step": 8194
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.823203011369097e-05,
+      "loss": 0.7237,
+      "step": 8195
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.822467437000448e-05,
+      "loss": 0.7612,
+      "step": 8196
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.821731817142597e-05,
+      "loss": 0.8023,
+      "step": 8197
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.820996151813903e-05,
+      "loss": 0.8091,
+      "step": 8198
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.820260441032728e-05,
+      "loss": 0.7747,
+      "step": 8199
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.819524684817438e-05,
+      "loss": 0.8091,
+      "step": 8200
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.818788883186396e-05,
+      "loss": 0.7828,
+      "step": 8201
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.818053036157967e-05,
+      "loss": 0.7438,
+      "step": 8202
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.817317143750521e-05,
+      "loss": 21.2275,
+      "step": 8203
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.816581205982425e-05,
+      "loss": 0.7545,
+      "step": 8204
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.815845222872046e-05,
+      "loss": 19.0997,
+      "step": 8205
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.815109194437755e-05,
+      "loss": 16.6081,
+      "step": 8206
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.814373120697924e-05,
+      "loss": 0.7338,
+      "step": 8207
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.813637001670927e-05,
+      "loss": 0.7292,
+      "step": 8208
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.812900837375136e-05,
+      "loss": 0.7927,
+      "step": 8209
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.812164627828925e-05,
+      "loss": 20.8966,
+      "step": 8210
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.811428373050672e-05,
+      "loss": 20.8147,
+      "step": 8211
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.810692073058753e-05,
+      "loss": 0.7868,
+      "step": 8212
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.809955727871547e-05,
+      "loss": 0.7656,
+      "step": 8213
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.809219337507432e-05,
+      "loss": 0.797,
+      "step": 8214
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.80848290198479e-05,
+      "loss": 19.3287,
+      "step": 8215
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.807746421322e-05,
+      "loss": 0.7436,
+      "step": 8216
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.807009895537448e-05,
+      "loss": 18.6527,
+      "step": 8217
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.806273324649516e-05,
+      "loss": 19.1275,
+      "step": 8218
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.80553670867659e-05,
+      "loss": 20.2279,
+      "step": 8219
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.804800047637054e-05,
+      "loss": 0.6939,
+      "step": 8220
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.8040633415493e-05,
+      "loss": 0.7528,
+      "step": 8221
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.803326590431711e-05,
+      "loss": 21.1992,
+      "step": 8222
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.802589794302678e-05,
+      "loss": 0.8519,
+      "step": 8223
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.801852953180593e-05,
+      "loss": 0.7498,
+      "step": 8224
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.801116067083846e-05,
+      "loss": 19.7456,
+      "step": 8225
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.800379136030832e-05,
+      "loss": 0.7483,
+      "step": 8226
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.799642160039941e-05,
+      "loss": 17.6559,
+      "step": 8227
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.798905139129575e-05,
+      "loss": 18.8403,
+      "step": 8228
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.798168073318122e-05,
+      "loss": 0.7985,
+      "step": 8229
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.797430962623985e-05,
+      "loss": 18.8184,
+      "step": 8230
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.796693807065561e-05,
+      "loss": 20.4075,
+      "step": 8231
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.795956606661249e-05,
+      "loss": 18.6957,
+      "step": 8232
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.795219361429451e-05,
+      "loss": 0.7514,
+      "step": 8233
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.794482071388566e-05,
+      "loss": 18.6011,
+      "step": 8234
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.793744736557e-05,
+      "loss": 0.7456,
+      "step": 8235
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.793007356953157e-05,
+      "loss": 0.7619,
+      "step": 8236
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.792269932595438e-05,
+      "loss": 19.4176,
+      "step": 8237
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.791532463502255e-05,
+      "loss": 0.7213,
+      "step": 8238
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.790794949692009e-05,
+      "loss": 17.3392,
+      "step": 8239
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.790057391183117e-05,
+      "loss": 0.772,
+      "step": 8240
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.789319787993979e-05,
+      "loss": 19.3988,
+      "step": 8241
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.788582140143015e-05,
+      "loss": 0.752,
+      "step": 8242
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.787844447648629e-05,
+      "loss": 0.754,
+      "step": 8243
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.787106710529238e-05,
+      "loss": 0.7531,
+      "step": 8244
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.786368928803257e-05,
+      "loss": 20.5907,
+      "step": 8245
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.785631102489099e-05,
+      "loss": 0.7693,
+      "step": 8246
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.78489323160518e-05,
+      "loss": 18.3545,
+      "step": 8247
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.78415531616992e-05,
+      "loss": 0.7529,
+      "step": 8248
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.783417356201735e-05,
+      "loss": 17.958,
+      "step": 8249
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.782679351719046e-05,
+      "loss": 18.3939,
+      "step": 8250
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.781941302740272e-05,
+      "loss": 0.7233,
+      "step": 8251
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.781203209283839e-05,
+      "loss": 0.7592,
+      "step": 8252
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.780465071368164e-05,
+      "loss": 0.7983,
+      "step": 8253
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.779726889011674e-05,
+      "loss": 0.8191,
+      "step": 8254
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.778988662232799e-05,
+      "loss": 0.8129,
+      "step": 8255
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.778250391049959e-05,
+      "loss": 0.7754,
+      "step": 8256
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.77751207548158e-05,
+      "loss": 0.7768,
+      "step": 8257
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.776773715546097e-05,
+      "loss": 0.7999,
+      "step": 8258
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.776035311261935e-05,
+      "loss": 19.9431,
+      "step": 8259
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.775296862647526e-05,
+      "loss": 0.7526,
+      "step": 8260
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.774558369721302e-05,
+      "loss": 0.8029,
+      "step": 8261
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.773819832501698e-05,
+      "loss": 19.8744,
+      "step": 8262
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.773081251007143e-05,
+      "loss": 0.815,
+      "step": 8263
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.772342625256076e-05,
+      "loss": 0.7805,
+      "step": 8264
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.771603955266933e-05,
+      "loss": 0.7587,
+      "step": 8265
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.770865241058151e-05,
+      "loss": 17.9984,
+      "step": 8266
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.770126482648167e-05,
+      "loss": 18.7934,
+      "step": 8267
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.769387680055422e-05,
+      "loss": 0.763,
+      "step": 8268
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.768648833298359e-05,
+      "loss": 0.79,
+      "step": 8269
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.767909942395415e-05,
+      "loss": 0.762,
+      "step": 8270
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.767171007365035e-05,
+      "loss": 20.9965,
+      "step": 8271
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.766432028225663e-05,
+      "loss": 0.7863,
+      "step": 8272
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.765693004995746e-05,
+      "loss": 17.7695,
+      "step": 8273
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.764953937693728e-05,
+      "loss": 18.377,
+      "step": 8274
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.764214826338055e-05,
+      "loss": 0.7052,
+      "step": 8275
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.763475670947179e-05,
+      "loss": 0.7416,
+      "step": 8276
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.762736471539545e-05,
+      "loss": 0.7766,
+      "step": 8277
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.76199722813361e-05,
+      "loss": 0.7422,
+      "step": 8278
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.761257940747819e-05,
+      "loss": 0.7891,
+      "step": 8279
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.76051860940063e-05,
+      "loss": 18.1067,
+      "step": 8280
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.759779234110491e-05,
+      "loss": 17.8802,
+      "step": 8281
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.759039814895862e-05,
+      "loss": 18.0156,
+      "step": 8282
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.758300351775201e-05,
+      "loss": 19.1412,
+      "step": 8283
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.757560844766958e-05,
+      "loss": 0.7718,
+      "step": 8284
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.756821293889594e-05,
+      "loss": 0.8014,
+      "step": 8285
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.756081699161571e-05,
+      "loss": 19.6289,
+      "step": 8286
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.75534206060135e-05,
+      "loss": 0.7786,
+      "step": 8287
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.754602378227389e-05,
+      "loss": 18.2496,
+      "step": 8288
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.753862652058151e-05,
+      "loss": 19.4688,
+      "step": 8289
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.753122882112101e-05,
+      "loss": 0.7731,
+      "step": 8290
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.752383068407704e-05,
+      "loss": 18.9017,
+      "step": 8291
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.751643210963426e-05,
+      "loss": 18.8981,
+      "step": 8292
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.750903309797733e-05,
+      "loss": 0.8063,
+      "step": 8293
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.750163364929093e-05,
+      "loss": 19.197,
+      "step": 8294
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.749423376375975e-05,
+      "loss": 20.7555,
+      "step": 8295
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.748683344156851e-05,
+      "loss": 18.6008,
+      "step": 8296
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.747943268290193e-05,
+      "loss": 19.0345,
+      "step": 8297
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.74720314879447e-05,
+      "loss": 0.7595,
+      "step": 8298
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.746462985688158e-05,
+      "loss": 0.7567,
+      "step": 8299
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.745722778989731e-05,
+      "loss": 0.7975,
+      "step": 8300
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.744982528717665e-05,
+      "loss": 0.7448,
+      "step": 8301
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.744242234890438e-05,
+      "loss": 20.8809,
+      "step": 8302
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.743501897526525e-05,
+      "loss": 20.5183,
+      "step": 8303
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.742761516644407e-05,
+      "loss": 19.2841,
+      "step": 8304
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.742021092262563e-05,
+      "loss": 18.9999,
+      "step": 8305
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.741280624399475e-05,
+      "loss": 18.4308,
+      "step": 8306
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.740540113073626e-05,
+      "loss": 18.8837,
+      "step": 8307
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.739799558303496e-05,
+      "loss": 0.7746,
+      "step": 8308
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.739058960107574e-05,
+      "loss": 17.3399,
+      "step": 8309
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.738318318504342e-05,
+      "loss": 19.1441,
+      "step": 8310
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.737577633512291e-05,
+      "loss": 18.5361,
+      "step": 8311
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.736836905149902e-05,
+      "loss": 19.6238,
+      "step": 8312
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.736096133435668e-05,
+      "loss": 0.7666,
+      "step": 8313
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.735355318388079e-05,
+      "loss": 20.4851,
+      "step": 8314
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.734614460025624e-05,
+      "loss": 17.2891,
+      "step": 8315
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.733873558366796e-05,
+      "loss": 0.7577,
+      "step": 8316
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.733132613430089e-05,
+      "loss": 0.7699,
+      "step": 8317
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.732391625233996e-05,
+      "loss": 18.9687,
+      "step": 8318
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.731650593797013e-05,
+      "loss": 18.9366,
+      "step": 8319
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.730909519137635e-05,
+      "loss": 0.7915,
+      "step": 8320
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.73016840127436e-05,
+      "loss": 19.3835,
+      "step": 8321
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.729427240225686e-05,
+      "loss": 0.7666,
+      "step": 8322
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.728686036010115e-05,
+      "loss": 18.4032,
+      "step": 8323
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.727944788646143e-05,
+      "loss": 0.7834,
+      "step": 8324
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.72720349815228e-05,
+      "loss": 0.7261,
+      "step": 8325
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.726462164547018e-05,
+      "loss": 0.7991,
+      "step": 8326
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.725720787848867e-05,
+      "loss": 20.221,
+      "step": 8327
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.724979368076333e-05,
+      "loss": 0.7739,
+      "step": 8328
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.724237905247919e-05,
+      "loss": 18.7262,
+      "step": 8329
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.723496399382133e-05,
+      "loss": 0.768,
+      "step": 8330
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.722754850497485e-05,
+      "loss": 20.4619,
+      "step": 8331
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.72201325861248e-05,
+      "loss": 0.7155,
+      "step": 8332
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.721271623745631e-05,
+      "loss": 0.7331,
+      "step": 8333
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.720529945915453e-05,
+      "loss": 18.9863,
+      "step": 8334
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.719788225140452e-05,
+      "loss": 19.11,
+      "step": 8335
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.719046461439143e-05,
+      "loss": 0.7143,
+      "step": 8336
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.718304654830044e-05,
+      "loss": 0.7426,
+      "step": 8337
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.717562805331668e-05,
+      "loss": 0.7888,
+      "step": 8338
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.716820912962531e-05,
+      "loss": 0.7627,
+      "step": 8339
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.716078977741153e-05,
+      "loss": 19.3719,
+      "step": 8340
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.715336999686051e-05,
+      "loss": 20.1652,
+      "step": 8341
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.714594978815747e-05,
+      "loss": 19.8121,
+      "step": 8342
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.713852915148761e-05,
+      "loss": 0.775,
+      "step": 8343
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.713110808703614e-05,
+      "loss": 0.7808,
+      "step": 8344
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.712368659498832e-05,
+      "loss": 0.7631,
+      "step": 8345
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.711626467552934e-05,
+      "loss": 19.8939,
+      "step": 8346
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.71088423288445e-05,
+      "loss": 19.0071,
+      "step": 8347
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.710141955511906e-05,
+      "loss": 0.8362,
+      "step": 8348
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.709399635453828e-05,
+      "loss": 18.5184,
+      "step": 8349
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.708657272728743e-05,
+      "loss": 19.8239,
+      "step": 8350
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.707914867355182e-05,
+      "loss": 19.2919,
+      "step": 8351
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.70717241935168e-05,
+      "loss": 0.7849,
+      "step": 8352
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.706429928736761e-05,
+      "loss": 0.7384,
+      "step": 8353
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.705687395528962e-05,
+      "loss": 0.8642,
+      "step": 8354
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.704944819746817e-05,
+      "loss": 0.8013,
+      "step": 8355
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.70420220140886e-05,
+      "loss": 0.8049,
+      "step": 8356
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.703459540533625e-05,
+      "loss": 0.775,
+      "step": 8357
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.702716837139653e-05,
+      "loss": 0.7602,
+      "step": 8358
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.70197409124548e-05,
+      "loss": 20.36,
+      "step": 8359
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.701231302869644e-05,
+      "loss": 21.3102,
+      "step": 8360
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.700488472030687e-05,
+      "loss": 0.7628,
+      "step": 8361
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.699745598747149e-05,
+      "loss": 0.7739,
+      "step": 8362
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.699002683037574e-05,
+      "loss": 0.7649,
+      "step": 8363
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.698259724920502e-05,
+      "loss": 19.8529,
+      "step": 8364
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.69751672441448e-05,
+      "loss": 0.8005,
+      "step": 8365
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.696773681538054e-05,
+      "loss": 19.4911,
+      "step": 8366
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.696030596309771e-05,
+      "loss": 0.783,
+      "step": 8367
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.695287468748174e-05,
+      "loss": 18.2366,
+      "step": 8368
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.694544298871817e-05,
+      "loss": 19.4558,
+      "step": 8369
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.693801086699245e-05,
+      "loss": 18.8737,
+      "step": 8370
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.693057832249011e-05,
+      "loss": 0.7429,
+      "step": 8371
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.69231453553967e-05,
+      "loss": 20.4436,
+      "step": 8372
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.69157119658977e-05,
+      "loss": 19.4601,
+      "step": 8373
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.690827815417866e-05,
+      "loss": 0.7364,
+      "step": 8374
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.690084392042513e-05,
+      "loss": 17.1902,
+      "step": 8375
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.68934092648227e-05,
+      "loss": 0.7361,
+      "step": 8376
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.688597418755691e-05,
+      "loss": 0.7819,
+      "step": 8377
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.687853868881336e-05,
+      "loss": 0.7385,
+      "step": 8378
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.68711027687776e-05,
+      "loss": 21.0297,
+      "step": 8379
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.686366642763529e-05,
+      "loss": 0.7707,
+      "step": 8380
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.685622966557201e-05,
+      "loss": 19.6355,
+      "step": 8381
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.684879248277337e-05,
+      "loss": 18.6265,
+      "step": 8382
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.684135487942505e-05,
+      "loss": 0.7939,
+      "step": 8383
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.683391685571264e-05,
+      "loss": 0.7251,
+      "step": 8384
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.682647841182184e-05,
+      "loss": 17.9295,
+      "step": 8385
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.681903954793829e-05,
+      "loss": 21.5099,
+      "step": 8386
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.681160026424769e-05,
+      "loss": 19.8679,
+      "step": 8387
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.680416056093567e-05,
+      "loss": 17.3859,
+      "step": 8388
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.679672043818799e-05,
+      "loss": 19.4811,
+      "step": 8389
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.678927989619033e-05,
+      "loss": 19.7962,
+      "step": 8390
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.678183893512842e-05,
+      "loss": 0.8065,
+      "step": 8391
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.677439755518797e-05,
+      "loss": 18.273,
+      "step": 8392
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.676695575655471e-05,
+      "loss": 0.8018,
+      "step": 8393
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.675951353941445e-05,
+      "loss": 18.4367,
+      "step": 8394
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.675207090395288e-05,
+      "loss": 17.3907,
+      "step": 8395
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.674462785035578e-05,
+      "loss": 20.0687,
+      "step": 8396
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.673718437880897e-05,
+      "loss": 19.808,
+      "step": 8397
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.672974048949819e-05,
+      "loss": 0.7116,
+      "step": 8398
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.672229618260928e-05,
+      "loss": 21.5713,
+      "step": 8399
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.671485145832805e-05,
+      "loss": 0.7832,
+      "step": 8400
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.670740631684032e-05,
+      "loss": 0.8138,
+      "step": 8401
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.669996075833188e-05,
+      "loss": 17.1832,
+      "step": 8402
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.669251478298862e-05,
+      "loss": 19.6465,
+      "step": 8403
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.668506839099639e-05,
+      "loss": 19.1593,
+      "step": 8404
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.667762158254104e-05,
+      "loss": 0.7137,
+      "step": 8405
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.667017435780843e-05,
+      "loss": 0.7902,
+      "step": 8406
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.666272671698449e-05,
+      "loss": 18.8682,
+      "step": 8407
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.665527866025506e-05,
+      "loss": 18.5909,
+      "step": 8408
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.664783018780609e-05,
+      "loss": 0.7413,
+      "step": 8409
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.664038129982346e-05,
+      "loss": 0.8179,
+      "step": 8410
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.663293199649315e-05,
+      "loss": 20.5108,
+      "step": 8411
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.662548227800102e-05,
+      "loss": 0.6979,
+      "step": 8412
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.661803214453307e-05,
+      "loss": 0.7287,
+      "step": 8413
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.661058159627524e-05,
+      "loss": 17.208,
+      "step": 8414
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.66031306334135e-05,
+      "loss": 20.4011,
+      "step": 8415
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.659567925613384e-05,
+      "loss": 16.6119,
+      "step": 8416
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.65882274646222e-05,
+      "loss": 17.6416,
+      "step": 8417
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.658077525906467e-05,
+      "loss": 0.7121,
+      "step": 8418
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.657332263964718e-05,
+      "loss": 0.7433,
+      "step": 8419
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.656586960655576e-05,
+      "loss": 20.5285,
+      "step": 8420
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.655841615997647e-05,
+      "loss": 18.6505,
+      "step": 8421
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.655096230009533e-05,
+      "loss": 19.6757,
+      "step": 8422
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.654350802709838e-05,
+      "loss": 16.2322,
+      "step": 8423
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.653605334117169e-05,
+      "loss": 0.8337,
+      "step": 8424
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.652859824250135e-05,
+      "loss": 0.7593,
+      "step": 8425
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.652114273127341e-05,
+      "loss": 18.4457,
+      "step": 8426
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.651368680767397e-05,
+      "loss": 0.8361,
+      "step": 8427
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.650623047188915e-05,
+      "loss": 17.4909,
+      "step": 8428
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.649877372410506e-05,
+      "loss": 0.7978,
+      "step": 8429
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.649131656450779e-05,
+      "loss": 19.6715,
+      "step": 8430
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.648385899328349e-05,
+      "loss": 20.8107,
+      "step": 8431
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.647640101061831e-05,
+      "loss": 19.206,
+      "step": 8432
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.646894261669842e-05,
+      "loss": 23.9675,
+      "step": 8433
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.646148381170993e-05,
+      "loss": 0.7794,
+      "step": 8434
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.645402459583908e-05,
+      "loss": 19.1655,
+      "step": 8435
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.6446564969272e-05,
+      "loss": 0.7639,
+      "step": 8436
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.64391049321949e-05,
+      "loss": 19.5886,
+      "step": 8437
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.643164448479398e-05,
+      "loss": 18.3367,
+      "step": 8438
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.642418362725549e-05,
+      "loss": 18.4574,
+      "step": 8439
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.641672235976559e-05,
+      "loss": 0.8174,
+      "step": 8440
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.640926068251057e-05,
+      "loss": 0.822,
+      "step": 8441
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.640179859567665e-05,
+      "loss": 19.2122,
+      "step": 8442
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.639433609945011e-05,
+      "loss": 0.7313,
+      "step": 8443
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.638687319401719e-05,
+      "loss": 0.7621,
+      "step": 8444
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.637940987956416e-05,
+      "loss": 0.7602,
+      "step": 8445
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.637194615627733e-05,
+      "loss": 19.9031,
+      "step": 8446
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.6364482024343e-05,
+      "loss": 0.8065,
+      "step": 8447
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.635701748394744e-05,
+      "loss": 0.761,
+      "step": 8448
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.634955253527701e-05,
+      "loss": 18.2869,
+      "step": 8449
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.634208717851798e-05,
+      "loss": 0.7689,
+      "step": 8450
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.633462141385675e-05,
+      "loss": 18.2851,
+      "step": 8451
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.632715524147966e-05,
+      "loss": 18.2072,
+      "step": 8452
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.631968866157301e-05,
+      "loss": 0.7435,
+      "step": 8453
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.631222167432321e-05,
+      "loss": 0.7082,
+      "step": 8454
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.630475427991662e-05,
+      "loss": 19.0328,
+      "step": 8455
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.629728647853968e-05,
+      "loss": 18.5603,
+      "step": 8456
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.628981827037872e-05,
+      "loss": 19.4617,
+      "step": 8457
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.628234965562019e-05,
+      "loss": 18.0835,
+      "step": 8458
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.627488063445049e-05,
+      "loss": 17.8827,
+      "step": 8459
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.626741120705604e-05,
+      "loss": 0.8221,
+      "step": 8460
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.625994137362332e-05,
+      "loss": 0.7072,
+      "step": 8461
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.625247113433872e-05,
+      "loss": 0.7603,
+      "step": 8462
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.624500048938875e-05,
+      "loss": 18.2492,
+      "step": 8463
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.623752943895986e-05,
+      "loss": 0.7135,
+      "step": 8464
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.623005798323851e-05,
+      "loss": 0.8442,
+      "step": 8465
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.622258612241123e-05,
+      "loss": 17.2981,
+      "step": 8466
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.621511385666448e-05,
+      "loss": 0.7824,
+      "step": 8467
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.620764118618479e-05,
+      "loss": 17.6548,
+      "step": 8468
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.620016811115867e-05,
+      "loss": 17.0078,
+      "step": 8469
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.619269463177267e-05,
+      "loss": 19.2062,
+      "step": 8470
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.618522074821329e-05,
+      "loss": 0.7543,
+      "step": 8471
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.617774646066712e-05,
+      "loss": 0.7258,
+      "step": 8472
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.617027176932071e-05,
+      "loss": 18.9768,
+      "step": 8473
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.616279667436062e-05,
+      "loss": 18.1026,
+      "step": 8474
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.615532117597342e-05,
+      "loss": 0.7942,
+      "step": 8475
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.614784527434575e-05,
+      "loss": 0.72,
+      "step": 8476
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.614036896966415e-05,
+      "loss": 19.3812,
+      "step": 8477
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.613289226211525e-05,
+      "loss": 0.7949,
+      "step": 8478
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.612541515188567e-05,
+      "loss": 0.8001,
+      "step": 8479
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.611793763916208e-05,
+      "loss": 0.7863,
+      "step": 8480
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.611045972413105e-05,
+      "loss": 17.6329,
+      "step": 8481
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.610298140697928e-05,
+      "loss": 0.8047,
+      "step": 8482
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.609550268789342e-05,
+      "loss": 18.0521,
+      "step": 8483
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.608802356706014e-05,
+      "loss": 22.039,
+      "step": 8484
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.608054404466612e-05,
+      "loss": 18.4357,
+      "step": 8485
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.607306412089805e-05,
+      "loss": 18.4282,
+      "step": 8486
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.606558379594262e-05,
+      "loss": 0.7771,
+      "step": 8487
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.605810306998656e-05,
+      "loss": 18.7142,
+      "step": 8488
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.605062194321658e-05,
+      "loss": 19.4953,
+      "step": 8489
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.60431404158194e-05,
+      "loss": 19.3657,
+      "step": 8490
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.603565848798178e-05,
+      "loss": 18.8076,
+      "step": 8491
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.602817615989045e-05,
+      "loss": 18.2268,
+      "step": 8492
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.60206934317322e-05,
+      "loss": 18.4206,
+      "step": 8493
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.601321030369379e-05,
+      "loss": 18.3885,
+      "step": 8494
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.600572677596198e-05,
+      "loss": 21.298,
+      "step": 8495
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.599824284872357e-05,
+      "loss": 18.5839,
+      "step": 8496
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.599075852216538e-05,
+      "loss": 0.7978,
+      "step": 8497
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.59832737964742e-05,
+      "loss": 0.7599,
+      "step": 8498
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.597578867183685e-05,
+      "loss": 0.7335,
+      "step": 8499
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.596830314844018e-05,
+      "loss": 0.7513,
+      "step": 8500
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.596081722647101e-05,
+      "loss": 17.7034,
+      "step": 8501
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.59533309061162e-05,
+      "loss": 21.8131,
+      "step": 8502
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.59458441875626e-05,
+      "loss": 0.6856,
+      "step": 8503
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.59383570709971e-05,
+      "loss": 0.7557,
+      "step": 8504
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.593086955660657e-05,
+      "loss": 19.7607,
+      "step": 8505
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.592338164457788e-05,
+      "loss": 0.7816,
+      "step": 8506
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.591589333509796e-05,
+      "loss": 21.5501,
+      "step": 8507
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.590840462835372e-05,
+      "loss": 0.7904,
+      "step": 8508
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.590091552453204e-05,
+      "loss": 19.7794,
+      "step": 8509
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.589342602381989e-05,
+      "loss": 0.7924,
+      "step": 8510
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.588593612640421e-05,
+      "loss": 18.113,
+      "step": 8511
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.587844583247194e-05,
+      "loss": 0.7774,
+      "step": 8512
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.587095514221003e-05,
+      "loss": 0.7433,
+      "step": 8513
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.586346405580544e-05,
+      "loss": 20.2437,
+      "step": 8514
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.585597257344519e-05,
+      "loss": 0.775,
+      "step": 8515
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.584848069531622e-05,
+      "loss": 0.7133,
+      "step": 8516
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.584098842160557e-05,
+      "loss": 0.7659,
+      "step": 8517
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.583349575250021e-05,
+      "loss": 0.7766,
+      "step": 8518
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.58260026881872e-05,
+      "loss": 0.7832,
+      "step": 8519
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.581850922885354e-05,
+      "loss": 0.7432,
+      "step": 8520
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.581101537468629e-05,
+      "loss": 18.3574,
+      "step": 8521
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.58035211258725e-05,
+      "loss": 0.7549,
+      "step": 8522
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.579602648259917e-05,
+      "loss": 0.7665,
+      "step": 8523
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.578853144505344e-05,
+      "loss": 20.0216,
+      "step": 8524
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.578103601342236e-05,
+      "loss": 19.9923,
+      "step": 8525
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.577354018789301e-05,
+      "loss": 19.8487,
+      "step": 8526
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.57660439686525e-05,
+      "loss": 20.57,
+      "step": 8527
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.575854735588794e-05,
+      "loss": 18.432,
+      "step": 8528
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.575105034978642e-05,
+      "loss": 0.7428,
+      "step": 8529
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.57435529505351e-05,
+      "loss": 19.0235,
+      "step": 8530
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.57360551583211e-05,
+      "loss": 19.2592,
+      "step": 8531
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.57285569733316e-05,
+      "loss": 19.9469,
+      "step": 8532
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.572105839575369e-05,
+      "loss": 18.6587,
+      "step": 8533
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.571355942577457e-05,
+      "loss": 0.8089,
+      "step": 8534
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.570606006358143e-05,
+      "loss": 19.6758,
+      "step": 8535
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.569856030936146e-05,
+      "loss": 19.3349,
+      "step": 8536
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.569106016330182e-05,
+      "loss": 0.7579,
+      "step": 8537
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.568355962558973e-05,
+      "loss": 19.8301,
+      "step": 8538
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.567605869641244e-05,
+      "loss": 18.9313,
+      "step": 8539
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.566855737595712e-05,
+      "loss": 0.7862,
+      "step": 8540
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.566105566441104e-05,
+      "loss": 0.6908,
+      "step": 8541
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.565355356196144e-05,
+      "loss": 0.7498,
+      "step": 8542
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.564605106879556e-05,
+      "loss": 20.4915,
+      "step": 8543
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.563854818510066e-05,
+      "loss": 19.45,
+      "step": 8544
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.563104491106404e-05,
+      "loss": 0.729,
+      "step": 8545
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.562354124687298e-05,
+      "loss": 17.2787,
+      "step": 8546
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.561603719271476e-05,
+      "loss": 0.777,
+      "step": 8547
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.560853274877668e-05,
+      "loss": 16.6914,
+      "step": 8548
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.560102791524605e-05,
+      "loss": 19.4169,
+      "step": 8549
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.559352269231023e-05,
+      "loss": 0.8168,
+      "step": 8550
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.55860170801565e-05,
+      "loss": 19.4386,
+      "step": 8551
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.557851107897222e-05,
+      "loss": 0.7816,
+      "step": 8552
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.557100468894475e-05,
+      "loss": 20.6204,
+      "step": 8553
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.556349791026145e-05,
+      "loss": 0.7446,
+      "step": 8554
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.55559907431097e-05,
+      "loss": 0.8085,
+      "step": 8555
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.554848318767686e-05,
+      "loss": 0.7682,
+      "step": 8556
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.554097524415033e-05,
+      "loss": 0.7744,
+      "step": 8557
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.55334669127175e-05,
+      "loss": 16.6484,
+      "step": 8558
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.55259581935658e-05,
+      "loss": 19.3494,
+      "step": 8559
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.551844908688265e-05,
+      "loss": 18.2087,
+      "step": 8560
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.551093959285545e-05,
+      "loss": 18.7636,
+      "step": 8561
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.550342971167165e-05,
+      "loss": 0.7507,
+      "step": 8562
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.549591944351871e-05,
+      "loss": 17.6005,
+      "step": 8563
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.54884087885841e-05,
+      "loss": 20.5198,
+      "step": 8564
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.548089774705526e-05,
+      "loss": 18.6282,
+      "step": 8565
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.547338631911968e-05,
+      "loss": 0.7805,
+      "step": 8566
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.546587450496483e-05,
+      "loss": 16.5091,
+      "step": 8567
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.545836230477822e-05,
+      "loss": 17.1757,
+      "step": 8568
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.545084971874738e-05,
+      "loss": 21.2918,
+      "step": 8569
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.544333674705979e-05,
+      "loss": 0.7524,
+      "step": 8570
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.543582338990298e-05,
+      "loss": 21.0909,
+      "step": 8571
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.54283096474645e-05,
+      "loss": 18.069,
+      "step": 8572
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.54207955199319e-05,
+      "loss": 0.7947,
+      "step": 8573
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.541328100749273e-05,
+      "loss": 0.789,
+      "step": 8574
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.540576611033452e-05,
+      "loss": 19.1035,
+      "step": 8575
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.539825082864488e-05,
+      "loss": 18.3517,
+      "step": 8576
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.539073516261138e-05,
+      "loss": 0.7641,
+      "step": 8577
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.538321911242163e-05,
+      "loss": 19.3443,
+      "step": 8578
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.537570267826321e-05,
+      "loss": 17.6317,
+      "step": 8579
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.536818586032374e-05,
+      "loss": 19.9074,
+      "step": 8580
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.536066865879084e-05,
+      "loss": 0.7637,
+      "step": 8581
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.535315107385214e-05,
+      "loss": 0.7933,
+      "step": 8582
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.534563310569529e-05,
+      "loss": 17.7501,
+      "step": 8583
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.533811475450794e-05,
+      "loss": 21.762,
+      "step": 8584
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.533059602047774e-05,
+      "loss": 0.8323,
+      "step": 8585
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.532307690379235e-05,
+      "loss": 0.7576,
+      "step": 8586
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.531555740463947e-05,
+      "loss": 0.7466,
+      "step": 8587
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.530803752320678e-05,
+      "loss": 18.5645,
+      "step": 8588
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.530051725968199e-05,
+      "loss": 0.6974,
+      "step": 8589
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.529299661425276e-05,
+      "loss": 19.1476,
+      "step": 8590
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.528547558710688e-05,
+      "loss": 0.7751,
+      "step": 8591
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.527795417843203e-05,
+      "loss": 0.7942,
+      "step": 8592
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.527043238841594e-05,
+      "loss": 0.7997,
+      "step": 8593
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.52629102172464e-05,
+      "loss": 0.7949,
+      "step": 8594
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.525538766511111e-05,
+      "loss": 0.8127,
+      "step": 8595
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.524786473219786e-05,
+      "loss": 0.7522,
+      "step": 8596
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.524034141869444e-05,
+      "loss": 0.8332,
+      "step": 8597
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.523281772478861e-05,
+      "loss": 17.7273,
+      "step": 8598
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.522529365066818e-05,
+      "loss": 0.747,
+      "step": 8599
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.521776919652093e-05,
+      "loss": 19.2308,
+      "step": 8600
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.52102443625347e-05,
+      "loss": 17.0272,
+      "step": 8601
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.52027191488973e-05,
+      "loss": 0.7753,
+      "step": 8602
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.519519355579654e-05,
+      "loss": 0.7874,
+      "step": 8603
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.518766758342029e-05,
+      "loss": 20.1652,
+      "step": 8604
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.51801412319564e-05,
+      "loss": 18.3261,
+      "step": 8605
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.517261450159271e-05,
+      "loss": 0.7605,
+      "step": 8606
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.516508739251708e-05,
+      "loss": 18.4103,
+      "step": 8607
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.515755990491743e-05,
+      "loss": 0.7563,
+      "step": 8608
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.515003203898161e-05,
+      "loss": 19.9572,
+      "step": 8609
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.514250379489753e-05,
+      "loss": 0.7471,
+      "step": 8610
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.513497517285312e-05,
+      "loss": 0.7786,
+      "step": 8611
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.512744617303625e-05,
+      "loss": 0.7442,
+      "step": 8612
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.511991679563489e-05,
+      "loss": 17.7845,
+      "step": 8613
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.511238704083693e-05,
+      "loss": 18.6206,
+      "step": 8614
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.510485690883036e-05,
+      "loss": 21.1329,
+      "step": 8615
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.50973263998031e-05,
+      "loss": 0.7054,
+      "step": 8616
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.508979551394313e-05,
+      "loss": 16.715,
+      "step": 8617
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.508226425143842e-05,
+      "loss": 0.7906,
+      "step": 8618
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.507473261247696e-05,
+      "loss": 20.6043,
+      "step": 8619
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.506720059724672e-05,
+      "loss": 0.7153,
+      "step": 8620
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.50596682059357e-05,
+      "loss": 0.8125,
+      "step": 8621
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.505213543873194e-05,
+      "loss": 21.227,
+      "step": 8622
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.504460229582344e-05,
+      "loss": 20.5881,
+      "step": 8623
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.503706877739823e-05,
+      "loss": 0.7256,
+      "step": 8624
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.502953488364435e-05,
+      "loss": 21.1957,
+      "step": 8625
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.502200061474986e-05,
+      "loss": 20.9827,
+      "step": 8626
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.501446597090278e-05,
+      "loss": 0.7626,
+      "step": 8627
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.500693095229122e-05,
+      "loss": 20.6603,
+      "step": 8628
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.499939555910323e-05,
+      "loss": 0.7832,
+      "step": 8629
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.49918597915269e-05,
+      "loss": 21.2474,
+      "step": 8630
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.498432364975034e-05,
+      "loss": 0.7877,
+      "step": 8631
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.497678713396163e-05,
+      "loss": 0.7425,
+      "step": 8632
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.496925024434889e-05,
+      "loss": 0.7468,
+      "step": 8633
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.496171298110026e-05,
+      "loss": 21.2446,
+      "step": 8634
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.495417534440385e-05,
+      "loss": 0.7542,
+      "step": 8635
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.494663733444779e-05,
+      "loss": 18.9947,
+      "step": 8636
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.493909895142027e-05,
+      "loss": 0.7487,
+      "step": 8637
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.493156019550941e-05,
+      "loss": 18.9252,
+      "step": 8638
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.492402106690343e-05,
+      "loss": 0.7634,
+      "step": 8639
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.491648156579046e-05,
+      "loss": 0.7322,
+      "step": 8640
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.490894169235869e-05,
+      "loss": 18.0572,
+      "step": 8641
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.490140144679635e-05,
+      "loss": 18.9564,
+      "step": 8642
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.489386082929163e-05,
+      "loss": 0.7849,
+      "step": 8643
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.488631984003273e-05,
+      "loss": 0.7276,
+      "step": 8644
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.487877847920789e-05,
+      "loss": 19.009,
+      "step": 8645
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.487123674700535e-05,
+      "loss": 21.1562,
+      "step": 8646
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.486369464361333e-05,
+      "loss": 18.4211,
+      "step": 8647
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.485615216922011e-05,
+      "loss": 0.8017,
+      "step": 8648
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.484860932401395e-05,
+      "loss": 18.8227,
+      "step": 8649
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.484106610818311e-05,
+      "loss": 18.0882,
+      "step": 8650
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.483352252191584e-05,
+      "loss": 0.7542,
+      "step": 8651
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.482597856540048e-05,
+      "loss": 0.7411,
+      "step": 8652
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.481843423882532e-05,
+      "loss": 18.3687,
+      "step": 8653
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.481088954237865e-05,
+      "loss": 20.0184,
+      "step": 8654
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.480334447624878e-05,
+      "loss": 20.7429,
+      "step": 8655
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.479579904062408e-05,
+      "loss": 18.5439,
+      "step": 8656
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.478825323569284e-05,
+      "loss": 0.7011,
+      "step": 8657
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.478070706164343e-05,
+      "loss": 0.7183,
+      "step": 8658
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.47731605186642e-05,
+      "loss": 18.6415,
+      "step": 8659
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.47656136069435e-05,
+      "loss": 0.7787,
+      "step": 8660
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.475806632666971e-05,
+      "loss": 17.9361,
+      "step": 8661
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.475051867803122e-05,
+      "loss": 17.3653,
+      "step": 8662
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.474297066121643e-05,
+      "loss": 17.7997,
+      "step": 8663
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.47354222764137e-05,
+      "loss": 0.6887,
+      "step": 8664
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.472787352381148e-05,
+      "loss": 0.7542,
+      "step": 8665
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.472032440359816e-05,
+      "loss": 0.7483,
+      "step": 8666
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.471277491596221e-05,
+      "loss": 0.7873,
+      "step": 8667
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.470522506109202e-05,
+      "loss": 0.7214,
+      "step": 8668
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.469767483917606e-05,
+      "loss": 21.3722,
+      "step": 8669
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.469012425040278e-05,
+      "loss": 21.3566,
+      "step": 8670
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.468257329496063e-05,
+      "loss": 0.7336,
+      "step": 8671
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.467502197303812e-05,
+      "loss": 0.7635,
+      "step": 8672
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.466747028482371e-05,
+      "loss": 18.5139,
+      "step": 8673
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.465991823050588e-05,
+      "loss": 18.8024,
+      "step": 8674
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.465236581027314e-05,
+      "loss": 17.8421,
+      "step": 8675
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.4644813024314e-05,
+      "loss": 18.9109,
+      "step": 8676
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.463725987281702e-05,
+      "loss": 19.5277,
+      "step": 8677
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.462970635597066e-05,
+      "loss": 18.825,
+      "step": 8678
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.462215247396349e-05,
+      "loss": 18.6252,
+      "step": 8679
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.461459822698405e-05,
+      "loss": 17.0338,
+      "step": 8680
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.460704361522092e-05,
+      "loss": 0.741,
+      "step": 8681
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.459948863886263e-05,
+      "loss": 18.3019,
+      "step": 8682
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.459193329809779e-05,
+      "loss": 0.7705,
+      "step": 8683
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.458437759311495e-05,
+      "loss": 0.7562,
+      "step": 8684
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.457682152410272e-05,
+      "loss": 19.9467,
+      "step": 8685
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.456926509124969e-05,
+      "loss": 19.7437,
+      "step": 8686
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.456170829474448e-05,
+      "loss": 0.7809,
+      "step": 8687
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.455415113477571e-05,
+      "loss": 0.7808,
+      "step": 8688
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.454659361153199e-05,
+      "loss": 0.8029,
+      "step": 8689
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.453903572520198e-05,
+      "loss": 20.3153,
+      "step": 8690
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.453147747597434e-05,
+      "loss": 0.781,
+      "step": 8691
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.452391886403767e-05,
+      "loss": 21.0495,
+      "step": 8692
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.451635988958068e-05,
+      "loss": 18.2894,
+      "step": 8693
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.450880055279205e-05,
+      "loss": 19.6384,
+      "step": 8694
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.450124085386043e-05,
+      "loss": 17.2809,
+      "step": 8695
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.449368079297453e-05,
+      "loss": 0.7271,
+      "step": 8696
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.448612037032304e-05,
+      "loss": 0.7695,
+      "step": 8697
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.44785595860947e-05,
+      "loss": 0.7299,
+      "step": 8698
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.447099844047817e-05,
+      "loss": 0.7267,
+      "step": 8699
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.446343693366225e-05,
+      "loss": 0.7539,
+      "step": 8700
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.445587506583563e-05,
+      "loss": 19.7671,
+      "step": 8701
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.444831283718707e-05,
+      "loss": 19.3742,
+      "step": 8702
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.44407502479053e-05,
+      "loss": 18.0728,
+      "step": 8703
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.443318729817913e-05,
+      "loss": 0.8237,
+      "step": 8704
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.442562398819732e-05,
+      "loss": 0.7612,
+      "step": 8705
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.441806031814862e-05,
+      "loss": 0.7308,
+      "step": 8706
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.441049628822185e-05,
+      "loss": 16.0229,
+      "step": 8707
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.44029318986058e-05,
+      "loss": 0.7881,
+      "step": 8708
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.439536714948927e-05,
+      "loss": 17.7982,
+      "step": 8709
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.438780204106111e-05,
+      "loss": 17.9135,
+      "step": 8710
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.438023657351012e-05,
+      "loss": 0.7238,
+      "step": 8711
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.437267074702513e-05,
+      "loss": 0.7507,
+      "step": 8712
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.436510456179501e-05,
+      "loss": 18.81,
+      "step": 8713
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.43575380180086e-05,
+      "loss": 0.8279,
+      "step": 8714
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.434997111585477e-05,
+      "loss": 19.7725,
+      "step": 8715
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.434240385552236e-05,
+      "loss": 0.7846,
+      "step": 8716
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.433483623720028e-05,
+      "loss": 18.9818,
+      "step": 8717
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.432726826107742e-05,
+      "loss": 17.2523,
+      "step": 8718
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.431969992734269e-05,
+      "loss": 20.4059,
+      "step": 8719
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.431213123618495e-05,
+      "loss": 0.7075,
+      "step": 8720
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.430456218779317e-05,
+      "loss": 0.7793,
+      "step": 8721
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.429699278235625e-05,
+      "loss": 19.0962,
+      "step": 8722
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.42894230200631e-05,
+      "loss": 19.1132,
+      "step": 8723
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.428185290110272e-05,
+      "loss": 18.6041,
+      "step": 8724
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.427428242566402e-05,
+      "loss": 19.1262,
+      "step": 8725
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.426671159393597e-05,
+      "loss": 0.7284,
+      "step": 8726
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.425914040610753e-05,
+      "loss": 19.6379,
+      "step": 8727
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.425156886236771e-05,
+      "loss": 19.8743,
+      "step": 8728
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.424399696290547e-05,
+      "loss": 18.1701,
+      "step": 8729
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.42364247079098e-05,
+      "loss": 0.7648,
+      "step": 8730
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.422885209756973e-05,
+      "loss": 0.7886,
+      "step": 8731
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.422127913207427e-05,
+      "loss": 17.8577,
+      "step": 8732
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.421370581161243e-05,
+      "loss": 0.7442,
+      "step": 8733
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.420613213637326e-05,
+      "loss": 0.7751,
+      "step": 8734
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.419855810654576e-05,
+      "loss": 0.7745,
+      "step": 8735
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.419098372231902e-05,
+      "loss": 0.7117,
+      "step": 8736
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.418340898388209e-05,
+      "loss": 19.5606,
+      "step": 8737
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.417583389142403e-05,
+      "loss": 0.7255,
+      "step": 8738
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.416825844513392e-05,
+      "loss": 0.7753,
+      "step": 8739
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.416068264520085e-05,
+      "loss": 0.8041,
+      "step": 8740
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.41531064918139e-05,
+      "loss": 19.6962,
+      "step": 8741
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.414552998516219e-05,
+      "loss": 0.7347,
+      "step": 8742
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.413795312543482e-05,
+      "loss": 19.859,
+      "step": 8743
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.413037591282089e-05,
+      "loss": 0.7931,
+      "step": 8744
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.412279834750955e-05,
+      "loss": 19.1761,
+      "step": 8745
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.411522042968998e-05,
+      "loss": 0.7393,
+      "step": 8746
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.410764215955126e-05,
+      "loss": 18.0302,
+      "step": 8747
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.410006353728257e-05,
+      "loss": 0.7713,
+      "step": 8748
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.409248456307308e-05,
+      "loss": 0.755,
+      "step": 8749
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.408490523711195e-05,
+      "loss": 18.022,
+      "step": 8750
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.407732555958836e-05,
+      "loss": 17.9551,
+      "step": 8751
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.406974553069154e-05,
+      "loss": 20.1277,
+      "step": 8752
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.406216515061065e-05,
+      "loss": 0.8354,
+      "step": 8753
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.40545844195349e-05,
+      "loss": 20.4832,
+      "step": 8754
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.40470033376535e-05,
+      "loss": 0.7607,
+      "step": 8755
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.403942190515571e-05,
+      "loss": 0.7038,
+      "step": 8756
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.403184012223073e-05,
+      "loss": 20.3079,
+      "step": 8757
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.402425798906782e-05,
+      "loss": 0.7785,
+      "step": 8758
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.401667550585623e-05,
+      "loss": 17.8121,
+      "step": 8759
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.400909267278522e-05,
+      "loss": 18.0738,
+      "step": 8760
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.400150949004407e-05,
+      "loss": 18.5046,
+      "step": 8761
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.399392595782202e-05,
+      "loss": 19.129,
+      "step": 8762
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.398634207630841e-05,
+      "loss": 17.8576,
+      "step": 8763
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.397875784569248e-05,
+      "loss": 18.6897,
+      "step": 8764
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.397117326616357e-05,
+      "loss": 0.7548,
+      "step": 8765
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.3963588337911e-05,
+      "loss": 17.6313,
+      "step": 8766
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.395600306112408e-05,
+      "loss": 0.7454,
+      "step": 8767
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.394841743599212e-05,
+      "loss": 20.1451,
+      "step": 8768
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.394083146270449e-05,
+      "loss": 19.569,
+      "step": 8769
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.393324514145051e-05,
+      "loss": 19.0906,
+      "step": 8770
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.392565847241956e-05,
+      "loss": 20.4216,
+      "step": 8771
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.391807145580099e-05,
+      "loss": 0.7637,
+      "step": 8772
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.391048409178417e-05,
+      "loss": 19.7456,
+      "step": 8773
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.390289638055851e-05,
+      "loss": 0.6857,
+      "step": 8774
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.389530832231337e-05,
+      "loss": 0.7373,
+      "step": 8775
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.388771991723816e-05,
+      "loss": 0.7473,
+      "step": 8776
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.388013116552231e-05,
+      "loss": 20.5168,
+      "step": 8777
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.38725420673552e-05,
+      "loss": 0.78,
+      "step": 8778
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.386495262292626e-05,
+      "loss": 0.7747,
+      "step": 8779
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.385736283242495e-05,
+      "loss": 0.7769,
+      "step": 8780
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.384977269604072e-05,
+      "loss": 0.7851,
+      "step": 8781
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.384218221396297e-05,
+      "loss": 0.7317,
+      "step": 8782
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.383459138638119e-05,
+      "loss": 17.8853,
+      "step": 8783
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.382700021348487e-05,
+      "loss": 0.752,
+      "step": 8784
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.381940869546347e-05,
+      "loss": 19.7464,
+      "step": 8785
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.381181683250645e-05,
+      "loss": 19.8769,
+      "step": 8786
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.380422462480335e-05,
+      "loss": 0.7457,
+      "step": 8787
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.379663207254363e-05,
+      "loss": 0.7871,
+      "step": 8788
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.378903917591686e-05,
+      "loss": 21.6159,
+      "step": 8789
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.378144593511248e-05,
+      "loss": 19.87,
+      "step": 8790
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.37738523503201e-05,
+      "loss": 20.0049,
+      "step": 8791
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.376625842172919e-05,
+      "loss": 0.8105,
+      "step": 8792
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.375866414952934e-05,
+      "loss": 18.248,
+      "step": 8793
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.375106953391009e-05,
+      "loss": 19.5684,
+      "step": 8794
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.374347457506101e-05,
+      "loss": 18.5553,
+      "step": 8795
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.373587927317166e-05,
+      "loss": 0.7169,
+      "step": 8796
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.372828362843162e-05,
+      "loss": 18.5271,
+      "step": 8797
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.372068764103051e-05,
+      "loss": 19.0434,
+      "step": 8798
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.371309131115789e-05,
+      "loss": 0.7424,
+      "step": 8799
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.370549463900337e-05,
+      "loss": 18.2416,
+      "step": 8800
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.369789762475658e-05,
+      "loss": 17.8736,
+      "step": 8801
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.369030026860713e-05,
+      "loss": 0.8341,
+      "step": 8802
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.36827025707447e-05,
+      "loss": 17.5504,
+      "step": 8803
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.367510453135885e-05,
+      "loss": 0.7475,
+      "step": 8804
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.366750615063928e-05,
+      "loss": 0.7312,
+      "step": 8805
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.365990742877563e-05,
+      "loss": 19.5577,
+      "step": 8806
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.365230836595757e-05,
+      "loss": 0.7668,
+      "step": 8807
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.364470896237478e-05,
+      "loss": 0.7285,
+      "step": 8808
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.363710921821695e-05,
+      "loss": 17.5749,
+      "step": 8809
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.362950913367375e-05,
+      "loss": 21.5351,
+      "step": 8810
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.362190870893489e-05,
+      "loss": 20.4768,
+      "step": 8811
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.361430794419008e-05,
+      "loss": 18.6483,
+      "step": 8812
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.360670683962904e-05,
+      "loss": 17.5849,
+      "step": 8813
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.359910539544148e-05,
+      "loss": 18.6859,
+      "step": 8814
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.359150361181715e-05,
+      "loss": 0.7582,
+      "step": 8815
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.358390148894578e-05,
+      "loss": 18.2328,
+      "step": 8816
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.357629902701714e-05,
+      "loss": 20.3406,
+      "step": 8817
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.356869622622097e-05,
+      "loss": 18.0976,
+      "step": 8818
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.356109308674705e-05,
+      "loss": 17.558,
+      "step": 8819
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.355348960878513e-05,
+      "loss": 15.8624,
+      "step": 8820
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.354588579252503e-05,
+      "loss": 0.7951,
+      "step": 8821
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.353828163815652e-05,
+      "loss": 19.2772,
+      "step": 8822
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.353067714586942e-05,
+      "loss": 20.0393,
+      "step": 8823
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.352307231585352e-05,
+      "loss": 0.753,
+      "step": 8824
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.351546714829866e-05,
+      "loss": 0.8215,
+      "step": 8825
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.350786164339464e-05,
+      "loss": 17.8683,
+      "step": 8826
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.350025580133131e-05,
+      "loss": 17.8803,
+      "step": 8827
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.349264962229852e-05,
+      "loss": 16.4691,
+      "step": 8828
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.348504310648612e-05,
+      "loss": 0.7883,
+      "step": 8829
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.347743625408395e-05,
+      "loss": 0.7828,
+      "step": 8830
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.34698290652819e-05,
+      "loss": 0.7382,
+      "step": 8831
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.346222154026985e-05,
+      "loss": 18.2337,
+      "step": 8832
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.345461367923767e-05,
+      "loss": 18.159,
+      "step": 8833
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.344700548237525e-05,
+      "loss": 22.1083,
+      "step": 8834
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.343939694987251e-05,
+      "loss": 0.7196,
+      "step": 8835
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.343178808191935e-05,
+      "loss": 0.7345,
+      "step": 8836
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.34241788787057e-05,
+      "loss": 21.6161,
+      "step": 8837
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.341656934042148e-05,
+      "loss": 17.5268,
+      "step": 8838
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.340895946725662e-05,
+      "loss": 17.6108,
+      "step": 8839
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.340134925940108e-05,
+      "loss": 0.7855,
+      "step": 8840
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.339373871704479e-05,
+      "loss": 18.1705,
+      "step": 8841
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.338612784037773e-05,
+      "loss": 0.7445,
+      "step": 8842
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.337851662958987e-05,
+      "loss": 19.909,
+      "step": 8843
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.337090508487116e-05,
+      "loss": 18.0483,
+      "step": 8844
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.336329320641161e-05,
+      "loss": 0.7239,
+      "step": 8845
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.335568099440123e-05,
+      "loss": 0.8467,
+      "step": 8846
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.334806844902999e-05,
+      "loss": 20.3526,
+      "step": 8847
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.334045557048791e-05,
+      "loss": 21.1835,
+      "step": 8848
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.333284235896502e-05,
+      "loss": 0.7818,
+      "step": 8849
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.332522881465134e-05,
+      "loss": 0.7449,
+      "step": 8850
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.331761493773691e-05,
+      "loss": 0.7528,
+      "step": 8851
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.331000072841177e-05,
+      "loss": 15.0469,
+      "step": 8852
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.330238618686597e-05,
+      "loss": 0.7689,
+      "step": 8853
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.329477131328956e-05,
+      "loss": 17.4622,
+      "step": 8854
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.328715610787263e-05,
+      "loss": 0.756,
+      "step": 8855
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.327954057080526e-05,
+      "loss": 18.4637,
+      "step": 8856
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.327192470227752e-05,
+      "loss": 0.7954,
+      "step": 8857
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.326430850247951e-05,
+      "loss": 18.321,
+      "step": 8858
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.325669197160132e-05,
+      "loss": 0.7682,
+      "step": 8859
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.32490751098331e-05,
+      "loss": 0.7805,
+      "step": 8860
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.324145791736492e-05,
+      "loss": 0.7943,
+      "step": 8861
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.323384039438692e-05,
+      "loss": 20.9079,
+      "step": 8862
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.322622254108926e-05,
+      "loss": 18.226,
+      "step": 8863
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.321860435766204e-05,
+      "loss": 20.4452,
+      "step": 8864
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.321098584429545e-05,
+      "loss": 17.9266,
+      "step": 8865
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.320336700117963e-05,
+      "loss": 0.7387,
+      "step": 8866
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.319574782850478e-05,
+      "loss": 20.6974,
+      "step": 8867
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.318812832646102e-05,
+      "loss": 0.7907,
+      "step": 8868
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.318050849523859e-05,
+      "loss": 18.4823,
+      "step": 8869
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.317288833502766e-05,
+      "loss": 0.757,
+      "step": 8870
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.316526784601842e-05,
+      "loss": 20.0182,
+      "step": 8871
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.315764702840108e-05,
+      "loss": 17.7608,
+      "step": 8872
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.315002588236586e-05,
+      "loss": 18.541,
+      "step": 8873
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.314240440810302e-05,
+      "loss": 0.7299,
+      "step": 8874
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.313478260580275e-05,
+      "loss": 0.782,
+      "step": 8875
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.31271604756553e-05,
+      "loss": 18.5994,
+      "step": 8876
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.311953801785095e-05,
+      "loss": 0.8314,
+      "step": 8877
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.311191523257991e-05,
+      "loss": 20.4562,
+      "step": 8878
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.310429212003248e-05,
+      "loss": 0.8061,
+      "step": 8879
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.309666868039891e-05,
+      "loss": 0.7721,
+      "step": 8880
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.308904491386953e-05,
+      "loss": 0.8038,
+      "step": 8881
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.308142082063458e-05,
+      "loss": 0.7846,
+      "step": 8882
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.30737964008844e-05,
+      "loss": 17.8953,
+      "step": 8883
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.306617165480927e-05,
+      "loss": 0.7804,
+      "step": 8884
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.30585465825995e-05,
+      "loss": 18.533,
+      "step": 8885
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.305092118444542e-05,
+      "loss": 18.6109,
+      "step": 8886
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.304329546053738e-05,
+      "loss": 0.7631,
+      "step": 8887
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.303566941106572e-05,
+      "loss": 18.7714,
+      "step": 8888
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.302804303622077e-05,
+      "loss": 21.01,
+      "step": 8889
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.302041633619287e-05,
+      "loss": 0.7865,
+      "step": 8890
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.301278931117241e-05,
+      "loss": 18.4928,
+      "step": 8891
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.300516196134977e-05,
+      "loss": 18.9709,
+      "step": 8892
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.299753428691532e-05,
+      "loss": 18.4346,
+      "step": 8893
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.298990628805943e-05,
+      "loss": 0.7536,
+      "step": 8894
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.298227796497253e-05,
+      "loss": 0.7734,
+      "step": 8895
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.297464931784499e-05,
+      "loss": 19.3113,
+      "step": 8896
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.296702034686725e-05,
+      "loss": 20.2114,
+      "step": 8897
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.295939105222975e-05,
+      "loss": 0.7431,
+      "step": 8898
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.295176143412285e-05,
+      "loss": 0.7568,
+      "step": 8899
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.294413149273704e-05,
+      "loss": 0.7841,
+      "step": 8900
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.293650122826275e-05,
+      "loss": 18.1624,
+      "step": 8901
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.292887064089044e-05,
+      "loss": 0.7465,
+      "step": 8902
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.292123973081056e-05,
+      "loss": 0.7497,
+      "step": 8903
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.291360849821358e-05,
+      "loss": 0.7717,
+      "step": 8904
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.290597694329001e-05,
+      "loss": 18.9638,
+      "step": 8905
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.289834506623029e-05,
+      "loss": 19.847,
+      "step": 8906
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.289071286722493e-05,
+      "loss": 17.7419,
+      "step": 8907
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.288308034646445e-05,
+      "loss": 18.0717,
+      "step": 8908
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.287544750413932e-05,
+      "loss": 0.8082,
+      "step": 8909
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.286781434044008e-05,
+      "loss": 0.8389,
+      "step": 8910
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.286018085555729e-05,
+      "loss": 20.7159,
+      "step": 8911
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.285254704968142e-05,
+      "loss": 0.7981,
+      "step": 8912
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.284491292300305e-05,
+      "loss": 18.0632,
+      "step": 8913
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.283727847571271e-05,
+      "loss": 17.6793,
+      "step": 8914
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.282964370800098e-05,
+      "loss": 0.7546,
+      "step": 8915
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.282200862005843e-05,
+      "loss": 18.2471,
+      "step": 8916
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.281437321207558e-05,
+      "loss": 0.8075,
+      "step": 8917
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.280673748424306e-05,
+      "loss": 18.7983,
+      "step": 8918
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.279910143675146e-05,
+      "loss": 0.7842,
+      "step": 8919
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.279146506979137e-05,
+      "loss": 0.7417,
+      "step": 8920
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.278382838355337e-05,
+      "loss": 0.7399,
+      "step": 8921
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.277619137822812e-05,
+      "loss": 17.3895,
+      "step": 8922
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.276855405400619e-05,
+      "loss": 17.2728,
+      "step": 8923
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.276091641107824e-05,
+      "loss": 17.7424,
+      "step": 8924
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.275327844963492e-05,
+      "loss": 19.6649,
+      "step": 8925
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.274564016986686e-05,
+      "loss": 18.8374,
+      "step": 8926
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.27380015719647e-05,
+      "loss": 16.9236,
+      "step": 8927
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.27303626561191e-05,
+      "loss": 18.8322,
+      "step": 8928
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.272272342252075e-05,
+      "loss": 0.7291,
+      "step": 8929
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.271508387136034e-05,
+      "loss": 0.7703,
+      "step": 8930
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.27074440028285e-05,
+      "loss": 0.7709,
+      "step": 8931
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.269980381711597e-05,
+      "loss": 18.1601,
+      "step": 8932
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.269216331441344e-05,
+      "loss": 19.9987,
+      "step": 8933
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.26845224949116e-05,
+      "loss": 19.6335,
+      "step": 8934
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.267688135880122e-05,
+      "loss": 16.6757,
+      "step": 8935
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.266923990627295e-05,
+      "loss": 17.3407,
+      "step": 8936
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.266159813751757e-05,
+      "loss": 19.4682,
+      "step": 8937
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.265395605272581e-05,
+      "loss": 0.753,
+      "step": 8938
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.264631365208843e-05,
+      "loss": 18.0854,
+      "step": 8939
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.263867093579616e-05,
+      "loss": 19.1007,
+      "step": 8940
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.263102790403979e-05,
+      "loss": 19.2139,
+      "step": 8941
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.262338455701006e-05,
+      "loss": 0.7248,
+      "step": 8942
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.261574089489778e-05,
+      "loss": 0.7566,
+      "step": 8943
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.260809691789373e-05,
+      "loss": 0.7319,
+      "step": 8944
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.26004526261887e-05,
+      "loss": 0.6887,
+      "step": 8945
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.25928080199735e-05,
+      "loss": 0.7099,
+      "step": 8946
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.258516309943891e-05,
+      "loss": 0.7823,
+      "step": 8947
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.25775178647758e-05,
+      "loss": 0.7334,
+      "step": 8948
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.256987231617497e-05,
+      "loss": 0.7248,
+      "step": 8949
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.256222645382725e-05,
+      "loss": 0.7393,
+      "step": 8950
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.255458027792349e-05,
+      "loss": 20.9878,
+      "step": 8951
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.254693378865455e-05,
+      "loss": 0.7714,
+      "step": 8952
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.253928698621128e-05,
+      "loss": 21.3214,
+      "step": 8953
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.253163987078453e-05,
+      "loss": 0.7933,
+      "step": 8954
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.252399244256519e-05,
+      "loss": 17.3346,
+      "step": 8955
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.251634470174414e-05,
+      "loss": 17.1451,
+      "step": 8956
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.250869664851227e-05,
+      "loss": 17.0376,
+      "step": 8957
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.250104828306048e-05,
+      "loss": 0.8284,
+      "step": 8958
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.249339960557966e-05,
+      "loss": 0.8043,
+      "step": 8959
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.248575061626075e-05,
+      "loss": 18.2342,
+      "step": 8960
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.247810131529463e-05,
+      "loss": 16.7648,
+      "step": 8961
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.247045170287227e-05,
+      "loss": 0.8074,
+      "step": 8962
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.246280177918459e-05,
+      "loss": 18.4507,
+      "step": 8963
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.245515154442253e-05,
+      "loss": 0.8018,
+      "step": 8964
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.244750099877703e-05,
+      "loss": 19.1413,
+      "step": 8965
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.243985014243906e-05,
+      "loss": 0.7802,
+      "step": 8966
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.243219897559962e-05,
+      "loss": 17.4752,
+      "step": 8967
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.242454749844964e-05,
+      "loss": 17.2841,
+      "step": 8968
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.24168957111801e-05,
+      "loss": 0.7408,
+      "step": 8969
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.240924361398202e-05,
+      "loss": 16.8239,
+      "step": 8970
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.24015912070464e-05,
+      "loss": 19.9207,
+      "step": 8971
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.239393849056422e-05,
+      "loss": 18.554,
+      "step": 8972
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.23862854647265e-05,
+      "loss": 19.2837,
+      "step": 8973
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.237863212972428e-05,
+      "loss": 20.5398,
+      "step": 8974
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.237097848574855e-05,
+      "loss": 16.9975,
+      "step": 8975
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.236332453299039e-05,
+      "loss": 0.7458,
+      "step": 8976
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.235567027164082e-05,
+      "loss": 0.7613,
+      "step": 8977
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.234801570189093e-05,
+      "loss": 0.7151,
+      "step": 8978
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.23403608239317e-05,
+      "loss": 20.2221,
+      "step": 8979
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.233270563795428e-05,
+      "loss": 18.8402,
+      "step": 8980
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.232505014414971e-05,
+      "loss": 18.2191,
+      "step": 8981
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.231739434270907e-05,
+      "loss": 0.8109,
+      "step": 8982
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.230973823382344e-05,
+      "loss": 0.7489,
+      "step": 8983
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.230208181768395e-05,
+      "loss": 0.7946,
+      "step": 8984
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.229442509448168e-05,
+      "loss": 0.7683,
+      "step": 8985
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.228676806440776e-05,
+      "loss": 0.7448,
+      "step": 8986
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.227911072765329e-05,
+      "loss": 0.782,
+      "step": 8987
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.227145308440943e-05,
+      "loss": 18.9401,
+      "step": 8988
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.226379513486729e-05,
+      "loss": 17.2512,
+      "step": 8989
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.225613687921801e-05,
+      "loss": 18.4429,
+      "step": 8990
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.224847831765278e-05,
+      "loss": 20.1199,
+      "step": 8991
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.224081945036274e-05,
+      "loss": 16.8592,
+      "step": 8992
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.223316027753902e-05,
+      "loss": 0.7544,
+      "step": 8993
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.222550079937283e-05,
+      "loss": 17.6365,
+      "step": 8994
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.221784101605537e-05,
+      "loss": 18.7956,
+      "step": 8995
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.22101809277778e-05,
+      "loss": 19.5838,
+      "step": 8996
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.220252053473132e-05,
+      "loss": 20.1508,
+      "step": 8997
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.219485983710716e-05,
+      "loss": 18.0268,
+      "step": 8998
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.218719883509648e-05,
+      "loss": 0.7409,
+      "step": 8999
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.217953752889057e-05,
+      "loss": 19.0899,
+      "step": 9000
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.217187591868059e-05,
+      "loss": 0.7818,
+      "step": 9001
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.216421400465784e-05,
+      "loss": 21.4477,
+      "step": 9002
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.215655178701351e-05,
+      "loss": 18.9342,
+      "step": 9003
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.214888926593887e-05,
+      "loss": 18.8181,
+      "step": 9004
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.214122644162519e-05,
+      "loss": 20.0513,
+      "step": 9005
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.213356331426372e-05,
+      "loss": 18.6886,
+      "step": 9006
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.212589988404574e-05,
+      "loss": 0.7738,
+      "step": 9007
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.211823615116253e-05,
+      "loss": 17.4565,
+      "step": 9008
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.211057211580539e-05,
+      "loss": 0.7695,
+      "step": 9009
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.21029077781656e-05,
+      "loss": 0.7669,
+      "step": 9010
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.209524313843446e-05,
+      "loss": 19.6308,
+      "step": 9011
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.208757819680329e-05,
+      "loss": 18.93,
+      "step": 9012
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.20799129534634e-05,
+      "loss": 0.7796,
+      "step": 9013
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.207224740860616e-05,
+      "loss": 0.768,
+      "step": 9014
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.206458156242285e-05,
+      "loss": 19.4764,
+      "step": 9015
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.20569154151048e-05,
+      "loss": 20.9433,
+      "step": 9016
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.204924896684343e-05,
+      "loss": 20.112,
+      "step": 9017
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.204158221783002e-05,
+      "loss": 0.7468,
+      "step": 9018
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.203391516825599e-05,
+      "loss": 0.7242,
+      "step": 9019
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.202624781831268e-05,
+      "loss": 19.2711,
+      "step": 9020
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.201858016819148e-05,
+      "loss": 0.7938,
+      "step": 9021
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.20109122180838e-05,
+      "loss": 16.812,
+      "step": 9022
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.200324396818097e-05,
+      "loss": 0.7195,
+      "step": 9023
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.199557541867445e-05,
+      "loss": 0.7141,
+      "step": 9024
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.198790656975563e-05,
+      "loss": 17.929,
+      "step": 9025
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.198023742161592e-05,
+      "loss": 0.7422,
+      "step": 9026
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.197256797444675e-05,
+      "loss": 0.816,
+      "step": 9027
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.196489822843955e-05,
+      "loss": 18.27,
+      "step": 9028
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.195722818378575e-05,
+      "loss": 17.1846,
+      "step": 9029
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.194955784067682e-05,
+      "loss": 15.9734,
+      "step": 9030
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.194188719930418e-05,
+      "loss": 18.1582,
+      "step": 9031
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.193421625985931e-05,
+      "loss": 0.7317,
+      "step": 9032
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.192654502253369e-05,
+      "loss": 18.5325,
+      "step": 9033
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.191887348751878e-05,
+      "loss": 20.202,
+      "step": 9034
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.191120165500606e-05,
+      "loss": 17.3449,
+      "step": 9035
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.190352952518704e-05,
+      "loss": 20.3376,
+      "step": 9036
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.189585709825319e-05,
+      "loss": 17.9604,
+      "step": 9037
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.188818437439604e-05,
+      "loss": 17.686,
+      "step": 9038
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.188051135380709e-05,
+      "loss": 19.5663,
+      "step": 9039
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.187283803667787e-05,
+      "loss": 0.7767,
+      "step": 9040
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.186516442319989e-05,
+      "loss": 18.5319,
+      "step": 9041
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.18574905135647e-05,
+      "loss": 21.9795,
+      "step": 9042
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.184981630796385e-05,
+      "loss": 0.7449,
+      "step": 9043
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.184214180658887e-05,
+      "loss": 0.7456,
+      "step": 9044
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.183446700963133e-05,
+      "loss": 17.7542,
+      "step": 9045
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.182679191728278e-05,
+      "loss": 20.6558,
+      "step": 9046
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.181911652973482e-05,
+      "loss": 0.8274,
+      "step": 9047
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.1811440847179e-05,
+      "loss": 0.7923,
+      "step": 9048
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.180376486980693e-05,
+      "loss": 17.5503,
+      "step": 9049
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.179608859781019e-05,
+      "loss": 0.7755,
+      "step": 9050
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.178841203138038e-05,
+      "loss": 19.9668,
+      "step": 9051
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.178073517070912e-05,
+      "loss": 0.7755,
+      "step": 9052
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.177305801598803e-05,
+      "loss": 0.8066,
+      "step": 9053
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.17653805674087e-05,
+      "loss": 19.2864,
+      "step": 9054
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.17577028251628e-05,
+      "loss": 0.8045,
+      "step": 9055
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.175002478944194e-05,
+      "loss": 0.8077,
+      "step": 9056
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.17423464604378e-05,
+      "loss": 0.7609,
+      "step": 9057
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.1734667838342e-05,
+      "loss": 0.7648,
+      "step": 9058
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.17269889233462e-05,
+      "loss": 18.6729,
+      "step": 9059
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.17193097156421e-05,
+      "loss": 19.0304,
+      "step": 9060
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.171163021542133e-05,
+      "loss": 0.7369,
+      "step": 9061
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.170395042287562e-05,
+      "loss": 19.7134,
+      "step": 9062
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.169627033819661e-05,
+      "loss": 0.7398,
+      "step": 9063
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.168858996157605e-05,
+      "loss": 18.6533,
+      "step": 9064
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.16809092932056e-05,
+      "loss": 17.0302,
+      "step": 9065
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.167322833327699e-05,
+      "loss": 0.7508,
+      "step": 9066
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.166554708198193e-05,
+      "loss": 0.7212,
+      "step": 9067
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.165786553951216e-05,
+      "loss": 0.822,
+      "step": 9068
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.165018370605939e-05,
+      "loss": 18.8077,
+      "step": 9069
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.164250158181538e-05,
+      "loss": 16.7643,
+      "step": 9070
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.163481916697188e-05,
+      "loss": 0.7481,
+      "step": 9071
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.162713646172065e-05,
+      "loss": 20.107,
+      "step": 9072
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.16194534662534e-05,
+      "loss": 0.7229,
+      "step": 9073
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.161177018076197e-05,
+      "loss": 17.5655,
+      "step": 9074
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.16040866054381e-05,
+      "loss": 0.7629,
+      "step": 9075
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.159640274047358e-05,
+      "loss": 0.8427,
+      "step": 9076
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.158871858606018e-05,
+      "loss": 16.7014,
+      "step": 9077
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.158103414238974e-05,
+      "loss": 0.7614,
+      "step": 9078
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.157334940965403e-05,
+      "loss": 18.174,
+      "step": 9079
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.156566438804489e-05,
+      "loss": 18.1904,
+      "step": 9080
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.155797907775413e-05,
+      "loss": 0.7865,
+      "step": 9081
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.155029347897354e-05,
+      "loss": 0.7377,
+      "step": 9082
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.154260759189501e-05,
+      "loss": 19.4083,
+      "step": 9083
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.153492141671036e-05,
+      "loss": 17.4618,
+      "step": 9084
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.152723495361144e-05,
+      "loss": 19.6128,
+      "step": 9085
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.151954820279011e-05,
+      "loss": 0.7511,
+      "step": 9086
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.15118611644382e-05,
+      "loss": 0.7573,
+      "step": 9087
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.150417383874762e-05,
+      "loss": 0.7646,
+      "step": 9088
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.149648622591024e-05,
+      "loss": 0.7314,
+      "step": 9089
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.148879832611793e-05,
+      "loss": 16.5444,
+      "step": 9090
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.148111013956262e-05,
+      "loss": 17.9595,
+      "step": 9091
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.147342166643614e-05,
+      "loss": 0.7213,
+      "step": 9092
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.146573290693045e-05,
+      "loss": 0.7641,
+      "step": 9093
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.145804386123744e-05,
+      "loss": 0.7695,
+      "step": 9094
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.145035452954906e-05,
+      "loss": 0.7592,
+      "step": 9095
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.14426649120572e-05,
+      "loss": 20.1586,
+      "step": 9096
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.14349750089538e-05,
+      "loss": 0.7691,
+      "step": 9097
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.142728482043085e-05,
+      "loss": 0.7975,
+      "step": 9098
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.141959434668025e-05,
+      "loss": 0.7311,
+      "step": 9099
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.141190358789395e-05,
+      "loss": 0.7695,
+      "step": 9100
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.140421254426392e-05,
+      "loss": 0.795,
+      "step": 9101
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.139652121598218e-05,
+      "loss": 0.7022,
+      "step": 9102
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.138882960324064e-05,
+      "loss": 0.8168,
+      "step": 9103
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.138113770623133e-05,
+      "loss": 18.3033,
+      "step": 9104
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.137344552514622e-05,
+      "loss": 0.7602,
+      "step": 9105
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.13657530601773e-05,
+      "loss": 19.6377,
+      "step": 9106
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.13580603115166e-05,
+      "loss": 16.8259,
+      "step": 9107
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.135036727935611e-05,
+      "loss": 18.9642,
+      "step": 9108
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.134267396388788e-05,
+      "loss": 18.5204,
+      "step": 9109
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.133498036530389e-05,
+      "loss": 0.7357,
+      "step": 9110
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.132728648379622e-05,
+      "loss": 0.7693,
+      "step": 9111
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.13195923195569e-05,
+      "loss": 0.7797,
+      "step": 9112
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.131189787277797e-05,
+      "loss": 0.7659,
+      "step": 9113
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.130420314365147e-05,
+      "loss": 0.8085,
+      "step": 9114
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.129650813236949e-05,
+      "loss": 0.7799,
+      "step": 9115
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.12888128391241e-05,
+      "loss": 0.7545,
+      "step": 9116
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.128111726410733e-05,
+      "loss": 0.7994,
+      "step": 9117
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.127342140751132e-05,
+      "loss": 20.5208,
+      "step": 9118
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.126572526952814e-05,
+      "loss": 0.7895,
+      "step": 9119
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.125802885034989e-05,
+      "loss": 0.7676,
+      "step": 9120
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.125033215016864e-05,
+      "loss": 17.6576,
+      "step": 9121
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.124263516917655e-05,
+      "loss": 19.6559,
+      "step": 9122
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.123493790756573e-05,
+      "loss": 0.8146,
+      "step": 9123
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.122724036552828e-05,
+      "loss": 0.7508,
+      "step": 9124
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.121954254325634e-05,
+      "loss": 18.0766,
+      "step": 9125
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.121184444094208e-05,
+      "loss": 0.7878,
+      "step": 9126
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.120414605877761e-05,
+      "loss": 0.7176,
+      "step": 9127
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.11964473969551e-05,
+      "loss": 20.1683,
+      "step": 9128
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.118874845566669e-05,
+      "loss": 0.7357,
+      "step": 9129
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.118104923510457e-05,
+      "loss": 17.0349,
+      "step": 9130
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.117334973546092e-05,
+      "loss": 0.7536,
+      "step": 9131
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.11656499569279e-05,
+      "loss": 0.7445,
+      "step": 9132
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.115794989969772e-05,
+      "loss": 21.2615,
+      "step": 9133
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.115024956396255e-05,
+      "loss": 0.7656,
+      "step": 9134
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.11425489499146e-05,
+      "loss": 0.8222,
+      "step": 9135
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.11348480577461e-05,
+      "loss": 0.7498,
+      "step": 9136
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.112714688764925e-05,
+      "loss": 18.3487,
+      "step": 9137
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.111944543981625e-05,
+      "loss": 0.8013,
+      "step": 9138
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.111174371443938e-05,
+      "loss": 18.0732,
+      "step": 9139
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.110404171171083e-05,
+      "loss": 0.694,
+      "step": 9140
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.109633943182289e-05,
+      "loss": 0.7667,
+      "step": 9141
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.108863687496776e-05,
+      "loss": 17.6139,
+      "step": 9142
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.108093404133773e-05,
+      "loss": 0.7429,
+      "step": 9143
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.107323093112504e-05,
+      "loss": 20.2619,
+      "step": 9144
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.106552754452198e-05,
+      "loss": 0.7012,
+      "step": 9145
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.105782388172086e-05,
+      "loss": 0.7598,
+      "step": 9146
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.105011994291391e-05,
+      "loss": 0.7869,
+      "step": 9147
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.104241572829344e-05,
+      "loss": 18.4876,
+      "step": 9148
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.1034711238051736e-05,
+      "loss": 0.7397,
+      "step": 9149
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.102700647238114e-05,
+      "loss": 0.75,
+      "step": 9150
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.101930143147395e-05,
+      "loss": 17.6817,
+      "step": 9151
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.101159611552247e-05,
+      "loss": 20.721,
+      "step": 9152
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.1003890524719035e-05,
+      "loss": 18.5721,
+      "step": 9153
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.099618465925598e-05,
+      "loss": 0.8216,
+      "step": 9154
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0988478519325665e-05,
+      "loss": 0.7562,
+      "step": 9155
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.09807721051204e-05,
+      "loss": 18.1877,
+      "step": 9156
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.097306541683256e-05,
+      "loss": 18.1824,
+      "step": 9157
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.09653584546545e-05,
+      "loss": 0.8029,
+      "step": 9158
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0957651218778575e-05,
+      "loss": 0.7637,
+      "step": 9159
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0949943709397197e-05,
+      "loss": 0.7816,
+      "step": 9160
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0942235926702715e-05,
+      "loss": 20.2754,
+      "step": 9161
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0934527870887514e-05,
+      "loss": 16.3253,
+      "step": 9162
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0926819542144007e-05,
+      "loss": 0.7496,
+      "step": 9163
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0919110940664604e-05,
+      "loss": 0.7808,
+      "step": 9164
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.09114020666417e-05,
+      "loss": 0.7793,
+      "step": 9165
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.090369292026768e-05,
+      "loss": 20.5008,
+      "step": 9166
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.089598350173501e-05,
+      "loss": 19.4321,
+      "step": 9167
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.088827381123612e-05,
+      "loss": 0.7866,
+      "step": 9168
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.088056384896341e-05,
+      "loss": 0.7776,
+      "step": 9169
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0872853615109346e-05,
+      "loss": 20.2282,
+      "step": 9170
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0865143109866386e-05,
+      "loss": 0.7654,
+      "step": 9171
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0857432333426956e-05,
+      "loss": 19.6073,
+      "step": 9172
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.084972128598354e-05,
+      "loss": 18.3485,
+      "step": 9173
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0842009967728606e-05,
+      "loss": 0.787,
+      "step": 9174
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.083429837885464e-05,
+      "loss": 0.7372,
+      "step": 9175
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0826586519554085e-05,
+      "loss": 17.9298,
+      "step": 9176
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0818874390019475e-05,
+      "loss": 0.7979,
+      "step": 9177
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.081116199044329e-05,
+      "loss": 17.0645,
+      "step": 9178
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0803449321018035e-05,
+      "loss": 20.4446,
+      "step": 9179
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0795736381936205e-05,
+      "loss": 21.5235,
+      "step": 9180
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.078802317339035e-05,
+      "loss": 18.5682,
+      "step": 9181
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0780309695572956e-05,
+      "loss": 20.7919,
+      "step": 9182
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.077259594867659e-05,
+      "loss": 0.74,
+      "step": 9183
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.076488193289375e-05,
+      "loss": 0.8092,
+      "step": 9184
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0757167648417014e-05,
+      "loss": 18.4275,
+      "step": 9185
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0749453095438905e-05,
+      "loss": 19.5518,
+      "step": 9186
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.074173827415199e-05,
+      "loss": 19.3695,
+      "step": 9187
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0734023184748844e-05,
+      "loss": 17.5056,
+      "step": 9188
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0726307827422035e-05,
+      "loss": 0.7543,
+      "step": 9189
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.071859220236412e-05,
+      "loss": 17.9572,
+      "step": 9190
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.07108763097677e-05,
+      "loss": 0.7364,
+      "step": 9191
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.070316014982536e-05,
+      "loss": 21.2941,
+      "step": 9192
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.06954437227297e-05,
+      "loss": 17.8581,
+      "step": 9193
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0687727028673325e-05,
+      "loss": 20.7276,
+      "step": 9194
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0680010067848846e-05,
+      "loss": 21.106,
+      "step": 9195
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0672292840448875e-05,
+      "loss": 18.4323,
+      "step": 9196
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.066457534666604e-05,
+      "loss": 0.7673,
+      "step": 9197
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.065685758669296e-05,
+      "loss": 0.7906,
+      "step": 9198
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.064913956072229e-05,
+      "loss": 0.7386,
+      "step": 9199
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.064142126894665e-05,
+      "loss": 19.412,
+      "step": 9200
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0633702711558715e-05,
+      "loss": 0.7777,
+      "step": 9201
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0625983888751135e-05,
+      "loss": 19.1407,
+      "step": 9202
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0618264800716586e-05,
+      "loss": 0.7446,
+      "step": 9203
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0610545447647705e-05,
+      "loss": 0.7784,
+      "step": 9204
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0602825829737175e-05,
+      "loss": 0.8005,
+      "step": 9205
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.059510594717771e-05,
+      "loss": 19.7118,
+      "step": 9206
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.058738580016199e-05,
+      "loss": 22.0589,
+      "step": 9207
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0579665388882675e-05,
+      "loss": 19.28,
+      "step": 9208
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.057194471353253e-05,
+      "loss": 19.9481,
+      "step": 9209
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.05642237743042e-05,
+      "loss": 0.7545,
+      "step": 9210
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.055650257139045e-05,
+      "loss": 0.7358,
+      "step": 9211
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.054878110498399e-05,
+      "loss": 17.9803,
+      "step": 9212
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.054105937527753e-05,
+      "loss": 0.8035,
+      "step": 9213
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0533337382463815e-05,
+      "loss": 16.4397,
+      "step": 9214
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.05256151267356e-05,
+      "loss": 0.7643,
+      "step": 9215
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.051789260828564e-05,
+      "loss": 19.5667,
+      "step": 9216
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0510169827306686e-05,
+      "loss": 0.7176,
+      "step": 9217
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0502446783991474e-05,
+      "loss": 0.7275,
+      "step": 9218
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.04947234785328e-05,
+      "loss": 19.5771,
+      "step": 9219
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.048699991112342e-05,
+      "loss": 17.47,
+      "step": 9220
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0479276081956146e-05,
+      "loss": 19.4254,
+      "step": 9221
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0471551991223743e-05,
+      "loss": 19.9155,
+      "step": 9222
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.046382763911901e-05,
+      "loss": 20.6752,
+      "step": 9223
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.045610302583474e-05,
+      "loss": 0.7151,
+      "step": 9224
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.044837815156377e-05,
+      "loss": 0.8198,
+      "step": 9225
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0440653016498884e-05,
+      "loss": 19.9951,
+      "step": 9226
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0432927620832916e-05,
+      "loss": 16.7253,
+      "step": 9227
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0425201964758684e-05,
+      "loss": 20.0769,
+      "step": 9228
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.041747604846904e-05,
+      "loss": 18.0924,
+      "step": 9229
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0409749872156804e-05,
+      "loss": 19.1887,
+      "step": 9230
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0402023436014856e-05,
+      "loss": 18.3544,
+      "step": 9231
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.039429674023601e-05,
+      "loss": 16.8394,
+      "step": 9232
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.038656978501316e-05,
+      "loss": 0.7767,
+      "step": 9233
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.037884257053914e-05,
+      "loss": 0.7878,
+      "step": 9234
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0371115097006844e-05,
+      "loss": 19.2748,
+      "step": 9235
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.036338736460915e-05,
+      "loss": 19.8895,
+      "step": 9236
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0355659373538945e-05,
+      "loss": 0.8143,
+      "step": 9237
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.034793112398911e-05,
+      "loss": 19.2704,
+      "step": 9238
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.034020261615255e-05,
+      "loss": 0.7517,
+      "step": 9239
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0332473850222194e-05,
+      "loss": 0.7502,
+      "step": 9240
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.03247448263909e-05,
+      "loss": 17.9719,
+      "step": 9241
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.031701554485163e-05,
+      "loss": 18.6955,
+      "step": 9242
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0309286005797294e-05,
+      "loss": 17.6661,
+      "step": 9243
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.030155620942083e-05,
+      "loss": 0.7764,
+      "step": 9244
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.029382615591518e-05,
+      "loss": 18.7527,
+      "step": 9245
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.028609584547325e-05,
+      "loss": 0.7587,
+      "step": 9246
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.027836527828805e-05,
+      "loss": 19.8512,
+      "step": 9247
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.027063445455249e-05,
+      "loss": 18.7141,
+      "step": 9248
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0262903374459554e-05,
+      "loss": 17.2208,
+      "step": 9249
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.025517203820221e-05,
+      "loss": 19.9049,
+      "step": 9250
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.024744044597342e-05,
+      "loss": 19.4992,
+      "step": 9251
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.023970859796618e-05,
+      "loss": 20.5258,
+      "step": 9252
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.023197649437348e-05,
+      "loss": 18.7445,
+      "step": 9253
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.022424413538832e-05,
+      "loss": 18.6367,
+      "step": 9254
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.021651152120368e-05,
+      "loss": 0.7559,
+      "step": 9255
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.020877865201257e-05,
+      "loss": 21.3325,
+      "step": 9256
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.020104552800803e-05,
+      "loss": 19.1425,
+      "step": 9257
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0193312149383054e-05,
+      "loss": 0.7818,
+      "step": 9258
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0185578516330687e-05,
+      "loss": 0.7935,
+      "step": 9259
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.017784462904394e-05,
+      "loss": 19.5177,
+      "step": 9260
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.017011048771588e-05,
+      "loss": 17.7038,
+      "step": 9261
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.016237609253954e-05,
+      "loss": 0.8371,
+      "step": 9262
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.015464144370797e-05,
+      "loss": 0.7822,
+      "step": 9263
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.014690654141424e-05,
+      "loss": 17.8725,
+      "step": 9264
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.013917138585139e-05,
+      "loss": 17.4567,
+      "step": 9265
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.013143597721251e-05,
+      "loss": 0.7326,
+      "step": 9266
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.012370031569068e-05,
+      "loss": 0.8429,
+      "step": 9267
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.011596440147899e-05,
+      "loss": 18.253,
+      "step": 9268
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.01082282347705e-05,
+      "loss": 18.4603,
+      "step": 9269
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.010049181575832e-05,
+      "loss": 0.7162,
+      "step": 9270
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0092755144635584e-05,
+      "loss": 0.7527,
+      "step": 9271
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.008501822159536e-05,
+      "loss": 0.7277,
+      "step": 9272
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.007728104683078e-05,
+      "loss": 0.7509,
+      "step": 9273
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.006954362053495e-05,
+      "loss": 16.7669,
+      "step": 9274
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.006180594290104e-05,
+      "loss": 0.7201,
+      "step": 9275
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.005406801412213e-05,
+      "loss": 18.3229,
+      "step": 9276
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.00463298343914e-05,
+      "loss": 17.2748,
+      "step": 9277
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.003859140390199e-05,
+      "loss": 19.2743,
+      "step": 9278
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.003085272284703e-05,
+      "loss": 0.7505,
+      "step": 9279
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.0023113791419705e-05,
+      "loss": 17.6291,
+      "step": 9280
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.001537460981317e-05,
+      "loss": 0.7728,
+      "step": 9281
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 6.000763517822061e-05,
+      "loss": 18.5303,
+      "step": 9282
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.9999895496835165e-05,
+      "loss": 0.7511,
+      "step": 9283
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.999215556585006e-05,
+      "loss": 0.7729,
+      "step": 9284
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.998441538545847e-05,
+      "loss": 0.7756,
+      "step": 9285
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.9976674955853594e-05,
+      "loss": 0.7468,
+      "step": 9286
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.996893427722864e-05,
+      "loss": 19.4601,
+      "step": 9287
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.99611933497768e-05,
+      "loss": 20.4298,
+      "step": 9288
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.995345217369129e-05,
+      "loss": 0.7587,
+      "step": 9289
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.994571074916535e-05,
+      "loss": 17.7624,
+      "step": 9290
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.993796907639221e-05,
+      "loss": 17.7908,
+      "step": 9291
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.993022715556509e-05,
+      "loss": 20.2768,
+      "step": 9292
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.992248498687722e-05,
+      "loss": 16.3089,
+      "step": 9293
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.991474257052185e-05,
+      "loss": 0.7762,
+      "step": 9294
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.9906999906692274e-05,
+      "loss": 18.8494,
+      "step": 9295
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.98992569955817e-05,
+      "loss": 0.8106,
+      "step": 9296
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.989151383738341e-05,
+      "loss": 0.747,
+      "step": 9297
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.988377043229067e-05,
+      "loss": 17.0561,
+      "step": 9298
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.9876026780496787e-05,
+      "loss": 0.7216,
+      "step": 9299
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.9868282882194994e-05,
+      "loss": 0.7671,
+      "step": 9300
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.986053873757862e-05,
+      "loss": 18.4789,
+      "step": 9301
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.985279434684096e-05,
+      "loss": 17.5596,
+      "step": 9302
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.984504971017528e-05,
+      "loss": 21.7199,
+      "step": 9303
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.983730482777492e-05,
+      "loss": 0.8547,
+      "step": 9304
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.9829559699833185e-05,
+      "loss": 0.7962,
+      "step": 9305
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.98218143265434e-05,
+      "loss": 18.9864,
+      "step": 9306
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.981406870809888e-05,
+      "loss": 0.7691,
+      "step": 9307
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.9806322844692965e-05,
+      "loss": 19.9518,
+      "step": 9308
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.9798576736518995e-05,
+      "loss": 18.7031,
+      "step": 9309
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.9790830383770326e-05,
+      "loss": 19.3587,
+      "step": 9310
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.978308378664027e-05,
+      "loss": 0.7543,
+      "step": 9311
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.977533694532221e-05,
+      "loss": 0.7678,
+      "step": 9312
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.976758986000953e-05,
+      "loss": 0.8005,
+      "step": 9313
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.975984253089556e-05,
+      "loss": 0.8135,
+      "step": 9314
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.975209495817371e-05,
+      "loss": 0.8356,
+      "step": 9315
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.974434714203733e-05,
+      "loss": 18.9151,
+      "step": 9316
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.973659908267981e-05,
+      "loss": 0.7355,
+      "step": 9317
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.9728850780294564e-05,
+      "loss": 0.8097,
+      "step": 9318
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.9721102235074986e-05,
+      "loss": 0.7982,
+      "step": 9319
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.9713353447214485e-05,
+      "loss": 0.7258,
+      "step": 9320
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.970560441690645e-05,
+      "loss": 18.214,
+      "step": 9321
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.9697855144344294e-05,
+      "loss": 18.5531,
+      "step": 9322
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.9690105629721495e-05,
+      "loss": 0.7724,
+      "step": 9323
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.968235587323143e-05,
+      "loss": 17.7826,
+      "step": 9324
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.967460587506755e-05,
+      "loss": 19.3211,
+      "step": 9325
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.9666855635423314e-05,
+      "loss": 0.8164,
+      "step": 9326
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.9659105154492135e-05,
+      "loss": 0.8619,
+      "step": 9327
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9651354432467474e-05,
+      "loss": 18.1887,
+      "step": 9328
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.964360346954283e-05,
+      "loss": 18.8193,
+      "step": 9329
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.963585226591164e-05,
+      "loss": 17.8751,
+      "step": 9330
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.962810082176736e-05,
+      "loss": 0.7523,
+      "step": 9331
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.96203491373035e-05,
+      "loss": 0.7749,
+      "step": 9332
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9612597212713525e-05,
+      "loss": 17.7379,
+      "step": 9333
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9604845048190935e-05,
+      "loss": 17.6215,
+      "step": 9334
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.959709264392921e-05,
+      "loss": 0.7502,
+      "step": 9335
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.958934000012185e-05,
+      "loss": 0.7582,
+      "step": 9336
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.95815871169624e-05,
+      "loss": 0.8431,
+      "step": 9337
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9573833994644334e-05,
+      "loss": 18.3244,
+      "step": 9338
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9566080633361186e-05,
+      "loss": 17.093,
+      "step": 9339
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.955832703330648e-05,
+      "loss": 16.6603,
+      "step": 9340
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9550573194673757e-05,
+      "loss": 18.7727,
+      "step": 9341
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9542819117656536e-05,
+      "loss": 19.0292,
+      "step": 9342
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.953506480244837e-05,
+      "loss": 18.56,
+      "step": 9343
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.952731024924283e-05,
+      "loss": 0.724,
+      "step": 9344
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.951955545823342e-05,
+      "loss": 0.723,
+      "step": 9345
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.951180042961375e-05,
+      "loss": 18.1002,
+      "step": 9346
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.950404516357737e-05,
+      "loss": 20.5833,
+      "step": 9347
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.949628966031785e-05,
+      "loss": 18.1199,
+      "step": 9348
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9488533920028764e-05,
+      "loss": 0.7553,
+      "step": 9349
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.948077794290371e-05,
+      "loss": 18.1346,
+      "step": 9350
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.947302172913628e-05,
+      "loss": 21.524,
+      "step": 9351
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.946526527892007e-05,
+      "loss": 19.8434,
+      "step": 9352
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.945750859244866e-05,
+      "loss": 0.7909,
+      "step": 9353
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.944975166991569e-05,
+      "loss": 0.7436,
+      "step": 9354
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9441994511514755e-05,
+      "loss": 17.4487,
+      "step": 9355
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9434237117439474e-05,
+      "loss": 17.3023,
+      "step": 9356
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.942647948788349e-05,
+      "loss": 18.0943,
+      "step": 9357
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.941872162304043e-05,
+      "loss": 18.2561,
+      "step": 9358
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9410963523103916e-05,
+      "loss": 0.7125,
+      "step": 9359
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9403205188267595e-05,
+      "loss": 18.4345,
+      "step": 9360
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.939544661872515e-05,
+      "loss": 0.7414,
+      "step": 9361
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.938768781467021e-05,
+      "loss": 18.4987,
+      "step": 9362
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.937992877629642e-05,
+      "loss": 0.7154,
+      "step": 9363
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.937216950379748e-05,
+      "loss": 0.7018,
+      "step": 9364
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.936440999736704e-05,
+      "loss": 19.3091,
+      "step": 9365
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.93566502571988e-05,
+      "loss": 0.7936,
+      "step": 9366
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9348890283486426e-05,
+      "loss": 0.7164,
+      "step": 9367
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.934113007642361e-05,
+      "loss": 19.865,
+      "step": 9368
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.933336963620406e-05,
+      "loss": 18.5384,
+      "step": 9369
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9325608963021474e-05,
+      "loss": 0.7465,
+      "step": 9370
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.931784805706956e-05,
+      "loss": 0.7485,
+      "step": 9371
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.931008691854204e-05,
+      "loss": 0.7281,
+      "step": 9372
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.930232554763261e-05,
+      "loss": 0.793,
+      "step": 9373
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.929456394453501e-05,
+      "loss": 0.7982,
+      "step": 9374
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9286802109442974e-05,
+      "loss": 0.7444,
+      "step": 9375
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.927904004255025e-05,
+      "loss": 20.4289,
+      "step": 9376
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9271277744050546e-05,
+      "loss": 0.8146,
+      "step": 9377
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9263515214137654e-05,
+      "loss": 15.3076,
+      "step": 9378
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.925575245300529e-05,
+      "loss": 20.4688,
+      "step": 9379
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9247989460847244e-05,
+      "loss": 18.7322,
+      "step": 9380
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.924022623785725e-05,
+      "loss": 0.7811,
+      "step": 9381
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.923246278422913e-05,
+      "loss": 0.7434,
+      "step": 9382
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.92246991001566e-05,
+      "loss": 19.2821,
+      "step": 9383
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.921693518583349e-05,
+      "loss": 18.8249,
+      "step": 9384
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.920917104145357e-05,
+      "loss": 0.7884,
+      "step": 9385
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9201406667210646e-05,
+      "loss": 17.0835,
+      "step": 9386
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.919364206329849e-05,
+      "loss": 0.7581,
+      "step": 9387
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.918587722991095e-05,
+      "loss": 18.9322,
+      "step": 9388
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9178112167241805e-05,
+      "loss": 0.8231,
+      "step": 9389
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.91703468754849e-05,
+      "loss": 0.707,
+      "step": 9390
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.916258135483404e-05,
+      "loss": 0.7776,
+      "step": 9391
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9154815605483036e-05,
+      "loss": 0.7285,
+      "step": 9392
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.914704962762576e-05,
+      "loss": 0.7745,
+      "step": 9393
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.913928342145604e-05,
+      "loss": 0.7942,
+      "step": 9394
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.913151698716771e-05,
+      "loss": 0.7196,
+      "step": 9395
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9123750324954644e-05,
+      "loss": 0.7452,
+      "step": 9396
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.911598343501067e-05,
+      "loss": 18.179,
+      "step": 9397
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.910821631752966e-05,
+      "loss": 0.7912,
+      "step": 9398
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9100448972705524e-05,
+      "loss": 17.5244,
+      "step": 9399
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9092681400732086e-05,
+      "loss": 0.7509,
+      "step": 9400
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.908491360180324e-05,
+      "loss": 19.2247,
+      "step": 9401
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.907714557611289e-05,
+      "loss": 0.76,
+      "step": 9402
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.906937732385489e-05,
+      "loss": 18.7089,
+      "step": 9403
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.906160884522318e-05,
+      "loss": 0.7869,
+      "step": 9404
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9053840140411645e-05,
+      "loss": 0.7462,
+      "step": 9405
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9046071209614184e-05,
+      "loss": 18.3724,
+      "step": 9406
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.9038302053024716e-05,
+      "loss": 17.3076,
+      "step": 9407
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.903053267083718e-05,
+      "loss": 19.1135,
+      "step": 9408
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.902276306324549e-05,
+      "loss": 17.054,
+      "step": 9409
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.901499323044356e-05,
+      "loss": 18.9436,
+      "step": 9410
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.900722317262534e-05,
+      "loss": 0.7527,
+      "step": 9411
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.899945288998476e-05,
+      "loss": 0.7687,
+      "step": 9412
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.899168238271581e-05,
+      "loss": 18.6403,
+      "step": 9413
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.898391165101241e-05,
+      "loss": 0.7527,
+      "step": 9414
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8976140695068517e-05,
+      "loss": 18.061,
+      "step": 9415
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8968369515078104e-05,
+      "loss": 0.7631,
+      "step": 9416
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.896059811123513e-05,
+      "loss": 0.7374,
+      "step": 9417
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.895282648373359e-05,
+      "loss": 0.7828,
+      "step": 9418
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.894505463276745e-05,
+      "loss": 17.3736,
+      "step": 9419
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8937282558530727e-05,
+      "loss": 0.7628,
+      "step": 9420
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.892951026121736e-05,
+      "loss": 18.4906,
+      "step": 9421
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.89217377410214e-05,
+      "loss": 18.7281,
+      "step": 9422
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.891396499813683e-05,
+      "loss": 18.6286,
+      "step": 9423
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.890619203275764e-05,
+      "loss": 19.0884,
+      "step": 9424
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.889841884507786e-05,
+      "loss": 18.7909,
+      "step": 9425
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8890645435291524e-05,
+      "loss": 17.4188,
+      "step": 9426
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.888287180359265e-05,
+      "loss": 16.5915,
+      "step": 9427
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.887509795017528e-05,
+      "loss": 0.8182,
+      "step": 9428
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.886732387523341e-05,
+      "loss": 18.8713,
+      "step": 9429
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.885954957896115e-05,
+      "loss": 17.1896,
+      "step": 9430
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.885177506155247e-05,
+      "loss": 0.7896,
+      "step": 9431
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8844000323201495e-05,
+      "loss": 17.3394,
+      "step": 9432
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.883622536410223e-05,
+      "loss": 0.7337,
+      "step": 9433
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.882845018444878e-05,
+      "loss": 0.7608,
+      "step": 9434
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.882067478443518e-05,
+      "loss": 18.0513,
+      "step": 9435
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.881289916425554e-05,
+      "loss": 16.2845,
+      "step": 9436
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8805123324103916e-05,
+      "loss": 19.868,
+      "step": 9437
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.879734726417439e-05,
+      "loss": 0.795,
+      "step": 9438
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.878957098466109e-05,
+      "loss": 17.5676,
+      "step": 9439
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.878179448575809e-05,
+      "loss": 17.9689,
+      "step": 9440
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.877401776765947e-05,
+      "loss": 0.7653,
+      "step": 9441
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.876624083055939e-05,
+      "loss": 0.7192,
+      "step": 9442
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.875846367465192e-05,
+      "loss": 17.5359,
+      "step": 9443
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8750686300131205e-05,
+      "loss": 0.7527,
+      "step": 9444
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.874290870719137e-05,
+      "loss": 0.7331,
+      "step": 9445
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.873513089602653e-05,
+      "loss": 0.7663,
+      "step": 9446
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.872735286683083e-05,
+      "loss": 0.7554,
+      "step": 9447
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8719574619798404e-05,
+      "loss": 22.6443,
+      "step": 9448
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8711796155123396e-05,
+      "loss": 0.7394,
+      "step": 9449
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8704017472999975e-05,
+      "loss": 0.7771,
+      "step": 9450
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8696238573622306e-05,
+      "loss": 0.7476,
+      "step": 9451
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.868845945718452e-05,
+      "loss": 0.7622,
+      "step": 9452
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8680680123880805e-05,
+      "loss": 0.7535,
+      "step": 9453
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.867290057390534e-05,
+      "loss": 20.2638,
+      "step": 9454
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.866512080745228e-05,
+      "loss": 0.7754,
+      "step": 9455
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8657340824715846e-05,
+      "loss": 18.8922,
+      "step": 9456
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.864956062589019e-05,
+      "loss": 18.5342,
+      "step": 9457
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.864178021116954e-05,
+      "loss": 18.1556,
+      "step": 9458
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.863399958074807e-05,
+      "loss": 0.7531,
+      "step": 9459
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8626218734820015e-05,
+      "loss": 0.7889,
+      "step": 9460
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.861843767357956e-05,
+      "loss": 16.2799,
+      "step": 9461
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.861065639722093e-05,
+      "loss": 18.6296,
+      "step": 9462
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.860287490593835e-05,
+      "loss": 0.71,
+      "step": 9463
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8595093199926045e-05,
+      "loss": 19.8727,
+      "step": 9464
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.858731127937828e-05,
+      "loss": 0.7334,
+      "step": 9465
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8579529144489235e-05,
+      "loss": 0.7603,
+      "step": 9466
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.857174679545319e-05,
+      "loss": 19.2898,
+      "step": 9467
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8563964232464396e-05,
+      "loss": 0.7671,
+      "step": 9468
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.855618145571708e-05,
+      "loss": 0.7565,
+      "step": 9469
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8548398465405544e-05,
+      "loss": 17.8728,
+      "step": 9470
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.854061526172402e-05,
+      "loss": 0.7732,
+      "step": 9471
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.85328318448668e-05,
+      "loss": 0.7421,
+      "step": 9472
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.852504821502812e-05,
+      "loss": 18.7148,
+      "step": 9473
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.851726437240232e-05,
+      "loss": 0.7082,
+      "step": 9474
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.850948031718365e-05,
+      "loss": 0.7564,
+      "step": 9475
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.85016960495664e-05,
+      "loss": 17.6774,
+      "step": 9476
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8493911569744876e-05,
+      "loss": 0.7168,
+      "step": 9477
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.848612687791338e-05,
+      "loss": 18.5702,
+      "step": 9478
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.847834197426624e-05,
+      "loss": 16.7396,
+      "step": 9479
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8470556858997716e-05,
+      "loss": 0.7564,
+      "step": 9480
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8462771532302175e-05,
+      "loss": 0.8113,
+      "step": 9481
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.845498599437392e-05,
+      "loss": 18.2195,
+      "step": 9482
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.844720024540728e-05,
+      "loss": 0.7578,
+      "step": 9483
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.843941428559662e-05,
+      "loss": 0.7307,
+      "step": 9484
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.843162811513623e-05,
+      "loss": 16.4453,
+      "step": 9485
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.842384173422048e-05,
+      "loss": 0.7161,
+      "step": 9486
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.84160551430437e-05,
+      "loss": 0.7501,
+      "step": 9487
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8408268341800285e-05,
+      "loss": 17.1004,
+      "step": 9488
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8400481330684575e-05,
+      "loss": 17.6194,
+      "step": 9489
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.839269410989092e-05,
+      "loss": 16.6737,
+      "step": 9490
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.838490667961371e-05,
+      "loss": 0.7762,
+      "step": 9491
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.837711904004731e-05,
+      "loss": 0.7903,
+      "step": 9492
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8369331191386114e-05,
+      "loss": 0.7738,
+      "step": 9493
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8361543133824495e-05,
+      "loss": 18.5385,
+      "step": 9494
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.835375486755683e-05,
+      "loss": 19.4912,
+      "step": 9495
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.834596639277756e-05,
+      "loss": 17.088,
+      "step": 9496
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8338177709681066e-05,
+      "loss": 0.7369,
+      "step": 9497
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.833038881846175e-05,
+      "loss": 0.7826,
+      "step": 9498
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.832259971931402e-05,
+      "loss": 0.7281,
+      "step": 9499
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.831481041243231e-05,
+      "loss": 19.1917,
+      "step": 9500
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.830702089801101e-05,
+      "loss": 0.7984,
+      "step": 9501
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8299231176244596e-05,
+      "loss": 19.1349,
+      "step": 9502
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.829144124732748e-05,
+      "loss": 17.1306,
+      "step": 9503
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8283651111454084e-05,
+      "loss": 18.4698,
+      "step": 9504
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.827586076881886e-05,
+      "loss": 0.7584,
+      "step": 9505
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8268070219616266e-05,
+      "loss": 0.7433,
+      "step": 9506
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.826027946404077e-05,
+      "loss": 18.5825,
+      "step": 9507
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.825248850228678e-05,
+      "loss": 0.7973,
+      "step": 9508
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8244697334548815e-05,
+      "loss": 0.7312,
+      "step": 9509
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8236905961021314e-05,
+      "loss": 0.765,
+      "step": 9510
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.822911438189876e-05,
+      "loss": 20.0183,
+      "step": 9511
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.822132259737565e-05,
+      "loss": 0.7404,
+      "step": 9512
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.821353060764644e-05,
+      "loss": 0.731,
+      "step": 9513
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.820573841290562e-05,
+      "loss": 20.8781,
+      "step": 9514
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8197946013347694e-05,
+      "loss": 0.7844,
+      "step": 9515
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.819015340916717e-05,
+      "loss": 0.7274,
+      "step": 9516
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.818236060055856e-05,
+      "loss": 17.0462,
+      "step": 9517
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8174567587716345e-05,
+      "loss": 21.1949,
+      "step": 9518
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.816677437083505e-05,
+      "loss": 0.7604,
+      "step": 9519
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.815898095010922e-05,
+      "loss": 19.0286,
+      "step": 9520
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.815118732573336e-05,
+      "loss": 0.7644,
+      "step": 9521
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.814339349790199e-05,
+      "loss": 18.8242,
+      "step": 9522
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.813559946680967e-05,
+      "loss": 16.8887,
+      "step": 9523
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.812780523265093e-05,
+      "loss": 0.7988,
+      "step": 9524
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.8120010795620304e-05,
+      "loss": 17.6674,
+      "step": 9525
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.811221615591237e-05,
+      "loss": 0.7489,
+      "step": 9526
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.810442131372167e-05,
+      "loss": 0.8031,
+      "step": 9527
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.809662626924275e-05,
+      "loss": 19.601,
+      "step": 9528
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.808883102267019e-05,
+      "loss": 0.7551,
+      "step": 9529
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.808103557419857e-05,
+      "loss": 0.7321,
+      "step": 9530
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.807323992402245e-05,
+      "loss": 17.5729,
+      "step": 9531
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.806544407233642e-05,
+      "loss": 16.5029,
+      "step": 9532
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.805764801933507e-05,
+      "loss": 0.7636,
+      "step": 9533
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.804985176521298e-05,
+      "loss": 19.9473,
+      "step": 9534
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.804205531016476e-05,
+      "loss": 0.7823,
+      "step": 9535
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.8034258654384986e-05,
+      "loss": 19.9578,
+      "step": 9536
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.80264617980683e-05,
+      "loss": 18.976,
+      "step": 9537
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.8018664741409276e-05,
+      "loss": 19.504,
+      "step": 9538
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.801086748460255e-05,
+      "loss": 19.5029,
+      "step": 9539
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.800307002784276e-05,
+      "loss": 20.0772,
+      "step": 9540
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.799527237132451e-05,
+      "loss": 18.2383,
+      "step": 9541
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7987474515242426e-05,
+      "loss": 0.6786,
+      "step": 9542
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.797967645979115e-05,
+      "loss": 17.3896,
+      "step": 9543
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.797187820516535e-05,
+      "loss": 0.77,
+      "step": 9544
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.796407975155964e-05,
+      "loss": 18.4908,
+      "step": 9545
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.795628109916869e-05,
+      "loss": 19.5304,
+      "step": 9546
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.794848224818712e-05,
+      "loss": 0.741,
+      "step": 9547
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.794068319880964e-05,
+      "loss": 0.7529,
+      "step": 9548
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.79328839512309e-05,
+      "loss": 0.7649,
+      "step": 9549
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.792508450564554e-05,
+      "loss": 20.7687,
+      "step": 9550
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7917284862248286e-05,
+      "loss": 18.2129,
+      "step": 9551
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.790948502123378e-05,
+      "loss": 17.9459,
+      "step": 9552
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7901684982796714e-05,
+      "loss": 0.7712,
+      "step": 9553
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.789388474713181e-05,
+      "loss": 16.275,
+      "step": 9554
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.788608431443373e-05,
+      "loss": 0.7392,
+      "step": 9555
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7878283684897185e-05,
+      "loss": 20.702,
+      "step": 9556
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.787048285871687e-05,
+      "loss": 20.7549,
+      "step": 9557
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.786268183608752e-05,
+      "loss": 19.3354,
+      "step": 9558
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.785488061720383e-05,
+      "loss": 0.7743,
+      "step": 9559
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7847079202260524e-05,
+      "loss": 0.7736,
+      "step": 9560
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7839277591452333e-05,
+      "loss": 18.6131,
+      "step": 9561
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7831475784973976e-05,
+      "loss": 19.4846,
+      "step": 9562
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.782367378302021e-05,
+      "loss": 16.688,
+      "step": 9563
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7815871585785744e-05,
+      "loss": 19.8363,
+      "step": 9564
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7808069193465356e-05,
+      "loss": 0.7683,
+      "step": 9565
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.780026660625376e-05,
+      "loss": 0.7397,
+      "step": 9566
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.779246382434573e-05,
+      "loss": 16.4079,
+      "step": 9567
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.778466084793603e-05,
+      "loss": 18.4867,
+      "step": 9568
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.777685767721943e-05,
+      "loss": 0.7755,
+      "step": 9569
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.776905431239067e-05,
+      "loss": 20.0284,
+      "step": 9570
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7761250753644545e-05,
+      "loss": 18.7245,
+      "step": 9571
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.775344700117583e-05,
+      "loss": 18.3077,
+      "step": 9572
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.774564305517931e-05,
+      "loss": 17.1384,
+      "step": 9573
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7737838915849765e-05,
+      "loss": 17.7636,
+      "step": 9574
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.773003458338201e-05,
+      "loss": 17.0038,
+      "step": 9575
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.77222300579708e-05,
+      "loss": 0.7587,
+      "step": 9576
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.771442533981098e-05,
+      "loss": 0.7716,
+      "step": 9577
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.770662042909734e-05,
+      "loss": 0.7897,
+      "step": 9578
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.76988153260247e-05,
+      "loss": 0.7644,
+      "step": 9579
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.769101003078786e-05,
+      "loss": 0.7599,
+      "step": 9580
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.768320454358166e-05,
+      "loss": 16.7007,
+      "step": 9581
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.767539886460091e-05,
+      "loss": 19.0275,
+      "step": 9582
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.766759299404048e-05,
+      "loss": 0.8071,
+      "step": 9583
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.765978693209515e-05,
+      "loss": 0.7704,
+      "step": 9584
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.765198067895979e-05,
+      "loss": 18.5079,
+      "step": 9585
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.764417423482926e-05,
+      "loss": 18.8651,
+      "step": 9586
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.76363675998984e-05,
+      "loss": 17.2349,
+      "step": 9587
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.762856077436205e-05,
+      "loss": 0.6857,
+      "step": 9588
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7620753758415094e-05,
+      "loss": 19.2645,
+      "step": 9589
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7612946552252354e-05,
+      "loss": 0.7794,
+      "step": 9590
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7605139156068756e-05,
+      "loss": 0.7656,
+      "step": 9591
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.759733157005915e-05,
+      "loss": 0.7418,
+      "step": 9592
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7589523794418396e-05,
+      "loss": 19.9892,
+      "step": 9593
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7581715829341396e-05,
+      "loss": 0.7635,
+      "step": 9594
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.757390767502304e-05,
+      "loss": 0.7711,
+      "step": 9595
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.756609933165823e-05,
+      "loss": 17.4841,
+      "step": 9596
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.755829079944185e-05,
+      "loss": 0.7508,
+      "step": 9597
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.75504820785688e-05,
+      "loss": 16.4628,
+      "step": 9598
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7542673169234005e-05,
+      "loss": 0.7729,
+      "step": 9599
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.753486407163234e-05,
+      "loss": 0.7812,
+      "step": 9600
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7527054785958776e-05,
+      "loss": 17.5376,
+      "step": 9601
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.751924531240819e-05,
+      "loss": 0.8017,
+      "step": 9602
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.751143565117554e-05,
+      "loss": 19.4353,
+      "step": 9603
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7503625802455715e-05,
+      "loss": 17.2502,
+      "step": 9604
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.749581576644369e-05,
+      "loss": 18.9028,
+      "step": 9605
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.74880055433344e-05,
+      "loss": 0.7856,
+      "step": 9606
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.748019513332278e-05,
+      "loss": 0.7912,
+      "step": 9607
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.747238453660376e-05,
+      "loss": 0.7452,
+      "step": 9608
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.746457375337233e-05,
+      "loss": 0.7935,
+      "step": 9609
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.745676278382345e-05,
+      "loss": 0.7704,
+      "step": 9610
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.744895162815205e-05,
+      "loss": 20.1498,
+      "step": 9611
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.744114028655312e-05,
+      "loss": 0.7618,
+      "step": 9612
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.743332875922164e-05,
+      "loss": 19.6551,
+      "step": 9613
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7425517046352564e-05,
+      "loss": 17.4305,
+      "step": 9614
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.741770514814089e-05,
+      "loss": 0.8055,
+      "step": 9615
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7409893064781596e-05,
+      "loss": 0.7475,
+      "step": 9616
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.740208079646969e-05,
+      "loss": 0.783,
+      "step": 9617
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7394268343400146e-05,
+      "loss": 0.7576,
+      "step": 9618
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.738645570576799e-05,
+      "loss": 20.0051,
+      "step": 9619
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.737864288376822e-05,
+      "loss": 0.6996,
+      "step": 9620
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.737082987759581e-05,
+      "loss": 19.3612,
+      "step": 9621
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.736301668744581e-05,
+      "loss": 19.9816,
+      "step": 9622
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.735520331351324e-05,
+      "loss": 0.7524,
+      "step": 9623
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.734738975599312e-05,
+      "loss": 18.5287,
+      "step": 9624
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7339576015080476e-05,
+      "loss": 0.7535,
+      "step": 9625
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.733176209097033e-05,
+      "loss": 0.7685,
+      "step": 9626
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.732394798385774e-05,
+      "loss": 19.1878,
+      "step": 9627
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.731613369393772e-05,
+      "loss": 0.7537,
+      "step": 9628
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7308319221405346e-05,
+      "loss": 17.1859,
+      "step": 9629
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.730050456645566e-05,
+      "loss": 0.7303,
+      "step": 9630
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.729268972928369e-05,
+      "loss": 19.5333,
+      "step": 9631
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7284874710084525e-05,
+      "loss": 21.2325,
+      "step": 9632
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7277059509053244e-05,
+      "loss": 0.776,
+      "step": 9633
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.726924412638488e-05,
+      "loss": 0.7406,
+      "step": 9634
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.726142856227452e-05,
+      "loss": 17.6224,
+      "step": 9635
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.725361281691724e-05,
+      "loss": 17.4674,
+      "step": 9636
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.724579689050815e-05,
+      "loss": 17.7839,
+      "step": 9637
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7237980783242295e-05,
+      "loss": 0.7526,
+      "step": 9638
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.723016449531481e-05,
+      "loss": 0.8143,
+      "step": 9639
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.722234802692075e-05,
+      "loss": 18.8318,
+      "step": 9640
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.721453137825523e-05,
+      "loss": 19.1402,
+      "step": 9641
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.720671454951337e-05,
+      "loss": 18.6352,
+      "step": 9642
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7198897540890274e-05,
+      "loss": 18.2459,
+      "step": 9643
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.719108035258105e-05,
+      "loss": 0.7745,
+      "step": 9644
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.718326298478081e-05,
+      "loss": 18.1998,
+      "step": 9645
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.717544543768468e-05,
+      "loss": 0.746,
+      "step": 9646
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.716762771148779e-05,
+      "loss": 0.7793,
+      "step": 9647
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.715980980638531e-05,
+      "loss": 0.7702,
+      "step": 9648
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.715199172257232e-05,
+      "loss": 17.274,
+      "step": 9649
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.714417346024398e-05,
+      "loss": 18.351,
+      "step": 9650
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7136355019595444e-05,
+      "loss": 0.7281,
+      "step": 9651
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.712853640082185e-05,
+      "loss": 17.7119,
+      "step": 9652
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.712071760411837e-05,
+      "loss": 0.8179,
+      "step": 9653
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.711289862968015e-05,
+      "loss": 17.6389,
+      "step": 9654
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.710507947770236e-05,
+      "loss": 17.1226,
+      "step": 9655
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.709726014838015e-05,
+      "loss": 0.7536,
+      "step": 9656
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.708944064190872e-05,
+      "loss": 0.755,
+      "step": 9657
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.708162095848324e-05,
+      "loss": 16.7936,
+      "step": 9658
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.707380109829886e-05,
+      "loss": 0.7597,
+      "step": 9659
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.70659810615508e-05,
+      "loss": 18.3482,
+      "step": 9660
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.705816084843423e-05,
+      "loss": 19.1775,
+      "step": 9661
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.705034045914438e-05,
+      "loss": 21.1251,
+      "step": 9662
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.7042519893876404e-05,
+      "loss": 17.4325,
+      "step": 9663
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.703469915282552e-05,
+      "loss": 0.8076,
+      "step": 9664
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.702687823618695e-05,
+      "loss": 0.8185,
+      "step": 9665
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.701905714415589e-05,
+      "loss": 17.9301,
+      "step": 9666
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.701123587692756e-05,
+      "loss": 0.7816,
+      "step": 9667
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.700341443469719e-05,
+      "loss": 18.7004,
+      "step": 9668
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.6995592817659984e-05,
+      "loss": 0.6675,
+      "step": 9669
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.698777102601118e-05,
+      "loss": 18.4758,
+      "step": 9670
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.697994905994603e-05,
+      "loss": 17.5422,
+      "step": 9671
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.697212691965976e-05,
+      "loss": 18.8051,
+      "step": 9672
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.69643046053476e-05,
+      "loss": 19.531,
+      "step": 9673
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.69564821172048e-05,
+      "loss": 19.8018,
+      "step": 9674
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.694865945542663e-05,
+      "loss": 21.328,
+      "step": 9675
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.6940836620208346e-05,
+      "loss": 17.8978,
+      "step": 9676
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.6933013611745166e-05,
+      "loss": 0.7449,
+      "step": 9677
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.69251904302324e-05,
+      "loss": 0.7489,
+      "step": 9678
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.69173670758653e-05,
+      "loss": 19.6673,
+      "step": 9679
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.690954354883913e-05,
+      "loss": 17.9022,
+      "step": 9680
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.6901719849349186e-05,
+      "loss": 0.7766,
+      "step": 9681
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.689389597759074e-05,
+      "loss": 18.8453,
+      "step": 9682
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.688607193375906e-05,
+      "loss": 18.0606,
+      "step": 9683
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.687824771804946e-05,
+      "loss": 17.2836,
+      "step": 9684
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.687042333065724e-05,
+      "loss": 0.7552,
+      "step": 9685
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.6862598771777676e-05,
+      "loss": 16.8684,
+      "step": 9686
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.6854774041606074e-05,
+      "loss": 17.098,
+      "step": 9687
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.684694914033775e-05,
+      "loss": 19.5129,
+      "step": 9688
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.683912406816801e-05,
+      "loss": 15.5218,
+      "step": 9689
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.683129882529218e-05,
+      "loss": 16.7703,
+      "step": 9690
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.682347341190556e-05,
+      "loss": 17.2375,
+      "step": 9691
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.681564782820348e-05,
+      "loss": 18.851,
+      "step": 9692
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.680782207438129e-05,
+      "loss": 18.835,
+      "step": 9693
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.6799996150634294e-05,
+      "loss": 18.3328,
+      "step": 9694
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.679217005715786e-05,
+      "loss": 19.014,
+      "step": 9695
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.67843437941473e-05,
+      "loss": 0.7674,
+      "step": 9696
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.677651736179797e-05,
+      "loss": 19.4252,
+      "step": 9697
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.6768690760305207e-05,
+      "loss": 0.7953,
+      "step": 9698
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.67608639898644e-05,
+      "loss": 18.9679,
+      "step": 9699
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.675303705067088e-05,
+      "loss": 17.9695,
+      "step": 9700
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.6745209942919995e-05,
+      "loss": 18.5126,
+      "step": 9701
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.673738266680714e-05,
+      "loss": 19.249,
+      "step": 9702
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.6729555222527675e-05,
+      "loss": 17.981,
+      "step": 9703
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.672172761027699e-05,
+      "loss": 0.7738,
+      "step": 9704
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.671389983025043e-05,
+      "loss": 0.7712,
+      "step": 9705
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.670607188264341e-05,
+      "loss": 17.8034,
+      "step": 9706
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.669824376765128e-05,
+      "loss": 17.5047,
+      "step": 9707
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.669041548546947e-05,
+      "loss": 18.5963,
+      "step": 9708
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.668258703629338e-05,
+      "loss": 20.2338,
+      "step": 9709
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.6674758420318375e-05,
+      "loss": 20.108,
+      "step": 9710
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.666692963773987e-05,
+      "loss": 0.7474,
+      "step": 9711
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.665910068875329e-05,
+      "loss": 0.787,
+      "step": 9712
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.665127157355403e-05,
+      "loss": 18.3365,
+      "step": 9713
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.664344229233752e-05,
+      "loss": 0.7978,
+      "step": 9714
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.663561284529917e-05,
+      "loss": 18.9655,
+      "step": 9715
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.66277832326344e-05,
+      "loss": 18.9909,
+      "step": 9716
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.6619953454538664e-05,
+      "loss": 0.7545,
+      "step": 9717
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.661212351120737e-05,
+      "loss": 18.1875,
+      "step": 9718
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.6604293402835965e-05,
+      "loss": 16.1229,
+      "step": 9719
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.659646312961988e-05,
+      "loss": 0.7403,
+      "step": 9720
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.658863269175457e-05,
+      "loss": 0.7553,
+      "step": 9721
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.658080208943549e-05,
+      "loss": 0.7624,
+      "step": 9722
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.657297132285809e-05,
+      "loss": 0.7949,
+      "step": 9723
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.656514039221783e-05,
+      "loss": 17.8642,
+      "step": 9724
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.6557309297710146e-05,
+      "loss": 0.7945,
+      "step": 9725
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.654947803953054e-05,
+      "loss": 18.4728,
+      "step": 9726
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.6541646617874465e-05,
+      "loss": 16.7301,
+      "step": 9727
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.6533815032937396e-05,
+      "loss": 18.8113,
+      "step": 9728
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.6525983284914806e-05,
+      "loss": 19.9617,
+      "step": 9729
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.651815137400218e-05,
+      "loss": 19.3832,
+      "step": 9730
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.651031930039503e-05,
+      "loss": 17.2381,
+      "step": 9731
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.650248706428881e-05,
+      "loss": 18.0008,
+      "step": 9732
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.649465466587902e-05,
+      "loss": 19.1997,
+      "step": 9733
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.6486822105361184e-05,
+      "loss": 18.8178,
+      "step": 9734
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.647898938293077e-05,
+      "loss": 0.7182,
+      "step": 9735
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.647115649878331e-05,
+      "loss": 17.8417,
+      "step": 9736
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.64633234531143e-05,
+      "loss": 0.7656,
+      "step": 9737
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.645549024611928e-05,
+      "loss": 0.7941,
+      "step": 9738
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.644765687799373e-05,
+      "loss": 19.1491,
+      "step": 9739
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.643982334893319e-05,
+      "loss": 16.2733,
+      "step": 9740
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.64319896591332e-05,
+      "loss": 0.7507,
+      "step": 9741
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.642415580878928e-05,
+      "loss": 0.8009,
+      "step": 9742
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.641632179809696e-05,
+      "loss": 18.3091,
+      "step": 9743
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.640848762725178e-05,
+      "loss": 0.7617,
+      "step": 9744
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.640065329644929e-05,
+      "loss": 18.1067,
+      "step": 9745
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.639281880588504e-05,
+      "loss": 0.7209,
+      "step": 9746
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.638498415575455e-05,
+      "loss": 19.1537,
+      "step": 9747
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.637714934625342e-05,
+      "loss": 18.0682,
+      "step": 9748
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.6369314377577175e-05,
+      "loss": 17.532,
+      "step": 9749
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.636147924992139e-05,
+      "loss": 19.5281,
+      "step": 9750
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.635364396348163e-05,
+      "loss": 0.7571,
+      "step": 9751
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.634580851845348e-05,
+      "loss": 0.797,
+      "step": 9752
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.633797291503248e-05,
+      "loss": 18.6884,
+      "step": 9753
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.6330137153414234e-05,
+      "loss": 18.7994,
+      "step": 9754
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.6322301233794325e-05,
+      "loss": 0.7842,
+      "step": 9755
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.6314465156368335e-05,
+      "loss": 0.7326,
+      "step": 9756
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.630662892133184e-05,
+      "loss": 0.7695,
+      "step": 9757
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.629879252888046e-05,
+      "loss": 18.8739,
+      "step": 9758
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.629095597920978e-05,
+      "loss": 18.2692,
+      "step": 9759
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.62831192725154e-05,
+      "loss": 18.953,
+      "step": 9760
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.6275282408992925e-05,
+      "loss": 0.7284,
+      "step": 9761
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.626744538883797e-05,
+      "loss": 0.7891,
+      "step": 9762
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.625960821224614e-05,
+      "loss": 0.8512,
+      "step": 9763
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.6251770879413066e-05,
+      "loss": 0.7725,
+      "step": 9764
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.6243933390534375e-05,
+      "loss": 0.7986,
+      "step": 9765
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.6236095745805674e-05,
+      "loss": 20.4182,
+      "step": 9766
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.6228257945422594e-05,
+      "loss": 17.2322,
+      "step": 9767
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.6220419989580766e-05,
+      "loss": 18.5261,
+      "step": 9768
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.6212581878475846e-05,
+      "loss": 0.7336,
+      "step": 9769
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.620474361230347e-05,
+      "loss": 19.7207,
+      "step": 9770
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.6196905191259265e-05,
+      "loss": 0.7774,
+      "step": 9771
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.61890666155389e-05,
+      "loss": 0.773,
+      "step": 9772
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.6181227885338004e-05,
+      "loss": 17.6546,
+      "step": 9773
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.617338900085226e-05,
+      "loss": 17.8194,
+      "step": 9774
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.616554996227732e-05,
+      "loss": 18.9879,
+      "step": 9775
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.6157710769808845e-05,
+      "loss": 0.7221,
+      "step": 9776
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.614987142364249e-05,
+      "loss": 18.4507,
+      "step": 9777
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.614203192397395e-05,
+      "loss": 0.7748,
+      "step": 9778
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.613419227099889e-05,
+      "loss": 0.6885,
+      "step": 9779
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.6126352464913e-05,
+      "loss": 0.7309,
+      "step": 9780
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.6118512505911937e-05,
+      "loss": 0.7346,
+      "step": 9781
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.611067239419141e-05,
+      "loss": 19.7375,
+      "step": 9782
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.61028321299471e-05,
+      "loss": 0.7727,
+      "step": 9783
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.6094991713374726e-05,
+      "loss": 0.7338,
+      "step": 9784
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.608715114466995e-05,
+      "loss": 19.1594,
+      "step": 9785
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.60793104240285e-05,
+      "loss": 0.7869,
+      "step": 9786
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.607146955164606e-05,
+      "loss": 18.8943,
+      "step": 9787
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.606362852771837e-05,
+      "loss": 19.4666,
+      "step": 9788
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.605578735244112e-05,
+      "loss": 0.784,
+      "step": 9789
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.604794602601003e-05,
+      "loss": 18.6296,
+      "step": 9790
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.604010454862083e-05,
+      "loss": 21.2812,
+      "step": 9791
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.603226292046924e-05,
+      "loss": 19.2841,
+      "step": 9792
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.602442114175101e-05,
+      "loss": 0.771,
+      "step": 9793
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.6016579212661836e-05,
+      "loss": 0.7129,
+      "step": 9794
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.600873713339746e-05,
+      "loss": 18.2102,
+      "step": 9795
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.6000894904153656e-05,
+      "loss": 0.763,
+      "step": 9796
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.599305252512612e-05,
+      "loss": 17.9953,
+      "step": 9797
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5985209996510646e-05,
+      "loss": 0.8008,
+      "step": 9798
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.597736731850295e-05,
+      "loss": 17.3303,
+      "step": 9799
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.596952449129882e-05,
+      "loss": 0.7287,
+      "step": 9800
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.596168151509397e-05,
+      "loss": 0.7788,
+      "step": 9801
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5953838390084215e-05,
+      "loss": 18.4885,
+      "step": 9802
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.594599511646528e-05,
+      "loss": 0.7596,
+      "step": 9803
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.593815169443294e-05,
+      "loss": 18.6185,
+      "step": 9804
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.593030812418298e-05,
+      "loss": 0.7467,
+      "step": 9805
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.592246440591118e-05,
+      "loss": 0.7695,
+      "step": 9806
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.591462053981332e-05,
+      "loss": 16.8893,
+      "step": 9807
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.590677652608519e-05,
+      "loss": 16.8407,
+      "step": 9808
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.589893236492255e-05,
+      "loss": 19.2494,
+      "step": 9809
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5891088056521226e-05,
+      "loss": 21.5268,
+      "step": 9810
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5883243601076987e-05,
+      "loss": 18.6177,
+      "step": 9811
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5875398998785654e-05,
+      "loss": 18.067,
+      "step": 9812
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5867554249843014e-05,
+      "loss": 19.8009,
+      "step": 9813
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.585970935444489e-05,
+      "loss": 0.7792,
+      "step": 9814
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.585186431278707e-05,
+      "loss": 17.6929,
+      "step": 9815
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5844019125065385e-05,
+      "loss": 16.7751,
+      "step": 9816
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5836173791475655e-05,
+      "loss": 0.7468,
+      "step": 9817
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.582832831221369e-05,
+      "loss": 0.8117,
+      "step": 9818
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.582048268747531e-05,
+      "loss": 19.5983,
+      "step": 9819
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5812636917456364e-05,
+      "loss": 17.8027,
+      "step": 9820
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5804791002352665e-05,
+      "loss": 0.8196,
+      "step": 9821
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5796944942360076e-05,
+      "loss": 17.6771,
+      "step": 9822
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.578909873767439e-05,
+      "loss": 18.0501,
+      "step": 9823
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.57812523884915e-05,
+      "loss": 19.9715,
+      "step": 9824
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5773405895007224e-05,
+      "loss": 21.1729,
+      "step": 9825
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.576555925741742e-05,
+      "loss": 20.1159,
+      "step": 9826
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5757712475917946e-05,
+      "loss": 0.7438,
+      "step": 9827
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.574986555070464e-05,
+      "loss": 18.3699,
+      "step": 9828
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.574201848197337e-05,
+      "loss": 16.8573,
+      "step": 9829
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.573417126992003e-05,
+      "loss": 18.6088,
+      "step": 9830
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.572632391474046e-05,
+      "loss": 17.6244,
+      "step": 9831
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.571847641663053e-05,
+      "loss": 18.1726,
+      "step": 9832
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5710628775786124e-05,
+      "loss": 18.8505,
+      "step": 9833
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5702780992403124e-05,
+      "loss": 18.6682,
+      "step": 9834
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.56949330666774e-05,
+      "loss": 18.4034,
+      "step": 9835
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.568708499880486e-05,
+      "loss": 0.7177,
+      "step": 9836
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5679236788981374e-05,
+      "loss": 18.4534,
+      "step": 9837
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.567138843740284e-05,
+      "loss": 0.8132,
+      "step": 9838
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5663539944265153e-05,
+      "loss": 0.7243,
+      "step": 9839
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.565569130976422e-05,
+      "loss": 18.8572,
+      "step": 9840
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.564784253409594e-05,
+      "loss": 19.9523,
+      "step": 9841
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5639993617456224e-05,
+      "loss": 17.4529,
+      "step": 9842
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5632144560040965e-05,
+      "loss": 18.3925,
+      "step": 9843
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5624295362046095e-05,
+      "loss": 0.7611,
+      "step": 9844
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5616446023667555e-05,
+      "loss": 0.7465,
+      "step": 9845
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.560859654510121e-05,
+      "loss": 19.7007,
+      "step": 9846
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5600746926543026e-05,
+      "loss": 0.773,
+      "step": 9847
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.559289716818893e-05,
+      "loss": 20.7843,
+      "step": 9848
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.558504727023482e-05,
+      "loss": 0.724,
+      "step": 9849
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.557719723287668e-05,
+      "loss": 0.793,
+      "step": 9850
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.55693470563104e-05,
+      "loss": 21.5303,
+      "step": 9851
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.556149674073197e-05,
+      "loss": 0.6801,
+      "step": 9852
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.555364628633729e-05,
+      "loss": 0.774,
+      "step": 9853
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5545795693322345e-05,
+      "loss": 17.0718,
+      "step": 9854
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.553794496188307e-05,
+      "loss": 0.7703,
+      "step": 9855
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5530094092215416e-05,
+      "loss": 19.5232,
+      "step": 9856
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.552224308451535e-05,
+      "loss": 18.126,
+      "step": 9857
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.551439193897885e-05,
+      "loss": 18.8877,
+      "step": 9858
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.550654065580187e-05,
+      "loss": 18.081,
+      "step": 9859
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.549868923518036e-05,
+      "loss": 20.592,
+      "step": 9860
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5490837677310315e-05,
+      "loss": 17.4988,
+      "step": 9861
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.548298598238772e-05,
+      "loss": 0.8108,
+      "step": 9862
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.547513415060852e-05,
+      "loss": 21.1412,
+      "step": 9863
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.546728218216875e-05,
+      "loss": 0.7497,
+      "step": 9864
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5459430077264355e-05,
+      "loss": 19.0965,
+      "step": 9865
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.545157783609134e-05,
+      "loss": 18.7831,
+      "step": 9866
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.544372545884568e-05,
+      "loss": 0.8121,
+      "step": 9867
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5435872945723425e-05,
+      "loss": 18.6481,
+      "step": 9868
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.542802029692052e-05,
+      "loss": 19.7333,
+      "step": 9869
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5420167512632984e-05,
+      "loss": 18.0913,
+      "step": 9870
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.541231459305682e-05,
+      "loss": 0.7123,
+      "step": 9871
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5404461538388074e-05,
+      "loss": 16.6968,
+      "step": 9872
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.539660834882273e-05,
+      "loss": 16.2802,
+      "step": 9873
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5388755024556804e-05,
+      "loss": 18.0884,
+      "step": 9874
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5380901565786316e-05,
+      "loss": 0.7528,
+      "step": 9875
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.53730479727073e-05,
+      "loss": 0.744,
+      "step": 9876
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5365194245515786e-05,
+      "loss": 17.0612,
+      "step": 9877
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5357340384407786e-05,
+      "loss": 0.7501,
+      "step": 9878
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5349486389579364e-05,
+      "loss": 0.7324,
+      "step": 9879
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5341632261226526e-05,
+      "loss": 0.8236,
+      "step": 9880
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.533377799954532e-05,
+      "loss": 0.7467,
+      "step": 9881
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.532592360473182e-05,
+      "loss": 17.0611,
+      "step": 9882
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.531806907698205e-05,
+      "loss": 0.7313,
+      "step": 9883
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5310214416492046e-05,
+      "loss": 18.1424,
+      "step": 9884
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.530235962345788e-05,
+      "loss": 0.7696,
+      "step": 9885
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.529450469807561e-05,
+      "loss": 0.7925,
+      "step": 9886
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.52866496405413e-05,
+      "loss": 18.5332,
+      "step": 9887
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5278794451050994e-05,
+      "loss": 17.1973,
+      "step": 9888
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.527093912980078e-05,
+      "loss": 20.7612,
+      "step": 9889
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.526308367698673e-05,
+      "loss": 16.7157,
+      "step": 9890
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.525522809280489e-05,
+      "loss": 0.751,
+      "step": 9891
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.524737237745138e-05,
+      "loss": 18.7379,
+      "step": 9892
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.523951653112225e-05,
+      "loss": 21.6086,
+      "step": 9893
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.523166055401359e-05,
+      "loss": 19.5654,
+      "step": 9894
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.522380444632147e-05,
+      "loss": 18.3589,
+      "step": 9895
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.521594820824202e-05,
+      "loss": 0.7134,
+      "step": 9896
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.520809183997131e-05,
+      "loss": 18.0966,
+      "step": 9897
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5200235341705433e-05,
+      "loss": 0.7475,
+      "step": 9898
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5192378713640495e-05,
+      "loss": 15.7047,
+      "step": 9899
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.51845219559726e-05,
+      "loss": 0.7422,
+      "step": 9900
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.517666506889786e-05,
+      "loss": 0.7302,
+      "step": 9901
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.516880805261236e-05,
+      "loss": 0.7463,
+      "step": 9902
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.516095090731225e-05,
+      "loss": 17.4737,
+      "step": 9903
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.515309363319361e-05,
+      "loss": 0.7962,
+      "step": 9904
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.514523623045258e-05,
+      "loss": 0.7798,
+      "step": 9905
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.513737869928529e-05,
+      "loss": 0.748,
+      "step": 9906
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.512952103988786e-05,
+      "loss": 0.7632,
+      "step": 9907
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.512166325245639e-05,
+      "loss": 19.6104,
+      "step": 9908
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.511380533718704e-05,
+      "loss": 18.8069,
+      "step": 9909
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.510594729427596e-05,
+      "loss": 16.5005,
+      "step": 9910
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5098089123919264e-05,
+      "loss": 18.9969,
+      "step": 9911
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.509023082631308e-05,
+      "loss": 18.3135,
+      "step": 9912
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5082372401653594e-05,
+      "loss": 18.4617,
+      "step": 9913
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.507451385013692e-05,
+      "loss": 0.7941,
+      "step": 9914
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.506665517195924e-05,
+      "loss": 0.8,
+      "step": 9915
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.505879636731667e-05,
+      "loss": 17.3964,
+      "step": 9916
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5050937436405394e-05,
+      "loss": 17.2869,
+      "step": 9917
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5043078379421566e-05,
+      "loss": 0.7544,
+      "step": 9918
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.503521919656134e-05,
+      "loss": 20.2216,
+      "step": 9919
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5027359888020914e-05,
+      "loss": 19.7779,
+      "step": 9920
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.501950045399643e-05,
+      "loss": 0.7971,
+      "step": 9921
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.5011640894684056e-05,
+      "loss": 0.7821,
+      "step": 9922
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.500378121027999e-05,
+      "loss": 0.7462,
+      "step": 9923
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.499592140098041e-05,
+      "loss": 18.1578,
+      "step": 9924
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.498806146698149e-05,
+      "loss": 0.7192,
+      "step": 9925
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.4980201408479405e-05,
+      "loss": 0.7476,
+      "step": 9926
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.4972341225670354e-05,
+      "loss": 0.7877,
+      "step": 9927
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.496448091875054e-05,
+      "loss": 20.3167,
+      "step": 9928
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.4956620487916154e-05,
+      "loss": 0.7337,
+      "step": 9929
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.494875993336338e-05,
+      "loss": 0.7728,
+      "step": 9930
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.494089925528844e-05,
+      "loss": 16.7049,
+      "step": 9931
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.4933038453887505e-05,
+      "loss": 0.7646,
+      "step": 9932
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.492517752935681e-05,
+      "loss": 18.9929,
+      "step": 9933
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.491731648189257e-05,
+      "loss": 19.6389,
+      "step": 9934
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.4909455311690984e-05,
+      "loss": 0.7795,
+      "step": 9935
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.490159401894826e-05,
+      "loss": 16.039,
+      "step": 9936
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.489373260386065e-05,
+      "loss": 0.7532,
+      "step": 9937
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.488587106662435e-05,
+      "loss": 0.8338,
+      "step": 9938
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.487800940743559e-05,
+      "loss": 18.999,
+      "step": 9939
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.4870147626490596e-05,
+      "loss": 0.7886,
+      "step": 9940
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.486228572398562e-05,
+      "loss": 0.8005,
+      "step": 9941
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.485442370011686e-05,
+      "loss": 17.752,
+      "step": 9942
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4846561555080603e-05,
+      "loss": 17.9345,
+      "step": 9943
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.483869928907305e-05,
+      "loss": 0.7916,
+      "step": 9944
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.483083690229046e-05,
+      "loss": 0.7722,
+      "step": 9945
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.482297439492906e-05,
+      "loss": 0.7842,
+      "step": 9946
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4815111767185137e-05,
+      "loss": 19.706,
+      "step": 9947
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.480724901925492e-05,
+      "loss": 0.7043,
+      "step": 9948
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.479938615133468e-05,
+      "loss": 0.7711,
+      "step": 9949
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.479152316362065e-05,
+      "loss": 17.7988,
+      "step": 9950
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.47836600563091e-05,
+      "loss": 18.5234,
+      "step": 9951
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.477579682959633e-05,
+      "loss": 18.21,
+      "step": 9952
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4767933483678566e-05,
+      "loss": 0.804,
+      "step": 9953
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.47600700187521e-05,
+      "loss": 0.7177,
+      "step": 9954
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4752206435013195e-05,
+      "loss": 0.8624,
+      "step": 9955
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.474434273265813e-05,
+      "loss": 0.7422,
+      "step": 9956
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.473647891188319e-05,
+      "loss": 0.7576,
+      "step": 9957
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.472861497288465e-05,
+      "loss": 0.7588,
+      "step": 9958
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.472075091585881e-05,
+      "loss": 18.1555,
+      "step": 9959
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.471288674100194e-05,
+      "loss": 17.3089,
+      "step": 9960
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4705022448510335e-05,
+      "loss": 19.4103,
+      "step": 9961
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.469715803858031e-05,
+      "loss": 20.6815,
+      "step": 9962
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4689293511408155e-05,
+      "loss": 0.7665,
+      "step": 9963
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.468142886719014e-05,
+      "loss": 18.2782,
+      "step": 9964
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4673564106122596e-05,
+      "loss": 0.7527,
+      "step": 9965
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.466569922840184e-05,
+      "loss": 0.76,
+      "step": 9966
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.465783423422417e-05,
+      "loss": 15.9307,
+      "step": 9967
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.464996912378587e-05,
+      "loss": 18.8213,
+      "step": 9968
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4642103897283304e-05,
+      "loss": 0.79,
+      "step": 9969
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.463423855491274e-05,
+      "loss": 0.7372,
+      "step": 9970
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.462637309687054e-05,
+      "loss": 19.5228,
+      "step": 9971
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.461850752335301e-05,
+      "loss": 17.9161,
+      "step": 9972
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4610641834556464e-05,
+      "loss": 17.0796,
+      "step": 9973
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.460277603067725e-05,
+      "loss": 19.8954,
+      "step": 9974
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.45949101119117e-05,
+      "loss": 17.9477,
+      "step": 9975
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4587044078456154e-05,
+      "loss": 20.2613,
+      "step": 9976
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.457917793050693e-05,
+      "loss": 0.7805,
+      "step": 9977
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.457131166826037e-05,
+      "loss": 18.9901,
+      "step": 9978
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.456344529191285e-05,
+      "loss": 18.9309,
+      "step": 9979
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4555578801660675e-05,
+      "loss": 0.7334,
+      "step": 9980
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.454771219770022e-05,
+      "loss": 0.7488,
+      "step": 9981
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.453984548022782e-05,
+      "loss": 19.1975,
+      "step": 9982
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4531978649439865e-05,
+      "loss": 18.568,
+      "step": 9983
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.452411170553266e-05,
+      "loss": 0.7912,
+      "step": 9984
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.451624464870262e-05,
+      "loss": 0.8237,
+      "step": 9985
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4508377479146066e-05,
+      "loss": 0.7781,
+      "step": 9986
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.450051019705939e-05,
+      "loss": 17.647,
+      "step": 9987
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.449264280263893e-05,
+      "loss": 0.7096,
+      "step": 9988
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4484775296081094e-05,
+      "loss": 17.9913,
+      "step": 9989
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4476907677582244e-05,
+      "loss": 0.7735,
+      "step": 9990
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.446903994733876e-05,
+      "loss": 0.7342,
+      "step": 9991
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.446117210554701e-05,
+      "loss": 0.7507,
+      "step": 9992
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.445330415240339e-05,
+      "loss": 19.5255,
+      "step": 9993
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4445436088104285e-05,
+      "loss": 16.2774,
+      "step": 9994
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.443756791284608e-05,
+      "loss": 16.1284,
+      "step": 9995
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.442969962682516e-05,
+      "loss": 0.8506,
+      "step": 9996
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.442183123023793e-05,
+      "loss": 0.741,
+      "step": 9997
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.441396272328078e-05,
+      "loss": 19.2252,
+      "step": 9998
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.440609410615012e-05,
+      "loss": 17.9272,
+      "step": 9999
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.439822537904234e-05,
+      "loss": 0.7657,
+      "step": 10000
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.439035654215384e-05,
+      "loss": 18.9162,
+      "step": 10001
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.438248759568105e-05,
+      "loss": 0.7739,
+      "step": 10002
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4374618539820354e-05,
+      "loss": 0.8509,
+      "step": 10003
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.43667493747682e-05,
+      "loss": 0.7683,
+      "step": 10004
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.435888010072099e-05,
+      "loss": 18.0648,
+      "step": 10005
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4351010717875116e-05,
+      "loss": 0.774,
+      "step": 10006
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.434314122642704e-05,
+      "loss": 0.7682,
+      "step": 10007
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.433527162657315e-05,
+      "loss": 20.2962,
+      "step": 10008
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4327401918509904e-05,
+      "loss": 0.7491,
+      "step": 10009
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4319532102433715e-05,
+      "loss": 0.7991,
+      "step": 10010
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.431166217854101e-05,
+      "loss": 18.4476,
+      "step": 10011
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4303792147028235e-05,
+      "loss": 18.6903,
+      "step": 10012
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.429592200809184e-05,
+      "loss": 0.7723,
+      "step": 10013
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.428805176192824e-05,
+      "loss": 17.332,
+      "step": 10014
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4280181408733876e-05,
+      "loss": 17.3725,
+      "step": 10015
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.427231094870522e-05,
+      "loss": 17.7335,
+      "step": 10016
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4264440382038705e-05,
+      "loss": 0.7645,
+      "step": 10017
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4256569708930785e-05,
+      "loss": 0.7411,
+      "step": 10018
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.424869892957792e-05,
+      "loss": 18.9312,
+      "step": 10019
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.424082804417655e-05,
+      "loss": 17.4426,
+      "step": 10020
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.423295705292315e-05,
+      "loss": 19.8214,
+      "step": 10021
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.422508595601417e-05,
+      "loss": 0.7905,
+      "step": 10022
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4217214753646086e-05,
+      "loss": 0.792,
+      "step": 10023
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4209343446015356e-05,
+      "loss": 18.1456,
+      "step": 10024
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4201472033318444e-05,
+      "loss": 0.7421,
+      "step": 10025
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.419360051575184e-05,
+      "loss": 0.7697,
+      "step": 10026
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.418572889351201e-05,
+      "loss": 0.7927,
+      "step": 10027
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.417785716679543e-05,
+      "loss": 0.7841,
+      "step": 10028
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.416998533579858e-05,
+      "loss": 0.7826,
+      "step": 10029
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.416211340071794e-05,
+      "loss": 16.9081,
+      "step": 10030
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.415424136175e-05,
+      "loss": 0.7998,
+      "step": 10031
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4146369219091254e-05,
+      "loss": 0.7184,
+      "step": 10032
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.413849697293818e-05,
+      "loss": 17.5842,
+      "step": 10033
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4130624623487283e-05,
+      "loss": 21.9492,
+      "step": 10034
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4122752170935055e-05,
+      "loss": 19.1175,
+      "step": 10035
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4114879615477973e-05,
+      "loss": 0.7622,
+      "step": 10036
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.410700695731257e-05,
+      "loss": 0.7674,
+      "step": 10037
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4099134196635336e-05,
+      "loss": 15.547,
+      "step": 10038
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4091261333642776e-05,
+      "loss": 0.773,
+      "step": 10039
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.408338836853138e-05,
+      "loss": 20.8552,
+      "step": 10040
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.40755153014977e-05,
+      "loss": 19.9165,
+      "step": 10041
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.406764213273824e-05,
+      "loss": 17.7363,
+      "step": 10042
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4059768862449475e-05,
+      "loss": 0.8001,
+      "step": 10043
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4051895490827956e-05,
+      "loss": 17.4403,
+      "step": 10044
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4044022018070214e-05,
+      "loss": 17.9106,
+      "step": 10045
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.403614844437275e-05,
+      "loss": 0.7354,
+      "step": 10046
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.4028274769932107e-05,
+      "loss": 0.8577,
+      "step": 10047
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.402040099494481e-05,
+      "loss": 20.0818,
+      "step": 10048
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.401252711960736e-05,
+      "loss": 0.8114,
+      "step": 10049
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.400465314411633e-05,
+      "loss": 19.1915,
+      "step": 10050
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.399677906866826e-05,
+      "loss": 18.5506,
+      "step": 10051
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3988904893459655e-05,
+      "loss": 0.7497,
+      "step": 10052
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.398103061868708e-05,
+      "loss": 0.7242,
+      "step": 10053
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3973156244547065e-05,
+      "loss": 19.3714,
+      "step": 10054
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.396528177123617e-05,
+      "loss": 19.7215,
+      "step": 10055
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3957407198950935e-05,
+      "loss": 18.8952,
+      "step": 10056
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3949532527887904e-05,
+      "loss": 0.7113,
+      "step": 10057
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3941657758243635e-05,
+      "loss": 0.792,
+      "step": 10058
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.393378289021471e-05,
+      "loss": 18.9906,
+      "step": 10059
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.392590792399765e-05,
+      "loss": 0.7569,
+      "step": 10060
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.391803285978905e-05,
+      "loss": 0.7278,
+      "step": 10061
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.391015769778544e-05,
+      "loss": 0.7696,
+      "step": 10062
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3902282438183396e-05,
+      "loss": 17.316,
+      "step": 10063
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.38944070811795e-05,
+      "loss": 0.7237,
+      "step": 10064
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3886531626970316e-05,
+      "loss": 20.7247,
+      "step": 10065
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.387865607575241e-05,
+      "loss": 0.8169,
+      "step": 10066
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.387078042772237e-05,
+      "loss": 0.7816,
+      "step": 10067
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.386290468307675e-05,
+      "loss": 0.7573,
+      "step": 10068
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3855028842012165e-05,
+      "loss": 20.0783,
+      "step": 10069
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.384715290472517e-05,
+      "loss": 19.6846,
+      "step": 10070
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.383927687141237e-05,
+      "loss": 17.1501,
+      "step": 10071
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.383140074227032e-05,
+      "loss": 0.7623,
+      "step": 10072
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.382352451749566e-05,
+      "loss": 0.7361,
+      "step": 10073
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.381564819728494e-05,
+      "loss": 19.9236,
+      "step": 10074
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.380777178183478e-05,
+      "loss": 0.7548,
+      "step": 10075
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.379989527134176e-05,
+      "loss": 17.4538,
+      "step": 10076
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.379201866600249e-05,
+      "loss": 17.1485,
+      "step": 10077
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3784141966013556e-05,
+      "loss": 0.7606,
+      "step": 10078
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3776265171571585e-05,
+      "loss": 18.4002,
+      "step": 10079
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3768388282873184e-05,
+      "loss": 18.2152,
+      "step": 10080
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3760511300114925e-05,
+      "loss": 0.7593,
+      "step": 10081
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3752634223493455e-05,
+      "loss": 0.7995,
+      "step": 10082
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.374475705320539e-05,
+      "loss": 0.7807,
+      "step": 10083
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3736879789447324e-05,
+      "loss": 17.0785,
+      "step": 10084
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.372900243241589e-05,
+      "loss": 0.7687,
+      "step": 10085
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.37211249823077e-05,
+      "loss": 0.7259,
+      "step": 10086
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.371324743931938e-05,
+      "loss": 17.7116,
+      "step": 10087
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.370536980364755e-05,
+      "loss": 0.7429,
+      "step": 10088
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.369749207548887e-05,
+      "loss": 0.7635,
+      "step": 10089
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.368961425503992e-05,
+      "loss": 16.0936,
+      "step": 10090
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3681736342497355e-05,
+      "loss": 19.1832,
+      "step": 10091
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3673858338057815e-05,
+      "loss": 18.5682,
+      "step": 10092
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.366598024191793e-05,
+      "loss": 0.7843,
+      "step": 10093
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3658102054274354e-05,
+      "loss": 19.6971,
+      "step": 10094
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.365022377532371e-05,
+      "loss": 0.8492,
+      "step": 10095
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3642345405262625e-05,
+      "loss": 18.068,
+      "step": 10096
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3634466944287796e-05,
+      "loss": 0.7776,
+      "step": 10097
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3626588392595834e-05,
+      "loss": 17.5336,
+      "step": 10098
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.361870975038339e-05,
+      "loss": 20.6241,
+      "step": 10099
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.361083101784713e-05,
+      "loss": 0.7267,
+      "step": 10100
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3602952195183696e-05,
+      "loss": 0.7924,
+      "step": 10101
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.359507328258976e-05,
+      "loss": 0.7427,
+      "step": 10102
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.358719428026198e-05,
+      "loss": 18.5644,
+      "step": 10103
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3579315188397005e-05,
+      "loss": 0.7791,
+      "step": 10104
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.35714360071915e-05,
+      "loss": 17.9298,
+      "step": 10105
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3563556736842136e-05,
+      "loss": 17.4422,
+      "step": 10106
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.35556773775456e-05,
+      "loss": 0.7304,
+      "step": 10107
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.354779792949853e-05,
+      "loss": 18.1261,
+      "step": 10108
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3539918392897616e-05,
+      "loss": 19.1853,
+      "step": 10109
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.353203876793952e-05,
+      "loss": 18.4935,
+      "step": 10110
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3524159054820954e-05,
+      "loss": 0.8045,
+      "step": 10111
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.351627925373857e-05,
+      "loss": 19.8501,
+      "step": 10112
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.350839936488903e-05,
+      "loss": 17.532,
+      "step": 10113
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.350051938846906e-05,
+      "loss": 18.0891,
+      "step": 10114
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.349263932467532e-05,
+      "loss": 0.7689,
+      "step": 10115
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.348475917370449e-05,
+      "loss": 18.2094,
+      "step": 10116
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.34768789357533e-05,
+      "loss": 0.8012,
+      "step": 10117
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.346899861101842e-05,
+      "loss": 19.1957,
+      "step": 10118
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3461118199696526e-05,
+      "loss": 18.1597,
+      "step": 10119
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.345323770198433e-05,
+      "loss": 0.8054,
+      "step": 10120
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3445357118078545e-05,
+      "loss": 18.7338,
+      "step": 10121
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.343747644817586e-05,
+      "loss": 18.158,
+      "step": 10122
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.342959569247297e-05,
+      "loss": 19.2481,
+      "step": 10123
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.342171485116659e-05,
+      "loss": 17.9859,
+      "step": 10124
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.341383392445343e-05,
+      "loss": 20.0479,
+      "step": 10125
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.340595291253019e-05,
+      "loss": 17.354,
+      "step": 10126
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3398071815593585e-05,
+      "loss": 0.734,
+      "step": 10127
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.339019063384034e-05,
+      "loss": 20.1577,
+      "step": 10128
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3382309367467155e-05,
+      "loss": 19.3636,
+      "step": 10129
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.337442801667075e-05,
+      "loss": 0.7744,
+      "step": 10130
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.336654658164787e-05,
+      "loss": 0.8174,
+      "step": 10131
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.335866506259521e-05,
+      "loss": 21.0133,
+      "step": 10132
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3350783459709505e-05,
+      "loss": 20.022,
+      "step": 10133
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.334290177318747e-05,
+      "loss": 19.7176,
+      "step": 10134
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3335020003225866e-05,
+      "loss": 19.0177,
+      "step": 10135
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.33271381500214e-05,
+      "loss": 18.1517,
+      "step": 10136
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3319256213770776e-05,
+      "loss": 0.7128,
+      "step": 10137
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3311374194670785e-05,
+      "loss": 0.748,
+      "step": 10138
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.330349209291812e-05,
+      "loss": 0.6932,
+      "step": 10139
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.329560990870955e-05,
+      "loss": 0.7772,
+      "step": 10140
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.32877276422418e-05,
+      "loss": 0.722,
+      "step": 10141
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.327984529371163e-05,
+      "loss": 17.9531,
+      "step": 10142
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.327196286331575e-05,
+      "loss": 18.9815,
+      "step": 10143
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.326408035125093e-05,
+      "loss": 18.3696,
+      "step": 10144
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3256197757713945e-05,
+      "loss": 17.7043,
+      "step": 10145
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3248315082901504e-05,
+      "loss": 19.424,
+      "step": 10146
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.324043232701037e-05,
+      "loss": 0.7599,
+      "step": 10147
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.3232549490237316e-05,
+      "loss": 19.8396,
+      "step": 10148
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.3224666572779094e-05,
+      "loss": 16.9808,
+      "step": 10149
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.3216783574832454e-05,
+      "loss": 20.1022,
+      "step": 10150
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.320890049659415e-05,
+      "loss": 0.7447,
+      "step": 10151
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.3201017338260975e-05,
+      "loss": 16.7981,
+      "step": 10152
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.319313410002965e-05,
+      "loss": 18.3639,
+      "step": 10153
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.318525078209699e-05,
+      "loss": 0.7646,
+      "step": 10154
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.317736738465974e-05,
+      "loss": 0.8116,
+      "step": 10155
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.3169483907914674e-05,
+      "loss": 0.7573,
+      "step": 10156
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.316160035205856e-05,
+      "loss": 19.2991,
+      "step": 10157
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.3153716717288184e-05,
+      "loss": 18.8759,
+      "step": 10158
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.314583300380032e-05,
+      "loss": 0.7978,
+      "step": 10159
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.313794921179175e-05,
+      "loss": 18.2532,
+      "step": 10160
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.313006534145925e-05,
+      "loss": 18.4389,
+      "step": 10161
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.312218139299959e-05,
+      "loss": 19.9583,
+      "step": 10162
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.311429736660958e-05,
+      "loss": 18.4227,
+      "step": 10163
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.3106413262485996e-05,
+      "loss": 17.0216,
+      "step": 10164
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.309852908082563e-05,
+      "loss": 0.8113,
+      "step": 10165
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.309064482182528e-05,
+      "loss": 0.7604,
+      "step": 10166
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.30827604856817e-05,
+      "loss": 18.732,
+      "step": 10167
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.307487607259175e-05,
+      "loss": 18.4732,
+      "step": 10168
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.306699158275218e-05,
+      "loss": 17.1018,
+      "step": 10169
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.3059107016359785e-05,
+      "loss": 18.203,
+      "step": 10170
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.3051222373611384e-05,
+      "loss": 0.7492,
+      "step": 10171
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.304333765470377e-05,
+      "loss": 17.5386,
+      "step": 10172
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.303545285983377e-05,
+      "loss": 0.7881,
+      "step": 10173
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.302756798919818e-05,
+      "loss": 0.7612,
+      "step": 10174
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.301968304299378e-05,
+      "loss": 0.823,
+      "step": 10175
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.301179802141741e-05,
+      "loss": 19.6361,
+      "step": 10176
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.300391292466588e-05,
+      "loss": 0.7495,
+      "step": 10177
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2996027752935995e-05,
+      "loss": 0.7611,
+      "step": 10178
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.298814250642456e-05,
+      "loss": 16.3214,
+      "step": 10179
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.298025718532843e-05,
+      "loss": 19.5132,
+      "step": 10180
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2972371789844375e-05,
+      "loss": 18.1579,
+      "step": 10181
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.296448632016926e-05,
+      "loss": 0.7901,
+      "step": 10182
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2956600776499884e-05,
+      "loss": 19.5479,
+      "step": 10183
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.294871515903308e-05,
+      "loss": 0.792,
+      "step": 10184
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.294082946796566e-05,
+      "loss": 18.1539,
+      "step": 10185
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2932943703494466e-05,
+      "loss": 0.702,
+      "step": 10186
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.292505786581634e-05,
+      "loss": 17.6263,
+      "step": 10187
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2917171955128096e-05,
+      "loss": 0.7964,
+      "step": 10188
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.290928597162658e-05,
+      "loss": 0.7121,
+      "step": 10189
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.290139991550862e-05,
+      "loss": 17.894,
+      "step": 10190
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.289351378697105e-05,
+      "loss": 17.5829,
+      "step": 10191
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2885627586210714e-05,
+      "loss": 19.7545,
+      "step": 10192
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.287774131342445e-05,
+      "loss": 18.9583,
+      "step": 10193
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.286985496880913e-05,
+      "loss": 18.4363,
+      "step": 10194
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2861968552561546e-05,
+      "loss": 18.3634,
+      "step": 10195
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2854082064878594e-05,
+      "loss": 18.6371,
+      "step": 10196
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2846195505957095e-05,
+      "loss": 16.4352,
+      "step": 10197
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2838308875993905e-05,
+      "loss": 0.7845,
+      "step": 10198
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.283042217518587e-05,
+      "loss": 17.1154,
+      "step": 10199
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2822535403729866e-05,
+      "loss": 0.73,
+      "step": 10200
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.281464856182272e-05,
+      "loss": 0.6974,
+      "step": 10201
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.280676164966132e-05,
+      "loss": 0.7821,
+      "step": 10202
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.279887466744249e-05,
+      "loss": 20.6476,
+      "step": 10203
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.279098761536312e-05,
+      "loss": 0.7784,
+      "step": 10204
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.278310049362005e-05,
+      "loss": 0.7882,
+      "step": 10205
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2775213302410174e-05,
+      "loss": 19.0545,
+      "step": 10206
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.276732604193034e-05,
+      "loss": 0.7948,
+      "step": 10207
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2759438712377395e-05,
+      "loss": 16.0596,
+      "step": 10208
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2751551313948246e-05,
+      "loss": 0.7139,
+      "step": 10209
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.274366384683974e-05,
+      "loss": 0.8106,
+      "step": 10210
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2735776311248774e-05,
+      "loss": 19.6337,
+      "step": 10211
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2727888707372184e-05,
+      "loss": 17.7723,
+      "step": 10212
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2720001035406885e-05,
+      "loss": 20.1,
+      "step": 10213
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.271211329554974e-05,
+      "loss": 17.8125,
+      "step": 10214
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.270422548799761e-05,
+      "loss": 19.4004,
+      "step": 10215
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.269633761294741e-05,
+      "loss": 16.1799,
+      "step": 10216
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.268844967059601e-05,
+      "loss": 0.7953,
+      "step": 10217
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.268056166114028e-05,
+      "loss": 19.0737,
+      "step": 10218
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.267267358477712e-05,
+      "loss": 19.4687,
+      "step": 10219
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.266478544170343e-05,
+      "loss": 18.2314,
+      "step": 10220
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.265689723211609e-05,
+      "loss": 0.7678,
+      "step": 10221
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.264900895621196e-05,
+      "loss": 15.14,
+      "step": 10222
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2641120614187976e-05,
+      "loss": 0.737,
+      "step": 10223
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2633232206241034e-05,
+      "loss": 0.729,
+      "step": 10224
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.262534373256801e-05,
+      "loss": 18.2416,
+      "step": 10225
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2617455193365796e-05,
+      "loss": 0.7346,
+      "step": 10226
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2609566588831297e-05,
+      "loss": 0.788,
+      "step": 10227
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2601677919161427e-05,
+      "loss": 0.8527,
+      "step": 10228
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.259378918455309e-05,
+      "loss": 0.7033,
+      "step": 10229
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.258590038520318e-05,
+      "loss": 0.7232,
+      "step": 10230
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.25780115213086e-05,
+      "loss": 0.7342,
+      "step": 10231
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.257012259306627e-05,
+      "loss": 17.6191,
+      "step": 10232
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.256223360067308e-05,
+      "loss": 0.7286,
+      "step": 10233
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.255434454432597e-05,
+      "loss": 0.7728,
+      "step": 10234
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.254645542422184e-05,
+      "loss": 18.3833,
+      "step": 10235
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2538566240557586e-05,
+      "loss": 0.7679,
+      "step": 10236
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.253067699353014e-05,
+      "loss": 16.7635,
+      "step": 10237
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2522787683336426e-05,
+      "loss": 0.8209,
+      "step": 10238
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2514898310173364e-05,
+      "loss": 19.2266,
+      "step": 10239
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.250700887423785e-05,
+      "loss": 20.7382,
+      "step": 10240
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2499119375726826e-05,
+      "loss": 18.7308,
+      "step": 10241
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2491229814837214e-05,
+      "loss": 19.0251,
+      "step": 10242
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.248334019176593e-05,
+      "loss": 0.8089,
+      "step": 10243
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.247545050670992e-05,
+      "loss": 0.7467,
+      "step": 10244
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.24675607598661e-05,
+      "loss": 17.1946,
+      "step": 10245
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.24596709514314e-05,
+      "loss": 17.1407,
+      "step": 10246
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.245178108160273e-05,
+      "loss": 17.7068,
+      "step": 10247
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2443891150577064e-05,
+      "loss": 15.5269,
+      "step": 10248
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.243600115855132e-05,
+      "loss": 17.5588,
+      "step": 10249
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.242811110572242e-05,
+      "loss": 20.6849,
+      "step": 10250
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2420220992287305e-05,
+      "loss": 16.4364,
+      "step": 10251
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2412330818442944e-05,
+      "loss": 0.7888,
+      "step": 10252
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.240444058438625e-05,
+      "loss": 0.8216,
+      "step": 10253
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.239655029031415e-05,
+      "loss": 18.1637,
+      "step": 10254
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2388659936423624e-05,
+      "loss": 18.5409,
+      "step": 10255
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.23807695229116e-05,
+      "loss": 18.9928,
+      "step": 10256
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.237287904997502e-05,
+      "loss": 19.8639,
+      "step": 10257
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.236498851781084e-05,
+      "loss": 0.7446,
+      "step": 10258
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2357097926616015e-05,
+      "loss": 19.098,
+      "step": 10259
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.234920727658748e-05,
+      "loss": 0.747,
+      "step": 10260
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2341316567922185e-05,
+      "loss": 0.7325,
+      "step": 10261
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2333425800817106e-05,
+      "loss": 17.4923,
+      "step": 10262
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.232553497546918e-05,
+      "loss": 19.2138,
+      "step": 10263
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.231764409207538e-05,
+      "loss": 20.4411,
+      "step": 10264
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2309753150832644e-05,
+      "loss": 16.7331,
+      "step": 10265
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.230186215193794e-05,
+      "loss": 0.7491,
+      "step": 10266
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.229397109558824e-05,
+      "loss": 18.6897,
+      "step": 10267
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.22860799819805e-05,
+      "loss": 0.7552,
+      "step": 10268
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.227818881131167e-05,
+      "loss": 15.2086,
+      "step": 10269
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2270297583778724e-05,
+      "loss": 17.8248,
+      "step": 10270
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2262406299578646e-05,
+      "loss": 19.6752,
+      "step": 10271
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2254514958908386e-05,
+      "loss": 0.7305,
+      "step": 10272
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.224662356196492e-05,
+      "loss": 0.8124,
+      "step": 10273
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.223873210894521e-05,
+      "loss": 17.4104,
+      "step": 10274
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2230840600046235e-05,
+      "loss": 0.7764,
+      "step": 10275
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.222294903546498e-05,
+      "loss": 17.1728,
+      "step": 10276
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2215057415398405e-05,
+      "loss": 0.6925,
+      "step": 10277
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2207165740043485e-05,
+      "loss": 18.3371,
+      "step": 10278
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.219927400959721e-05,
+      "loss": 18.5691,
+      "step": 10279
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.219138222425656e-05,
+      "loss": 0.7392,
+      "step": 10280
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.21834903842185e-05,
+      "loss": 0.7708,
+      "step": 10281
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.217559848968001e-05,
+      "loss": 0.7553,
+      "step": 10282
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2167706540838114e-05,
+      "loss": 18.216,
+      "step": 10283
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.215981453788974e-05,
+      "loss": 0.7944,
+      "step": 10284
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.215192248103191e-05,
+      "loss": 18.4434,
+      "step": 10285
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.214403037046162e-05,
+      "loss": 0.6865,
+      "step": 10286
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.213613820637583e-05,
+      "loss": 0.7382,
+      "step": 10287
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2128245988971535e-05,
+      "loss": 0.7595,
+      "step": 10288
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.212035371844574e-05,
+      "loss": 20.3934,
+      "step": 10289
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2112461394995424e-05,
+      "loss": 0.7198,
+      "step": 10290
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.210456901881761e-05,
+      "loss": 18.8339,
+      "step": 10291
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.209667659010926e-05,
+      "loss": 18.2805,
+      "step": 10292
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.208878410906738e-05,
+      "loss": 0.7256,
+      "step": 10293
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.208089157588898e-05,
+      "loss": 18.7809,
+      "step": 10294
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.207299899077105e-05,
+      "loss": 15.7602,
+      "step": 10295
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2065106353910586e-05,
+      "loss": 17.1951,
+      "step": 10296
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.20572136655046e-05,
+      "loss": 19.193,
+      "step": 10297
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.204932092575009e-05,
+      "loss": 16.3784,
+      "step": 10298
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.204142813484405e-05,
+      "loss": 16.0635,
+      "step": 10299
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.203353529298352e-05,
+      "loss": 21.1091,
+      "step": 10300
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.2025642400365485e-05,
+      "loss": 15.9508,
+      "step": 10301
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.201774945718694e-05,
+      "loss": 19.8896,
+      "step": 10302
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.20098564636449e-05,
+      "loss": 19.4812,
+      "step": 10303
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.20019634199364e-05,
+      "loss": 0.7273,
+      "step": 10304
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.199407032625844e-05,
+      "loss": 0.7488,
+      "step": 10305
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.1986177182808e-05,
+      "loss": 0.7595,
+      "step": 10306
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.197828398978215e-05,
+      "loss": 0.7796,
+      "step": 10307
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.197039074737787e-05,
+      "loss": 19.1006,
+      "step": 10308
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.1962497455792204e-05,
+      "loss": 0.7667,
+      "step": 10309
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.1954604115222125e-05,
+      "loss": 20.0198,
+      "step": 10310
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.194671072586471e-05,
+      "loss": 18.4207,
+      "step": 10311
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.193881728791693e-05,
+      "loss": 0.7706,
+      "step": 10312
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.1930923801575824e-05,
+      "loss": 0.7947,
+      "step": 10313
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.192303026703844e-05,
+      "loss": 16.4834,
+      "step": 10314
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.191513668450178e-05,
+      "loss": 0.7224,
+      "step": 10315
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.190724305416286e-05,
+      "loss": 16.6847,
+      "step": 10316
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.189934937621871e-05,
+      "loss": 19.5465,
+      "step": 10317
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.1891455650866384e-05,
+      "loss": 17.9517,
+      "step": 10318
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.188356187830289e-05,
+      "loss": 0.8051,
+      "step": 10319
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.187566805872525e-05,
+      "loss": 18.0677,
+      "step": 10320
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.186777419233052e-05,
+      "loss": 0.7608,
+      "step": 10321
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.185988027931571e-05,
+      "loss": 0.7608,
+      "step": 10322
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.1851986319877865e-05,
+      "loss": 0.7594,
+      "step": 10323
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.184409231421402e-05,
+      "loss": 19.2521,
+      "step": 10324
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.1836198262521204e-05,
+      "loss": 18.8033,
+      "step": 10325
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.1828304164996466e-05,
+      "loss": 0.788,
+      "step": 10326
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.182041002183684e-05,
+      "loss": 0.774,
+      "step": 10327
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.181251583323936e-05,
+      "loss": 0.7536,
+      "step": 10328
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.180462159940108e-05,
+      "loss": 16.7753,
+      "step": 10329
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.179672732051902e-05,
+      "loss": 0.7736,
+      "step": 10330
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.178883299679024e-05,
+      "loss": 15.2717,
+      "step": 10331
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.1780938628411794e-05,
+      "loss": 0.7555,
+      "step": 10332
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.17730442155807e-05,
+      "loss": 19.3785,
+      "step": 10333
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.1765149758494014e-05,
+      "loss": 17.6103,
+      "step": 10334
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.17572552573488e-05,
+      "loss": 0.6884,
+      "step": 10335
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.174936071234208e-05,
+      "loss": 0.7491,
+      "step": 10336
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.1741466123670937e-05,
+      "loss": 20.117,
+      "step": 10337
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.1733571491532375e-05,
+      "loss": 17.8942,
+      "step": 10338
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.1725676816123495e-05,
+      "loss": 16.8764,
+      "step": 10339
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.171778209764131e-05,
+      "loss": 19.0536,
+      "step": 10340
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.170988733628289e-05,
+      "loss": 0.7877,
+      "step": 10341
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.1701992532245315e-05,
+      "loss": 16.4039,
+      "step": 10342
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.169409768572561e-05,
+      "loss": 0.8008,
+      "step": 10343
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.168620279692082e-05,
+      "loss": 17.0769,
+      "step": 10344
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.167830786602804e-05,
+      "loss": 18.2145,
+      "step": 10345
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.167041289324431e-05,
+      "loss": 17.1648,
+      "step": 10346
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.16625178787667e-05,
+      "loss": 16.4812,
+      "step": 10347
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.165462282279225e-05,
+      "loss": 17.0099,
+      "step": 10348
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.164672772551805e-05,
+      "loss": 0.7415,
+      "step": 10349
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.163883258714114e-05,
+      "loss": 0.7107,
+      "step": 10350
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.1630937407858604e-05,
+      "loss": 0.753,
+      "step": 10351
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.16230421878675e-05,
+      "loss": 0.8073,
+      "step": 10352
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.161514692736488e-05,
+      "loss": 19.8732,
+      "step": 10353
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.1607251626547814e-05,
+      "loss": 16.9792,
+      "step": 10354
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.159935628561339e-05,
+      "loss": 0.7834,
+      "step": 10355
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.159146090475868e-05,
+      "loss": 0.7893,
+      "step": 10356
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.158356548418074e-05,
+      "loss": 18.1593,
+      "step": 10357
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.157567002407664e-05,
+      "loss": 17.0985,
+      "step": 10358
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.156777452464345e-05,
+      "loss": 17.4511,
+      "step": 10359
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.155987898607826e-05,
+      "loss": 0.7809,
+      "step": 10360
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.155198340857813e-05,
+      "loss": 0.7617,
+      "step": 10361
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.1544087792340136e-05,
+      "loss": 0.7143,
+      "step": 10362
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.153619213756137e-05,
+      "loss": 17.592,
+      "step": 10363
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.1528296444438874e-05,
+      "loss": 15.7494,
+      "step": 10364
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.152040071316977e-05,
+      "loss": 0.7549,
+      "step": 10365
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.1512504943951124e-05,
+      "loss": 18.5953,
+      "step": 10366
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.1504609136979986e-05,
+      "loss": 16.3205,
+      "step": 10367
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.1496713292453467e-05,
+      "loss": 17.2392,
+      "step": 10368
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.148881741056865e-05,
+      "loss": 0.7645,
+      "step": 10369
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.14809214915226e-05,
+      "loss": 16.4557,
+      "step": 10370
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.147302553551243e-05,
+      "loss": 18.9787,
+      "step": 10371
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.146512954273519e-05,
+      "loss": 0.7784,
+      "step": 10372
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.1457233513387994e-05,
+      "loss": 0.7435,
+      "step": 10373
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.14493374476679e-05,
+      "loss": 17.2518,
+      "step": 10374
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.144144134577203e-05,
+      "loss": 0.7027,
+      "step": 10375
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.143354520789744e-05,
+      "loss": 0.7525,
+      "step": 10376
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.142564903424125e-05,
+      "loss": 19.9673,
+      "step": 10377
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.141775282500052e-05,
+      "loss": 17.9119,
+      "step": 10378
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.140985658037236e-05,
+      "loss": 0.8076,
+      "step": 10379
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.140196030055388e-05,
+      "loss": 0.8049,
+      "step": 10380
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.1394063985742116e-05,
+      "loss": 0.7793,
+      "step": 10381
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.138616763613421e-05,
+      "loss": 17.5782,
+      "step": 10382
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.137827125192723e-05,
+      "loss": 0.7274,
+      "step": 10383
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.137037483331831e-05,
+      "loss": 0.7522,
+      "step": 10384
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.136247838050452e-05,
+      "loss": 17.6667,
+      "step": 10385
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.135458189368294e-05,
+      "loss": 18.8889,
+      "step": 10386
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.13466853730507e-05,
+      "loss": 19.1472,
+      "step": 10387
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.1338788818804874e-05,
+      "loss": 20.641,
+      "step": 10388
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.133089223114259e-05,
+      "loss": 19.1808,
+      "step": 10389
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.132299561026093e-05,
+      "loss": 0.7911,
+      "step": 10390
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.1315098956356976e-05,
+      "loss": 17.3095,
+      "step": 10391
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.130720226962786e-05,
+      "loss": 0.7266,
+      "step": 10392
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.129930555027069e-05,
+      "loss": 18.9593,
+      "step": 10393
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.1291408798482554e-05,
+      "loss": 16.9213,
+      "step": 10394
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.128351201446056e-05,
+      "loss": 0.7495,
+      "step": 10395
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.1275615198401796e-05,
+      "loss": 16.8254,
+      "step": 10396
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.12677183505034e-05,
+      "loss": 0.7516,
+      "step": 10397
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.1259821470962466e-05,
+      "loss": 17.4992,
+      "step": 10398
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.125192455997609e-05,
+      "loss": 0.7505,
+      "step": 10399
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.12440276177414e-05,
+      "loss": 19.3254,
+      "step": 10400
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.12361306444555e-05,
+      "loss": 0.7925,
+      "step": 10401
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.122823364031548e-05,
+      "loss": 17.7848,
+      "step": 10402
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.122033660551849e-05,
+      "loss": 18.7353,
+      "step": 10403
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.121243954026161e-05,
+      "loss": 0.7215,
+      "step": 10404
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.1204542444741956e-05,
+      "loss": 18.489,
+      "step": 10405
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.1196645319156655e-05,
+      "loss": 0.7462,
+      "step": 10406
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.1188748163702817e-05,
+      "loss": 17.9803,
+      "step": 10407
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.118085097857754e-05,
+      "loss": 0.7665,
+      "step": 10408
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.117295376397796e-05,
+      "loss": 21.0637,
+      "step": 10409
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.116505652010118e-05,
+      "loss": 18.9135,
+      "step": 10410
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.115715924714434e-05,
+      "loss": 19.5746,
+      "step": 10411
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.1149261945304526e-05,
+      "loss": 17.8223,
+      "step": 10412
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.114136461477888e-05,
+      "loss": 18.2367,
+      "step": 10413
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.11334672557645e-05,
+      "loss": 17.0181,
+      "step": 10414
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.1125569868458536e-05,
+      "loss": 0.7534,
+      "step": 10415
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.111767245305807e-05,
+      "loss": 0.7588,
+      "step": 10416
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.1109775009760274e-05,
+      "loss": 16.5312,
+      "step": 10417
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.110187753876222e-05,
+      "loss": 0.7152,
+      "step": 10418
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.1093980040261056e-05,
+      "loss": 18.0941,
+      "step": 10419
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.108608251445389e-05,
+      "loss": 0.7744,
+      "step": 10420
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.107818496153787e-05,
+      "loss": 0.7557,
+      "step": 10421
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.1070287381710115e-05,
+      "loss": 19.5627,
+      "step": 10422
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.1062389775167715e-05,
+      "loss": 0.7698,
+      "step": 10423
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.105449214210784e-05,
+      "loss": 0.7405,
+      "step": 10424
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.1046594482727604e-05,
+      "loss": 0.6896,
+      "step": 10425
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.1038696797224115e-05,
+      "loss": 17.8043,
+      "step": 10426
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.1030799085794534e-05,
+      "loss": 0.753,
+      "step": 10427
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.1022901348635974e-05,
+      "loss": 0.7111,
+      "step": 10428
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.101500358594555e-05,
+      "loss": 20.0216,
+      "step": 10429
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.100710579792041e-05,
+      "loss": 19.6321,
+      "step": 10430
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.099920798475768e-05,
+      "loss": 19.6911,
+      "step": 10431
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.099131014665449e-05,
+      "loss": 17.1855,
+      "step": 10432
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.098341228380796e-05,
+      "loss": 18.8785,
+      "step": 10433
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0975514396415246e-05,
+      "loss": 18.8132,
+      "step": 10434
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.096761648467347e-05,
+      "loss": 16.7218,
+      "step": 10435
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.095971854877978e-05,
+      "loss": 0.783,
+      "step": 10436
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.095182058893128e-05,
+      "loss": 0.7848,
+      "step": 10437
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0943922605325115e-05,
+      "loss": 0.7392,
+      "step": 10438
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.093602459815844e-05,
+      "loss": 17.1585,
+      "step": 10439
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0928126567628364e-05,
+      "loss": 0.7672,
+      "step": 10440
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.092022851393205e-05,
+      "loss": 19.0087,
+      "step": 10441
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.091233043726662e-05,
+      "loss": 18.4237,
+      "step": 10442
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.090443233782921e-05,
+      "loss": 17.6909,
+      "step": 10443
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.089653421581695e-05,
+      "loss": 19.5292,
+      "step": 10444
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0888636071427006e-05,
+      "loss": 0.7455,
+      "step": 10445
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0880737904856515e-05,
+      "loss": 0.777,
+      "step": 10446
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.087283971630259e-05,
+      "loss": 18.1072,
+      "step": 10447
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.086494150596237e-05,
+      "loss": 0.7354,
+      "step": 10448
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.085704327403304e-05,
+      "loss": 15.9784,
+      "step": 10449
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.084914502071171e-05,
+      "loss": 19.0806,
+      "step": 10450
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0841246746195514e-05,
+      "loss": 16.8187,
+      "step": 10451
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0833348450681604e-05,
+      "loss": 17.9742,
+      "step": 10452
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.082545013436715e-05,
+      "loss": 17.4246,
+      "step": 10453
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.081755179744925e-05,
+      "loss": 0.7295,
+      "step": 10454
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.080965344012508e-05,
+      "loss": 18.6245,
+      "step": 10455
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.080175506259178e-05,
+      "loss": 17.001,
+      "step": 10456
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0793856665046477e-05,
+      "loss": 17.7394,
+      "step": 10457
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0785958247686335e-05,
+      "loss": 0.7953,
+      "step": 10458
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.07780598107085e-05,
+      "loss": 0.7653,
+      "step": 10459
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.077016135431012e-05,
+      "loss": 0.6994,
+      "step": 10460
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0762262878688326e-05,
+      "loss": 20.8383,
+      "step": 10461
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.075436438404028e-05,
+      "loss": 17.7817,
+      "step": 10462
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.074646587056312e-05,
+      "loss": 19.3265,
+      "step": 10463
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0738567338454015e-05,
+      "loss": 0.76,
+      "step": 10464
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.07306687879101e-05,
+      "loss": 0.8005,
+      "step": 10465
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0722770219128514e-05,
+      "loss": 19.2209,
+      "step": 10466
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0714871632306414e-05,
+      "loss": 17.0044,
+      "step": 10467
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0706973027640955e-05,
+      "loss": 20.4537,
+      "step": 10468
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.069907440532931e-05,
+      "loss": 16.6968,
+      "step": 10469
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0691175765568596e-05,
+      "loss": 17.9968,
+      "step": 10470
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.068327710855597e-05,
+      "loss": 0.7087,
+      "step": 10471
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.06753784344886e-05,
+      "loss": 16.4109,
+      "step": 10472
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.066747974356363e-05,
+      "loss": 17.8904,
+      "step": 10473
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.065958103597822e-05,
+      "loss": 17.9903,
+      "step": 10474
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.065168231192949e-05,
+      "loss": 18.3036,
+      "step": 10475
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0643783571614646e-05,
+      "loss": 17.2357,
+      "step": 10476
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.063588481523081e-05,
+      "loss": 20.3945,
+      "step": 10477
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.062798604297515e-05,
+      "loss": 19.426,
+      "step": 10478
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.06200872550448e-05,
+      "loss": 18.629,
+      "step": 10479
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.061218845163694e-05,
+      "loss": 0.7608,
+      "step": 10480
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0604289632948706e-05,
+      "loss": 17.0197,
+      "step": 10481
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0596390799177275e-05,
+      "loss": 0.7259,
+      "step": 10482
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.05884919505198e-05,
+      "loss": 18.2592,
+      "step": 10483
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0580593087173435e-05,
+      "loss": 16.816,
+      "step": 10484
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.057269420933531e-05,
+      "loss": 0.7322,
+      "step": 10485
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.056479531720262e-05,
+      "loss": 16.2643,
+      "step": 10486
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.055689641097251e-05,
+      "loss": 0.7585,
+      "step": 10487
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.054899749084214e-05,
+      "loss": 18.3963,
+      "step": 10488
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.054109855700866e-05,
+      "loss": 0.7517,
+      "step": 10489
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.053319960966923e-05,
+      "loss": 18.4874,
+      "step": 10490
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.052530064902105e-05,
+      "loss": 0.7567,
+      "step": 10491
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.051740167526122e-05,
+      "loss": 0.775,
+      "step": 10492
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.050950268858693e-05,
+      "loss": 19.5515,
+      "step": 10493
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.050160368919533e-05,
+      "loss": 19.3682,
+      "step": 10494
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.049370467728359e-05,
+      "loss": 0.7329,
+      "step": 10495
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.048580565304887e-05,
+      "loss": 0.7486,
+      "step": 10496
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0477906616688344e-05,
+      "loss": 0.7461,
+      "step": 10497
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0470007568399144e-05,
+      "loss": 18.5715,
+      "step": 10498
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.046210850837845e-05,
+      "loss": 18.6303,
+      "step": 10499
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0454209436823416e-05,
+      "loss": 0.797,
+      "step": 10500
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0446310353931225e-05,
+      "loss": 0.7591,
+      "step": 10501
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.043841125989902e-05,
+      "loss": 0.7997,
+      "step": 10502
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.043051215492397e-05,
+      "loss": 0.7435,
+      "step": 10503
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0422613039203236e-05,
+      "loss": 0.7388,
+      "step": 10504
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.041471391293399e-05,
+      "loss": 0.7509,
+      "step": 10505
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0406814776313385e-05,
+      "loss": 0.8454,
+      "step": 10506
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.039891562953859e-05,
+      "loss": 17.0621,
+      "step": 10507
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0391016472806774e-05,
+      "loss": 0.7828,
+      "step": 10508
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.038311730631509e-05,
+      "loss": 15.5691,
+      "step": 10509
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0375218130260705e-05,
+      "loss": 0.7753,
+      "step": 10510
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.03673189448408e-05,
+      "loss": 0.7806,
+      "step": 10511
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.035941975025254e-05,
+      "loss": 0.7762,
+      "step": 10512
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0351520546693064e-05,
+      "loss": 18.4593,
+      "step": 10513
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0343621334359555e-05,
+      "loss": 0.7401,
+      "step": 10514
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.033572211344919e-05,
+      "loss": 0.8364,
+      "step": 10515
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.032782288415912e-05,
+      "loss": 0.7852,
+      "step": 10516
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0319923646686516e-05,
+      "loss": 0.7884,
+      "step": 10517
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.031202440122855e-05,
+      "loss": 18.8093,
+      "step": 10518
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0304125147982375e-05,
+      "loss": 0.7966,
+      "step": 10519
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.029622588714518e-05,
+      "loss": 19.1603,
+      "step": 10520
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0288326618914095e-05,
+      "loss": 0.7886,
+      "step": 10521
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0280427343486335e-05,
+      "loss": 0.755,
+      "step": 10522
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0272528061059034e-05,
+      "loss": 19.0858,
+      "step": 10523
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.026462877182936e-05,
+      "loss": 16.1254,
+      "step": 10524
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.025672947599451e-05,
+      "loss": 18.0922,
+      "step": 10525
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.024883017375164e-05,
+      "loss": 19.489,
+      "step": 10526
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.024093086529789e-05,
+      "loss": 18.994,
+      "step": 10527
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0233031550830455e-05,
+      "loss": 0.715,
+      "step": 10528
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.02251322305465e-05,
+      "loss": 18.2141,
+      "step": 10529
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.021723290464321e-05,
+      "loss": 0.7548,
+      "step": 10530
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.020933357331772e-05,
+      "loss": 15.9698,
+      "step": 10531
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.020143423676722e-05,
+      "loss": 15.3932,
+      "step": 10532
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.019353489518888e-05,
+      "loss": 17.3415,
+      "step": 10533
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0185635548779865e-05,
+      "loss": 0.7787,
+      "step": 10534
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0177736197737325e-05,
+      "loss": 18.275,
+      "step": 10535
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.016983684225848e-05,
+      "loss": 18.8278,
+      "step": 10536
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0161937482540446e-05,
+      "loss": 0.7178,
+      "step": 10537
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.015403811878041e-05,
+      "loss": 19.5026,
+      "step": 10538
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0146138751175565e-05,
+      "loss": 17.8674,
+      "step": 10539
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.013823937992307e-05,
+      "loss": 0.8117,
+      "step": 10540
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.013034000522007e-05,
+      "loss": 18.5415,
+      "step": 10541
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.012244062726375e-05,
+      "loss": 19.2689,
+      "step": 10542
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0114541246251304e-05,
+      "loss": 0.7132,
+      "step": 10543
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0106641862379874e-05,
+      "loss": 18.1009,
+      "step": 10544
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.009874247584663e-05,
+      "loss": 19.5903,
+      "step": 10545
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.009084308684876e-05,
+      "loss": 20.5564,
+      "step": 10546
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0082943695583414e-05,
+      "loss": 0.7604,
+      "step": 10547
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0075044302247794e-05,
+      "loss": 21.7325,
+      "step": 10548
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.006714490703905e-05,
+      "loss": 0.8163,
+      "step": 10549
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.005924551015434e-05,
+      "loss": 17.2838,
+      "step": 10550
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.005134611179085e-05,
+      "loss": 0.7811,
+      "step": 10551
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0043446712145744e-05,
+      "loss": 19.8025,
+      "step": 10552
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.003554731141622e-05,
+      "loss": 0.776,
+      "step": 10553
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0027647909799414e-05,
+      "loss": 0.7658,
+      "step": 10554
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.00197485074925e-05,
+      "loss": 20.8431,
+      "step": 10555
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0011849104692685e-05,
+      "loss": 17.3374,
+      "step": 10556
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0003949701597083e-05,
+      "loss": 15.2459,
+      "step": 10557
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.999605029840292e-05,
+      "loss": 19.9998,
+      "step": 10558
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.998815089530733e-05,
+      "loss": 0.7491,
+      "step": 10559
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.998025149250749e-05,
+      "loss": 19.2081,
+      "step": 10560
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9972352090200604e-05,
+      "loss": 0.7971,
+      "step": 10561
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9964452688583794e-05,
+      "loss": 0.7823,
+      "step": 10562
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.995655328785426e-05,
+      "loss": 19.9501,
+      "step": 10563
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.994865388820916e-05,
+      "loss": 0.7384,
+      "step": 10564
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.994075448984567e-05,
+      "loss": 0.743,
+      "step": 10565
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.993285509296096e-05,
+      "loss": 0.7404,
+      "step": 10566
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.992495569775221e-05,
+      "loss": 0.7963,
+      "step": 10567
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.99170563044166e-05,
+      "loss": 20.3999,
+      "step": 10568
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9909156913151254e-05,
+      "loss": 0.7325,
+      "step": 10569
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9901257524153385e-05,
+      "loss": 18.1504,
+      "step": 10570
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.989335813762014e-05,
+      "loss": 0.7731,
+      "step": 10571
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9885458753748715e-05,
+      "loss": 19.6371,
+      "step": 10572
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.987755937273625e-05,
+      "loss": 18.5332,
+      "step": 10573
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.986965999477993e-05,
+      "loss": 0.7723,
+      "step": 10574
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9861760620076955e-05,
+      "loss": 17.5467,
+      "step": 10575
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9853861248824447e-05,
+      "loss": 17.6025,
+      "step": 10576
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.98459618812196e-05,
+      "loss": 16.251,
+      "step": 10577
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.983806251745957e-05,
+      "loss": 18.2397,
+      "step": 10578
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.983016315774154e-05,
+      "loss": 17.4218,
+      "step": 10579
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.982226380226267e-05,
+      "loss": 0.8325,
+      "step": 10580
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9814364451220153e-05,
+      "loss": 17.5399,
+      "step": 10581
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.980646510481115e-05,
+      "loss": 16.7282,
+      "step": 10582
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.97985657632328e-05,
+      "loss": 0.7665,
+      "step": 10583
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9790666426682294e-05,
+      "loss": 0.7918,
+      "step": 10584
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9782767095356804e-05,
+      "loss": 15.1289,
+      "step": 10585
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9774867769453504e-05,
+      "loss": 0.8122,
+      "step": 10586
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.976696844916955e-05,
+      "loss": 17.9977,
+      "step": 10587
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9759069134702116e-05,
+      "loss": 18.963,
+      "step": 10588
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.975116982624839e-05,
+      "loss": 0.7758,
+      "step": 10589
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9743270524005494e-05,
+      "loss": 0.6674,
+      "step": 10590
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9735371228170644e-05,
+      "loss": 0.7691,
+      "step": 10591
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.972747193894098e-05,
+      "loss": 0.7547,
+      "step": 10592
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9719572656513676e-05,
+      "loss": 0.7936,
+      "step": 10593
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9711673381085896e-05,
+      "loss": 17.5305,
+      "step": 10594
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9703774112854833e-05,
+      "loss": 0.7449,
+      "step": 10595
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.969587485201765e-05,
+      "loss": 0.7853,
+      "step": 10596
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.968797559877147e-05,
+      "loss": 0.8084,
+      "step": 10597
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9680076353313495e-05,
+      "loss": 17.5109,
+      "step": 10598
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.967217711584088e-05,
+      "loss": 17.8691,
+      "step": 10599
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.966427788655082e-05,
+      "loss": 17.4222,
+      "step": 10600
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.965637866564045e-05,
+      "loss": 0.7745,
+      "step": 10601
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9648479453306934e-05,
+      "loss": 0.7275,
+      "step": 10602
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.964058024974748e-05,
+      "loss": 19.4813,
+      "step": 10603
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.963268105515921e-05,
+      "loss": 0.7498,
+      "step": 10604
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9624781869739306e-05,
+      "loss": 17.8985,
+      "step": 10605
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9616882693684923e-05,
+      "loss": 0.7343,
+      "step": 10606
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9608983527193244e-05,
+      "loss": 17.0781,
+      "step": 10607
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.960108437046142e-05,
+      "loss": 0.7479,
+      "step": 10608
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9593185223686626e-05,
+      "loss": 18.7221,
+      "step": 10609
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.958528608706604e-05,
+      "loss": 17.9971,
+      "step": 10610
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9577386960796776e-05,
+      "loss": 0.718,
+      "step": 10611
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.956948784507604e-05,
+      "loss": 0.7619,
+      "step": 10612
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9561588740100984e-05,
+      "loss": 16.0981,
+      "step": 10613
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9553689646068786e-05,
+      "loss": 14.6437,
+      "step": 10614
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.954579056317658e-05,
+      "loss": 0.745,
+      "step": 10615
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.953789149162155e-05,
+      "loss": 16.3487,
+      "step": 10616
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9529992431600874e-05,
+      "loss": 17.9401,
+      "step": 10617
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9522093383311674e-05,
+      "loss": 17.3156,
+      "step": 10618
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9514194346951144e-05,
+      "loss": 20.4368,
+      "step": 10619
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.950629532271642e-05,
+      "loss": 17.9228,
+      "step": 10620
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9498396310804675e-05,
+      "loss": 17.0826,
+      "step": 10621
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9490497311413074e-05,
+      "loss": 16.4185,
+      "step": 10622
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9482598324738785e-05,
+      "loss": 17.454,
+      "step": 10623
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9474699350978984e-05,
+      "loss": 0.793,
+      "step": 10624
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.946680039033077e-05,
+      "loss": 19.5218,
+      "step": 10625
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9458901442991345e-05,
+      "loss": 18.3006,
+      "step": 10626
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.945100250915787e-05,
+      "loss": 19.8526,
+      "step": 10627
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.94431035890275e-05,
+      "loss": 17.0702,
+      "step": 10628
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.943520468279739e-05,
+      "loss": 0.7685,
+      "step": 10629
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9427305790664694e-05,
+      "loss": 0.7645,
+      "step": 10630
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.941940691282659e-05,
+      "loss": 0.7816,
+      "step": 10631
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.941150804948021e-05,
+      "loss": 0.7721,
+      "step": 10632
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.940360920082274e-05,
+      "loss": 19.1169,
+      "step": 10633
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.93957103670513e-05,
+      "loss": 20.054,
+      "step": 10634
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9387811548363064e-05,
+      "loss": 0.7421,
+      "step": 10635
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9379912744955195e-05,
+      "loss": 17.0359,
+      "step": 10636
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9372013957024875e-05,
+      "loss": 0.7807,
+      "step": 10637
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.936411518476921e-05,
+      "loss": 0.7502,
+      "step": 10638
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.935621642838537e-05,
+      "loss": 0.7307,
+      "step": 10639
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9348317688070514e-05,
+      "loss": 0.7544,
+      "step": 10640
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.934041896402179e-05,
+      "loss": 0.7791,
+      "step": 10641
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.933252025643638e-05,
+      "loss": 0.7373,
+      "step": 10642
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9324621565511406e-05,
+      "loss": 0.7666,
+      "step": 10643
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9316722891444045e-05,
+      "loss": 0.7474,
+      "step": 10644
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.930882423443142e-05,
+      "loss": 0.7331,
+      "step": 10645
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.93009255946707e-05,
+      "loss": 0.7813,
+      "step": 10646
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.929302697235905e-05,
+      "loss": 17.6954,
+      "step": 10647
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.92851283676936e-05,
+      "loss": 0.7699,
+      "step": 10648
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.92772297808715e-05,
+      "loss": 18.713,
+      "step": 10649
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9269331212089905e-05,
+      "loss": 0.7564,
+      "step": 10650
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9261432661546e-05,
+      "loss": 0.7305,
+      "step": 10651
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.92535341294369e-05,
+      "loss": 18.3703,
+      "step": 10652
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.924563561595974e-05,
+      "loss": 19.2747,
+      "step": 10653
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.923773712131169e-05,
+      "loss": 18.0276,
+      "step": 10654
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9229838645689893e-05,
+      "loss": 0.7784,
+      "step": 10655
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.922194018929151e-05,
+      "loss": 0.73,
+      "step": 10656
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.921404175231367e-05,
+      "loss": 19.4478,
+      "step": 10657
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.920614333495354e-05,
+      "loss": 0.7622,
+      "step": 10658
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9198244937408245e-05,
+      "loss": 0.7573,
+      "step": 10659
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.919034655987493e-05,
+      "loss": 17.6297,
+      "step": 10660
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.918244820255076e-05,
+      "loss": 0.7628,
+      "step": 10661
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.917454986563287e-05,
+      "loss": 0.7258,
+      "step": 10662
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9166651549318394e-05,
+      "loss": 0.7464,
+      "step": 10663
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9158753253804484e-05,
+      "loss": 17.478,
+      "step": 10664
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.915085497928831e-05,
+      "loss": 17.3647,
+      "step": 10665
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.914295672596698e-05,
+      "loss": 0.7421,
+      "step": 10666
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.913505849403763e-05,
+      "loss": 18.4716,
+      "step": 10667
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9127160283697426e-05,
+      "loss": 0.752,
+      "step": 10668
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9119262095143496e-05,
+      "loss": 0.7734,
+      "step": 10669
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9111363928573e-05,
+      "loss": 18.4606,
+      "step": 10670
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.910346578418305e-05,
+      "loss": 19.7497,
+      "step": 10671
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.909556766217081e-05,
+      "loss": 18.1263,
+      "step": 10672
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.90876695627334e-05,
+      "loss": 16.4634,
+      "step": 10673
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.907977148606796e-05,
+      "loss": 0.7687,
+      "step": 10674
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.907187343237165e-05,
+      "loss": 18.2766,
+      "step": 10675
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.906397540184157e-05,
+      "loss": 16.7357,
+      "step": 10676
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.905607739467489e-05,
+      "loss": 16.5318,
+      "step": 10677
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9048179411068726e-05,
+      "loss": 20.4425,
+      "step": 10678
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9040281451220245e-05,
+      "loss": 19.2759,
+      "step": 10679
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.903238351532653e-05,
+      "loss": 19.8382,
+      "step": 10680
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9024485603584765e-05,
+      "loss": 0.7728,
+      "step": 10681
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9016587716192044e-05,
+      "loss": 0.7664,
+      "step": 10682
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.900868985334552e-05,
+      "loss": 0.7008,
+      "step": 10683
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.9000792015242336e-05,
+      "loss": 18.5201,
+      "step": 10684
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.89928942020796e-05,
+      "loss": 17.0545,
+      "step": 10685
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8984996414054474e-05,
+      "loss": 0.7708,
+      "step": 10686
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.897709865136405e-05,
+      "loss": 18.6328,
+      "step": 10687
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.896920091420548e-05,
+      "loss": 0.753,
+      "step": 10688
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8961303202775896e-05,
+      "loss": 15.8914,
+      "step": 10689
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.895340551727241e-05,
+      "loss": 0.7849,
+      "step": 10690
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.894550785789217e-05,
+      "loss": 0.7647,
+      "step": 10691
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.893761022483228e-05,
+      "loss": 20.3301,
+      "step": 10692
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.892971261828991e-05,
+      "loss": 0.7489,
+      "step": 10693
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.892181503846214e-05,
+      "loss": 18.5148,
+      "step": 10694
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.891391748554612e-05,
+      "loss": 18.0108,
+      "step": 10695
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8906019959738956e-05,
+      "loss": 20.0381,
+      "step": 10696
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8898122461237784e-05,
+      "loss": 18.5222,
+      "step": 10697
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8890224990239744e-05,
+      "loss": 20.236,
+      "step": 10698
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.888232754694192e-05,
+      "loss": 19.2858,
+      "step": 10699
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.887443013154149e-05,
+      "loss": 0.7454,
+      "step": 10700
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.88665327442355e-05,
+      "loss": 0.7699,
+      "step": 10701
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.885863538522113e-05,
+      "loss": 0.7985,
+      "step": 10702
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8850738054695486e-05,
+      "loss": 0.7679,
+      "step": 10703
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.884284075285567e-05,
+      "loss": 0.7833,
+      "step": 10704
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8834943479898815e-05,
+      "loss": 20.9485,
+      "step": 10705
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.882704623602204e-05,
+      "loss": 16.7614,
+      "step": 10706
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8819149021422475e-05,
+      "loss": 18.0429,
+      "step": 10707
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.88112518362972e-05,
+      "loss": 19.4981,
+      "step": 10708
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8803354680843364e-05,
+      "loss": 16.3093,
+      "step": 10709
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.879545755525805e-05,
+      "loss": 18.7632,
+      "step": 10710
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.87875604597384e-05,
+      "loss": 17.4851,
+      "step": 10711
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.877966339448152e-05,
+      "loss": 0.7699,
+      "step": 10712
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.877176635968453e-05,
+      "loss": 0.7802,
+      "step": 10713
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8763869355544523e-05,
+      "loss": 17.1339,
+      "step": 10714
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.875597238225861e-05,
+      "loss": 0.7613,
+      "step": 10715
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.874807544002392e-05,
+      "loss": 18.0424,
+      "step": 10716
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.874017852903755e-05,
+      "loss": 18.3528,
+      "step": 10717
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8732281649496606e-05,
+      "loss": 18.3753,
+      "step": 10718
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.87243848015982e-05,
+      "loss": 0.7358,
+      "step": 10719
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.871648798553947e-05,
+      "loss": 0.7916,
+      "step": 10720
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8708591201517465e-05,
+      "loss": 0.7145,
+      "step": 10721
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8700694449729314e-05,
+      "loss": 0.811,
+      "step": 10722
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8692797730372144e-05,
+      "loss": 0.7695,
+      "step": 10723
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.868490104364303e-05,
+      "loss": 17.905,
+      "step": 10724
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.867700438973908e-05,
+      "loss": 17.5479,
+      "step": 10725
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.866910776885743e-05,
+      "loss": 17.9587,
+      "step": 10726
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.866121118119513e-05,
+      "loss": 0.7866,
+      "step": 10727
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.865331462694932e-05,
+      "loss": 0.8108,
+      "step": 10728
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.864541810631707e-05,
+      "loss": 0.7445,
+      "step": 10729
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.86375216194955e-05,
+      "loss": 0.7873,
+      "step": 10730
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8629625166681706e-05,
+      "loss": 19.0877,
+      "step": 10731
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8621728748072766e-05,
+      "loss": 18.2224,
+      "step": 10732
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.861383236386579e-05,
+      "loss": 0.8245,
+      "step": 10733
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.860593601425791e-05,
+      "loss": 0.7734,
+      "step": 10734
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8598039699446155e-05,
+      "loss": 0.7412,
+      "step": 10735
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.859014341962764e-05,
+      "loss": 17.612,
+      "step": 10736
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.858224717499949e-05,
+      "loss": 18.2772,
+      "step": 10737
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.857435096575876e-05,
+      "loss": 18.1035,
+      "step": 10738
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.856645479210255e-05,
+      "loss": 20.0562,
+      "step": 10739
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.855855865422798e-05,
+      "loss": 16.9492,
+      "step": 10740
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.855066255233211e-05,
+      "loss": 18.7146,
+      "step": 10741
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.854276648661203e-05,
+      "loss": 0.7423,
+      "step": 10742
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.853487045726482e-05,
+      "loss": 17.0641,
+      "step": 10743
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.852697446448758e-05,
+      "loss": 16.9293,
+      "step": 10744
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.85190785084774e-05,
+      "loss": 0.8,
+      "step": 10745
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.851118258943136e-05,
+      "loss": 0.7667,
+      "step": 10746
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.850328670754653e-05,
+      "loss": 0.7563,
+      "step": 10747
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.849539086302003e-05,
+      "loss": 0.7265,
+      "step": 10748
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.84874950560489e-05,
+      "loss": 0.7665,
+      "step": 10749
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.847959928683024e-05,
+      "loss": 18.5783,
+      "step": 10750
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.847170355556113e-05,
+      "loss": 18.5824,
+      "step": 10751
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.846380786243865e-05,
+      "loss": 0.7204,
+      "step": 10752
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.845591220765986e-05,
+      "loss": 19.4337,
+      "step": 10753
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.844801659142188e-05,
+      "loss": 0.7897,
+      "step": 10754
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.844012101392176e-05,
+      "loss": 18.672,
+      "step": 10755
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8432225475356566e-05,
+      "loss": 0.7219,
+      "step": 10756
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.842432997592338e-05,
+      "loss": 18.7678,
+      "step": 10757
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.841643451581927e-05,
+      "loss": 0.7745,
+      "step": 10758
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.840853909524133e-05,
+      "loss": 19.5124,
+      "step": 10759
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.840064371438661e-05,
+      "loss": 0.7519,
+      "step": 10760
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8392748373452184e-05,
+      "loss": 0.7727,
+      "step": 10761
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.838485307263515e-05,
+      "loss": 16.1062,
+      "step": 10762
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.837695781213253e-05,
+      "loss": 0.7728,
+      "step": 10763
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.836906259214141e-05,
+      "loss": 0.7844,
+      "step": 10764
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.836116741285887e-05,
+      "loss": 19.453,
+      "step": 10765
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.835327227448196e-05,
+      "loss": 18.2402,
+      "step": 10766
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.8345377177207746e-05,
+      "loss": 0.8015,
+      "step": 10767
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.833748212123331e-05,
+      "loss": 0.7309,
+      "step": 10768
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.83295871067557e-05,
+      "loss": 0.7975,
+      "step": 10769
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.8321692133971976e-05,
+      "loss": 16.6534,
+      "step": 10770
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.831379720307919e-05,
+      "loss": 0.7192,
+      "step": 10771
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.8305902314274404e-05,
+      "loss": 0.7706,
+      "step": 10772
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.8298007467754704e-05,
+      "loss": 17.2859,
+      "step": 10773
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.829011266371711e-05,
+      "loss": 0.7342,
+      "step": 10774
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.828221790235869e-05,
+      "loss": 20.0548,
+      "step": 10775
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.827432318387653e-05,
+      "loss": 0.7517,
+      "step": 10776
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.8266428508467636e-05,
+      "loss": 0.7649,
+      "step": 10777
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.825853387632908e-05,
+      "loss": 18.2523,
+      "step": 10778
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.825063928765793e-05,
+      "loss": 17.2082,
+      "step": 10779
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.8242744742651205e-05,
+      "loss": 0.7924,
+      "step": 10780
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.823485024150598e-05,
+      "loss": 18.5958,
+      "step": 10781
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.822695578441931e-05,
+      "loss": 19.4064,
+      "step": 10782
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.821906137158822e-05,
+      "loss": 18.6864,
+      "step": 10783
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.821116700320977e-05,
+      "loss": 16.5525,
+      "step": 10784
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.820327267948099e-05,
+      "loss": 20.9235,
+      "step": 10785
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.819537840059893e-05,
+      "loss": 0.7688,
+      "step": 10786
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.8187484166760643e-05,
+      "loss": 16.992,
+      "step": 10787
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.817958997816317e-05,
+      "loss": 19.4756,
+      "step": 10788
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.817169583500355e-05,
+      "loss": 0.7184,
+      "step": 10789
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.816380173747881e-05,
+      "loss": 17.5544,
+      "step": 10790
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.815590768578599e-05,
+      "loss": 0.7545,
+      "step": 10791
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.8148013680122147e-05,
+      "loss": 19.4523,
+      "step": 10792
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.8140119720684304e-05,
+      "loss": 0.7948,
+      "step": 10793
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.81322258076695e-05,
+      "loss": 17.2895,
+      "step": 10794
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.8124331941274755e-05,
+      "loss": 0.8011,
+      "step": 10795
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.8116438121697136e-05,
+      "loss": 0.7687,
+      "step": 10796
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.8108544349133634e-05,
+      "loss": 0.7525,
+      "step": 10797
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.81006506237813e-05,
+      "loss": 16.6267,
+      "step": 10798
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.809275694583716e-05,
+      "loss": 0.7585,
+      "step": 10799
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.8084863315498234e-05,
+      "loss": 0.7223,
+      "step": 10800
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.8076969732961566e-05,
+      "loss": 0.743,
+      "step": 10801
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.806907619842417e-05,
+      "loss": 17.093,
+      "step": 10802
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.806118271208309e-05,
+      "loss": 17.9403,
+      "step": 10803
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.8053289274135315e-05,
+      "loss": 0.8042,
+      "step": 10804
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.804539588477788e-05,
+      "loss": 17.6109,
+      "step": 10805
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.803750254420781e-05,
+      "loss": 0.7901,
+      "step": 10806
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.802960925262214e-05,
+      "loss": 0.7413,
+      "step": 10807
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.802171601021785e-05,
+      "loss": 0.7871,
+      "step": 10808
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.801382281719199e-05,
+      "loss": 19.3249,
+      "step": 10809
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.800592967374159e-05,
+      "loss": 0.7488,
+      "step": 10810
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.799803658006361e-05,
+      "loss": 16.6818,
+      "step": 10811
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.799014353635512e-05,
+      "loss": 18.1946,
+      "step": 10812
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.798225054281308e-05,
+      "loss": 21.0083,
+      "step": 10813
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.797435759963453e-05,
+      "loss": 17.2181,
+      "step": 10814
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.796646470701649e-05,
+      "loss": 0.8045,
+      "step": 10815
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.795857186515594e-05,
+      "loss": 0.7275,
+      "step": 10816
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.795067907424993e-05,
+      "loss": 0.7265,
+      "step": 10817
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.794278633449542e-05,
+      "loss": 0.7407,
+      "step": 10818
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7934893646089425e-05,
+      "loss": 19.5419,
+      "step": 10819
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.792700100922896e-05,
+      "loss": 0.7582,
+      "step": 10820
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.791910842411103e-05,
+      "loss": 0.7266,
+      "step": 10821
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.791121589093262e-05,
+      "loss": 18.224,
+      "step": 10822
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7903323409890745e-05,
+      "loss": 18.1352,
+      "step": 10823
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.789543098118241e-05,
+      "loss": 0.7897,
+      "step": 10824
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.788753860500458e-05,
+      "loss": 0.7474,
+      "step": 10825
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.787964628155428e-05,
+      "loss": 18.0951,
+      "step": 10826
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.787175401102848e-05,
+      "loss": 17.5736,
+      "step": 10827
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.786386179362419e-05,
+      "loss": 18.4549,
+      "step": 10828
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.78559696295384e-05,
+      "loss": 16.5353,
+      "step": 10829
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.784807751896809e-05,
+      "loss": 0.7778,
+      "step": 10830
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7840185462110285e-05,
+      "loss": 18.1613,
+      "step": 10831
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.783229345916192e-05,
+      "loss": 0.7472,
+      "step": 10832
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7824401510319995e-05,
+      "loss": 18.0466,
+      "step": 10833
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.781650961578151e-05,
+      "loss": 17.9899,
+      "step": 10834
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.780861777574346e-05,
+      "loss": 16.104,
+      "step": 10835
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7800725990402794e-05,
+      "loss": 17.0172,
+      "step": 10836
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.779283425995651e-05,
+      "loss": 0.7781,
+      "step": 10837
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7784942584601614e-05,
+      "loss": 17.1942,
+      "step": 10838
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.777705096453503e-05,
+      "loss": 0.7779,
+      "step": 10839
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.776915939995377e-05,
+      "loss": 17.9359,
+      "step": 10840
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.77612678910548e-05,
+      "loss": 0.7514,
+      "step": 10841
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.775337643803508e-05,
+      "loss": 16.9776,
+      "step": 10842
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.774548504109161e-05,
+      "loss": 0.8283,
+      "step": 10843
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.773759370042136e-05,
+      "loss": 18.1165,
+      "step": 10844
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.772970241622129e-05,
+      "loss": 22.8965,
+      "step": 10845
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.772181118868835e-05,
+      "loss": 0.7349,
+      "step": 10846
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.771392001801952e-05,
+      "loss": 17.6923,
+      "step": 10847
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.770602890441176e-05,
+      "loss": 17.7465,
+      "step": 10848
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.769813784806207e-05,
+      "loss": 0.7759,
+      "step": 10849
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.769024684916736e-05,
+      "loss": 17.9024,
+      "step": 10850
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.768235590792463e-05,
+      "loss": 0.7317,
+      "step": 10851
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7674465024530836e-05,
+      "loss": 0.8062,
+      "step": 10852
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7666574199182905e-05,
+      "loss": 17.3181,
+      "step": 10853
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7658683432077834e-05,
+      "loss": 17.6961,
+      "step": 10854
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.765079272341254e-05,
+      "loss": 0.7698,
+      "step": 10855
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7642902073384e-05,
+      "loss": 0.7891,
+      "step": 10856
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7635011482189156e-05,
+      "loss": 19.6886,
+      "step": 10857
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.762712095002498e-05,
+      "loss": 0.7452,
+      "step": 10858
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.761923047708842e-05,
+      "loss": 0.7842,
+      "step": 10859
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7611340063576394e-05,
+      "loss": 16.9384,
+      "step": 10860
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7603449709685857e-05,
+      "loss": 17.6163,
+      "step": 10861
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.759555941561377e-05,
+      "loss": 0.7547,
+      "step": 10862
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7587669181557074e-05,
+      "loss": 0.7743,
+      "step": 10863
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.757977900771269e-05,
+      "loss": 18.8923,
+      "step": 10864
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7571888894277604e-05,
+      "loss": 0.8094,
+      "step": 10865
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7563998841448705e-05,
+      "loss": 0.817,
+      "step": 10866
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.755610884942295e-05,
+      "loss": 0.757,
+      "step": 10867
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.754821891839728e-05,
+      "loss": 0.7719,
+      "step": 10868
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.754032904856862e-05,
+      "loss": 20.4181,
+      "step": 10869
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.753243924013391e-05,
+      "loss": 0.7674,
+      "step": 10870
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.752454949329008e-05,
+      "loss": 18.1361,
+      "step": 10871
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.751665980823408e-05,
+      "loss": 0.7448,
+      "step": 10872
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.750877018516281e-05,
+      "loss": 0.7348,
+      "step": 10873
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7500880624273185e-05,
+      "loss": 0.8329,
+      "step": 10874
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.749299112576216e-05,
+      "loss": 0.7534,
+      "step": 10875
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.748510168982665e-05,
+      "loss": 0.8101,
+      "step": 10876
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.747721231666358e-05,
+      "loss": 19.9238,
+      "step": 10877
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7469323006469854e-05,
+      "loss": 17.8434,
+      "step": 10878
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.746143375944244e-05,
+      "loss": 18.2688,
+      "step": 10879
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7453544575778184e-05,
+      "loss": 0.7804,
+      "step": 10880
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7445655455674034e-05,
+      "loss": 0.7189,
+      "step": 10881
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.743776639932693e-05,
+      "loss": 0.7317,
+      "step": 10882
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7429877406933745e-05,
+      "loss": 0.781,
+      "step": 10883
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.74219884786914e-05,
+      "loss": 18.9954,
+      "step": 10884
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.741409961479682e-05,
+      "loss": 17.0755,
+      "step": 10885
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.740621081544693e-05,
+      "loss": 17.2255,
+      "step": 10886
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7398322080838585e-05,
+      "loss": 18.7433,
+      "step": 10887
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7390433411168715e-05,
+      "loss": 0.7156,
+      "step": 10888
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.738254480663422e-05,
+      "loss": 0.7488,
+      "step": 10889
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7374656267432e-05,
+      "loss": 0.7846,
+      "step": 10890
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.736676779375898e-05,
+      "loss": 19.1119,
+      "step": 10891
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7358879385812016e-05,
+      "loss": 17.6194,
+      "step": 10892
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.735099104378805e-05,
+      "loss": 0.7292,
+      "step": 10893
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7343102767883935e-05,
+      "loss": 20.1575,
+      "step": 10894
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.733521455829658e-05,
+      "loss": 19.2417,
+      "step": 10895
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.732732641522289e-05,
+      "loss": 0.743,
+      "step": 10896
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.731943833885973e-05,
+      "loss": 19.3209,
+      "step": 10897
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7311550329404e-05,
+      "loss": 0.7297,
+      "step": 10898
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.730366238705259e-05,
+      "loss": 16.3685,
+      "step": 10899
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.72957745120024e-05,
+      "loss": 0.7246,
+      "step": 10900
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7287886704450285e-05,
+      "loss": 18.0094,
+      "step": 10901
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7279998964593134e-05,
+      "loss": 17.244,
+      "step": 10902
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.727211129262782e-05,
+      "loss": 17.5588,
+      "step": 10903
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.726422368875124e-05,
+      "loss": 0.8103,
+      "step": 10904
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.725633615316026e-05,
+      "loss": 16.4095,
+      "step": 10905
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.724844868605176e-05,
+      "loss": 0.7566,
+      "step": 10906
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.724056128762262e-05,
+      "loss": 17.7779,
+      "step": 10907
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.723267395806968e-05,
+      "loss": 0.746,
+      "step": 10908
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.722478669758984e-05,
+      "loss": 0.7655,
+      "step": 10909
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.721689950637996e-05,
+      "loss": 18.2056,
+      "step": 10910
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.720901238463689e-05,
+      "loss": 0.733,
+      "step": 10911
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.720112533255751e-05,
+      "loss": 0.7062,
+      "step": 10912
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7193238350338686e-05,
+      "loss": 19.1462,
+      "step": 10913
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7185351438177296e-05,
+      "loss": 16.467,
+      "step": 10914
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.717746459627016e-05,
+      "loss": 19.6682,
+      "step": 10915
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.716957782481414e-05,
+      "loss": 0.7759,
+      "step": 10916
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7161691124006106e-05,
+      "loss": 0.7306,
+      "step": 10917
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.715380449404292e-05,
+      "loss": 19.9413,
+      "step": 10918
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.714591793512142e-05,
+      "loss": 0.7802,
+      "step": 10919
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.713803144743845e-05,
+      "loss": 17.6322,
+      "step": 10920
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.71301450311909e-05,
+      "loss": 19.2325,
+      "step": 10921
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.712225868657555e-05,
+      "loss": 16.9134,
+      "step": 10922
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.711437241378929e-05,
+      "loss": 17.5061,
+      "step": 10923
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7106486213028965e-05,
+      "loss": 16.8067,
+      "step": 10924
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.70986000844914e-05,
+      "loss": 19.9336,
+      "step": 10925
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.709071402837342e-05,
+      "loss": 17.9291,
+      "step": 10926
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7082828044871895e-05,
+      "loss": 0.7557,
+      "step": 10927
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.707494213418367e-05,
+      "loss": 18.3583,
+      "step": 10928
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7067056296505545e-05,
+      "loss": 17.2863,
+      "step": 10929
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.705917053203435e-05,
+      "loss": 17.3037,
+      "step": 10930
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7051284840966935e-05,
+      "loss": 19.045,
+      "step": 10931
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.704339922350012e-05,
+      "loss": 17.3701,
+      "step": 10932
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7035513679830754e-05,
+      "loss": 16.3653,
+      "step": 10933
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7027628210155623e-05,
+      "loss": 18.8944,
+      "step": 10934
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7019742814671594e-05,
+      "loss": 0.7725,
+      "step": 10935
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.701185749357545e-05,
+      "loss": 0.771,
+      "step": 10936
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.7003972247064016e-05,
+      "loss": 19.048,
+      "step": 10937
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6996087075334135e-05,
+      "loss": 0.746,
+      "step": 10938
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.69882019785826e-05,
+      "loss": 18.047,
+      "step": 10939
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.698031695700622e-05,
+      "loss": 0.7993,
+      "step": 10940
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.697243201080185e-05,
+      "loss": 0.7673,
+      "step": 10941
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6964547140166235e-05,
+      "loss": 19.823,
+      "step": 10942
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.695666234529624e-05,
+      "loss": 19.5921,
+      "step": 10943
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.694877762638863e-05,
+      "loss": 0.8047,
+      "step": 10944
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.694089298364023e-05,
+      "loss": 0.7457,
+      "step": 10945
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6933008417247835e-05,
+      "loss": 16.0428,
+      "step": 10946
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.692512392740827e-05,
+      "loss": 18.2401,
+      "step": 10947
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6917239514318316e-05,
+      "loss": 17.8733,
+      "step": 10948
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.690935517817475e-05,
+      "loss": 0.7793,
+      "step": 10949
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.690147091917438e-05,
+      "loss": 0.7991,
+      "step": 10950
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.689358673751401e-05,
+      "loss": 0.7794,
+      "step": 10951
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6885702633390424e-05,
+      "loss": 18.9088,
+      "step": 10952
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6877818607000415e-05,
+      "loss": 21.4187,
+      "step": 10953
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.686993465854076e-05,
+      "loss": 0.7721,
+      "step": 10954
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6862050788208274e-05,
+      "loss": 20.1121,
+      "step": 10955
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.685416699619969e-05,
+      "loss": 17.1903,
+      "step": 10956
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.684628328271183e-05,
+      "loss": 18.6239,
+      "step": 10957
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.683839964794145e-05,
+      "loss": 0.7179,
+      "step": 10958
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.683051609208534e-05,
+      "loss": 18.634,
+      "step": 10959
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.682263261534026e-05,
+      "loss": 18.3208,
+      "step": 10960
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.681474921790301e-05,
+      "loss": 19.1938,
+      "step": 10961
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6806865899970363e-05,
+      "loss": 17.8278,
+      "step": 10962
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6798982661739057e-05,
+      "loss": 16.8913,
+      "step": 10963
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.679109950340586e-05,
+      "loss": 0.7101,
+      "step": 10964
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.678321642516756e-05,
+      "loss": 0.7822,
+      "step": 10965
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6775333427220924e-05,
+      "loss": 16.6026,
+      "step": 10966
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.676745050976269e-05,
+      "loss": 0.7641,
+      "step": 10967
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.675956767298963e-05,
+      "loss": 17.4496,
+      "step": 10968
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6751684917098514e-05,
+      "loss": 0.7552,
+      "step": 10969
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6743802242286073e-05,
+      "loss": 19.0273,
+      "step": 10970
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.673591964874908e-05,
+      "loss": 0.7686,
+      "step": 10971
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.672803713668426e-05,
+      "loss": 0.7707,
+      "step": 10972
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.672015470628839e-05,
+      "loss": 17.9629,
+      "step": 10973
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.67122723577582e-05,
+      "loss": 0.7822,
+      "step": 10974
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.670439009129046e-05,
+      "loss": 0.7778,
+      "step": 10975
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.66965079070819e-05,
+      "loss": 0.7768,
+      "step": 10976
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.668862580532924e-05,
+      "loss": 0.7223,
+      "step": 10977
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6680743786229235e-05,
+      "loss": 0.7624,
+      "step": 10978
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.667286184997862e-05,
+      "loss": 0.7484,
+      "step": 10979
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.666497999677415e-05,
+      "loss": 0.7572,
+      "step": 10980
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6657098226812525e-05,
+      "loss": 19.0637,
+      "step": 10981
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.664921654029049e-05,
+      "loss": 0.7596,
+      "step": 10982
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.66413349374048e-05,
+      "loss": 0.7284,
+      "step": 10983
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.663345341835213e-05,
+      "loss": 18.1379,
+      "step": 10984
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.662557198332925e-05,
+      "loss": 16.9601,
+      "step": 10985
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.661769063253285e-05,
+      "loss": 19.095,
+      "step": 10986
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6609809366159665e-05,
+      "loss": 19.4606,
+      "step": 10987
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.660192818440641e-05,
+      "loss": 17.0462,
+      "step": 10988
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.659404708746982e-05,
+      "loss": 0.7339,
+      "step": 10989
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.65861660755466e-05,
+      "loss": 0.7861,
+      "step": 10990
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6578285148833426e-05,
+      "loss": 21.2548,
+      "step": 10991
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.657040430752704e-05,
+      "loss": 0.778,
+      "step": 10992
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6562523551824153e-05,
+      "loss": 0.8182,
+      "step": 10993
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6554642881921467e-05,
+      "loss": 18.2147,
+      "step": 10994
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.654676229801567e-05,
+      "loss": 0.817,
+      "step": 10995
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.653888180030348e-05,
+      "loss": 18.2926,
+      "step": 10996
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.653100138898161e-05,
+      "loss": 17.552,
+      "step": 10997
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.652312106424671e-05,
+      "loss": 18.986,
+      "step": 10998
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6515240826295515e-05,
+      "loss": 18.3868,
+      "step": 10999
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.65073606753247e-05,
+      "loss": 16.9625,
+      "step": 11000
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.649948061153096e-05,
+      "loss": 19.0168,
+      "step": 11001
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.649160063511097e-05,
+      "loss": 0.6911,
+      "step": 11002
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.648372074626145e-05,
+      "loss": 0.7237,
+      "step": 11003
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.647584094517907e-05,
+      "loss": 0.6983,
+      "step": 11004
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.646796123206049e-05,
+      "loss": 17.4747,
+      "step": 11005
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6460081607102395e-05,
+      "loss": 20.47,
+      "step": 11006
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6452202070501474e-05,
+      "loss": 17.7656,
+      "step": 11007
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.644432262245442e-05,
+      "loss": 16.9704,
+      "step": 11008
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.643644326315786e-05,
+      "loss": 18.4711,
+      "step": 11009
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.64285639928085e-05,
+      "loss": 0.7632,
+      "step": 11010
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.642068481160301e-05,
+      "loss": 20.2097,
+      "step": 11011
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6412805719738035e-05,
+      "loss": 18.7197,
+      "step": 11012
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.640492671741025e-05,
+      "loss": 18.2899,
+      "step": 11013
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6397047804816315e-05,
+      "loss": 17.1069,
+      "step": 11014
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.638916898215288e-05,
+      "loss": 19.1483,
+      "step": 11015
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6381290249616613e-05,
+      "loss": 0.7899,
+      "step": 11016
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6373411607404185e-05,
+      "loss": 17.4707,
+      "step": 11017
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.636553305571223e-05,
+      "loss": 17.2363,
+      "step": 11018
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.635765459473738e-05,
+      "loss": 18.3554,
+      "step": 11019
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.634977622467631e-05,
+      "loss": 18.3234,
+      "step": 11020
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.634189794572565e-05,
+      "loss": 17.2578,
+      "step": 11021
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.633401975808207e-05,
+      "loss": 16.486,
+      "step": 11022
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.632614166194219e-05,
+      "loss": 18.7242,
+      "step": 11023
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.631826365750267e-05,
+      "loss": 17.0734,
+      "step": 11024
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.63103857449601e-05,
+      "loss": 0.7585,
+      "step": 11025
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.630250792451115e-05,
+      "loss": 15.1753,
+      "step": 11026
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.629463019635245e-05,
+      "loss": 19.1024,
+      "step": 11027
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.628675256068063e-05,
+      "loss": 19.314,
+      "step": 11028
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.627887501769231e-05,
+      "loss": 0.761,
+      "step": 11029
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.627099756758411e-05,
+      "loss": 0.7835,
+      "step": 11030
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.626312021055269e-05,
+      "loss": 16.9081,
+      "step": 11031
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.625524294679463e-05,
+      "loss": 17.2384,
+      "step": 11032
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6247365776506556e-05,
+      "loss": 17.3465,
+      "step": 11033
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.623948869988508e-05,
+      "loss": 19.5864,
+      "step": 11034
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6231611717126834e-05,
+      "loss": 16.2014,
+      "step": 11035
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.622373482842842e-05,
+      "loss": 0.7364,
+      "step": 11036
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.621585803398645e-05,
+      "loss": 17.513,
+      "step": 11037
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6207981333997536e-05,
+      "loss": 0.7444,
+      "step": 11038
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.620010472865826e-05,
+      "loss": 0.7928,
+      "step": 11039
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.619222821816523e-05,
+      "loss": 17.4617,
+      "step": 11040
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6184351802715076e-05,
+      "loss": 17.3447,
+      "step": 11041
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.617647548250435e-05,
+      "loss": 0.6732,
+      "step": 11042
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.616859925772967e-05,
+      "loss": 18.1603,
+      "step": 11043
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6160723128587634e-05,
+      "loss": 0.7272,
+      "step": 11044
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.615284709527484e-05,
+      "loss": 17.7535,
+      "step": 11045
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.614497115798786e-05,
+      "loss": 0.782,
+      "step": 11046
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.613709531692326e-05,
+      "loss": 18.3086,
+      "step": 11047
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.612921957227765e-05,
+      "loss": 19.2211,
+      "step": 11048
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6121343924247596e-05,
+      "loss": 0.8165,
+      "step": 11049
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.61134683730297e-05,
+      "loss": 17.3016,
+      "step": 11050
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.610559291882051e-05,
+      "loss": 18.6483,
+      "step": 11051
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.609771756181662e-05,
+      "loss": 0.7745,
+      "step": 11052
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.608984230221458e-05,
+      "loss": 19.3355,
+      "step": 11053
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.608196714021097e-05,
+      "loss": 17.4383,
+      "step": 11054
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.607409207600236e-05,
+      "loss": 0.7969,
+      "step": 11055
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6066217109785305e-05,
+      "loss": 16.325,
+      "step": 11056
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6058342241756357e-05,
+      "loss": 20.1523,
+      "step": 11057
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.6050467472112094e-05,
+      "loss": 0.7064,
+      "step": 11058
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.604259280104909e-05,
+      "loss": 0.7726,
+      "step": 11059
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.603471822876386e-05,
+      "loss": 18.7545,
+      "step": 11060
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.602684375545295e-05,
+      "loss": 0.8123,
+      "step": 11061
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.601896938131294e-05,
+      "loss": 16.9265,
+      "step": 11062
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.601109510654035e-05,
+      "loss": 19.7452,
+      "step": 11063
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.600322093133176e-05,
+      "loss": 20.0563,
+      "step": 11064
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.599534685588367e-05,
+      "loss": 18.2799,
+      "step": 11065
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.598747288039266e-05,
+      "loss": 0.7342,
+      "step": 11066
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.597959900505522e-05,
+      "loss": 0.7507,
+      "step": 11067
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5971725230067905e-05,
+      "loss": 18.3619,
+      "step": 11068
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.596385155562726e-05,
+      "loss": 18.6457,
+      "step": 11069
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.59559779819298e-05,
+      "loss": 18.7774,
+      "step": 11070
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.594810450917204e-05,
+      "loss": 0.7672,
+      "step": 11071
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.594023113755053e-05,
+      "loss": 17.8264,
+      "step": 11072
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.593235786726179e-05,
+      "loss": 0.7118,
+      "step": 11073
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5924484698502305e-05,
+      "loss": 17.5231,
+      "step": 11074
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.591661163146862e-05,
+      "loss": 0.7589,
+      "step": 11075
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5908738666357235e-05,
+      "loss": 0.7379,
+      "step": 11076
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.590086580336467e-05,
+      "loss": 20.027,
+      "step": 11077
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5892993042687435e-05,
+      "loss": 0.7849,
+      "step": 11078
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.588512038452203e-05,
+      "loss": 18.457,
+      "step": 11079
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.587724782906497e-05,
+      "loss": 0.759,
+      "step": 11080
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5869375376512735e-05,
+      "loss": 0.7787,
+      "step": 11081
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5861503027061825e-05,
+      "loss": 17.7101,
+      "step": 11082
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.585363078090876e-05,
+      "loss": 0.8095,
+      "step": 11083
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5845758638250004e-05,
+      "loss": 17.4581,
+      "step": 11084
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5837886599282057e-05,
+      "loss": 18.0172,
+      "step": 11085
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.583001466420142e-05,
+      "loss": 0.7484,
+      "step": 11086
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.582214283320458e-05,
+      "loss": 18.0259,
+      "step": 11087
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5814271106488e-05,
+      "loss": 21.0509,
+      "step": 11088
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.580639948424817e-05,
+      "loss": 0.8056,
+      "step": 11089
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.579852796668157e-05,
+      "loss": 17.5004,
+      "step": 11090
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.579065655398465e-05,
+      "loss": 0.7092,
+      "step": 11091
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5782785246353926e-05,
+      "loss": 0.7738,
+      "step": 11092
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5774914043985844e-05,
+      "loss": 18.9611,
+      "step": 11093
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.576704294707687e-05,
+      "loss": 17.2784,
+      "step": 11094
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.575917195582347e-05,
+      "loss": 0.7494,
+      "step": 11095
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5751301070422095e-05,
+      "loss": 20.0023,
+      "step": 11096
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.574343029106923e-05,
+      "loss": 16.934,
+      "step": 11097
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.57355596179613e-05,
+      "loss": 17.0217,
+      "step": 11098
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.572768905129478e-05,
+      "loss": 20.166,
+      "step": 11099
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5719818591266136e-05,
+      "loss": 0.7278,
+      "step": 11100
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5711948238071786e-05,
+      "loss": 0.7617,
+      "step": 11101
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.570407799190818e-05,
+      "loss": 0.7764,
+      "step": 11102
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.569620785297178e-05,
+      "loss": 0.7033,
+      "step": 11103
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5688337821459e-05,
+      "loss": 0.763,
+      "step": 11104
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.568046789756629e-05,
+      "loss": 0.7527,
+      "step": 11105
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.567259808149011e-05,
+      "loss": 17.5425,
+      "step": 11106
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.566472837342686e-05,
+      "loss": 16.762,
+      "step": 11107
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.565685877357298e-05,
+      "loss": 17.06,
+      "step": 11108
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5648989282124895e-05,
+      "loss": 19.1106,
+      "step": 11109
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.564111989927902e-05,
+      "loss": 18.3716,
+      "step": 11110
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5633250625231806e-05,
+      "loss": 0.7317,
+      "step": 11111
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.562538146017964e-05,
+      "loss": 17.201,
+      "step": 11112
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5617512404318954e-05,
+      "loss": 17.4479,
+      "step": 11113
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.560964345784618e-05,
+      "loss": 15.5599,
+      "step": 11114
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.560177462095768e-05,
+      "loss": 0.7751,
+      "step": 11115
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.55939058938499e-05,
+      "loss": 0.8063,
+      "step": 11116
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5586037276719235e-05,
+      "loss": 19.5788,
+      "step": 11117
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.557816876976208e-05,
+      "loss": 18.4332,
+      "step": 11118
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5570300373174846e-05,
+      "loss": 17.8514,
+      "step": 11119
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5562432087153934e-05,
+      "loss": 0.7048,
+      "step": 11120
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.555456391189573e-05,
+      "loss": 18.2506,
+      "step": 11121
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5546695847596626e-05,
+      "loss": 0.7887,
+      "step": 11122
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5538827894453e-05,
+      "loss": 17.0739,
+      "step": 11123
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.553096005266125e-05,
+      "loss": 17.3009,
+      "step": 11124
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.552309232241777e-05,
+      "loss": 0.746,
+      "step": 11125
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.551522470391891e-05,
+      "loss": 0.6905,
+      "step": 11126
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5507357197361075e-05,
+      "loss": 0.7924,
+      "step": 11127
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5499489802940644e-05,
+      "loss": 17.6747,
+      "step": 11128
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.549162252085395e-05,
+      "loss": 18.2017,
+      "step": 11129
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.54837553512974e-05,
+      "loss": 18.4763,
+      "step": 11130
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.547588829446735e-05,
+      "loss": 17.2899,
+      "step": 11131
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5468021350560154e-05,
+      "loss": 18.5177,
+      "step": 11132
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5460154519772176e-05,
+      "loss": 17.5626,
+      "step": 11133
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5452287802299785e-05,
+      "loss": 0.7654,
+      "step": 11134
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5444421198339336e-05,
+      "loss": 19.8243,
+      "step": 11135
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.543655470808717e-05,
+      "loss": 17.9857,
+      "step": 11136
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5428688331739635e-05,
+      "loss": 0.7158,
+      "step": 11137
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.542082206949308e-05,
+      "loss": 18.756,
+      "step": 11138
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5412955921543865e-05,
+      "loss": 0.7211,
+      "step": 11139
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.54050898880883e-05,
+      "loss": 17.9926,
+      "step": 11140
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.539722396932274e-05,
+      "loss": 19.744,
+      "step": 11141
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5389358165443554e-05,
+      "loss": 0.8099,
+      "step": 11142
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.538149247664701e-05,
+      "loss": 0.8366,
+      "step": 11143
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.537362690312947e-05,
+      "loss": 0.7568,
+      "step": 11144
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.536576144508727e-05,
+      "loss": 18.9331,
+      "step": 11145
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5357896102716715e-05,
+      "loss": 16.4638,
+      "step": 11146
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.535003087621413e-05,
+      "loss": 17.8493,
+      "step": 11147
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.534216576577585e-05,
+      "loss": 0.7459,
+      "step": 11148
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.533430077159817e-05,
+      "loss": 18.8763,
+      "step": 11149
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.532643589387741e-05,
+      "loss": 17.3826,
+      "step": 11150
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5318571132809864e-05,
+      "loss": 0.7723,
+      "step": 11151
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5310706488591856e-05,
+      "loss": 0.7805,
+      "step": 11152
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.530284196141969e-05,
+      "loss": 0.7396,
+      "step": 11153
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.529497755148966e-05,
+      "loss": 18.3764,
+      "step": 11154
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.528711325899806e-05,
+      "loss": 0.8046,
+      "step": 11155
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.527924908414121e-05,
+      "loss": 0.7627,
+      "step": 11156
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.527138502711536e-05,
+      "loss": 0.7797,
+      "step": 11157
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.526352108811682e-05,
+      "loss": 18.8068,
+      "step": 11158
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5255657267341884e-05,
+      "loss": 18.8752,
+      "step": 11159
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.524779356498681e-05,
+      "loss": 17.1561,
+      "step": 11160
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5239929981247906e-05,
+      "loss": 17.9856,
+      "step": 11161
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.523206651632144e-05,
+      "loss": 16.6407,
+      "step": 11162
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.522420317040368e-05,
+      "loss": 0.7735,
+      "step": 11163
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.52163399436909e-05,
+      "loss": 0.7193,
+      "step": 11164
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.520847683637936e-05,
+      "loss": 19.0654,
+      "step": 11165
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5200613848665327e-05,
+      "loss": 0.7457,
+      "step": 11166
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.519275098074509e-05,
+      "loss": 20.0977,
+      "step": 11167
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.518488823281486e-05,
+      "loss": 15.9578,
+      "step": 11168
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.517702560507095e-05,
+      "loss": 16.2274,
+      "step": 11169
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.516916309770957e-05,
+      "loss": 0.8156,
+      "step": 11170
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.516130071092697e-05,
+      "loss": 16.8613,
+      "step": 11171
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.515343844491941e-05,
+      "loss": 17.4832,
+      "step": 11172
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.514557629988314e-05,
+      "loss": 0.7555,
+      "step": 11173
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.513771427601439e-05,
+      "loss": 17.3389,
+      "step": 11174
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.51298523735094e-05,
+      "loss": 17.6852,
+      "step": 11175
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.512199059256443e-05,
+      "loss": 0.7792,
+      "step": 11176
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.511412893337566e-05,
+      "loss": 18.3887,
+      "step": 11177
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.510626739613937e-05,
+      "loss": 16.714,
+      "step": 11178
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.509840598105174e-05,
+      "loss": 19.0103,
+      "step": 11179
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.509054468830902e-05,
+      "loss": 17.8429,
+      "step": 11180
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.508268351810744e-05,
+      "loss": 0.7064,
+      "step": 11181
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.507482247064319e-05,
+      "loss": 0.7616,
+      "step": 11182
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.506696154611252e-05,
+      "loss": 0.732,
+      "step": 11183
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.505910074471159e-05,
+      "loss": 0.7738,
+      "step": 11184
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.5051240066636634e-05,
+      "loss": 0.787,
+      "step": 11185
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.504337951208386e-05,
+      "loss": 18.4698,
+      "step": 11186
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.5035519081249465e-05,
+      "loss": 20.3619,
+      "step": 11187
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.5027658774329644e-05,
+      "loss": 0.7751,
+      "step": 11188
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.5019798591520593e-05,
+      "loss": 0.8082,
+      "step": 11189
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.501193853301854e-05,
+      "loss": 16.6068,
+      "step": 11190
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.5004078599019605e-05,
+      "loss": 0.7986,
+      "step": 11191
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4996218789720025e-05,
+      "loss": 0.7274,
+      "step": 11192
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4988359105315956e-05,
+      "loss": 0.7519,
+      "step": 11193
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.498049954600358e-05,
+      "loss": 0.7451,
+      "step": 11194
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4972640111979104e-05,
+      "loss": 16.5219,
+      "step": 11195
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4964780803438654e-05,
+      "loss": 17.4331,
+      "step": 11196
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.495692162057846e-05,
+      "loss": 0.6851,
+      "step": 11197
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4949062563594624e-05,
+      "loss": 18.7935,
+      "step": 11198
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.494120363268334e-05,
+      "loss": 18.94,
+      "step": 11199
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4933344828040774e-05,
+      "loss": 0.7869,
+      "step": 11200
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.492548614986309e-05,
+      "loss": 0.7681,
+      "step": 11201
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.491762759834641e-05,
+      "loss": 0.8099,
+      "step": 11202
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.490976917368691e-05,
+      "loss": 0.749,
+      "step": 11203
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.490191087608076e-05,
+      "loss": 0.7192,
+      "step": 11204
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4894052705724056e-05,
+      "loss": 0.7056,
+      "step": 11205
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.488619466281297e-05,
+      "loss": 16.8044,
+      "step": 11206
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.487833674754362e-05,
+      "loss": 0.803,
+      "step": 11207
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4870478960112154e-05,
+      "loss": 0.7675,
+      "step": 11208
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.486262130071472e-05,
+      "loss": 0.7814,
+      "step": 11209
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.485476376954742e-05,
+      "loss": 0.773,
+      "step": 11210
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.484690636680641e-05,
+      "loss": 0.7727,
+      "step": 11211
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4839049092687776e-05,
+      "loss": 16.5922,
+      "step": 11212
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.483119194738764e-05,
+      "loss": 0.7147,
+      "step": 11213
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4823334931102154e-05,
+      "loss": 18.7879,
+      "step": 11214
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4815478044027415e-05,
+      "loss": 0.7341,
+      "step": 11215
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.480762128635951e-05,
+      "loss": 0.7482,
+      "step": 11216
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.479976465829457e-05,
+      "loss": 18.6252,
+      "step": 11217
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4791908160028704e-05,
+      "loss": 18.2649,
+      "step": 11218
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.478405179175799e-05,
+      "loss": 19.2505,
+      "step": 11219
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.477619555367854e-05,
+      "loss": 17.4339,
+      "step": 11220
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.476833944598643e-05,
+      "loss": 17.6114,
+      "step": 11221
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4760483468877764e-05,
+      "loss": 15.7313,
+      "step": 11222
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.475262762254863e-05,
+      "loss": 0.7429,
+      "step": 11223
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.474477190719511e-05,
+      "loss": 0.7504,
+      "step": 11224
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4736916323013296e-05,
+      "loss": 0.7839,
+      "step": 11225
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.472906087019923e-05,
+      "loss": 18.5595,
+      "step": 11226
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.472120554894902e-05,
+      "loss": 17.6098,
+      "step": 11227
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4713350359458714e-05,
+      "loss": 17.4124,
+      "step": 11228
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.47054953019244e-05,
+      "loss": 15.8371,
+      "step": 11229
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.469764037654213e-05,
+      "loss": 18.1249,
+      "step": 11230
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.468978558350796e-05,
+      "loss": 17.8206,
+      "step": 11231
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.468193092301798e-05,
+      "loss": 19.6786,
+      "step": 11232
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.467407639526819e-05,
+      "loss": 18.1381,
+      "step": 11233
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4666222000454683e-05,
+      "loss": 0.7523,
+      "step": 11234
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4658367738773486e-05,
+      "loss": 0.6941,
+      "step": 11235
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.465051361042065e-05,
+      "loss": 18.9916,
+      "step": 11236
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.464265961559222e-05,
+      "loss": 0.7448,
+      "step": 11237
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.463480575448422e-05,
+      "loss": 19.4539,
+      "step": 11238
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4626952027292716e-05,
+      "loss": 0.7594,
+      "step": 11239
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.46190984342137e-05,
+      "loss": 0.7674,
+      "step": 11240
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4611244975443214e-05,
+      "loss": 17.9472,
+      "step": 11241
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4603391651177276e-05,
+      "loss": 16.7343,
+      "step": 11242
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.459553846161193e-05,
+      "loss": 0.7835,
+      "step": 11243
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.458768540694317e-05,
+      "loss": 0.7769,
+      "step": 11244
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.457983248736704e-05,
+      "loss": 18.1601,
+      "step": 11245
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4571979703079503e-05,
+      "loss": 17.2812,
+      "step": 11246
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4564127054276594e-05,
+      "loss": 18.561,
+      "step": 11247
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.455627454115432e-05,
+      "loss": 15.4375,
+      "step": 11248
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.454842216390868e-05,
+      "loss": 0.7639,
+      "step": 11249
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.454056992273565e-05,
+      "loss": 16.3668,
+      "step": 11250
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.453271781783126e-05,
+      "loss": 0.8006,
+      "step": 11251
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.452486584939149e-05,
+      "loss": 18.0496,
+      "step": 11252
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.451701401761231e-05,
+      "loss": 0.7182,
+      "step": 11253
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4509162322689704e-05,
+      "loss": 0.7474,
+      "step": 11254
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.450131076481965e-05,
+      "loss": 18.05,
+      "step": 11255
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4493459344198145e-05,
+      "loss": 17.5394,
+      "step": 11256
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.448560806102116e-05,
+      "loss": 18.5743,
+      "step": 11257
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.447775691548465e-05,
+      "loss": 0.7146,
+      "step": 11258
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.44699059077846e-05,
+      "loss": 0.7508,
+      "step": 11259
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.446205503811695e-05,
+      "loss": 0.7181,
+      "step": 11260
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4454204306677673e-05,
+      "loss": 19.6174,
+      "step": 11261
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4446353713662724e-05,
+      "loss": 17.6233,
+      "step": 11262
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.443850325926805e-05,
+      "loss": 0.7372,
+      "step": 11263
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.44306529436896e-05,
+      "loss": 0.7414,
+      "step": 11264
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.442280276712333e-05,
+      "loss": 19.3397,
+      "step": 11265
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.441495272976519e-05,
+      "loss": 0.794,
+      "step": 11266
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.44071028318111e-05,
+      "loss": 0.7615,
+      "step": 11267
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4399253073456986e-05,
+      "loss": 17.781,
+      "step": 11268
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.439140345489879e-05,
+      "loss": 16.2891,
+      "step": 11269
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.438355397633246e-05,
+      "loss": 0.7818,
+      "step": 11270
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.43757046379539e-05,
+      "loss": 17.4242,
+      "step": 11271
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.436785543995903e-05,
+      "loss": 17.9622,
+      "step": 11272
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.43600063825438e-05,
+      "loss": 19.4893,
+      "step": 11273
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.435215746590408e-05,
+      "loss": 0.7608,
+      "step": 11274
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.434430869023579e-05,
+      "loss": 0.791,
+      "step": 11275
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4336460055734865e-05,
+      "loss": 0.7786,
+      "step": 11276
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4328611562597174e-05,
+      "loss": 19.0439,
+      "step": 11277
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.432076321101863e-05,
+      "loss": 0.7913,
+      "step": 11278
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.431291500119514e-05,
+      "loss": 18.312,
+      "step": 11279
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.430506693332261e-05,
+      "loss": 0.713,
+      "step": 11280
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4297219007596895e-05,
+      "loss": 0.7308,
+      "step": 11281
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4289371224213894e-05,
+      "loss": 0.7426,
+      "step": 11282
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.428152358336948e-05,
+      "loss": 0.7909,
+      "step": 11283
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4273676085259555e-05,
+      "loss": 18.1035,
+      "step": 11284
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4265828730079987e-05,
+      "loss": 0.7711,
+      "step": 11285
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4257981518026624e-05,
+      "loss": 18.2974,
+      "step": 11286
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.425013444929539e-05,
+      "loss": 19.2509,
+      "step": 11287
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.424228752408208e-05,
+      "loss": 16.5213,
+      "step": 11288
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.42344407425826e-05,
+      "loss": 18.1175,
+      "step": 11289
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.422659410499279e-05,
+      "loss": 18.0568,
+      "step": 11290
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.421874761150851e-05,
+      "loss": 16.4142,
+      "step": 11291
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4210901262325606e-05,
+      "loss": 17.966,
+      "step": 11292
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.420305505763993e-05,
+      "loss": 0.7623,
+      "step": 11293
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.419520899764734e-05,
+      "loss": 0.7378,
+      "step": 11294
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4187363082543654e-05,
+      "loss": 18.9923,
+      "step": 11295
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.41795173125247e-05,
+      "loss": 17.3612,
+      "step": 11296
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4171671687786325e-05,
+      "loss": 0.798,
+      "step": 11297
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.416382620852435e-05,
+      "loss": 19.739,
+      "step": 11298
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.415598087493462e-05,
+      "loss": 0.7873,
+      "step": 11299
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4148135687212935e-05,
+      "loss": 0.7917,
+      "step": 11300
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.414029064555513e-05,
+      "loss": 17.1467,
+      "step": 11301
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4132445750157e-05,
+      "loss": 18.0583,
+      "step": 11302
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.412460100121436e-05,
+      "loss": 0.7585,
+      "step": 11303
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.411675639892302e-05,
+      "loss": 18.5149,
+      "step": 11304
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4108911943478786e-05,
+      "loss": 0.7594,
+      "step": 11305
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.410106763507745e-05,
+      "loss": 0.7563,
+      "step": 11306
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4093223473914815e-05,
+      "loss": 0.7922,
+      "step": 11307
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.408537946018669e-05,
+      "loss": 17.5662,
+      "step": 11308
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.407753559408883e-05,
+      "loss": 0.7915,
+      "step": 11309
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.406969187581703e-05,
+      "loss": 17.5621,
+      "step": 11310
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.406184830556707e-05,
+      "loss": 16.2152,
+      "step": 11311
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4054004883534734e-05,
+      "loss": 0.7228,
+      "step": 11312
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.40461616099158e-05,
+      "loss": 0.7052,
+      "step": 11313
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4038318484906024e-05,
+      "loss": 0.7881,
+      "step": 11314
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4030475508701205e-05,
+      "loss": 18.5885,
+      "step": 11315
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.402263268149706e-05,
+      "loss": 17.5587,
+      "step": 11316
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4014790003489366e-05,
+      "loss": 16.4875,
+      "step": 11317
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.4006947474873886e-05,
+      "loss": 17.3034,
+      "step": 11318
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.399910509584636e-05,
+      "loss": 0.7531,
+      "step": 11319
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.399126286660254e-05,
+      "loss": 18.3432,
+      "step": 11320
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.398342078733819e-05,
+      "loss": 0.7752,
+      "step": 11321
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.397557885824901e-05,
+      "loss": 0.7858,
+      "step": 11322
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3967737079530766e-05,
+      "loss": 0.7533,
+      "step": 11323
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.395989545137918e-05,
+      "loss": 17.5763,
+      "step": 11324
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.395205397398998e-05,
+      "loss": 0.7803,
+      "step": 11325
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.394421264755889e-05,
+      "loss": 0.7719,
+      "step": 11326
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.393637147228164e-05,
+      "loss": 17.9946,
+      "step": 11327
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.392853044835396e-05,
+      "loss": 0.8044,
+      "step": 11328
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3920689575971526e-05,
+      "loss": 17.4878,
+      "step": 11329
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3912848855330064e-05,
+      "loss": 20.7485,
+      "step": 11330
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3905008286625286e-05,
+      "loss": 17.1481,
+      "step": 11331
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.389716787005291e-05,
+      "loss": 18.1389,
+      "step": 11332
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3889327605808596e-05,
+      "loss": 18.6816,
+      "step": 11333
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.388148749408806e-05,
+      "loss": 0.7332,
+      "step": 11334
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3873647535087025e-05,
+      "loss": 0.8086,
+      "step": 11335
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.386580772900112e-05,
+      "loss": 0.8046,
+      "step": 11336
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.385796807602607e-05,
+      "loss": 18.4394,
+      "step": 11337
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3850128576357516e-05,
+      "loss": 19.5928,
+      "step": 11338
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3842289230191166e-05,
+      "loss": 0.779,
+      "step": 11339
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3834450037722685e-05,
+      "loss": 0.7329,
+      "step": 11340
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3826610999147744e-05,
+      "loss": 0.7686,
+      "step": 11341
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3818772114662015e-05,
+      "loss": 0.7338,
+      "step": 11342
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.381093338446113e-05,
+      "loss": 19.4305,
+      "step": 11343
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3803094808740754e-05,
+      "loss": 18.0392,
+      "step": 11344
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3795256387696545e-05,
+      "loss": 0.8059,
+      "step": 11345
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3787418121524165e-05,
+      "loss": 16.3459,
+      "step": 11346
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.377958001041923e-05,
+      "loss": 17.0399,
+      "step": 11347
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.377174205457741e-05,
+      "loss": 0.7406,
+      "step": 11348
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.376390425419435e-05,
+      "loss": 15.9057,
+      "step": 11349
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.375606660946564e-05,
+      "loss": 0.7461,
+      "step": 11350
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3748229120586946e-05,
+      "loss": 0.7344,
+      "step": 11351
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.374039178775386e-05,
+      "loss": 0.7447,
+      "step": 11352
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.373255461116204e-05,
+      "loss": 19.264,
+      "step": 11353
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.372471759100708e-05,
+      "loss": 18.8964,
+      "step": 11354
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.371688072748461e-05,
+      "loss": 0.7456,
+      "step": 11355
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3709044020790246e-05,
+      "loss": 16.7646,
+      "step": 11356
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3701207471119556e-05,
+      "loss": 17.5779,
+      "step": 11357
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.369337107866816e-05,
+      "loss": 18.9256,
+      "step": 11358
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3685534843631676e-05,
+      "loss": 21.035,
+      "step": 11359
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.367769876620569e-05,
+      "loss": 0.753,
+      "step": 11360
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.366986284658577e-05,
+      "loss": 19.5448,
+      "step": 11361
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.366202708496752e-05,
+      "loss": 16.7387,
+      "step": 11362
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.365419148154655e-05,
+      "loss": 0.7932,
+      "step": 11363
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.364635603651838e-05,
+      "loss": 0.7873,
+      "step": 11364
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.363852075007863e-05,
+      "loss": 0.7445,
+      "step": 11365
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.363068562242284e-05,
+      "loss": 0.7572,
+      "step": 11366
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.362285065374659e-05,
+      "loss": 0.7242,
+      "step": 11367
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3615015844245446e-05,
+      "loss": 19.6615,
+      "step": 11368
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3607181194114974e-05,
+      "loss": 17.1776,
+      "step": 11369
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.359934670355073e-05,
+      "loss": 19.8809,
+      "step": 11370
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3591512372748236e-05,
+      "loss": 0.7507,
+      "step": 11371
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3583678201903057e-05,
+      "loss": 16.3338,
+      "step": 11372
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3575844191210726e-05,
+      "loss": 0.7527,
+      "step": 11373
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.356801034086681e-05,
+      "loss": 0.7425,
+      "step": 11374
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.356017665106681e-05,
+      "loss": 15.9197,
+      "step": 11375
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.355234312200627e-05,
+      "loss": 20.7448,
+      "step": 11376
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3544509753880744e-05,
+      "loss": 19.2663,
+      "step": 11377
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3536676546885705e-05,
+      "loss": 0.7207,
+      "step": 11378
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.35288435012167e-05,
+      "loss": 0.7559,
+      "step": 11379
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3521010617069234e-05,
+      "loss": 0.7659,
+      "step": 11380
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.351317789463882e-05,
+      "loss": 17.1646,
+      "step": 11381
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.350534533412097e-05,
+      "loss": 0.7893,
+      "step": 11382
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.34975129357112e-05,
+      "loss": 0.7865,
+      "step": 11383
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3489680699604996e-05,
+      "loss": 18.5457,
+      "step": 11384
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.348184862599782e-05,
+      "loss": 20.2983,
+      "step": 11385
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.34740167150852e-05,
+      "loss": 19.0835,
+      "step": 11386
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3466184967062615e-05,
+      "loss": 0.7947,
+      "step": 11387
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3458353382125547e-05,
+      "loss": 18.3483,
+      "step": 11388
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.345052196046947e-05,
+      "loss": 17.6361,
+      "step": 11389
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.344269070228985e-05,
+      "loss": 16.8758,
+      "step": 11390
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.34348596077822e-05,
+      "loss": 0.6767,
+      "step": 11391
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3427028677141915e-05,
+      "loss": 17.7601,
+      "step": 11392
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.341919791056452e-05,
+      "loss": 0.7507,
+      "step": 11393
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.341136730824543e-05,
+      "loss": 18.2832,
+      "step": 11394
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.340353687038013e-05,
+      "loss": 0.7818,
+      "step": 11395
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.339570659716404e-05,
+      "loss": 0.7914,
+      "step": 11396
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.338787648879265e-05,
+      "loss": 20.7584,
+      "step": 11397
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.338004654546136e-05,
+      "loss": 0.7571,
+      "step": 11398
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.337221676736561e-05,
+      "loss": 0.7687,
+      "step": 11399
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.336438715470084e-05,
+      "loss": 0.7588,
+      "step": 11400
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3356557707662485e-05,
+      "loss": 0.7171,
+      "step": 11401
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.334872842644598e-05,
+      "loss": 17.2895,
+      "step": 11402
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3340899311246716e-05,
+      "loss": 16.7698,
+      "step": 11403
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.333307036226015e-05,
+      "loss": 19.0721,
+      "step": 11404
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.332524157968164e-05,
+      "loss": 17.7097,
+      "step": 11405
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.331741296370664e-05,
+      "loss": 18.3344,
+      "step": 11406
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3309584514530536e-05,
+      "loss": 0.8113,
+      "step": 11407
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3301756232348726e-05,
+      "loss": 19.3134,
+      "step": 11408
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.329392811735661e-05,
+      "loss": 0.758,
+      "step": 11409
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.328610016974957e-05,
+      "loss": 0.775,
+      "step": 11410
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3278272389723036e-05,
+      "loss": 19.6969,
+      "step": 11411
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3270444777472344e-05,
+      "loss": 16.6457,
+      "step": 11412
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.326261733319287e-05,
+      "loss": 0.7687,
+      "step": 11413
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3254790057080016e-05,
+      "loss": 17.6326,
+      "step": 11414
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3246962949329134e-05,
+      "loss": 0.7668,
+      "step": 11415
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.323913601013561e-05,
+      "loss": 17.9829,
+      "step": 11416
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.323130923969479e-05,
+      "loss": 17.4587,
+      "step": 11417
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.322348263820205e-05,
+      "loss": 0.7231,
+      "step": 11418
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3215656205852716e-05,
+      "loss": 0.7606,
+      "step": 11419
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.320782994284215e-05,
+      "loss": 17.8901,
+      "step": 11420
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.320000384936571e-05,
+      "loss": 18.2561,
+      "step": 11421
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.319217792561871e-05,
+      "loss": 21.6808,
+      "step": 11422
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.318435217179652e-05,
+      "loss": 0.6794,
+      "step": 11423
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.317652658809445e-05,
+      "loss": 18.3591,
+      "step": 11424
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3168701174707845e-05,
+      "loss": 18.456,
+      "step": 11425
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.316087593183201e-05,
+      "loss": 18.8217,
+      "step": 11426
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.315305085966227e-05,
+      "loss": 16.0529,
+      "step": 11427
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3145225958393944e-05,
+      "loss": 18.3686,
+      "step": 11428
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3137401228222335e-05,
+      "loss": 19.3833,
+      "step": 11429
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.312957666934278e-05,
+      "loss": 17.0742,
+      "step": 11430
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.312175228195054e-05,
+      "loss": 17.5845,
+      "step": 11431
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3113928066240955e-05,
+      "loss": 0.724,
+      "step": 11432
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.310610402240928e-05,
+      "loss": 18.5675,
+      "step": 11433
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.309828015065082e-05,
+      "loss": 21.45,
+      "step": 11434
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3090456451160874e-05,
+      "loss": 0.795,
+      "step": 11435
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.308263292413471e-05,
+      "loss": 18.8682,
+      "step": 11436
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.30748095697676e-05,
+      "loss": 16.8999,
+      "step": 11437
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3066986388254825e-05,
+      "loss": 17.2134,
+      "step": 11438
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.305916337979168e-05,
+      "loss": 0.7901,
+      "step": 11439
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.305134054457338e-05,
+      "loss": 18.2413,
+      "step": 11440
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.30435178827952e-05,
+      "loss": 16.9947,
+      "step": 11441
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.303569539465241e-05,
+      "loss": 0.7261,
+      "step": 11442
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.302787308034025e-05,
+      "loss": 0.8202,
+      "step": 11443
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.302005094005397e-05,
+      "loss": 19.2076,
+      "step": 11444
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.301222897398882e-05,
+      "loss": 17.6364,
+      "step": 11445
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.3004407182340034e-05,
+      "loss": 17.0787,
+      "step": 11446
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.299658556530283e-05,
+      "loss": 0.7355,
+      "step": 11447
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.298876412307244e-05,
+      "loss": 0.7515,
+      "step": 11448
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2980942855844125e-05,
+      "loss": 0.7817,
+      "step": 11449
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.297312176381306e-05,
+      "loss": 18.7887,
+      "step": 11450
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2965300847174476e-05,
+      "loss": 16.4221,
+      "step": 11451
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2957480106123594e-05,
+      "loss": 18.2618,
+      "step": 11452
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.294965954085564e-05,
+      "loss": 0.7647,
+      "step": 11453
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2941839151565774e-05,
+      "loss": 0.77,
+      "step": 11454
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.293401893844922e-05,
+      "loss": 0.7709,
+      "step": 11455
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2926198901701155e-05,
+      "loss": 19.4382,
+      "step": 11456
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.291837904151678e-05,
+      "loss": 18.5428,
+      "step": 11457
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.291055935809129e-05,
+      "loss": 0.7549,
+      "step": 11458
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.290273985161985e-05,
+      "loss": 0.7794,
+      "step": 11459
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.289492052229767e-05,
+      "loss": 0.8217,
+      "step": 11460
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2887101370319864e-05,
+      "loss": 18.7901,
+      "step": 11461
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2879282395881636e-05,
+      "loss": 18.2513,
+      "step": 11462
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.287146359917816e-05,
+      "loss": 18.4605,
+      "step": 11463
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2863644980404574e-05,
+      "loss": 19.0196,
+      "step": 11464
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2855826539756026e-05,
+      "loss": 16.8693,
+      "step": 11465
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.284800827742769e-05,
+      "loss": 16.8279,
+      "step": 11466
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.284019019361471e-05,
+      "loss": 0.7994,
+      "step": 11467
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.283237228851221e-05,
+      "loss": 15.7031,
+      "step": 11468
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2824554562315335e-05,
+      "loss": 0.6649,
+      "step": 11469
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2816737015219204e-05,
+      "loss": 18.3887,
+      "step": 11470
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.280891964741897e-05,
+      "loss": 0.7483,
+      "step": 11471
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.280110245910974e-05,
+      "loss": 17.5837,
+      "step": 11472
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.279328545048664e-05,
+      "loss": 0.7366,
+      "step": 11473
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.278546862174479e-05,
+      "loss": 0.7941,
+      "step": 11474
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.277765197307927e-05,
+      "loss": 20.7202,
+      "step": 11475
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.276983550468521e-05,
+      "loss": 0.7597,
+      "step": 11476
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.276201921675771e-05,
+      "loss": 17.2297,
+      "step": 11477
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2754203109491856e-05,
+      "loss": 15.6086,
+      "step": 11478
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2746387183082755e-05,
+      "loss": 0.7556,
+      "step": 11479
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.27385714377255e-05,
+      "loss": 19.8244,
+      "step": 11480
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.273075587361514e-05,
+      "loss": 18.4794,
+      "step": 11481
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2722940490946774e-05,
+      "loss": 17.687,
+      "step": 11482
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.271512528991548e-05,
+      "loss": 0.7756,
+      "step": 11483
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.270731027071632e-05,
+      "loss": 0.7607,
+      "step": 11484
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2699495433544354e-05,
+      "loss": 17.3033,
+      "step": 11485
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2691680778594666e-05,
+      "loss": 17.0112,
+      "step": 11486
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.268386630606229e-05,
+      "loss": 17.8496,
+      "step": 11487
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2676052016142285e-05,
+      "loss": 0.7465,
+      "step": 11488
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.266823790902968e-05,
+      "loss": 20.3115,
+      "step": 11489
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.266042398491953e-05,
+      "loss": 18.7973,
+      "step": 11490
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.265261024400689e-05,
+      "loss": 0.7535,
+      "step": 11491
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.264479668648676e-05,
+      "loss": 0.766,
+      "step": 11492
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.263698331255419e-05,
+      "loss": 0.7711,
+      "step": 11493
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.26291701224042e-05,
+      "loss": 18.0305,
+      "step": 11494
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2621357116231806e-05,
+      "loss": 16.9438,
+      "step": 11495
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2613544294232014e-05,
+      "loss": 18.8182,
+      "step": 11496
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.260573165659986e-05,
+      "loss": 18.2564,
+      "step": 11497
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.259791920353032e-05,
+      "loss": 16.5104,
+      "step": 11498
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.25901069352184e-05,
+      "loss": 0.7555,
+      "step": 11499
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2582294851859115e-05,
+      "loss": 0.7255,
+      "step": 11500
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2574482953647454e-05,
+      "loss": 17.2043,
+      "step": 11501
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2566671240778386e-05,
+      "loss": 17.5681,
+      "step": 11502
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.255885971344689e-05,
+      "loss": 0.7484,
+      "step": 11503
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2551048371847956e-05,
+      "loss": 0.7257,
+      "step": 11504
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.254323721617657e-05,
+      "loss": 0.7426,
+      "step": 11505
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2535426246627667e-05,
+      "loss": 0.7827,
+      "step": 11506
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.252761546339624e-05,
+      "loss": 20.6125,
+      "step": 11507
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.251980486667725e-05,
+      "loss": 0.7708,
+      "step": 11508
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2511994456665615e-05,
+      "loss": 0.7571,
+      "step": 11509
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.250418423355631e-05,
+      "loss": 17.5002,
+      "step": 11510
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2496374197544296e-05,
+      "loss": 18.6269,
+      "step": 11511
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.248856434882448e-05,
+      "loss": 0.7425,
+      "step": 11512
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.248075468759181e-05,
+      "loss": 19.0121,
+      "step": 11513
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.247294521404123e-05,
+      "loss": 0.7541,
+      "step": 11514
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.246513592836766e-05,
+      "loss": 0.7701,
+      "step": 11515
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.245732683076602e-05,
+      "loss": 0.7631,
+      "step": 11516
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2449517921431206e-05,
+      "loss": 16.4999,
+      "step": 11517
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.244170920055815e-05,
+      "loss": 0.7154,
+      "step": 11518
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.243390066834178e-05,
+      "loss": 17.9964,
+      "step": 11519
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.242609232497696e-05,
+      "loss": 18.3258,
+      "step": 11520
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.24182841706586e-05,
+      "loss": 0.7407,
+      "step": 11521
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.241047620558163e-05,
+      "loss": 18.8354,
+      "step": 11522
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.240266842994087e-05,
+      "loss": 17.6758,
+      "step": 11523
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.239486084393125e-05,
+      "loss": 0.7163,
+      "step": 11524
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.238705344774765e-05,
+      "loss": 15.3577,
+      "step": 11525
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2379246241584925e-05,
+      "loss": 0.7335,
+      "step": 11526
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2371439225637955e-05,
+      "loss": 0.7733,
+      "step": 11527
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.236363240010161e-05,
+      "loss": 0.7573,
+      "step": 11528
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2355825765170744e-05,
+      "loss": 17.9281,
+      "step": 11529
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.234801932104021e-05,
+      "loss": 16.4397,
+      "step": 11530
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2340213067904857e-05,
+      "loss": 0.7899,
+      "step": 11531
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.233240700595953e-05,
+      "loss": 20.9714,
+      "step": 11532
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.232460113539909e-05,
+      "loss": 0.7451,
+      "step": 11533
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2316795456418344e-05,
+      "loss": 17.6155,
+      "step": 11534
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.230898996921214e-05,
+      "loss": 19.7913,
+      "step": 11535
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.230118467397532e-05,
+      "loss": 0.7189,
+      "step": 11536
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2293379570902666e-05,
+      "loss": 0.7758,
+      "step": 11537
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.228557466018903e-05,
+      "loss": 0.7876,
+      "step": 11538
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.227776994202921e-05,
+      "loss": 0.7335,
+      "step": 11539
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2269965416618005e-05,
+      "loss": 0.7465,
+      "step": 11540
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.226216108415023e-05,
+      "loss": 0.7557,
+      "step": 11541
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.225435694482069e-05,
+      "loss": 19.6178,
+      "step": 11542
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.224655299882418e-05,
+      "loss": 18.674,
+      "step": 11543
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.223874924635547e-05,
+      "loss": 19.0026,
+      "step": 11544
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.223094568760934e-05,
+      "loss": 0.723,
+      "step": 11545
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.222314232278058e-05,
+      "loss": 0.732,
+      "step": 11546
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.221533915206397e-05,
+      "loss": 19.6031,
+      "step": 11547
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.220753617565427e-05,
+      "loss": 0.7898,
+      "step": 11548
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.219973339374626e-05,
+      "loss": 0.7894,
+      "step": 11549
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.219193080653467e-05,
+      "loss": 19.1505,
+      "step": 11550
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.218412841421426e-05,
+      "loss": 0.788,
+      "step": 11551
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.21763262169798e-05,
+      "loss": 0.7651,
+      "step": 11552
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.216852421502603e-05,
+      "loss": 20.6648,
+      "step": 11553
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.216072240854768e-05,
+      "loss": 0.7857,
+      "step": 11554
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2152920797739474e-05,
+      "loss": 16.8482,
+      "step": 11555
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2145119382796194e-05,
+      "loss": 0.8117,
+      "step": 11556
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.21373181639125e-05,
+      "loss": 0.6897,
+      "step": 11557
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.212951714128315e-05,
+      "loss": 19.7176,
+      "step": 11558
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2121716315102833e-05,
+      "loss": 18.387,
+      "step": 11559
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.211391568556628e-05,
+      "loss": 18.1049,
+      "step": 11560
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.21061152528682e-05,
+      "loss": 19.3539,
+      "step": 11561
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2098315017203284e-05,
+      "loss": 18.0161,
+      "step": 11562
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.209051497876625e-05,
+      "loss": 19.2444,
+      "step": 11563
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.208271513775174e-05,
+      "loss": 0.7934,
+      "step": 11564
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.207491549435447e-05,
+      "loss": 0.7535,
+      "step": 11565
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2067116048769114e-05,
+      "loss": 19.4234,
+      "step": 11566
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.205931680119037e-05,
+      "loss": 0.7661,
+      "step": 11567
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.205151775181288e-05,
+      "loss": 17.8624,
+      "step": 11568
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.204371890083132e-05,
+      "loss": 0.7755,
+      "step": 11569
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.203592024844038e-05,
+      "loss": 0.7925,
+      "step": 11570
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.202812179483466e-05,
+      "loss": 17.1025,
+      "step": 11571
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.202032354020885e-05,
+      "loss": 18.19,
+      "step": 11572
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.201252548475758e-05,
+      "loss": 17.7254,
+      "step": 11573
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.20047276286755e-05,
+      "loss": 16.3435,
+      "step": 11574
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.199692997215725e-05,
+      "loss": 17.1802,
+      "step": 11575
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.1989132515397446e-05,
+      "loss": 0.8145,
+      "step": 11576
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.198133525859075e-05,
+      "loss": 17.572,
+      "step": 11577
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.197353820193173e-05,
+      "loss": 0.7317,
+      "step": 11578
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.1965741345615025e-05,
+      "loss": 0.8125,
+      "step": 11579
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.195794468983526e-05,
+      "loss": 0.716,
+      "step": 11580
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.195014823478703e-05,
+      "loss": 18.1476,
+      "step": 11581
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.194235198066494e-05,
+      "loss": 16.9272,
+      "step": 11582
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.193455592766358e-05,
+      "loss": 0.7993,
+      "step": 11583
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1926760075977565e-05,
+      "loss": 0.811,
+      "step": 11584
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.191896442580144e-05,
+      "loss": 0.799,
+      "step": 11585
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.191116897732982e-05,
+      "loss": 0.7772,
+      "step": 11586
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.190337373075727e-05,
+      "loss": 0.7569,
+      "step": 11587
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1895578686278344e-05,
+      "loss": 17.3865,
+      "step": 11588
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.188778384408764e-05,
+      "loss": 0.7312,
+      "step": 11589
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1879989204379694e-05,
+      "loss": 0.7065,
+      "step": 11590
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1872194767349094e-05,
+      "loss": 0.7799,
+      "step": 11591
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1864400533190346e-05,
+      "loss": 0.7614,
+      "step": 11592
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1856606502098015e-05,
+      "loss": 0.73,
+      "step": 11593
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.184881267426665e-05,
+      "loss": 17.3904,
+      "step": 11594
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1841019049890796e-05,
+      "loss": 0.7516,
+      "step": 11595
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.183322562916495e-05,
+      "loss": 20.6272,
+      "step": 11596
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.182543241228366e-05,
+      "loss": 20.6307,
+      "step": 11597
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1817639399441464e-05,
+      "loss": 18.624,
+      "step": 11598
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1809846590832836e-05,
+      "loss": 0.7721,
+      "step": 11599
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1802053986652324e-05,
+      "loss": 0.7578,
+      "step": 11600
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.17942615870944e-05,
+      "loss": 18.4461,
+      "step": 11601
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1786469392353575e-05,
+      "loss": 0.8491,
+      "step": 11602
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.177867740262437e-05,
+      "loss": 18.5006,
+      "step": 11603
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1770885618101235e-05,
+      "loss": 17.9709,
+      "step": 11604
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.17630940389787e-05,
+      "loss": 0.7712,
+      "step": 11605
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.17553026654512e-05,
+      "loss": 16.3544,
+      "step": 11606
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.174751149771322e-05,
+      "loss": 0.7937,
+      "step": 11607
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.173972053595924e-05,
+      "loss": 0.8137,
+      "step": 11608
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.173192978038374e-05,
+      "loss": 0.7399,
+      "step": 11609
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.172413923118114e-05,
+      "loss": 0.7518,
+      "step": 11610
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.171634888854592e-05,
+      "loss": 18.829,
+      "step": 11611
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.170855875267254e-05,
+      "loss": 17.4871,
+      "step": 11612
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.170076882375541e-05,
+      "loss": 17.2498,
+      "step": 11613
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.169297910198899e-05,
+      "loss": 0.7722,
+      "step": 11614
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.168518958756771e-05,
+      "loss": 0.7509,
+      "step": 11615
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1677400280685985e-05,
+      "loss": 0.7617,
+      "step": 11616
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1669611181538265e-05,
+      "loss": 21.159,
+      "step": 11617
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.166182229031894e-05,
+      "loss": 17.7563,
+      "step": 11618
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.165403360722245e-05,
+      "loss": 0.7585,
+      "step": 11619
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1646245132443174e-05,
+      "loss": 0.7762,
+      "step": 11620
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1638456866175523e-05,
+      "loss": 18.0619,
+      "step": 11621
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.16306688086139e-05,
+      "loss": 20.4747,
+      "step": 11622
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.16228809599527e-05,
+      "loss": 16.3713,
+      "step": 11623
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.16150933203863e-05,
+      "loss": 0.8299,
+      "step": 11624
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1607305890109103e-05,
+      "loss": 0.7523,
+      "step": 11625
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.159951866931544e-05,
+      "loss": 0.6915,
+      "step": 11626
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.159173165819972e-05,
+      "loss": 16.8023,
+      "step": 11627
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1583944856956305e-05,
+      "loss": 0.7612,
+      "step": 11628
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1576158265779534e-05,
+      "loss": 0.7452,
+      "step": 11629
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.156837188486378e-05,
+      "loss": 18.4938,
+      "step": 11630
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.15605857144034e-05,
+      "loss": 0.766,
+      "step": 11631
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.155279975459272e-05,
+      "loss": 0.8003,
+      "step": 11632
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1545014005626096e-05,
+      "loss": 16.4122,
+      "step": 11633
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1537228467697844e-05,
+      "loss": 0.7756,
+      "step": 11634
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.152944314100229e-05,
+      "loss": 17.2497,
+      "step": 11635
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.152165802573378e-05,
+      "loss": 18.2396,
+      "step": 11636
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.151387312208662e-05,
+      "loss": 0.8011,
+      "step": 11637
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.150608843025512e-05,
+      "loss": 0.7328,
+      "step": 11638
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.149830395043362e-05,
+      "loss": 18.5162,
+      "step": 11639
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.149051968281637e-05,
+      "loss": 0.7365,
+      "step": 11640
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.148273562759769e-05,
+      "loss": 17.3631,
+      "step": 11641
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1474951784971885e-05,
+      "loss": 0.7982,
+      "step": 11642
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.146716815513322e-05,
+      "loss": 17.7805,
+      "step": 11643
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.145938473827598e-05,
+      "loss": 0.718,
+      "step": 11644
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1451601534594454e-05,
+      "loss": 16.6879,
+      "step": 11645
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.144381854428293e-05,
+      "loss": 18.841,
+      "step": 11646
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.143603576753563e-05,
+      "loss": 18.6616,
+      "step": 11647
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.142825320454682e-05,
+      "loss": 16.6112,
+      "step": 11648
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.142047085551077e-05,
+      "loss": 19.1953,
+      "step": 11649
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.141268872062173e-05,
+      "loss": 16.2126,
+      "step": 11650
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.140490680007395e-05,
+      "loss": 18.8352,
+      "step": 11651
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.139712509406165e-05,
+      "loss": 0.7423,
+      "step": 11652
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.138934360277909e-05,
+      "loss": 0.7145,
+      "step": 11653
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1381562326420456e-05,
+      "loss": 0.738,
+      "step": 11654
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.137378126518e-05,
+      "loss": 14.7405,
+      "step": 11655
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.136600041925194e-05,
+      "loss": 19.4664,
+      "step": 11656
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.135821978883047e-05,
+      "loss": 16.8854,
+      "step": 11657
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.135043937410981e-05,
+      "loss": 0.7615,
+      "step": 11658
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.134265917528416e-05,
+      "loss": 0.7602,
+      "step": 11659
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1334879192547725e-05,
+      "loss": 18.4722,
+      "step": 11660
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.132709942609468e-05,
+      "loss": 0.7261,
+      "step": 11661
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1319319876119206e-05,
+      "loss": 0.7814,
+      "step": 11662
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.131154054281548e-05,
+      "loss": 18.1973,
+      "step": 11663
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.13037614263777e-05,
+      "loss": 18.9296,
+      "step": 11664
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.129598252700002e-05,
+      "loss": 0.7911,
+      "step": 11665
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.12882038448766e-05,
+      "loss": 17.8539,
+      "step": 11666
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.128042538020162e-05,
+      "loss": 18.8813,
+      "step": 11667
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.127264713316919e-05,
+      "loss": 0.7673,
+      "step": 11668
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.126486910397348e-05,
+      "loss": 0.8116,
+      "step": 11669
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.125709129280865e-05,
+      "loss": 0.7687,
+      "step": 11670
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.12493136998688e-05,
+      "loss": 0.6991,
+      "step": 11671
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.124153632534807e-05,
+      "loss": 17.4617,
+      "step": 11672
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.123375916944061e-05,
+      "loss": 18.6743,
+      "step": 11673
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.122598223234053e-05,
+      "loss": 0.7403,
+      "step": 11674
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1218205514241934e-05,
+      "loss": 0.7694,
+      "step": 11675
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.121042901533893e-05,
+      "loss": 0.7244,
+      "step": 11676
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.120265273582561e-05,
+      "loss": 17.7495,
+      "step": 11677
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1194876675896095e-05,
+      "loss": 18.6681,
+      "step": 11678
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1187100835744476e-05,
+      "loss": 0.7854,
+      "step": 11679
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.117932521556482e-05,
+      "loss": 0.779,
+      "step": 11680
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1171549815551244e-05,
+      "loss": 0.7678,
+      "step": 11681
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.116377463589778e-05,
+      "loss": 18.7555,
+      "step": 11682
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1155999676798516e-05,
+      "loss": 19.1357,
+      "step": 11683
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1148224938447534e-05,
+      "loss": 0.7444,
+      "step": 11684
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.114045042103887e-05,
+      "loss": 0.8028,
+      "step": 11685
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.113267612476658e-05,
+      "loss": 16.3764,
+      "step": 11686
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.112490204982472e-05,
+      "loss": 16.921,
+      "step": 11687
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.111712819640735e-05,
+      "loss": 0.7518,
+      "step": 11688
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.110935456470849e-05,
+      "loss": 18.1109,
+      "step": 11689
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1101581154922144e-05,
+      "loss": 0.8022,
+      "step": 11690
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.109380796724237e-05,
+      "loss": 18.2859,
+      "step": 11691
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.108603500186319e-05,
+      "loss": 18.1226,
+      "step": 11692
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1078262258978615e-05,
+      "loss": 0.7267,
+      "step": 11693
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1070489738782634e-05,
+      "loss": 17.1923,
+      "step": 11694
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.10627174414693e-05,
+      "loss": 18.9302,
+      "step": 11695
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.105494536723255e-05,
+      "loss": 18.8606,
+      "step": 11696
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.104717351626642e-05,
+      "loss": 19.9505,
+      "step": 11697
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1039401888764887e-05,
+      "loss": 0.7512,
+      "step": 11698
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1031630484921914e-05,
+      "loss": 0.8154,
+      "step": 11699
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.102385930493149e-05,
+      "loss": 0.7631,
+      "step": 11700
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.101608834898761e-05,
+      "loss": 0.778,
+      "step": 11701
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1008317617284196e-05,
+      "loss": 17.9929,
+      "step": 11702
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.1000547110015245e-05,
+      "loss": 0.7502,
+      "step": 11703
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0992776827374674e-05,
+      "loss": 0.7179,
+      "step": 11704
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0985006769556454e-05,
+      "loss": 19.6235,
+      "step": 11705
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.097723693675452e-05,
+      "loss": 0.8109,
+      "step": 11706
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.096946732916282e-05,
+      "loss": 0.7413,
+      "step": 11707
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0961697946975296e-05,
+      "loss": 18.7511,
+      "step": 11708
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0953928790385834e-05,
+      "loss": 0.7278,
+      "step": 11709
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0946159859588366e-05,
+      "loss": 18.8533,
+      "step": 11710
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.093839115477682e-05,
+      "loss": 17.9203,
+      "step": 11711
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.093062267614511e-05,
+      "loss": 17.2581,
+      "step": 11712
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.092285442388712e-05,
+      "loss": 0.7513,
+      "step": 11713
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0915086398196754e-05,
+      "loss": 0.7923,
+      "step": 11714
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.090731859926793e-05,
+      "loss": 0.7772,
+      "step": 11715
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0899551027294495e-05,
+      "loss": 0.7961,
+      "step": 11716
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.089178368247034e-05,
+      "loss": 19.6118,
+      "step": 11717
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.088401656498934e-05,
+      "loss": 17.4179,
+      "step": 11718
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.087624967504537e-05,
+      "loss": 16.9625,
+      "step": 11719
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0868483012832295e-05,
+      "loss": 15.6409,
+      "step": 11720
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0860716578543964e-05,
+      "loss": 19.0789,
+      "step": 11721
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.085295037237426e-05,
+      "loss": 0.7103,
+      "step": 11722
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0845184394516975e-05,
+      "loss": 18.9785,
+      "step": 11723
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.083741864516598e-05,
+      "loss": 0.7093,
+      "step": 11724
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.082965312451511e-05,
+      "loss": 0.8507,
+      "step": 11725
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.08218878327582e-05,
+      "loss": 18.0772,
+      "step": 11726
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.081412277008905e-05,
+      "loss": 17.6985,
+      "step": 11727
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.08063579367015e-05,
+      "loss": 0.7068,
+      "step": 11728
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.079859333278937e-05,
+      "loss": 16.8135,
+      "step": 11729
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.079082895854644e-05,
+      "loss": 17.6426,
+      "step": 11730
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0783064814166525e-05,
+      "loss": 17.985,
+      "step": 11731
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.077530089984341e-05,
+      "loss": 0.7239,
+      "step": 11732
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.076753721577088e-05,
+      "loss": 0.7865,
+      "step": 11733
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0759773762142745e-05,
+      "loss": 16.5971,
+      "step": 11734
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.075201053915277e-05,
+      "loss": 15.8207,
+      "step": 11735
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.074424754699473e-05,
+      "loss": 17.4297,
+      "step": 11736
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.073648478586237e-05,
+      "loss": 0.7558,
+      "step": 11737
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0728722255949466e-05,
+      "loss": 0.7987,
+      "step": 11738
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.072095995744977e-05,
+      "loss": 0.7136,
+      "step": 11739
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.071319789055704e-05,
+      "loss": 16.9734,
+      "step": 11740
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0705436055464995e-05,
+      "loss": 0.7808,
+      "step": 11741
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0697674452367396e-05,
+      "loss": 0.8095,
+      "step": 11742
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.068991308145799e-05,
+      "loss": 17.5345,
+      "step": 11743
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.068215194293045e-05,
+      "loss": 0.8244,
+      "step": 11744
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0674391036978545e-05,
+      "loss": 19.1583,
+      "step": 11745
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.066663036379595e-05,
+      "loss": 0.8211,
+      "step": 11746
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0658869923576394e-05,
+      "loss": 19.3938,
+      "step": 11747
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.065110971651358e-05,
+      "loss": 17.0111,
+      "step": 11748
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.064334974280121e-05,
+      "loss": 0.8047,
+      "step": 11749
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.063559000263298e-05,
+      "loss": 17.9802,
+      "step": 11750
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.062783049620254e-05,
+      "loss": 19.9407,
+      "step": 11751
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.062007122370359e-05,
+      "loss": 0.7514,
+      "step": 11752
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.06123121853298e-05,
+      "loss": 0.7702,
+      "step": 11753
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0604553381274865e-05,
+      "loss": 16.8519,
+      "step": 11754
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0596794811732396e-05,
+      "loss": 0.7583,
+      "step": 11755
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.058903647689609e-05,
+      "loss": 17.7637,
+      "step": 11756
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.058127837695959e-05,
+      "loss": 0.7318,
+      "step": 11757
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.057352051211652e-05,
+      "loss": 17.8534,
+      "step": 11758
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.056576288256054e-05,
+      "loss": 17.3716,
+      "step": 11759
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.055800548848526e-05,
+      "loss": 0.8321,
+      "step": 11760
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0550248330084316e-05,
+      "loss": 15.8931,
+      "step": 11761
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0542491407551335e-05,
+      "loss": 17.6528,
+      "step": 11762
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0534734721079944e-05,
+      "loss": 0.7653,
+      "step": 11763
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0526978270863735e-05,
+      "loss": 0.7653,
+      "step": 11764
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.05192220570963e-05,
+      "loss": 0.8179,
+      "step": 11765
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.051146607997124e-05,
+      "loss": 18.2556,
+      "step": 11766
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.050371033968216e-05,
+      "loss": 0.7046,
+      "step": 11767
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.049595483642265e-05,
+      "loss": 18.4596,
+      "step": 11768
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0488199570386254e-05,
+      "loss": 17.1404,
+      "step": 11769
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0480444541766576e-05,
+      "loss": 18.3767,
+      "step": 11770
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0472689750757197e-05,
+      "loss": 0.7628,
+      "step": 11771
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.046493519755163e-05,
+      "loss": 0.7592,
+      "step": 11772
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.045718088234348e-05,
+      "loss": 0.7154,
+      "step": 11773
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.044942680532626e-05,
+      "loss": 0.7679,
+      "step": 11774
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.044167296669352e-05,
+      "loss": 0.7183,
+      "step": 11775
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.043391936663882e-05,
+      "loss": 19.5479,
+      "step": 11776
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0426166005355684e-05,
+      "loss": 18.289,
+      "step": 11777
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0418412883037626e-05,
+      "loss": 0.7614,
+      "step": 11778
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0410659999878156e-05,
+      "loss": 17.1528,
+      "step": 11779
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.040290735607081e-05,
+      "loss": 0.7979,
+      "step": 11780
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0395154951809076e-05,
+      "loss": 0.7524,
+      "step": 11781
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.038740278728649e-05,
+      "loss": 18.6966,
+      "step": 11782
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.03796508626965e-05,
+      "loss": 17.7282,
+      "step": 11783
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0371899178232655e-05,
+      "loss": 0.8018,
+      "step": 11784
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.036414773408838e-05,
+      "loss": 0.7438,
+      "step": 11785
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0356396530457176e-05,
+      "loss": 18.1813,
+      "step": 11786
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.034864556753253e-05,
+      "loss": 16.7604,
+      "step": 11787
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0340894845507884e-05,
+      "loss": 19.4933,
+      "step": 11788
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0333144364576705e-05,
+      "loss": 17.0894,
+      "step": 11789
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.032539412493245e-05,
+      "loss": 0.8104,
+      "step": 11790
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0317644126768586e-05,
+      "loss": 15.9608,
+      "step": 11791
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.030989437027853e-05,
+      "loss": 0.744,
+      "step": 11792
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.030214485565571e-05,
+      "loss": 0.7984,
+      "step": 11793
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.029439558309357e-05,
+      "loss": 18.3408,
+      "step": 11794
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.028664655278553e-05,
+      "loss": 0.739,
+      "step": 11795
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.027889776492502e-05,
+      "loss": 17.6309,
+      "step": 11796
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0271149219705434e-05,
+      "loss": 0.7996,
+      "step": 11797
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.026340091732021e-05,
+      "loss": 16.0568,
+      "step": 11798
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.025565285796269e-05,
+      "loss": 17.9885,
+      "step": 11799
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.024790504182631e-05,
+      "loss": 0.7434,
+      "step": 11800
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.024015746910445e-05,
+      "loss": 0.7684,
+      "step": 11801
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0232410139990475e-05,
+      "loss": 15.9263,
+      "step": 11802
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.022466305467778e-05,
+      "loss": 16.4081,
+      "step": 11803
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.021691621335973e-05,
+      "loss": 0.8014,
+      "step": 11804
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.02091696162297e-05,
+      "loss": 0.7497,
+      "step": 11805
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.020142326348102e-05,
+      "loss": 18.7817,
+      "step": 11806
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0193677155307047e-05,
+      "loss": 0.7265,
+      "step": 11807
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.018593129190113e-05,
+      "loss": 17.2982,
+      "step": 11808
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.017818567345661e-05,
+      "loss": 0.7633,
+      "step": 11809
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.017044030016682e-05,
+      "loss": 17.6055,
+      "step": 11810
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.016269517222508e-05,
+      "loss": 18.445,
+      "step": 11811
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0154950289824745e-05,
+      "loss": 18.3865,
+      "step": 11812
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.014720565315907e-05,
+      "loss": 0.7775,
+      "step": 11813
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.013946126242138e-05,
+      "loss": 0.7832,
+      "step": 11814
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.013171711780502e-05,
+      "loss": 0.7339,
+      "step": 11815
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.012397321950323e-05,
+      "loss": 0.7476,
+      "step": 11816
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0116229567709325e-05,
+      "loss": 17.5834,
+      "step": 11817
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.010848616261659e-05,
+      "loss": 0.7581,
+      "step": 11818
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.010074300441832e-05,
+      "loss": 0.7093,
+      "step": 11819
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.009300009330775e-05,
+      "loss": 0.7143,
+      "step": 11820
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.008525742947815e-05,
+      "loss": 0.8046,
+      "step": 11821
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.007751501312279e-05,
+      "loss": 17.6324,
+      "step": 11822
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0069772844434925e-05,
+      "loss": 16.7229,
+      "step": 11823
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.00620309236078e-05,
+      "loss": 0.7505,
+      "step": 11824
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.005428925083465e-05,
+      "loss": 0.7496,
+      "step": 11825
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.004654782630872e-05,
+      "loss": 18.0171,
+      "step": 11826
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.003880665022322e-05,
+      "loss": 0.7172,
+      "step": 11827
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.003106572277138e-05,
+      "loss": 18.3516,
+      "step": 11828
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.002332504414642e-05,
+      "loss": 0.7711,
+      "step": 11829
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.001558461454153e-05,
+      "loss": 19.2028,
+      "step": 11830
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0007844434149936e-05,
+      "loss": 0.7378,
+      "step": 11831
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 4.0000104503164826e-05,
+      "loss": 17.9848,
+      "step": 11832
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9992364821779414e-05,
+      "loss": 19.6432,
+      "step": 11833
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.998462539018685e-05,
+      "loss": 19.1376,
+      "step": 11834
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9976886208580307e-05,
+      "loss": 18.9198,
+      "step": 11835
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9969147277152973e-05,
+      "loss": 0.6801,
+      "step": 11836
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9961408596098015e-05,
+      "loss": 15.6981,
+      "step": 11837
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9953670165608603e-05,
+      "loss": 15.7416,
+      "step": 11838
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9945931985877874e-05,
+      "loss": 0.783,
+      "step": 11839
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9938194057098986e-05,
+      "loss": 0.7564,
+      "step": 11840
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9930456379465054e-05,
+      "loss": 18.6087,
+      "step": 11841
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.992271895316924e-05,
+      "loss": 20.2194,
+      "step": 11842
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.991498177840466e-05,
+      "loss": 0.7585,
+      "step": 11843
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.990724485536443e-05,
+      "loss": 17.8042,
+      "step": 11844
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9899508184241675e-05,
+      "loss": 15.1031,
+      "step": 11845
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9891771765229504e-05,
+      "loss": 0.8082,
+      "step": 11846
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.988403559852103e-05,
+      "loss": 0.7304,
+      "step": 11847
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.987629968430934e-05,
+      "loss": 0.763,
+      "step": 11848
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9868564022787505e-05,
+      "loss": 0.7486,
+      "step": 11849
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.986082861414862e-05,
+      "loss": 18.8598,
+      "step": 11850
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.985309345858577e-05,
+      "loss": 14.3976,
+      "step": 11851
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9845358556292034e-05,
+      "loss": 0.7564,
+      "step": 11852
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.983762390746048e-05,
+      "loss": 18.8457,
+      "step": 11853
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.982988951228414e-05,
+      "loss": 16.8352,
+      "step": 11854
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.982215537095607e-05,
+      "loss": 17.5928,
+      "step": 11855
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.981442148366933e-05,
+      "loss": 0.7727,
+      "step": 11856
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9806687850616965e-05,
+      "loss": 0.777,
+      "step": 11857
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.979895447199198e-05,
+      "loss": 19.4826,
+      "step": 11858
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.979122134798743e-05,
+      "loss": 18.1185,
+      "step": 11859
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.978348847879635e-05,
+      "loss": 17.7247,
+      "step": 11860
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.97757558646117e-05,
+      "loss": 17.355,
+      "step": 11861
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.976802350562653e-05,
+      "loss": 16.7075,
+      "step": 11862
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.976029140203383e-05,
+      "loss": 0.7848,
+      "step": 11863
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.975255955402658e-05,
+      "loss": 0.8386,
+      "step": 11864
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9744827961797794e-05,
+      "loss": 18.8881,
+      "step": 11865
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.973709662554045e-05,
+      "loss": 20.211,
+      "step": 11866
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9729365545447514e-05,
+      "loss": 0.7688,
+      "step": 11867
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9721634721711965e-05,
+      "loss": 18.8886,
+      "step": 11868
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.971390415452675e-05,
+      "loss": 0.7423,
+      "step": 11869
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9706173844084834e-05,
+      "loss": 17.0512,
+      "step": 11870
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9698443790579174e-05,
+      "loss": 17.0785,
+      "step": 11871
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.969071399420271e-05,
+      "loss": 0.7736,
+      "step": 11872
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.968298445514837e-05,
+      "loss": 16.3794,
+      "step": 11873
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.967525517360912e-05,
+      "loss": 18.6263,
+      "step": 11874
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.966752614977783e-05,
+      "loss": 18.8847,
+      "step": 11875
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9659797383847455e-05,
+      "loss": 20.0071,
+      "step": 11876
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9652068876010905e-05,
+      "loss": 0.7623,
+      "step": 11877
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.964434062646107e-05,
+      "loss": 0.7427,
+      "step": 11878
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9636612635390854e-05,
+      "loss": 0.7922,
+      "step": 11879
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.962888490299316e-05,
+      "loss": 19.2441,
+      "step": 11880
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.962115742946088e-05,
+      "loss": 0.7811,
+      "step": 11881
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.961343021498687e-05,
+      "loss": 16.9783,
+      "step": 11882
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9605703259764006e-05,
+      "loss": 0.7802,
+      "step": 11883
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9597976563985155e-05,
+      "loss": 17.7005,
+      "step": 11884
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.95902501278432e-05,
+      "loss": 17.1322,
+      "step": 11885
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.958252395153097e-05,
+      "loss": 17.8153,
+      "step": 11886
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.957479803524132e-05,
+      "loss": 17.1563,
+      "step": 11887
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9567072379167115e-05,
+      "loss": 0.7157,
+      "step": 11888
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.955934698350114e-05,
+      "loss": 18.2529,
+      "step": 11889
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.955162184843625e-05,
+      "loss": 0.6803,
+      "step": 11890
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9543896974165276e-05,
+      "loss": 0.7364,
+      "step": 11891
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.953617236088101e-05,
+      "loss": 0.7128,
+      "step": 11892
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.952844800877626e-05,
+      "loss": 16.6609,
+      "step": 11893
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9520723918043865e-05,
+      "loss": 18.2304,
+      "step": 11894
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.951300008887659e-05,
+      "loss": 19.8281,
+      "step": 11895
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.950527652146722e-05,
+      "loss": 18.9831,
+      "step": 11896
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9497553216008544e-05,
+      "loss": 19.5178,
+      "step": 11897
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.948983017269333e-05,
+      "loss": 16.9259,
+      "step": 11898
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.948210739171437e-05,
+      "loss": 16.7775,
+      "step": 11899
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.94743848732644e-05,
+      "loss": 0.7542,
+      "step": 11900
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.946666261753618e-05,
+      "loss": 0.7281,
+      "step": 11901
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9458940624722495e-05,
+      "loss": 18.3507,
+      "step": 11902
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.945121889501604e-05,
+      "loss": 17.456,
+      "step": 11903
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.944349742860956e-05,
+      "loss": 17.2475,
+      "step": 11904
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9435776225695805e-05,
+      "loss": 16.077,
+      "step": 11905
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9428055286467484e-05,
+      "loss": 0.7776,
+      "step": 11906
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9420334611117316e-05,
+      "loss": 0.7355,
+      "step": 11907
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9412614199838024e-05,
+      "loss": 18.6188,
+      "step": 11908
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.940489405282229e-05,
+      "loss": 16.0615,
+      "step": 11909
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.939717417026283e-05,
+      "loss": 0.7573,
+      "step": 11910
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.938945455235231e-05,
+      "loss": 18.2851,
+      "step": 11911
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9381735199283425e-05,
+      "loss": 16.9785,
+      "step": 11912
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.937401611124887e-05,
+      "loss": 0.7411,
+      "step": 11913
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.936629728844128e-05,
+      "loss": 17.1251,
+      "step": 11914
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.935857873105334e-05,
+      "loss": 0.7443,
+      "step": 11915
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.935086043927773e-05,
+      "loss": 0.7954,
+      "step": 11916
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.934314241330706e-05,
+      "loss": 19.6802,
+      "step": 11917
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9335424653333977e-05,
+      "loss": 0.7482,
+      "step": 11918
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.932770715955114e-05,
+      "loss": 18.0343,
+      "step": 11919
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9319989932151166e-05,
+      "loss": 0.7617,
+      "step": 11920
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9312272971326673e-05,
+      "loss": 17.1393,
+      "step": 11921
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.93045562772703e-05,
+      "loss": 15.5672,
+      "step": 11922
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9296839850174646e-05,
+      "loss": 0.8094,
+      "step": 11923
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.928912369023232e-05,
+      "loss": 16.8597,
+      "step": 11924
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9281407797635895e-05,
+      "loss": 0.8069,
+      "step": 11925
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9273692172577984e-05,
+      "loss": 17.538,
+      "step": 11926
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.926597681525117e-05,
+      "loss": 19.5964,
+      "step": 11927
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9258261725848014e-05,
+      "loss": 0.7739,
+      "step": 11928
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.925054690456112e-05,
+      "loss": 18.384,
+      "step": 11929
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.924283235158301e-05,
+      "loss": 19.64,
+      "step": 11930
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9235118067106255e-05,
+      "loss": 18.3656,
+      "step": 11931
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.922740405132342e-05,
+      "loss": 17.8707,
+      "step": 11932
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.921969030442705e-05,
+      "loss": 18.3208,
+      "step": 11933
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.921197682660966e-05,
+      "loss": 0.7333,
+      "step": 11934
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9204263618063786e-05,
+      "loss": 0.7491,
+      "step": 11935
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.919655067898198e-05,
+      "loss": 17.5721,
+      "step": 11936
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.918883800955672e-05,
+      "loss": 19.0605,
+      "step": 11937
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.918112560998054e-05,
+      "loss": 17.0464,
+      "step": 11938
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.917341348044592e-05,
+      "loss": 21.1085,
+      "step": 11939
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.916570162114538e-05,
+      "loss": 0.7764,
+      "step": 11940
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.91579900322714e-05,
+      "loss": 0.7538,
+      "step": 11941
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.915027871401646e-05,
+      "loss": 17.4391,
+      "step": 11942
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.914256766657306e-05,
+      "loss": 0.7681,
+      "step": 11943
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.913485689013364e-05,
+      "loss": 20.7238,
+      "step": 11944
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.912714638489066e-05,
+      "loss": 18.4263,
+      "step": 11945
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9119436151036595e-05,
+      "loss": 0.7558,
+      "step": 11946
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.91117261887639e-05,
+      "loss": 0.7431,
+      "step": 11947
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9104016498264984e-05,
+      "loss": 0.7293,
+      "step": 11948
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9096307079732315e-05,
+      "loss": 0.762,
+      "step": 11949
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9088597933358326e-05,
+      "loss": 17.8923,
+      "step": 11950
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.908088905933541e-05,
+      "loss": 0.7758,
+      "step": 11951
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9073180457856e-05,
+      "loss": 18.6386,
+      "step": 11952
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.906547212911249e-05,
+      "loss": 19.2308,
+      "step": 11953
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.905776407329729e-05,
+      "loss": 0.715,
+      "step": 11954
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9050056290602815e-05,
+      "loss": 0.7976,
+      "step": 11955
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.904234878122143e-05,
+      "loss": 17.0301,
+      "step": 11956
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.903464154534553e-05,
+      "loss": 18.0984,
+      "step": 11957
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.902693458316746e-05,
+      "loss": 0.7814,
+      "step": 11958
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.901922789487962e-05,
+      "loss": 0.8039,
+      "step": 11959
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.9011521480674347e-05,
+      "loss": 17.6736,
+      "step": 11960
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.900381534074402e-05,
+      "loss": 15.8405,
+      "step": 11961
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.899610947528096e-05,
+      "loss": 16.0034,
+      "step": 11962
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.898840388447753e-05,
+      "loss": 0.7474,
+      "step": 11963
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.898069856852607e-05,
+      "loss": 17.2786,
+      "step": 11964
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.8972993527618864e-05,
+      "loss": 0.7728,
+      "step": 11965
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.896528876194827e-05,
+      "loss": 0.7374,
+      "step": 11966
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.8957584271706584e-05,
+      "loss": 15.8593,
+      "step": 11967
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.894988005708611e-05,
+      "loss": 18.5498,
+      "step": 11968
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.894217611827916e-05,
+      "loss": 0.7993,
+      "step": 11969
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.893447245547801e-05,
+      "loss": 0.7684,
+      "step": 11970
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.892676906887498e-05,
+      "loss": 0.7417,
+      "step": 11971
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.89190659586623e-05,
+      "loss": 20.0265,
+      "step": 11972
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.891136312503226e-05,
+      "loss": 0.7171,
+      "step": 11973
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.890366056817713e-05,
+      "loss": 0.8093,
+      "step": 11974
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.889595828828918e-05,
+      "loss": 0.7696,
+      "step": 11975
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.888825628556063e-05,
+      "loss": 14.9446,
+      "step": 11976
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.8880554560183745e-05,
+      "loss": 18.9388,
+      "step": 11977
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.887285311235078e-05,
+      "loss": 0.7481,
+      "step": 11978
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.886515194225391e-05,
+      "loss": 0.7128,
+      "step": 11979
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.8857451050085415e-05,
+      "loss": 19.7167,
+      "step": 11980
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.884975043603746e-05,
+      "loss": 17.6561,
+      "step": 11981
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.884205010030229e-05,
+      "loss": 0.8019,
+      "step": 11982
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.883435004307211e-05,
+      "loss": 16.5933,
+      "step": 11983
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.882665026453909e-05,
+      "loss": 20.9501,
+      "step": 11984
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.881895076489544e-05,
+      "loss": 0.7454,
+      "step": 11985
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.881125154433333e-05,
+      "loss": 17.5777,
+      "step": 11986
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.880355260304492e-05,
+      "loss": 0.7459,
+      "step": 11987
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.8795853941222405e-05,
+      "loss": 0.7516,
+      "step": 11988
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.878815555905794e-05,
+      "loss": 0.788,
+      "step": 11989
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.878045745674366e-05,
+      "loss": 0.7618,
+      "step": 11990
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.877275963447172e-05,
+      "loss": 17.0676,
+      "step": 11991
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.876506209243429e-05,
+      "loss": 0.7352,
+      "step": 11992
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.875736483082346e-05,
+      "loss": 16.0792,
+      "step": 11993
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.874966784983137e-05,
+      "loss": 0.8025,
+      "step": 11994
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.874197114965014e-05,
+      "loss": 0.771,
+      "step": 11995
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8734274730471864e-05,
+      "loss": 17.0492,
+      "step": 11996
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.872657859248869e-05,
+      "loss": 18.6737,
+      "step": 11997
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.871888273589267e-05,
+      "loss": 0.784,
+      "step": 11998
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.871118716087593e-05,
+      "loss": 18.4085,
+      "step": 11999
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.870349186763053e-05,
+      "loss": 18.9922,
+      "step": 12000
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.869579685634854e-05,
+      "loss": 0.791,
+      "step": 12001
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8688102127222037e-05,
+      "loss": 17.6709,
+      "step": 12002
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.868040768044311e-05,
+      "loss": 17.7922,
+      "step": 12003
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8672713516203774e-05,
+      "loss": 0.7654,
+      "step": 12004
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.86650196346961e-05,
+      "loss": 0.7626,
+      "step": 12005
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.865732603611214e-05,
+      "loss": 18.8804,
+      "step": 12006
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.86496327206439e-05,
+      "loss": 15.9475,
+      "step": 12007
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8641939688483416e-05,
+      "loss": 0.7627,
+      "step": 12008
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.863424693982271e-05,
+      "loss": 18.3554,
+      "step": 12009
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8626554474853793e-05,
+      "loss": 0.7621,
+      "step": 12010
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8618862293768685e-05,
+      "loss": 0.7544,
+      "step": 12011
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8611170396759365e-05,
+      "loss": 0.7551,
+      "step": 12012
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.860347878401785e-05,
+      "loss": 20.5289,
+      "step": 12013
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8595787455736074e-05,
+      "loss": 0.764,
+      "step": 12014
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.858809641210606e-05,
+      "loss": 0.7444,
+      "step": 12015
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.858040565331976e-05,
+      "loss": 17.3439,
+      "step": 12016
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.857271517956916e-05,
+      "loss": 0.7485,
+      "step": 12017
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.856502499104619e-05,
+      "loss": 0.7785,
+      "step": 12018
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8557335087942816e-05,
+      "loss": 19.7911,
+      "step": 12019
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.854964547045096e-05,
+      "loss": 17.1811,
+      "step": 12020
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.854195613876256e-05,
+      "loss": 0.7626,
+      "step": 12021
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8534267093069566e-05,
+      "loss": 17.3928,
+      "step": 12022
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.852657833356387e-05,
+      "loss": 0.7317,
+      "step": 12023
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8518889860437404e-05,
+      "loss": 18.5277,
+      "step": 12024
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8511201673882067e-05,
+      "loss": 0.8252,
+      "step": 12025
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8503513774089774e-05,
+      "loss": 17.9035,
+      "step": 12026
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8495826161252396e-05,
+      "loss": 0.7256,
+      "step": 12027
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.848813883556181e-05,
+      "loss": 19.8472,
+      "step": 12028
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.848045179720991e-05,
+      "loss": 0.7504,
+      "step": 12029
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8472765046388565e-05,
+      "loss": 0.7534,
+      "step": 12030
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.846507858328965e-05,
+      "loss": 18.1206,
+      "step": 12031
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.845739240810499e-05,
+      "loss": 0.8167,
+      "step": 12032
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8449706521026475e-05,
+      "loss": 0.7478,
+      "step": 12033
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8442020922245906e-05,
+      "loss": 17.668,
+      "step": 12034
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.843433561195512e-05,
+      "loss": 17.4621,
+      "step": 12035
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.842665059034598e-05,
+      "loss": 17.3231,
+      "step": 12036
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8418965857610265e-05,
+      "loss": 18.0383,
+      "step": 12037
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8411281413939815e-05,
+      "loss": 18.4945,
+      "step": 12038
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8403597259526434e-05,
+      "loss": 18.9978,
+      "step": 12039
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.839591339456192e-05,
+      "loss": 0.7698,
+      "step": 12040
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.838822981923805e-05,
+      "loss": 17.3878,
+      "step": 12041
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8380546533746606e-05,
+      "loss": 20.0783,
+      "step": 12042
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.837286353827937e-05,
+      "loss": 0.7467,
+      "step": 12043
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.836518083302812e-05,
+      "loss": 0.739,
+      "step": 12044
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.835749841818462e-05,
+      "loss": 0.7821,
+      "step": 12045
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.834981629394061e-05,
+      "loss": 0.8038,
+      "step": 12046
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.834213446048787e-05,
+      "loss": 0.8011,
+      "step": 12047
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.833445291801808e-05,
+      "loss": 17.3037,
+      "step": 12048
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.832677166672302e-05,
+      "loss": 19.0531,
+      "step": 12049
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8319090706794415e-05,
+      "loss": 18.9727,
+      "step": 12050
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8311410038423964e-05,
+      "loss": 18.2692,
+      "step": 12051
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.830372966180338e-05,
+      "loss": 0.7598,
+      "step": 12052
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.829604957712438e-05,
+      "loss": 0.8717,
+      "step": 12053
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.828836978457867e-05,
+      "loss": 20.5464,
+      "step": 12054
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8280690284357926e-05,
+      "loss": 21.0019,
+      "step": 12055
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8273011076653806e-05,
+      "loss": 17.3988,
+      "step": 12056
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8265332161658015e-05,
+      "loss": 18.0481,
+      "step": 12057
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8257653539562206e-05,
+      "loss": 18.5912,
+      "step": 12058
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.824997521055806e-05,
+      "loss": 0.7442,
+      "step": 12059
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8242297174837206e-05,
+      "loss": 18.6411,
+      "step": 12060
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.823461943259132e-05,
+      "loss": 19.035,
+      "step": 12061
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8226941984011995e-05,
+      "loss": 0.7553,
+      "step": 12062
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.821926482929089e-05,
+      "loss": 0.7396,
+      "step": 12063
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.821158796861963e-05,
+      "loss": 17.2393,
+      "step": 12064
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8203911402189816e-05,
+      "loss": 0.7864,
+      "step": 12065
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.819623513019307e-05,
+      "loss": 0.7675,
+      "step": 12066
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.818855915282099e-05,
+      "loss": 0.77,
+      "step": 12067
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.81808834702652e-05,
+      "loss": 0.7713,
+      "step": 12068
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.817320808271723e-05,
+      "loss": 18.9214,
+      "step": 12069
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.816553299036868e-05,
+      "loss": 0.7702,
+      "step": 12070
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.815785819341114e-05,
+      "loss": 0.766,
+      "step": 12071
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8150183692036156e-05,
+      "loss": 0.7651,
+      "step": 12072
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.814250948643531e-05,
+      "loss": 18.0242,
+      "step": 12073
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.813483557680011e-05,
+      "loss": 0.7777,
+      "step": 12074
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.812716196332216e-05,
+      "loss": 0.7668,
+      "step": 12075
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8119488646192925e-05,
+      "loss": 17.0432,
+      "step": 12076
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.811181562560397e-05,
+      "loss": 19.9188,
+      "step": 12077
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.810414290174682e-05,
+      "loss": 17.1349,
+      "step": 12078
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8096470474812973e-05,
+      "loss": 16.7268,
+      "step": 12079
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.808879834499394e-05,
+      "loss": 0.7367,
+      "step": 12080
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.808112651248121e-05,
+      "loss": 0.7313,
+      "step": 12081
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.807345497746632e-05,
+      "loss": 0.7469,
+      "step": 12082
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.80657837401407e-05,
+      "loss": 0.7874,
+      "step": 12083
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8058112800695834e-05,
+      "loss": 0.7582,
+      "step": 12084
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.8050442159323197e-05,
+      "loss": 18.08,
+      "step": 12085
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.804277181621425e-05,
+      "loss": 0.7826,
+      "step": 12086
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.803510177156047e-05,
+      "loss": 0.7635,
+      "step": 12087
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.802743202555328e-05,
+      "loss": 17.2136,
+      "step": 12088
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.80197625783841e-05,
+      "loss": 0.781,
+      "step": 12089
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.801209343024439e-05,
+      "loss": 19.06,
+      "step": 12090
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.800442458132556e-05,
+      "loss": 0.7776,
+      "step": 12091
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.799675603181904e-05,
+      "loss": 17.2674,
+      "step": 12092
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.798908778191622e-05,
+      "loss": 18.9699,
+      "step": 12093
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.798141983180851e-05,
+      "loss": 18.2001,
+      "step": 12094
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7973752181687335e-05,
+      "loss": 0.7364,
+      "step": 12095
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.796608483174402e-05,
+      "loss": 0.7783,
+      "step": 12096
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.795841778216999e-05,
+      "loss": 18.2568,
+      "step": 12097
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.79507510331566e-05,
+      "loss": 17.7823,
+      "step": 12098
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.79430845848952e-05,
+      "loss": 0.77,
+      "step": 12099
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.793541843757717e-05,
+      "loss": 19.7902,
+      "step": 12100
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.792775259139386e-05,
+      "loss": 16.2434,
+      "step": 12101
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7920087046536614e-05,
+      "loss": 0.7883,
+      "step": 12102
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7912421803196726e-05,
+      "loss": 19.6047,
+      "step": 12103
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7904756861565556e-05,
+      "loss": 0.754,
+      "step": 12104
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.789709222183442e-05,
+      "loss": 0.7671,
+      "step": 12105
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.788942788419463e-05,
+      "loss": 0.6978,
+      "step": 12106
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.788176384883747e-05,
+      "loss": 17.7227,
+      "step": 12107
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.787410011595426e-05,
+      "loss": 16.9914,
+      "step": 12108
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.78664366857363e-05,
+      "loss": 18.3961,
+      "step": 12109
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.785877355837482e-05,
+      "loss": 16.851,
+      "step": 12110
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.785111073406115e-05,
+      "loss": 0.7936,
+      "step": 12111
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.784344821298651e-05,
+      "loss": 0.729,
+      "step": 12112
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7835785995342176e-05,
+      "loss": 16.8508,
+      "step": 12113
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.78281240813194e-05,
+      "loss": 0.777,
+      "step": 12114
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.782046247110945e-05,
+      "loss": 17.2404,
+      "step": 12115
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.781280116490354e-05,
+      "loss": 0.7818,
+      "step": 12116
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.780514016289287e-05,
+      "loss": 18.2413,
+      "step": 12117
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7797479465268694e-05,
+      "loss": 19.1991,
+      "step": 12118
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.778981907222221e-05,
+      "loss": 0.7843,
+      "step": 12119
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7782158983944646e-05,
+      "loss": 19.1016,
+      "step": 12120
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7774499200627166e-05,
+      "loss": 17.0988,
+      "step": 12121
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.776683972246098e-05,
+      "loss": 16.7742,
+      "step": 12122
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7759180549637295e-05,
+      "loss": 0.702,
+      "step": 12123
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7751521682347235e-05,
+      "loss": 0.7336,
+      "step": 12124
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7743863120781994e-05,
+      "loss": 16.7992,
+      "step": 12125
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7736204865132724e-05,
+      "loss": 0.7354,
+      "step": 12126
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7728546915590577e-05,
+      "loss": 0.7877,
+      "step": 12127
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.77208892723467e-05,
+      "loss": 0.756,
+      "step": 12128
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.771323193559225e-05,
+      "loss": 0.8059,
+      "step": 12129
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7705574905518345e-05,
+      "loss": 0.7587,
+      "step": 12130
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.769791818231607e-05,
+      "loss": 0.7579,
+      "step": 12131
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7690261766176565e-05,
+      "loss": 0.7491,
+      "step": 12132
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7682605657290944e-05,
+      "loss": 0.7787,
+      "step": 12133
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7674949855850304e-05,
+      "loss": 16.8749,
+      "step": 12134
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.766729436204572e-05,
+      "loss": 19.6251,
+      "step": 12135
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.765963917606828e-05,
+      "loss": 0.7203,
+      "step": 12136
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.765198429810909e-05,
+      "loss": 19.486,
+      "step": 12137
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.764432972835917e-05,
+      "loss": 0.7544,
+      "step": 12138
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.763667546700962e-05,
+      "loss": 18.4516,
+      "step": 12139
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.762902151425145e-05,
+      "loss": 0.7699,
+      "step": 12140
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7621367870275736e-05,
+      "loss": 17.8338,
+      "step": 12141
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7613714535273506e-05,
+      "loss": 0.7806,
+      "step": 12142
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7606061509435795e-05,
+      "loss": 0.7782,
+      "step": 12143
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7598408792953625e-05,
+      "loss": 19.2013,
+      "step": 12144
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.759075638601799e-05,
+      "loss": 18.3273,
+      "step": 12145
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.758310428881991e-05,
+      "loss": 18.5691,
+      "step": 12146
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.757545250155038e-05,
+      "loss": 16.4331,
+      "step": 12147
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7567801024400396e-05,
+      "loss": 0.8115,
+      "step": 12148
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.756014985756093e-05,
+      "loss": 0.7572,
+      "step": 12149
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7552499001222974e-05,
+      "loss": 0.7791,
+      "step": 12150
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7544848455577495e-05,
+      "loss": 0.8113,
+      "step": 12151
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.753719822081543e-05,
+      "loss": 0.8093,
+      "step": 12152
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7529548297127745e-05,
+      "loss": 0.7601,
+      "step": 12153
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7521898684705377e-05,
+      "loss": 16.9048,
+      "step": 12154
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7514249383739266e-05,
+      "loss": 0.7695,
+      "step": 12155
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7506600394420334e-05,
+      "loss": 0.7725,
+      "step": 12156
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.749895171693952e-05,
+      "loss": 19.5689,
+      "step": 12157
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.749130335148775e-05,
+      "loss": 18.9977,
+      "step": 12158
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.748365529825587e-05,
+      "loss": 18.5719,
+      "step": 12159
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.747600755743482e-05,
+      "loss": 18.0975,
+      "step": 12160
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.746836012921547e-05,
+      "loss": 18.8742,
+      "step": 12161
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.746071301378874e-05,
+      "loss": 0.7067,
+      "step": 12162
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7453066211345457e-05,
+      "loss": 19.0085,
+      "step": 12163
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7445419722076523e-05,
+      "loss": 0.7325,
+      "step": 12164
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.743777354617276e-05,
+      "loss": 15.8717,
+      "step": 12165
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7430127683825046e-05,
+      "loss": 19.037,
+      "step": 12166
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7422482135224216e-05,
+      "loss": 0.7765,
+      "step": 12167
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.74148369005611e-05,
+      "loss": 18.1505,
+      "step": 12168
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.740719198002652e-05,
+      "loss": 17.1021,
+      "step": 12169
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.739954737381131e-05,
+      "loss": 0.7414,
+      "step": 12170
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.73919030821063e-05,
+      "loss": 19.8685,
+      "step": 12171
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.738425910510224e-05,
+      "loss": 18.0452,
+      "step": 12172
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.737661544298996e-05,
+      "loss": 0.7044,
+      "step": 12173
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.736897209596023e-05,
+      "loss": 15.5485,
+      "step": 12174
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.736132906420384e-05,
+      "loss": 19.2508,
+      "step": 12175
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.735368634791159e-05,
+      "loss": 19.0481,
+      "step": 12176
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.734604394727419e-05,
+      "loss": 18.6032,
+      "step": 12177
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.733840186248244e-05,
+      "loss": 0.7584,
+      "step": 12178
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.733076009372706e-05,
+      "loss": 17.6261,
+      "step": 12179
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7323118641198796e-05,
+      "loss": 17.7807,
+      "step": 12180
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7315477505088405e-05,
+      "loss": 16.1297,
+      "step": 12181
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.730783668558657e-05,
+      "loss": 0.7381,
+      "step": 12182
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.730019618288403e-05,
+      "loss": 18.5354,
+      "step": 12183
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7292555997171495e-05,
+      "loss": 0.7921,
+      "step": 12184
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.728491612863969e-05,
+      "loss": 0.7262,
+      "step": 12185
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7277276577479265e-05,
+      "loss": 0.7609,
+      "step": 12186
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.726963734388092e-05,
+      "loss": 0.7554,
+      "step": 12187
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.726199842803532e-05,
+      "loss": 17.5485,
+      "step": 12188
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.725435983013316e-05,
+      "loss": 20.92,
+      "step": 12189
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7246721550365086e-05,
+      "loss": 15.89,
+      "step": 12190
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7239083588921755e-05,
+      "loss": 18.5485,
+      "step": 12191
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7231445945993826e-05,
+      "loss": 17.4826,
+      "step": 12192
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7223808621771904e-05,
+      "loss": 0.7719,
+      "step": 12193
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.721617161644664e-05,
+      "loss": 18.3629,
+      "step": 12194
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.720853493020865e-05,
+      "loss": 0.7792,
+      "step": 12195
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7200898563248544e-05,
+      "loss": 0.7257,
+      "step": 12196
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.719326251575693e-05,
+      "loss": 0.7171,
+      "step": 12197
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7185626787924416e-05,
+      "loss": 20.6298,
+      "step": 12198
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.71779913799416e-05,
+      "loss": 0.7899,
+      "step": 12199
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7170356291999035e-05,
+      "loss": 16.2318,
+      "step": 12200
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.71627215242873e-05,
+      "loss": 18.3051,
+      "step": 12201
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7155087076996966e-05,
+      "loss": 0.7543,
+      "step": 12202
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7147452950318585e-05,
+      "loss": 18.5405,
+      "step": 12203
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.713981914444273e-05,
+      "loss": 17.9192,
+      "step": 12204
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7132185659559906e-05,
+      "loss": 18.1203,
+      "step": 12205
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.71245524958607e-05,
+      "loss": 17.3059,
+      "step": 12206
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.711691965353558e-05,
+      "loss": 19.5797,
+      "step": 12207
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.710928713277507e-05,
+      "loss": 0.7864,
+      "step": 12208
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.710165493376972e-05,
+      "loss": 0.7764,
+      "step": 12209
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7094023056710005e-05,
+      "loss": 0.7949,
+      "step": 12210
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.708639150178641e-05,
+      "loss": 18.4442,
+      "step": 12211
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.707876026918944e-05,
+      "loss": 0.7883,
+      "step": 12212
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7071129359109577e-05,
+      "loss": 19.1133,
+      "step": 12213
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.706349877173727e-05,
+      "loss": 19.4375,
+      "step": 12214
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.705586850726298e-05,
+      "loss": 18.4331,
+      "step": 12215
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7048238565877165e-05,
+      "loss": 16.4747,
+      "step": 12216
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.704060894777027e-05,
+      "loss": 0.7774,
+      "step": 12217
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.703297965313275e-05,
+      "loss": 17.134,
+      "step": 12218
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7025350682155005e-05,
+      "loss": 0.6907,
+      "step": 12219
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.701772203502749e-05,
+      "loss": 0.7354,
+      "step": 12220
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.701009371194058e-05,
+      "loss": 0.7868,
+      "step": 12221
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.7002465713084696e-05,
+      "loss": 19.5916,
+      "step": 12222
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.699483803865024e-05,
+      "loss": 0.7859,
+      "step": 12223
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.698721068882759e-05,
+      "loss": 17.7886,
+      "step": 12224
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.697958366380714e-05,
+      "loss": 0.8026,
+      "step": 12225
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6971956963779245e-05,
+      "loss": 16.1281,
+      "step": 12226
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.69643305889343e-05,
+      "loss": 18.0681,
+      "step": 12227
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.695670453946263e-05,
+      "loss": 17.6573,
+      "step": 12228
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.694907881555458e-05,
+      "loss": 17.8324,
+      "step": 12229
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6941453417400506e-05,
+      "loss": 0.774,
+      "step": 12230
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.693382834519074e-05,
+      "loss": 0.7732,
+      "step": 12231
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.692620359911561e-05,
+      "loss": 17.9534,
+      "step": 12232
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.691857917936541e-05,
+      "loss": 0.7219,
+      "step": 12233
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.691095508613048e-05,
+      "loss": 17.6041,
+      "step": 12234
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6903331319601084e-05,
+      "loss": 19.7694,
+      "step": 12235
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.689570787996753e-05,
+      "loss": 17.2225,
+      "step": 12236
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.68880847674201e-05,
+      "loss": 0.7319,
+      "step": 12237
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6880461982149075e-05,
+      "loss": 0.731,
+      "step": 12238
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.68728395243447e-05,
+      "loss": 0.7503,
+      "step": 12239
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.686521739419727e-05,
+      "loss": 17.6729,
+      "step": 12240
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6857595591897e-05,
+      "loss": 0.7068,
+      "step": 12241
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6849974117634143e-05,
+      "loss": 0.7219,
+      "step": 12242
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6842352971598934e-05,
+      "loss": 0.7571,
+      "step": 12243
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.68347321539816e-05,
+      "loss": 0.7538,
+      "step": 12244
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6827111664972345e-05,
+      "loss": 17.5595,
+      "step": 12245
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.681949150476142e-05,
+      "loss": 18.3325,
+      "step": 12246
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6811871673538976e-05,
+      "loss": 20.5077,
+      "step": 12247
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.680425217149524e-05,
+      "loss": 0.7314,
+      "step": 12248
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.679663299882037e-05,
+      "loss": 0.7262,
+      "step": 12249
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.678901415570455e-05,
+      "loss": 16.3223,
+      "step": 12250
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.678139564233797e-05,
+      "loss": 17.6122,
+      "step": 12251
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.677377745891076e-05,
+      "loss": 0.7511,
+      "step": 12252
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.676615960561308e-05,
+      "loss": 18.2237,
+      "step": 12253
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.67585420826351e-05,
+      "loss": 17.9723,
+      "step": 12254
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.675092489016693e-05,
+      "loss": 0.7493,
+      "step": 12255
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.674330802839869e-05,
+      "loss": 17.1384,
+      "step": 12256
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6735691497520506e-05,
+      "loss": 19.796,
+      "step": 12257
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.672807529772249e-05,
+      "loss": 0.7021,
+      "step": 12258
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.672045942919474e-05,
+      "loss": 0.7383,
+      "step": 12259
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.671284389212737e-05,
+      "loss": 16.0062,
+      "step": 12260
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6705228686710446e-05,
+      "loss": 17.2889,
+      "step": 12261
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.669761381313406e-05,
+      "loss": 0.7565,
+      "step": 12262
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6689999271588246e-05,
+      "loss": 0.7656,
+      "step": 12263
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.66823850622631e-05,
+      "loss": 0.8444,
+      "step": 12264
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.667477118534867e-05,
+      "loss": 17.8311,
+      "step": 12265
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.666715764103498e-05,
+      "loss": 16.9599,
+      "step": 12266
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.665954442951208e-05,
+      "loss": 16.823,
+      "step": 12267
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.665193155097003e-05,
+      "loss": 16.9173,
+      "step": 12268
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.664431900559878e-05,
+      "loss": 0.7328,
+      "step": 12269
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.663670679358839e-05,
+      "loss": 0.7754,
+      "step": 12270
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.662909491512885e-05,
+      "loss": 16.9527,
+      "step": 12271
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6621483370410146e-05,
+      "loss": 16.3211,
+      "step": 12272
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.661387215962227e-05,
+      "loss": 17.159,
+      "step": 12273
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6606261282955214e-05,
+      "loss": 16.1765,
+      "step": 12274
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6598650740598937e-05,
+      "loss": 0.7701,
+      "step": 12275
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.659104053274339e-05,
+      "loss": 0.704,
+      "step": 12276
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6583430659578533e-05,
+      "loss": 0.727,
+      "step": 12277
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6575821121294305e-05,
+      "loss": 18.4445,
+      "step": 12278
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.656821191808066e-05,
+      "loss": 18.874,
+      "step": 12279
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6560603050127496e-05,
+      "loss": 0.7298,
+      "step": 12280
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.655299451762475e-05,
+      "loss": 0.7093,
+      "step": 12281
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.654538632076236e-05,
+      "loss": 0.7506,
+      "step": 12282
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.653777845973017e-05,
+      "loss": 18.8266,
+      "step": 12283
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6530170934718106e-05,
+      "loss": 0.6922,
+      "step": 12284
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.652256374591606e-05,
+      "loss": 0.7654,
+      "step": 12285
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.65149568935139e-05,
+      "loss": 17.8416,
+      "step": 12286
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.650735037770148e-05,
+      "loss": 0.7693,
+      "step": 12287
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.64997441986687e-05,
+      "loss": 16.003,
+      "step": 12288
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.649213835660537e-05,
+      "loss": 0.6949,
+      "step": 12289
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.648453285170136e-05,
+      "loss": 18.5228,
+      "step": 12290
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.647692768414649e-05,
+      "loss": 19.215,
+      "step": 12291
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.646932285413058e-05,
+      "loss": 15.5721,
+      "step": 12292
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.646171836184349e-05,
+      "loss": 0.7717,
+      "step": 12293
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.645411420747498e-05,
+      "loss": 17.384,
+      "step": 12294
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.644651039121487e-05,
+      "loss": 15.3413,
+      "step": 12295
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.643890691325298e-05,
+      "loss": 17.7024,
+      "step": 12296
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6431303773779044e-05,
+      "loss": 16.593,
+      "step": 12297
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.642370097298287e-05,
+      "loss": 0.7635,
+      "step": 12298
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.641609851105423e-05,
+      "loss": 18.0169,
+      "step": 12299
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.640849638818286e-05,
+      "loss": 0.7914,
+      "step": 12300
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.640089460455852e-05,
+      "loss": 0.6872,
+      "step": 12301
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6393293160370974e-05,
+      "loss": 0.7328,
+      "step": 12302
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6385692055809935e-05,
+      "loss": 17.6653,
+      "step": 12303
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.637809129106513e-05,
+      "loss": 0.7023,
+      "step": 12304
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.637049086632626e-05,
+      "loss": 0.7647,
+      "step": 12305
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6362890781783055e-05,
+      "loss": 18.9127,
+      "step": 12306
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.635529103762523e-05,
+      "loss": 19.9827,
+      "step": 12307
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.634769163404243e-05,
+      "loss": 0.756,
+      "step": 12308
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.634009257122437e-05,
+      "loss": 17.3362,
+      "step": 12309
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6332493849360735e-05,
+      "loss": 0.7664,
+      "step": 12310
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.632489546864116e-05,
+      "loss": 0.7569,
+      "step": 12311
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.631729742925532e-05,
+      "loss": 0.7277,
+      "step": 12312
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6309699731392865e-05,
+      "loss": 18.054,
+      "step": 12313
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.630210237524342e-05,
+      "loss": 17.3825,
+      "step": 12314
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.629450536099663e-05,
+      "loss": 0.7596,
+      "step": 12315
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.628690868884213e-05,
+      "loss": 0.7503,
+      "step": 12316
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.627931235896951e-05,
+      "loss": 16.9148,
+      "step": 12317
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.627171637156839e-05,
+      "loss": 0.7505,
+      "step": 12318
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.626412072682835e-05,
+      "loss": 18.9678,
+      "step": 12319
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6256525424939004e-05,
+      "loss": 18.6864,
+      "step": 12320
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.624893046608992e-05,
+      "loss": 0.7282,
+      "step": 12321
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6241335850470665e-05,
+      "loss": 0.826,
+      "step": 12322
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.623374157827083e-05,
+      "loss": 0.7508,
+      "step": 12323
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.622614764967993e-05,
+      "loss": 18.3189,
+      "step": 12324
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.621855406488753e-05,
+      "loss": 17.7223,
+      "step": 12325
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6210960824083166e-05,
+      "loss": 17.6906,
+      "step": 12326
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.620336792745638e-05,
+      "loss": 0.7512,
+      "step": 12327
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.619577537519666e-05,
+      "loss": 0.7568,
+      "step": 12328
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.618818316749355e-05,
+      "loss": 19.0702,
+      "step": 12329
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.618059130453656e-05,
+      "loss": 18.6478,
+      "step": 12330
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.617299978651514e-05,
+      "loss": 15.9186,
+      "step": 12331
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.616540861361882e-05,
+      "loss": 18.3611,
+      "step": 12332
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.615781778603705e-05,
+      "loss": 17.5857,
+      "step": 12333
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.61502273039593e-05,
+      "loss": 17.3565,
+      "step": 12334
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.614263716757506e-05,
+      "loss": 19.941,
+      "step": 12335
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.613504737707374e-05,
+      "loss": 0.7136,
+      "step": 12336
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.612745793264483e-05,
+      "loss": 15.9669,
+      "step": 12337
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6119868834477715e-05,
+      "loss": 0.7497,
+      "step": 12338
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.611228008276184e-05,
+      "loss": 18.4002,
+      "step": 12339
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.610469167768663e-05,
+      "loss": 0.7718,
+      "step": 12340
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.60971036194415e-05,
+      "loss": 0.7459,
+      "step": 12341
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6089515908215826e-05,
+      "loss": 17.0086,
+      "step": 12342
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.608192854419901e-05,
+      "loss": 18.8101,
+      "step": 12343
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6074341527580455e-05,
+      "loss": 17.8843,
+      "step": 12344
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.60667548585495e-05,
+      "loss": 17.7779,
+      "step": 12345
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.605916853729553e-05,
+      "loss": 18.6081,
+      "step": 12346
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.605158256400789e-05,
+      "loss": 0.7854,
+      "step": 12347
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.604399693887593e-05,
+      "loss": 16.5695,
+      "step": 12348
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.6036411662089005e-05,
+      "loss": 19.7966,
+      "step": 12349
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.602882673383642e-05,
+      "loss": 18.3587,
+      "step": 12350
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.602124215430753e-05,
+      "loss": 0.7515,
+      "step": 12351
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.601365792369161e-05,
+      "loss": 0.7559,
+      "step": 12352
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.600607404217798e-05,
+      "loss": 0.8074,
+      "step": 12353
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5998490509955946e-05,
+      "loss": 16.4542,
+      "step": 12354
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.599090732721479e-05,
+      "loss": 17.7169,
+      "step": 12355
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.598332449414378e-05,
+      "loss": 18.7428,
+      "step": 12356
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5975742010932177e-05,
+      "loss": 0.7538,
+      "step": 12357
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.596815987776929e-05,
+      "loss": 0.6994,
+      "step": 12358
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.596057809484431e-05,
+      "loss": 0.7493,
+      "step": 12359
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5952996662346516e-05,
+      "loss": 17.9418,
+      "step": 12360
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5945415580465124e-05,
+      "loss": 16.2151,
+      "step": 12361
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5937834849389366e-05,
+      "loss": 17.4219,
+      "step": 12362
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5930254469308475e-05,
+      "loss": 0.7432,
+      "step": 12363
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.592267444041163e-05,
+      "loss": 17.7545,
+      "step": 12364
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5915094762888065e-05,
+      "loss": 0.756,
+      "step": 12365
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.590751543692694e-05,
+      "loss": 0.7493,
+      "step": 12366
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.589993646271744e-05,
+      "loss": 0.7397,
+      "step": 12367
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.589235784044874e-05,
+      "loss": 0.7354,
+      "step": 12368
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5884779570310036e-05,
+      "loss": 16.3595,
+      "step": 12369
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.587720165249043e-05,
+      "loss": 18.6158,
+      "step": 12370
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.58696240871791e-05,
+      "loss": 0.7308,
+      "step": 12371
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.586204687456521e-05,
+      "loss": 16.734,
+      "step": 12372
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.585447001483783e-05,
+      "loss": 16.9114,
+      "step": 12373
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.584689350818612e-05,
+      "loss": 18.0579,
+      "step": 12374
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5839317354799165e-05,
+      "loss": 0.7889,
+      "step": 12375
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5831741554866086e-05,
+      "loss": 0.7285,
+      "step": 12376
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.582416610857598e-05,
+      "loss": 18.3673,
+      "step": 12377
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.581659101611792e-05,
+      "loss": 0.7902,
+      "step": 12378
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5809016277680994e-05,
+      "loss": 0.7132,
+      "step": 12379
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.580144189345426e-05,
+      "loss": 17.6718,
+      "step": 12380
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.579386786362676e-05,
+      "loss": 0.7707,
+      "step": 12381
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.578629418838757e-05,
+      "loss": 17.5036,
+      "step": 12382
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.577872086792574e-05,
+      "loss": 18.478,
+      "step": 12383
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5771147902430266e-05,
+      "loss": 0.7412,
+      "step": 12384
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.576357529209019e-05,
+      "loss": 0.7138,
+      "step": 12385
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5756003037094544e-05,
+      "loss": 0.7707,
+      "step": 12386
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.57484311376323e-05,
+      "loss": 0.7454,
+      "step": 12387
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.574085959389247e-05,
+      "loss": 19.276,
+      "step": 12388
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5733288406064044e-05,
+      "loss": 18.5794,
+      "step": 12389
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5725717574335985e-05,
+      "loss": 17.3513,
+      "step": 12390
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.571814709889729e-05,
+      "loss": 0.7673,
+      "step": 12391
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.571057697993691e-05,
+      "loss": 0.725,
+      "step": 12392
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.570300721764378e-05,
+      "loss": 18.4025,
+      "step": 12393
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.569543781220685e-05,
+      "loss": 0.7101,
+      "step": 12394
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.568786876381505e-05,
+      "loss": 16.3988,
+      "step": 12395
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5680300072657325e-05,
+      "loss": 16.401,
+      "step": 12396
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.567273173892258e-05,
+      "loss": 17.6654,
+      "step": 12397
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.566516376279972e-05,
+      "loss": 16.7697,
+      "step": 12398
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.565759614447766e-05,
+      "loss": 18.2812,
+      "step": 12399
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.565002888414526e-05,
+      "loss": 0.7506,
+      "step": 12400
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5642461981991415e-05,
+      "loss": 17.7505,
+      "step": 12401
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.563489543820501e-05,
+      "loss": 18.4556,
+      "step": 12402
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.562732925297487e-05,
+      "loss": 18.0628,
+      "step": 12403
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.561976342648989e-05,
+      "loss": 0.7716,
+      "step": 12404
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.56121979589389e-05,
+      "loss": 0.7469,
+      "step": 12405
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.560463285051073e-05,
+      "loss": 0.746,
+      "step": 12406
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5597068101394226e-05,
+      "loss": 19.4393,
+      "step": 12407
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5589503711778174e-05,
+      "loss": 17.3482,
+      "step": 12408
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.55819396818514e-05,
+      "loss": 18.0415,
+      "step": 12409
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.557437601180269e-05,
+      "loss": 16.822,
+      "step": 12410
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.556681270182087e-05,
+      "loss": 0.7351,
+      "step": 12411
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.555924975209469e-05,
+      "loss": 0.7098,
+      "step": 12412
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5551687162812965e-05,
+      "loss": 0.7681,
+      "step": 12413
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.554412493416439e-05,
+      "loss": 0.7594,
+      "step": 12414
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5536563066337764e-05,
+      "loss": 18.2349,
+      "step": 12415
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.552900155952184e-05,
+      "loss": 18.5059,
+      "step": 12416
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5521440413905324e-05,
+      "loss": 19.4974,
+      "step": 12417
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.551387962967696e-05,
+      "loss": 17.5676,
+      "step": 12418
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5506319207025485e-05,
+      "loss": 20.2699,
+      "step": 12419
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.549875914613959e-05,
+      "loss": 0.7421,
+      "step": 12420
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.549119944720798e-05,
+      "loss": 16.0519,
+      "step": 12421
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5483640110419326e-05,
+      "loss": 17.6098,
+      "step": 12422
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.547608113596233e-05,
+      "loss": 0.7447,
+      "step": 12423
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5468522524025675e-05,
+      "loss": 0.7848,
+      "step": 12424
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5460964274798027e-05,
+      "loss": 0.7409,
+      "step": 12425
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.545340638846801e-05,
+      "loss": 17.1813,
+      "step": 12426
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.544584886522432e-05,
+      "loss": 16.7795,
+      "step": 12427
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5438291705255534e-05,
+      "loss": 0.7716,
+      "step": 12428
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.543073490875032e-05,
+      "loss": 0.7203,
+      "step": 12429
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.54231784758973e-05,
+      "loss": 0.7527,
+      "step": 12430
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5415622406885063e-05,
+      "loss": 16.5523,
+      "step": 12431
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.540806670190222e-05,
+      "loss": 0.7583,
+      "step": 12432
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.540051136113737e-05,
+      "loss": 18.9368,
+      "step": 12433
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.53929563847791e-05,
+      "loss": 0.7679,
+      "step": 12434
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.538540177301596e-05,
+      "loss": 19.0924,
+      "step": 12435
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5377847526036524e-05,
+      "loss": 16.4754,
+      "step": 12436
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5370293644029355e-05,
+      "loss": 16.4746,
+      "step": 12437
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.536274012718299e-05,
+      "loss": 19.5509,
+      "step": 12438
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5355186975685995e-05,
+      "loss": 17.4742,
+      "step": 12439
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.534763418972686e-05,
+      "loss": 17.4919,
+      "step": 12440
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.534008176949415e-05,
+      "loss": 0.7468,
+      "step": 12441
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.533252971517631e-05,
+      "loss": 18.5583,
+      "step": 12442
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.532497802696189e-05,
+      "loss": 0.7462,
+      "step": 12443
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.531742670503937e-05,
+      "loss": 18.8529,
+      "step": 12444
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.530987574959723e-05,
+      "loss": 17.4002,
+      "step": 12445
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.530232516082395e-05,
+      "loss": 15.2864,
+      "step": 12446
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5294774938907984e-05,
+      "loss": 0.7525,
+      "step": 12447
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.52872250840378e-05,
+      "loss": 18.1484,
+      "step": 12448
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5279675596401846e-05,
+      "loss": 16.3299,
+      "step": 12449
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5272126476188535e-05,
+      "loss": 17.2772,
+      "step": 12450
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.526457772358631e-05,
+      "loss": 16.8093,
+      "step": 12451
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5257029338783584e-05,
+      "loss": 16.3879,
+      "step": 12452
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.524948132196879e-05,
+      "loss": 0.7607,
+      "step": 12453
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.52419336733303e-05,
+      "loss": 18.0087,
+      "step": 12454
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.523438639305653e-05,
+      "loss": 0.7531,
+      "step": 12455
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.522683948133583e-05,
+      "loss": 18.1193,
+      "step": 12456
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.521929293835658e-05,
+      "loss": 0.7272,
+      "step": 12457
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5211746764307176e-05,
+      "loss": 19.0362,
+      "step": 12458
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5204200959375934e-05,
+      "loss": 0.7756,
+      "step": 12459
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5196655523751207e-05,
+      "loss": 0.6932,
+      "step": 12460
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5189110457621346e-05,
+      "loss": 16.2232,
+      "step": 12461
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.518156576117469e-05,
+      "loss": 19.4728,
+      "step": 12462
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.517402143459953e-05,
+      "loss": 18.5678,
+      "step": 12463
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.516647747808417e-05,
+      "loss": 17.4344,
+      "step": 12464
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.515893389181692e-05,
+      "loss": 0.7687,
+      "step": 12465
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.515139067598606e-05,
+      "loss": 0.7656,
+      "step": 12466
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5143847830779885e-05,
+      "loss": 0.7892,
+      "step": 12467
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5136305356386685e-05,
+      "loss": 17.9555,
+      "step": 12468
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.512876325299467e-05,
+      "loss": 20.0578,
+      "step": 12469
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.512122152079212e-05,
+      "loss": 0.7416,
+      "step": 12470
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5113680159967276e-05,
+      "loss": 19.169,
+      "step": 12471
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5106139170708383e-05,
+      "loss": 19.3968,
+      "step": 12472
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.509859855320365e-05,
+      "loss": 0.7721,
+      "step": 12473
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.50910583076413e-05,
+      "loss": 17.3519,
+      "step": 12474
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.508351843420956e-05,
+      "loss": 18.5137,
+      "step": 12475
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5075978933096584e-05,
+      "loss": 15.8125,
+      "step": 12476
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.506843980449059e-05,
+      "loss": 16.8932,
+      "step": 12477
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.506090104857974e-05,
+      "loss": 0.7521,
+      "step": 12478
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5053362665552216e-05,
+      "loss": 0.7522,
+      "step": 12479
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.5045824655596164e-05,
+      "loss": 0.7607,
+      "step": 12480
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.503828701889976e-05,
+      "loss": 0.7323,
+      "step": 12481
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.503074975565114e-05,
+      "loss": 17.7945,
+      "step": 12482
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.502321286603839e-05,
+      "loss": 0.7752,
+      "step": 12483
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.501567635024968e-05,
+      "loss": 19.201,
+      "step": 12484
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.500814020847311e-05,
+      "loss": 0.6825,
+      "step": 12485
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.500060444089678e-05,
+      "loss": 0.7686,
+      "step": 12486
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.499306904770879e-05,
+      "loss": 0.7184,
+      "step": 12487
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4985534029097214e-05,
+      "loss": 19.1346,
+      "step": 12488
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.497799938525017e-05,
+      "loss": 0.7483,
+      "step": 12489
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.497046511635565e-05,
+      "loss": 0.717,
+      "step": 12490
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4962931222601785e-05,
+      "loss": 16.7884,
+      "step": 12491
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.495539770417657e-05,
+      "loss": 18.1878,
+      "step": 12492
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.494786456126806e-05,
+      "loss": 0.766,
+      "step": 12493
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.49403317940643e-05,
+      "loss": 16.7547,
+      "step": 12494
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.493279940275329e-05,
+      "loss": 19.6575,
+      "step": 12495
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4925267387523075e-05,
+      "loss": 0.8057,
+      "step": 12496
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4917735748561595e-05,
+      "loss": 19.3165,
+      "step": 12497
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.491020448605688e-05,
+      "loss": 20.0677,
+      "step": 12498
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.49026736001969e-05,
+      "loss": 0.8132,
+      "step": 12499
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4895143091169654e-05,
+      "loss": 17.8315,
+      "step": 12500
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.488761295916307e-05,
+      "loss": 0.8037,
+      "step": 12501
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4880083204365115e-05,
+      "loss": 16.7981,
+      "step": 12502
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.487255382696376e-05,
+      "loss": 17.1762,
+      "step": 12503
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4865024827146895e-05,
+      "loss": 0.7954,
+      "step": 12504
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4857496205102474e-05,
+      "loss": 0.7075,
+      "step": 12505
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.484996796101839e-05,
+      "loss": 0.7388,
+      "step": 12506
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.484244009508257e-05,
+      "loss": 0.756,
+      "step": 12507
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.483491260748291e-05,
+      "loss": 17.3643,
+      "step": 12508
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.482738549840731e-05,
+      "loss": 0.75,
+      "step": 12509
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.481985876804363e-05,
+      "loss": 0.7486,
+      "step": 12510
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.481233241657972e-05,
+      "loss": 20.0286,
+      "step": 12511
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.480480644420346e-05,
+      "loss": 0.8131,
+      "step": 12512
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.479728085110271e-05,
+      "loss": 18.7558,
+      "step": 12513
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.478975563746531e-05,
+      "loss": 0.7634,
+      "step": 12514
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4782230803479065e-05,
+      "loss": 17.1094,
+      "step": 12515
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.477470634933182e-05,
+      "loss": 0.7607,
+      "step": 12516
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.47671822752114e-05,
+      "loss": 16.8508,
+      "step": 12517
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.475965858130556e-05,
+      "loss": 0.7055,
+      "step": 12518
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.475213526780214e-05,
+      "loss": 0.7974,
+      "step": 12519
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.47446123348889e-05,
+      "loss": 18.5226,
+      "step": 12520
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4737089782753616e-05,
+      "loss": 0.7614,
+      "step": 12521
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.472956761158405e-05,
+      "loss": 16.2956,
+      "step": 12522
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.472204582156798e-05,
+      "loss": 0.7808,
+      "step": 12523
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.471452441289314e-05,
+      "loss": 0.7648,
+      "step": 12524
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.470700338574725e-05,
+      "loss": 18.6759,
+      "step": 12525
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.469948274031804e-05,
+      "loss": 0.7503,
+      "step": 12526
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.469196247679323e-05,
+      "loss": 16.6283,
+      "step": 12527
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4684442595360544e-05,
+      "loss": 0.7442,
+      "step": 12528
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.467692309620766e-05,
+      "loss": 0.7844,
+      "step": 12529
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4669403979522275e-05,
+      "loss": 17.8504,
+      "step": 12530
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.466188524549208e-05,
+      "loss": 17.0302,
+      "step": 12531
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.465436689430472e-05,
+      "loss": 0.7877,
+      "step": 12532
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.464684892614787e-05,
+      "loss": 17.6976,
+      "step": 12533
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4639331341209175e-05,
+      "loss": 0.7553,
+      "step": 12534
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.463181413967627e-05,
+      "loss": 16.4547,
+      "step": 12535
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4624297321736796e-05,
+      "loss": 0.7305,
+      "step": 12536
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4616780887578384e-05,
+      "loss": 0.7442,
+      "step": 12537
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.460926483738863e-05,
+      "loss": 17.1879,
+      "step": 12538
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4601749171355136e-05,
+      "loss": 18.9213,
+      "step": 12539
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4594233889665496e-05,
+      "loss": 16.2304,
+      "step": 12540
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4586718992507286e-05,
+      "loss": 17.9854,
+      "step": 12541
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.457920448006811e-05,
+      "loss": 19.3233,
+      "step": 12542
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.45716903525355e-05,
+      "loss": 16.4784,
+      "step": 12543
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4564176610097033e-05,
+      "loss": 19.0678,
+      "step": 12544
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4556663252940226e-05,
+      "loss": 17.4655,
+      "step": 12545
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4549150281252636e-05,
+      "loss": 0.753,
+      "step": 12546
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.454163769522178e-05,
+      "loss": 15.5698,
+      "step": 12547
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4534125495035186e-05,
+      "loss": 16.573,
+      "step": 12548
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.452661368088034e-05,
+      "loss": 0.7831,
+      "step": 12549
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4519102252944745e-05,
+      "loss": 19.4036,
+      "step": 12550
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.451159121141592e-05,
+      "loss": 17.7649,
+      "step": 12551
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.45040805564813e-05,
+      "loss": 17.6998,
+      "step": 12552
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4496570288328364e-05,
+      "loss": 0.778,
+      "step": 12553
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4489060407144566e-05,
+      "loss": 0.7317,
+      "step": 12554
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.448155091311737e-05,
+      "loss": 16.5792,
+      "step": 12555
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.447404180643421e-05,
+      "loss": 0.7736,
+      "step": 12556
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4466533087282506e-05,
+      "loss": 16.9364,
+      "step": 12557
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.44590247558497e-05,
+      "loss": 18.9725,
+      "step": 12558
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.445151681232316e-05,
+      "loss": 16.7182,
+      "step": 12559
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4444009256890316e-05,
+      "loss": 18.3344,
+      "step": 12560
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.443650208973857e-05,
+      "loss": 16.424,
+      "step": 12561
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.442899531105526e-05,
+      "loss": 15.6744,
+      "step": 12562
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.442148892102779e-05,
+      "loss": 17.3729,
+      "step": 12563
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.441398291984351e-05,
+      "loss": 18.2458,
+      "step": 12564
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.44064773076898e-05,
+      "loss": 0.73,
+      "step": 12565
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4398972084753965e-05,
+      "loss": 0.7713,
+      "step": 12566
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.439146725122334e-05,
+      "loss": 0.7771,
+      "step": 12567
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.438396280728525e-05,
+      "loss": 18.1784,
+      "step": 12568
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.437645875312703e-05,
+      "loss": 0.7582,
+      "step": 12569
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.436895508893596e-05,
+      "loss": 0.7406,
+      "step": 12570
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4361451814899335e-05,
+      "loss": 0.7566,
+      "step": 12571
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.435394893120447e-05,
+      "loss": 0.6975,
+      "step": 12572
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.434644643803858e-05,
+      "loss": 0.7593,
+      "step": 12573
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4338944335588976e-05,
+      "loss": 0.7566,
+      "step": 12574
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.43314426240429e-05,
+      "loss": 0.8029,
+      "step": 12575
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.432394130358758e-05,
+      "loss": 0.7698,
+      "step": 12576
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4316440374410265e-05,
+      "loss": 17.438,
+      "step": 12577
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.430893983669818e-05,
+      "loss": 0.7173,
+      "step": 12578
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.430143969063856e-05,
+      "loss": 18.6014,
+      "step": 12579
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4293939936418586e-05,
+      "loss": 18.403,
+      "step": 12580
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.428644057422544e-05,
+      "loss": 0.7416,
+      "step": 12581
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.427894160424633e-05,
+      "loss": 19.1628,
+      "step": 12582
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.427144302666843e-05,
+      "loss": 0.7122,
+      "step": 12583
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4263944841678895e-05,
+      "loss": 0.7842,
+      "step": 12584
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.42564470494649e-05,
+      "loss": 17.4987,
+      "step": 12585
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.424894965021359e-05,
+      "loss": 0.7765,
+      "step": 12586
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.424145264411208e-05,
+      "loss": 17.1974,
+      "step": 12587
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.42339560313475e-05,
+      "loss": 16.8822,
+      "step": 12588
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4226459812107003e-05,
+      "loss": 18.61,
+      "step": 12589
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.421896398657765e-05,
+      "loss": 0.7432,
+      "step": 12590
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4211468554946566e-05,
+      "loss": 0.7518,
+      "step": 12591
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4203973517400826e-05,
+      "loss": 15.5003,
+      "step": 12592
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4196478874127536e-05,
+      "loss": 16.722,
+      "step": 12593
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.418898462531372e-05,
+      "loss": 0.8138,
+      "step": 12594
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.418149077114646e-05,
+      "loss": 0.8025,
+      "step": 12595
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4173997311812805e-05,
+      "loss": 18.4643,
+      "step": 12596
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4166504247499785e-05,
+      "loss": 0.781,
+      "step": 12597
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.415901157839445e-05,
+      "loss": 18.7639,
+      "step": 12598
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.415151930468379e-05,
+      "loss": 0.7715,
+      "step": 12599
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.414402742655484e-05,
+      "loss": 0.7932,
+      "step": 12600
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.413653594419457e-05,
+      "loss": 0.7416,
+      "step": 12601
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.412904485778999e-05,
+      "loss": 0.7622,
+      "step": 12602
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.412155416752808e-05,
+      "loss": 16.8372,
+      "step": 12603
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.411406387359579e-05,
+      "loss": 0.7392,
+      "step": 12604
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4106573976180104e-05,
+      "loss": 16.2071,
+      "step": 12605
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.409908447546795e-05,
+      "loss": 17.2592,
+      "step": 12606
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.40915953716463e-05,
+      "loss": 0.7492,
+      "step": 12607
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.408410666490206e-05,
+      "loss": 17.8064,
+      "step": 12608
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.407661835542213e-05,
+      "loss": 0.7537,
+      "step": 12609
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.406913044339345e-05,
+      "loss": 18.6757,
+      "step": 12610
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.40616429290029e-05,
+      "loss": 0.7736,
+      "step": 12611
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4054155812437405e-05,
+      "loss": 18.3806,
+      "step": 12612
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.404666909388381e-05,
+      "loss": 0.7666,
+      "step": 12613
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.403918277352901e-05,
+      "loss": 15.6775,
+      "step": 12614
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4031696851559834e-05,
+      "loss": 0.7679,
+      "step": 12615
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.402421132816316e-05,
+      "loss": 0.7387,
+      "step": 12616
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.4016726203525815e-05,
+      "loss": 16.3367,
+      "step": 12617
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.400924147783463e-05,
+      "loss": 0.7629,
+      "step": 12618
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.400175715127643e-05,
+      "loss": 18.2184,
+      "step": 12619
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.399427322403804e-05,
+      "loss": 0.726,
+      "step": 12620
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3986789696306234e-05,
+      "loss": 17.8169,
+      "step": 12621
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3979306568267816e-05,
+      "loss": 0.7822,
+      "step": 12622
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.397182384010956e-05,
+      "loss": 16.73,
+      "step": 12623
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.396434151201823e-05,
+      "loss": 0.7943,
+      "step": 12624
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.395685958418061e-05,
+      "loss": 17.9412,
+      "step": 12625
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.394937805678344e-05,
+      "loss": 0.7615,
+      "step": 12626
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.394189693001346e-05,
+      "loss": 0.7565,
+      "step": 12627
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3934416204057395e-05,
+      "loss": 0.8291,
+      "step": 12628
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3926935879101965e-05,
+      "loss": 18.067,
+      "step": 12629
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.391945595533389e-05,
+      "loss": 16.5373,
+      "step": 12630
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.391197643293987e-05,
+      "loss": 0.7493,
+      "step": 12631
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3904497312106574e-05,
+      "loss": 0.7576,
+      "step": 12632
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3897018593020716e-05,
+      "loss": 18.4675,
+      "step": 12633
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.388954027586896e-05,
+      "loss": 0.7335,
+      "step": 12634
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.388206236083794e-05,
+      "loss": 18.7314,
+      "step": 12635
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.387458484811433e-05,
+      "loss": 17.9952,
+      "step": 12636
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.386710773788476e-05,
+      "loss": 18.507,
+      "step": 12637
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.385963103033587e-05,
+      "loss": 0.7898,
+      "step": 12638
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.385215472565426e-05,
+      "loss": 19.7869,
+      "step": 12639
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3844678824026575e-05,
+      "loss": 0.7499,
+      "step": 12640
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3837203325639394e-05,
+      "loss": 0.8178,
+      "step": 12641
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.382972823067931e-05,
+      "loss": 0.756,
+      "step": 12642
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.382225353933288e-05,
+      "loss": 20.2787,
+      "step": 12643
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.381477925178671e-05,
+      "loss": 16.4904,
+      "step": 12644
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.380730536822735e-05,
+      "loss": 0.7349,
+      "step": 12645
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.379983188884134e-05,
+      "loss": 0.7546,
+      "step": 12646
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.379235881381522e-05,
+      "loss": 18.6631,
+      "step": 12647
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3784886143335536e-05,
+      "loss": 19.1013,
+      "step": 12648
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.377741387758879e-05,
+      "loss": 16.325,
+      "step": 12649
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.37699420167615e-05,
+      "loss": 17.179,
+      "step": 12650
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.376247056104016e-05,
+      "loss": 16.6765,
+      "step": 12651
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.375499951061125e-05,
+      "loss": 18.3192,
+      "step": 12652
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3747528865661274e-05,
+      "loss": 0.7468,
+      "step": 12653
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3740058626376695e-05,
+      "loss": 18.5319,
+      "step": 12654
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.373258879294396e-05,
+      "loss": 19.5888,
+      "step": 12655
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.372511936554953e-05,
+      "loss": 0.7664,
+      "step": 12656
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3717650344379826e-05,
+      "loss": 0.7723,
+      "step": 12657
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.371018172962128e-05,
+      "loss": 0.7912,
+      "step": 12658
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.370271352146034e-05,
+      "loss": 19.5461,
+      "step": 12659
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.369524572008337e-05,
+      "loss": 0.7957,
+      "step": 12660
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3687778325676795e-05,
+      "loss": 0.7782,
+      "step": 12661
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.368031133842702e-05,
+      "loss": 18.7233,
+      "step": 12662
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.367284475852037e-05,
+      "loss": 0.7794,
+      "step": 12663
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.366537858614326e-05,
+      "loss": 0.7348,
+      "step": 12664
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.365791282148202e-05,
+      "loss": 17.7045,
+      "step": 12665
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3650447464723015e-05,
+      "loss": 0.7758,
+      "step": 12666
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.364298251605256e-05,
+      "loss": 0.7483,
+      "step": 12667
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3635517975657014e-05,
+      "loss": 16.0241,
+      "step": 12668
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.362805384372267e-05,
+      "loss": 16.56,
+      "step": 12669
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.362059012043585e-05,
+      "loss": 18.5015,
+      "step": 12670
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.361312680598282e-05,
+      "loss": 17.3178,
+      "step": 12671
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.360566390054989e-05,
+      "loss": 16.9684,
+      "step": 12672
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3598201404323346e-05,
+      "loss": 18.268,
+      "step": 12673
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3590739317489426e-05,
+      "loss": 0.7914,
+      "step": 12674
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3583277640234405e-05,
+      "loss": 0.7598,
+      "step": 12675
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.357581637274454e-05,
+      "loss": 19.2374,
+      "step": 12676
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.356835551520603e-05,
+      "loss": 0.76,
+      "step": 12677
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.356089506780511e-05,
+      "loss": 0.7671,
+      "step": 12678
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.355343503072802e-05,
+      "loss": 18.4556,
+      "step": 12679
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3545975404160936e-05,
+      "loss": 0.7801,
+      "step": 12680
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.353851618829006e-05,
+      "loss": 0.7543,
+      "step": 12681
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.353105738330159e-05,
+      "loss": 16.668,
+      "step": 12682
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3523598989381686e-05,
+      "loss": 0.77,
+      "step": 12683
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.351614100671652e-05,
+      "loss": 19.5583,
+      "step": 12684
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3508683435492225e-05,
+      "loss": 20.5187,
+      "step": 12685
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3501226275894956e-05,
+      "loss": 0.7197,
+      "step": 12686
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.349376952811085e-05,
+      "loss": 0.7516,
+      "step": 12687
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.348631319232603e-05,
+      "loss": 0.7142,
+      "step": 12688
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3478857268726594e-05,
+      "loss": 19.0946,
+      "step": 12689
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3471401757498673e-05,
+      "loss": 0.7787,
+      "step": 12690
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3463946658828314e-05,
+      "loss": 18.0478,
+      "step": 12691
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.345649197290163e-05,
+      "loss": 0.8003,
+      "step": 12692
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.344903769990469e-05,
+      "loss": 15.7979,
+      "step": 12693
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.344158384002354e-05,
+      "loss": 0.7832,
+      "step": 12694
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3434130393444244e-05,
+      "loss": 0.78,
+      "step": 12695
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.342667736035284e-05,
+      "loss": 19.1086,
+      "step": 12696
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.341922474093535e-05,
+      "loss": 16.8541,
+      "step": 12697
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3411772535377794e-05,
+      "loss": 0.726,
+      "step": 12698
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.340432074386618e-05,
+      "loss": 0.7736,
+      "step": 12699
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.33968693665865e-05,
+      "loss": 0.786,
+      "step": 12700
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.338941840372477e-05,
+      "loss": 0.764,
+      "step": 12701
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.338196785546694e-05,
+      "loss": 0.8083,
+      "step": 12702
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3374517721999e-05,
+      "loss": 17.2582,
+      "step": 12703
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3367068003506886e-05,
+      "loss": 0.7598,
+      "step": 12704
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3359618700176544e-05,
+      "loss": 0.7757,
+      "step": 12705
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.335216981219392e-05,
+      "loss": 20.3524,
+      "step": 12706
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3344721339744946e-05,
+      "loss": 0.7283,
+      "step": 12707
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3337273283015523e-05,
+      "loss": 0.7855,
+      "step": 12708
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.332982564219156e-05,
+      "loss": 0.7249,
+      "step": 12709
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.332237841745898e-05,
+      "loss": 18.2079,
+      "step": 12710
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.331493160900362e-05,
+      "loss": 0.7459,
+      "step": 12711
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.330748521701139e-05,
+      "loss": 0.7252,
+      "step": 12712
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3300039241668125e-05,
+      "loss": 18.5374,
+      "step": 12713
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.32925936831597e-05,
+      "loss": 17.3407,
+      "step": 12714
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3285148541671954e-05,
+      "loss": 17.9702,
+      "step": 12715
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3277703817390714e-05,
+      "loss": 19.181,
+      "step": 12716
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3270259510501824e-05,
+      "loss": 17.1276,
+      "step": 12717
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.326281562119106e-05,
+      "loss": 17.3071,
+      "step": 12718
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3255372149644223e-05,
+      "loss": 0.7365,
+      "step": 12719
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3247929096047136e-05,
+      "loss": 0.6653,
+      "step": 12720
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3240486460585566e-05,
+      "loss": 0.7949,
+      "step": 12721
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.323304424344528e-05,
+      "loss": 17.3039,
+      "step": 12722
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.322560244481203e-05,
+      "loss": 19.5128,
+      "step": 12723
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.32181610648716e-05,
+      "loss": 0.7739,
+      "step": 12724
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.321072010380967e-05,
+      "loss": 17.8225,
+      "step": 12725
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3203279561812014e-05,
+      "loss": 18.4328,
+      "step": 12726
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.319583943906434e-05,
+      "loss": 0.7651,
+      "step": 12727
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3188399735752326e-05,
+      "loss": 17.9822,
+      "step": 12728
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.318096045206172e-05,
+      "loss": 0.7604,
+      "step": 12729
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.317352158817817e-05,
+      "loss": 18.5628,
+      "step": 12730
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3166083144287374e-05,
+      "loss": 0.7274,
+      "step": 12731
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.315864512057497e-05,
+      "loss": 17.6273,
+      "step": 12732
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.315120751722664e-05,
+      "loss": 0.7347,
+      "step": 12733
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3143770334428005e-05,
+      "loss": 17.9247,
+      "step": 12734
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3136333572364727e-05,
+      "loss": 17.743,
+      "step": 12735
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.31288972312224e-05,
+      "loss": 0.7427,
+      "step": 12736
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3121461311186655e-05,
+      "loss": 19.528,
+      "step": 12737
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.311402581244311e-05,
+      "loss": 18.5456,
+      "step": 12738
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3106590735177305e-05,
+      "loss": 17.1883,
+      "step": 12739
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.309915607957487e-05,
+      "loss": 17.6889,
+      "step": 12740
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.309172184582135e-05,
+      "loss": 0.751,
+      "step": 12741
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.308428803410231e-05,
+      "loss": 17.2126,
+      "step": 12742
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.307685464460332e-05,
+      "loss": 19.1859,
+      "step": 12743
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3069421677509884e-05,
+      "loss": 0.7622,
+      "step": 12744
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.306198913300758e-05,
+      "loss": 0.6888,
+      "step": 12745
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.305455701128186e-05,
+      "loss": 0.7646,
+      "step": 12746
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.304712531251827e-05,
+      "loss": 16.1698,
+      "step": 12747
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.303969403690231e-05,
+      "loss": 0.7628,
+      "step": 12748
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.303226318461946e-05,
+      "loss": 0.7719,
+      "step": 12749
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3024832755855194e-05,
+      "loss": 0.7995,
+      "step": 12750
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3017402750794974e-05,
+      "loss": 16.8137,
+      "step": 12751
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.3009973169624274e-05,
+      "loss": 17.6676,
+      "step": 12752
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.300254401252851e-05,
+      "loss": 17.8938,
+      "step": 12753
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.299511527969314e-05,
+      "loss": 16.0426,
+      "step": 12754
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.298768697130357e-05,
+      "loss": 18.1098,
+      "step": 12755
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.298025908754521e-05,
+      "loss": 17.4715,
+      "step": 12756
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2972831628603474e-05,
+      "loss": 0.7503,
+      "step": 12757
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.296540459466374e-05,
+      "loss": 17.6568,
+      "step": 12758
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.295797798591143e-05,
+      "loss": 16.0525,
+      "step": 12759
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2950551802531844e-05,
+      "loss": 19.0738,
+      "step": 12760
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.294312604471038e-05,
+      "loss": 0.719,
+      "step": 12761
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.29357007126324e-05,
+      "loss": 22.2477,
+      "step": 12762
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.292827580648322e-05,
+      "loss": 0.8004,
+      "step": 12763
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.292085132644817e-05,
+      "loss": 0.738,
+      "step": 12764
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2913427272712574e-05,
+      "loss": 16.8438,
+      "step": 12765
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2906003645461747e-05,
+      "loss": 0.7137,
+      "step": 12766
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.289858044488096e-05,
+      "loss": 17.5869,
+      "step": 12767
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.289115767115552e-05,
+      "loss": 17.8873,
+      "step": 12768
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.288373532447067e-05,
+      "loss": 19.0714,
+      "step": 12769
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.28763134050117e-05,
+      "loss": 17.706,
+      "step": 12770
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.286889191296387e-05,
+      "loss": 0.7618,
+      "step": 12771
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.286147084851241e-05,
+      "loss": 0.7346,
+      "step": 12772
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2854050211842545e-05,
+      "loss": 17.3867,
+      "step": 12773
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.28466300031395e-05,
+      "loss": 0.7562,
+      "step": 12774
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.283921022258848e-05,
+      "loss": 0.7751,
+      "step": 12775
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.283179087037469e-05,
+      "loss": 18.0936,
+      "step": 12776
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.282437194668334e-05,
+      "loss": 16.4561,
+      "step": 12777
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.281695345169957e-05,
+      "loss": 18.3171,
+      "step": 12778
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2809535385608585e-05,
+      "loss": 0.758,
+      "step": 12779
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.280211774859551e-05,
+      "loss": 18.1769,
+      "step": 12780
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.279470054084549e-05,
+      "loss": 0.7295,
+      "step": 12781
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2787283762543685e-05,
+      "loss": 0.7209,
+      "step": 12782
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.277986741387521e-05,
+      "loss": 0.7585,
+      "step": 12783
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.277245149502516e-05,
+      "loss": 0.7736,
+      "step": 12784
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.276503600617867e-05,
+      "loss": 0.7619,
+      "step": 12785
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2757620947520824e-05,
+      "loss": 17.8522,
+      "step": 12786
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2750206319236696e-05,
+      "loss": 17.6,
+      "step": 12787
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2742792121511334e-05,
+      "loss": 0.7806,
+      "step": 12788
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.273537835452983e-05,
+      "loss": 15.9978,
+      "step": 12789
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2727965018477224e-05,
+      "loss": 16.5096,
+      "step": 12790
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.272055211353856e-05,
+      "loss": 0.7671,
+      "step": 12791
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.271313963989886e-05,
+      "loss": 18.4238,
+      "step": 12792
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.270572759774315e-05,
+      "loss": 0.7407,
+      "step": 12793
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2698315987256425e-05,
+      "loss": 0.7668,
+      "step": 12794
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.269090480862367e-05,
+      "loss": 0.7572,
+      "step": 12795
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.268349406202989e-05,
+      "loss": 17.0256,
+      "step": 12796
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.267608374766005e-05,
+      "loss": 17.3741,
+      "step": 12797
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2668673865699115e-05,
+      "loss": 17.0983,
+      "step": 12798
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.266126441633204e-05,
+      "loss": 0.7615,
+      "step": 12799
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.265385539974377e-05,
+      "loss": 0.7554,
+      "step": 12800
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.264644681611924e-05,
+      "loss": 16.5292,
+      "step": 12801
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.263903866564333e-05,
+      "loss": 17.6,
+      "step": 12802
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.263163094850099e-05,
+      "loss": 0.8088,
+      "step": 12803
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2624223664877104e-05,
+      "loss": 15.6058,
+      "step": 12804
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2616816814956575e-05,
+      "loss": 16.9523,
+      "step": 12805
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2609410398924255e-05,
+      "loss": 0.7584,
+      "step": 12806
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.260200441696505e-05,
+      "loss": 20.4964,
+      "step": 12807
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2594598869263756e-05,
+      "loss": 17.4124,
+      "step": 12808
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.258719375600525e-05,
+      "loss": 0.7774,
+      "step": 12809
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.257978907737438e-05,
+      "loss": 18.7515,
+      "step": 12810
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2572384833555945e-05,
+      "loss": 17.1212,
+      "step": 12811
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.256498102473475e-05,
+      "loss": 16.2689,
+      "step": 12812
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.255757765109563e-05,
+      "loss": 17.4654,
+      "step": 12813
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2550174712823356e-05,
+      "loss": 0.7878,
+      "step": 12814
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.25427722101027e-05,
+      "loss": 0.7397,
+      "step": 12815
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2535370143118425e-05,
+      "loss": 0.776,
+      "step": 12816
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2527968512055306e-05,
+      "loss": 0.741,
+      "step": 12817
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.252056731709807e-05,
+      "loss": 18.8986,
+      "step": 12818
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.251316655843149e-05,
+      "loss": 19.5206,
+      "step": 12819
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2505766236240244e-05,
+      "loss": 18.6769,
+      "step": 12820
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.249836635070909e-05,
+      "loss": 0.7625,
+      "step": 12821
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.249096690202269e-05,
+      "loss": 19.8818,
+      "step": 12822
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.248356789036575e-05,
+      "loss": 16.9806,
+      "step": 12823
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2476169315922975e-05,
+      "loss": 18.6619,
+      "step": 12824
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2468771178879e-05,
+      "loss": 0.6944,
+      "step": 12825
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.246137347941849e-05,
+      "loss": 19.3506,
+      "step": 12826
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2453976217726125e-05,
+      "loss": 20.4648,
+      "step": 12827
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.244657939398651e-05,
+      "loss": 19.6841,
+      "step": 12828
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.243918300838429e-05,
+      "loss": 16.9558,
+      "step": 12829
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.243178706110406e-05,
+      "loss": 0.7264,
+      "step": 12830
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.242439155233043e-05,
+      "loss": 0.8272,
+      "step": 12831
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.241699648224801e-05,
+      "loss": 0.7561,
+      "step": 12832
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.240960185104137e-05,
+      "loss": 16.8897,
+      "step": 12833
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.240220765889508e-05,
+      "loss": 0.7736,
+      "step": 12834
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2394813905993724e-05,
+      "loss": 16.5021,
+      "step": 12835
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.238742059252182e-05,
+      "loss": 18.7746,
+      "step": 12836
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.238002771866391e-05,
+      "loss": 17.4096,
+      "step": 12837
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2372635284604546e-05,
+      "loss": 0.7935,
+      "step": 12838
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2365243290528226e-05,
+      "loss": 21.2707,
+      "step": 12839
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.235785173661945e-05,
+      "loss": 0.6847,
+      "step": 12840
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.235046062306272e-05,
+      "loss": 0.7518,
+      "step": 12841
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2343069950042556e-05,
+      "loss": 17.7933,
+      "step": 12842
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2335679717743375e-05,
+      "loss": 0.7431,
+      "step": 12843
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2328289926349665e-05,
+      "loss": 16.8102,
+      "step": 12844
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.232090057604586e-05,
+      "loss": 0.7862,
+      "step": 12845
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2313511667016415e-05,
+      "loss": 16.3333,
+      "step": 12846
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2306123199445775e-05,
+      "loss": 0.7837,
+      "step": 12847
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2298735173518344e-05,
+      "loss": 0.759,
+      "step": 12848
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.229134758941851e-05,
+      "loss": 18.0703,
+      "step": 12849
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.228396044733068e-05,
+      "loss": 16.1315,
+      "step": 12850
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.227657374743924e-05,
+      "loss": 0.7341,
+      "step": 12851
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2269187489928584e-05,
+      "loss": 18.8448,
+      "step": 12852
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.226180167498304e-05,
+      "loss": 15.9067,
+      "step": 12853
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2254416302786975e-05,
+      "loss": 19.4704,
+      "step": 12854
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2247031373524755e-05,
+      "loss": 17.3053,
+      "step": 12855
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2239646887380665e-05,
+      "loss": 18.9207,
+      "step": 12856
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2232262844539054e-05,
+      "loss": 18.7812,
+      "step": 12857
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.22248792451842e-05,
+      "loss": 0.7603,
+      "step": 12858
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.221749608950043e-05,
+      "loss": 19.1359,
+      "step": 12859
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2210113377672015e-05,
+      "loss": 0.7613,
+      "step": 12860
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.220273110988325e-05,
+      "loss": 0.7718,
+      "step": 12861
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.219534928631838e-05,
+      "loss": 16.0647,
+      "step": 12862
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2187967907161635e-05,
+      "loss": 17.0108,
+      "step": 12863
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2180586972597284e-05,
+      "loss": 17.743,
+      "step": 12864
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.217320648280955e-05,
+      "loss": 19.3287,
+      "step": 12865
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.216582643798266e-05,
+      "loss": 0.7439,
+      "step": 12866
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.21584468383008e-05,
+      "loss": 0.7599,
+      "step": 12867
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.215106768394819e-05,
+      "loss": 0.7557,
+      "step": 12868
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.214368897510903e-05,
+      "loss": 0.7702,
+      "step": 12869
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.213631071196744e-05,
+      "loss": 18.7253,
+      "step": 12870
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2128932894707625e-05,
+      "loss": 0.7302,
+      "step": 12871
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.212155552351371e-05,
+      "loss": 0.7771,
+      "step": 12872
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2114178598569864e-05,
+      "loss": 17.4323,
+      "step": 12873
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.21068021200602e-05,
+      "loss": 0.7482,
+      "step": 12874
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.209942608816885e-05,
+      "loss": 19.1205,
+      "step": 12875
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.209205050307992e-05,
+      "loss": 0.7321,
+      "step": 12876
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.208467536497748e-05,
+      "loss": 17.8632,
+      "step": 12877
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.207730067404563e-05,
+      "loss": 0.7329,
+      "step": 12878
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.206992643046845e-05,
+      "loss": 18.3492,
+      "step": 12879
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.206255263443001e-05,
+      "loss": 17.8172,
+      "step": 12880
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.205517928611435e-05,
+      "loss": 0.7694,
+      "step": 12881
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.20478063857055e-05,
+      "loss": 0.8124,
+      "step": 12882
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2040433933387525e-05,
+      "loss": 0.7773,
+      "step": 12883
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.20330619293444e-05,
+      "loss": 16.5717,
+      "step": 12884
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2025690373760166e-05,
+      "loss": 0.8189,
+      "step": 12885
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.2018319266818783e-05,
+      "loss": 17.335,
+      "step": 12886
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.201094860870427e-05,
+      "loss": 0.7185,
+      "step": 12887
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.200357839960058e-05,
+      "loss": 18.9213,
+      "step": 12888
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1996208639691695e-05,
+      "loss": 16.4034,
+      "step": 12889
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1988839329161565e-05,
+      "loss": 17.466,
+      "step": 12890
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.19814704681941e-05,
+      "loss": 0.7176,
+      "step": 12891
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.197410205697323e-05,
+      "loss": 0.7513,
+      "step": 12892
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.19667340956829e-05,
+      "loss": 18.0294,
+      "step": 12893
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.195936658450702e-05,
+      "loss": 19.2339,
+      "step": 12894
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.195199952362945e-05,
+      "loss": 0.8167,
+      "step": 12895
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1944632913234095e-05,
+      "loss": 18.2979,
+      "step": 12896
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1937266753504854e-05,
+      "loss": 0.792,
+      "step": 12897
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1929901044625535e-05,
+      "loss": 16.2404,
+      "step": 12898
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.192253578678002e-05,
+      "loss": 18.6067,
+      "step": 12899
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1915170980152124e-05,
+      "loss": 16.2974,
+      "step": 12900
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.190780662492569e-05,
+      "loss": 16.812,
+      "step": 12901
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.190044272128454e-05,
+      "loss": 18.5724,
+      "step": 12902
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1893079269412473e-05,
+      "loss": 0.7649,
+      "step": 12903
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1885716269493306e-05,
+      "loss": 0.7509,
+      "step": 12904
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1878353721710765e-05,
+      "loss": 0.7458,
+      "step": 12905
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1870991626248655e-05,
+      "loss": 14.8822,
+      "step": 12906
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.186362998329074e-05,
+      "loss": 16.2475,
+      "step": 12907
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.185626879302077e-05,
+      "loss": 0.7801,
+      "step": 12908
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.184890805562246e-05,
+      "loss": 0.8082,
+      "step": 12909
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1841547771279543e-05,
+      "loss": 0.7287,
+      "step": 12910
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.183418794017577e-05,
+      "loss": 0.8011,
+      "step": 12911
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1826828562494795e-05,
+      "loss": 17.7977,
+      "step": 12912
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.181946963842033e-05,
+      "loss": 18.7299,
+      "step": 12913
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.181211116813605e-05,
+      "loss": 0.8361,
+      "step": 12914
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.180475315182563e-05,
+      "loss": 18.4579,
+      "step": 12915
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.179739558967272e-05,
+      "loss": 18.2696,
+      "step": 12916
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.179003848186099e-05,
+      "loss": 19.5269,
+      "step": 12917
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.178268182857406e-05,
+      "loss": 19.563,
+      "step": 12918
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.177532562999553e-05,
+      "loss": 0.7608,
+      "step": 12919
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.176796988630903e-05,
+      "loss": 0.7467,
+      "step": 12920
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1760614597698174e-05,
+      "loss": 0.7601,
+      "step": 12921
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.175325976434655e-05,
+      "loss": 17.0087,
+      "step": 12922
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.174590538643772e-05,
+      "loss": 16.1097,
+      "step": 12923
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.173855146415527e-05,
+      "loss": 0.791,
+      "step": 12924
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1731197997682735e-05,
+      "loss": 0.7566,
+      "step": 12925
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.172384498720367e-05,
+      "loss": 17.819,
+      "step": 12926
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.171649243290161e-05,
+      "loss": 17.701,
+      "step": 12927
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1709140334960074e-05,
+      "loss": 0.7592,
+      "step": 12928
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.170178869356256e-05,
+      "loss": 0.713,
+      "step": 12929
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1694437508892584e-05,
+      "loss": 17.0231,
+      "step": 12930
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.168708678113365e-05,
+      "loss": 18.0596,
+      "step": 12931
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1679736510469196e-05,
+      "loss": 0.7466,
+      "step": 12932
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.167238669708269e-05,
+      "loss": 17.3358,
+      "step": 12933
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.166503734115761e-05,
+      "loss": 0.7365,
+      "step": 12934
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1657688442877364e-05,
+      "loss": 0.7356,
+      "step": 12935
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.165034000242543e-05,
+      "loss": 16.1583,
+      "step": 12936
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.164299201998519e-05,
+      "loss": 20.3142,
+      "step": 12937
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1635644495740066e-05,
+      "loss": 17.2089,
+      "step": 12938
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.162829742987343e-05,
+      "loss": 17.3877,
+      "step": 12939
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1620950822568695e-05,
+      "loss": 17.367,
+      "step": 12940
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.161360467400923e-05,
+      "loss": 18.703,
+      "step": 12941
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.160625898437838e-05,
+      "loss": 0.7829,
+      "step": 12942
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.15989137538595e-05,
+      "loss": 17.8607,
+      "step": 12943
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1591568982635946e-05,
+      "loss": 17.0634,
+      "step": 12944
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.158422467089105e-05,
+      "loss": 0.7186,
+      "step": 12945
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.15768808188081e-05,
+      "loss": 0.7432,
+      "step": 12946
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.15695374265704e-05,
+      "loss": 17.4413,
+      "step": 12947
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.156219449436125e-05,
+      "loss": 18.297,
+      "step": 12948
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.155485202236394e-05,
+      "loss": 0.7524,
+      "step": 12949
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1547510010761746e-05,
+      "loss": 17.6902,
+      "step": 12950
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.15401684597379e-05,
+      "loss": 0.7681,
+      "step": 12951
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1532827369475695e-05,
+      "loss": 16.8508,
+      "step": 12952
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1525486740158304e-05,
+      "loss": 17.2268,
+      "step": 12953
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1518146571968984e-05,
+      "loss": 0.7446,
+      "step": 12954
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.151080686509096e-05,
+      "loss": 0.8202,
+      "step": 12955
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.150346761970741e-05,
+      "loss": 0.7037,
+      "step": 12956
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1496128836001524e-05,
+      "loss": 19.0468,
+      "step": 12957
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.148879051415649e-05,
+      "loss": 17.2315,
+      "step": 12958
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.148145265435549e-05,
+      "loss": 16.4352,
+      "step": 12959
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1474115256781655e-05,
+      "loss": 0.8082,
+      "step": 12960
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.146677832161812e-05,
+      "loss": 18.8027,
+      "step": 12961
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.145944184904803e-05,
+      "loss": 0.7802,
+      "step": 12962
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.145210583925451e-05,
+      "loss": 0.7117,
+      "step": 12963
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.144477029242067e-05,
+      "loss": 18.1758,
+      "step": 12964
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.143743520872958e-05,
+      "loss": 0.7574,
+      "step": 12965
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.143010058836438e-05,
+      "loss": 17.5955,
+      "step": 12966
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.142276643150809e-05,
+      "loss": 18.7358,
+      "step": 12967
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.141543273834378e-05,
+      "loss": 0.7381,
+      "step": 12968
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.140809950905453e-05,
+      "loss": 18.1373,
+      "step": 12969
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.140076674382335e-05,
+      "loss": 16.3789,
+      "step": 12970
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.139343444283327e-05,
+      "loss": 16.7394,
+      "step": 12971
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.138610260626733e-05,
+      "loss": 0.7569,
+      "step": 12972
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.137877123430853e-05,
+      "loss": 15.52,
+      "step": 12973
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1371440327139844e-05,
+      "loss": 0.7942,
+      "step": 12974
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.136410988494425e-05,
+      "loss": 0.7475,
+      "step": 12975
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.135677990790473e-05,
+      "loss": 0.7557,
+      "step": 12976
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.134945039620424e-05,
+      "loss": 0.7106,
+      "step": 12977
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.134212135002573e-05,
+      "loss": 0.8116,
+      "step": 12978
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.133479276955213e-05,
+      "loss": 19.0071,
+      "step": 12979
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.132746465496638e-05,
+      "loss": 0.7197,
+      "step": 12980
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1320137006451355e-05,
+      "loss": 18.7003,
+      "step": 12981
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.131280982418998e-05,
+      "loss": 19.6582,
+      "step": 12982
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.130548310836515e-05,
+      "loss": 19.8426,
+      "step": 12983
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.129815685915971e-05,
+      "loss": 15.8306,
+      "step": 12984
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1290831076756554e-05,
+      "loss": 17.1666,
+      "step": 12985
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.128350576133852e-05,
+      "loss": 19.4236,
+      "step": 12986
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.127618091308848e-05,
+      "loss": 19.049,
+      "step": 12987
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.126885653218923e-05,
+      "loss": 0.7741,
+      "step": 12988
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.126153261882359e-05,
+      "loss": 14.2621,
+      "step": 12989
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.125420917317436e-05,
+      "loss": 19.2956,
+      "step": 12990
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.124688619542437e-05,
+      "loss": 0.8097,
+      "step": 12991
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.123956368575638e-05,
+      "loss": 0.7631,
+      "step": 12992
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1232241644353156e-05,
+      "loss": 19.1092,
+      "step": 12993
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.122492007139749e-05,
+      "loss": 0.7517,
+      "step": 12994
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1217598967072083e-05,
+      "loss": 0.7495,
+      "step": 12995
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.121027833155969e-05,
+      "loss": 0.778,
+      "step": 12996
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.120295816504305e-05,
+      "loss": 0.7731,
+      "step": 12997
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.119563846770486e-05,
+      "loss": 0.7828,
+      "step": 12998
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1188319239727833e-05,
+      "loss": 17.5399,
+      "step": 12999
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.118100048129466e-05,
+      "loss": 18.5457,
+      "step": 13000
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1173682192588004e-05,
+      "loss": 17.5746,
+      "step": 13001
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1166364373790546e-05,
+      "loss": 0.7575,
+      "step": 13002
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.115904702508492e-05,
+      "loss": 16.6057,
+      "step": 13003
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1151730146653784e-05,
+      "loss": 0.7589,
+      "step": 13004
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.114441373867977e-05,
+      "loss": 18.6802,
+      "step": 13005
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.11370978013455e-05,
+      "loss": 18.8987,
+      "step": 13006
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.112978233483358e-05,
+      "loss": 0.6932,
+      "step": 13007
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.11224673393266e-05,
+      "loss": 19.047,
+      "step": 13008
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.111515281500713e-05,
+      "loss": 0.7669,
+      "step": 13009
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.110783876205776e-05,
+      "loss": 17.0394,
+      "step": 13010
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.110052518066107e-05,
+      "loss": 0.7428,
+      "step": 13011
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1093212070999564e-05,
+      "loss": 0.7626,
+      "step": 13012
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.10858994332558e-05,
+      "loss": 0.75,
+      "step": 13013
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.107858726761233e-05,
+      "loss": 0.794,
+      "step": 13014
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1071275574251625e-05,
+      "loss": 0.7776,
+      "step": 13015
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.106396435335621e-05,
+      "loss": 0.7634,
+      "step": 13016
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1056653605108555e-05,
+      "loss": 0.6953,
+      "step": 13017
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1049343329691154e-05,
+      "loss": 16.9573,
+      "step": 13018
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1042033527286475e-05,
+      "loss": 0.8017,
+      "step": 13019
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.103472419807697e-05,
+      "loss": 0.7352,
+      "step": 13020
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.102741534224509e-05,
+      "loss": 19.3204,
+      "step": 13021
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.102010695997325e-05,
+      "loss": 0.7594,
+      "step": 13022
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.101279905144386e-05,
+      "loss": 17.5864,
+      "step": 13023
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.1005491616839345e-05,
+      "loss": 16.588,
+      "step": 13024
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.09981846563421e-05,
+      "loss": 0.7379,
+      "step": 13025
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0990878170134504e-05,
+      "loss": 17.2154,
+      "step": 13026
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0983572158398925e-05,
+      "loss": 0.7336,
+      "step": 13027
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0976266621317746e-05,
+      "loss": 0.7439,
+      "step": 13028
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0968961559073276e-05,
+      "loss": 17.6137,
+      "step": 13029
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0961656971847885e-05,
+      "loss": 0.727,
+      "step": 13030
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.095435285982387e-05,
+      "loss": 0.7565,
+      "step": 13031
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0947049223183554e-05,
+      "loss": 0.7806,
+      "step": 13032
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0939746062109244e-05,
+      "loss": 18.5259,
+      "step": 13033
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0932443376783226e-05,
+      "loss": 19.1406,
+      "step": 13034
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.092514116738778e-05,
+      "loss": 0.7517,
+      "step": 13035
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.091783943410517e-05,
+      "loss": 0.7458,
+      "step": 13036
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.091053817711763e-05,
+      "loss": 17.5177,
+      "step": 13037
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.090323739660742e-05,
+      "loss": 19.8136,
+      "step": 13038
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.089593709275677e-05,
+      "loss": 0.7251,
+      "step": 13039
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.088863726574788e-05,
+      "loss": 16.6516,
+      "step": 13040
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0881337915762976e-05,
+      "loss": 0.7953,
+      "step": 13041
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.087403904298426e-05,
+      "loss": 20.3453,
+      "step": 13042
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.086674064759387e-05,
+      "loss": 17.3587,
+      "step": 13043
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.085944272977402e-05,
+      "loss": 18.7151,
+      "step": 13044
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0852145289706835e-05,
+      "loss": 16.4744,
+      "step": 13045
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.084484832757448e-05,
+      "loss": 17.3202,
+      "step": 13046
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0837551843559086e-05,
+      "loss": 0.7566,
+      "step": 13047
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.083025583784278e-05,
+      "loss": 16.9063,
+      "step": 13048
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.082296031060766e-05,
+      "loss": 0.721,
+      "step": 13049
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.081566526203584e-05,
+      "loss": 18.0721,
+      "step": 13050
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0808370692309375e-05,
+      "loss": 0.7536,
+      "step": 13051
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.080107660161037e-05,
+      "loss": 15.4778,
+      "step": 13052
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0793782990120875e-05,
+      "loss": 0.7353,
+      "step": 13053
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.078648985802294e-05,
+      "loss": 0.7402,
+      "step": 13054
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.07791972054986e-05,
+      "loss": 19.2944,
+      "step": 13055
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.077190503272991e-05,
+      "loss": 0.7513,
+      "step": 13056
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.076461333989884e-05,
+      "loss": 18.4268,
+      "step": 13057
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.075732212718743e-05,
+      "loss": 14.9923,
+      "step": 13058
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0750031394777635e-05,
+      "loss": 16.879,
+      "step": 13059
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.074274114285145e-05,
+      "loss": 18.2183,
+      "step": 13060
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.073545137159085e-05,
+      "loss": 0.7382,
+      "step": 13061
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0728162081177785e-05,
+      "loss": 0.7466,
+      "step": 13062
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.07208732717942e-05,
+      "loss": 18.3561,
+      "step": 13063
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.071358494362202e-05,
+      "loss": 0.7307,
+      "step": 13064
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0706297096843153e-05,
+      "loss": 0.8039,
+      "step": 13065
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.069900973163952e-05,
+      "loss": 0.7799,
+      "step": 13066
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.069172284819301e-05,
+      "loss": 16.6729,
+      "step": 13067
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.068443644668551e-05,
+      "loss": 0.7381,
+      "step": 13068
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.067715052729887e-05,
+      "loss": 17.0784,
+      "step": 13069
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0669865090215e-05,
+      "loss": 17.0239,
+      "step": 13070
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.066258013561568e-05,
+      "loss": 18.3177,
+      "step": 13071
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.065529566368279e-05,
+      "loss": 19.2465,
+      "step": 13072
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.064801167459813e-05,
+      "loss": 0.7511,
+      "step": 13073
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0640728168543506e-05,
+      "loss": 0.7683,
+      "step": 13074
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0633445145700746e-05,
+      "loss": 0.7807,
+      "step": 13075
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0626162606251626e-05,
+      "loss": 0.7772,
+      "step": 13076
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.061888055037789e-05,
+      "loss": 0.7369,
+      "step": 13077
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.061159897826133e-05,
+      "loss": 0.7547,
+      "step": 13078
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0604317890083676e-05,
+      "loss": 18.0775,
+      "step": 13079
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.059703728602669e-05,
+      "loss": 0.7439,
+      "step": 13080
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.058975716627208e-05,
+      "loss": 17.5977,
+      "step": 13081
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.058247753100155e-05,
+      "loss": 0.7705,
+      "step": 13082
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.057519838039684e-05,
+      "loss": 0.7516,
+      "step": 13083
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.056791971463959e-05,
+      "loss": 19.1438,
+      "step": 13084
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.056064153391151e-05,
+      "loss": 18.4953,
+      "step": 13085
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.055336383839425e-05,
+      "loss": 0.7281,
+      "step": 13086
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0546086628269474e-05,
+      "loss": 0.7437,
+      "step": 13087
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0538809903718816e-05,
+      "loss": 15.9038,
+      "step": 13088
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.053153366492391e-05,
+      "loss": 18.634,
+      "step": 13089
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.052425791206638e-05,
+      "loss": 0.7443,
+      "step": 13090
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0516982645327802e-05,
+      "loss": 0.725,
+      "step": 13091
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.050970786488979e-05,
+      "loss": 16.434,
+      "step": 13092
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.050243357093392e-05,
+      "loss": 20.302,
+      "step": 13093
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0495159763641758e-05,
+      "loss": 0.7524,
+      "step": 13094
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.048788644319487e-05,
+      "loss": 18.8382,
+      "step": 13095
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0480613609774777e-05,
+      "loss": 19.0673,
+      "step": 13096
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.047334126356305e-05,
+      "loss": 18.1737,
+      "step": 13097
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.046606940474116e-05,
+      "loss": 0.7705,
+      "step": 13098
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0458798033490644e-05,
+      "loss": 17.1473,
+      "step": 13099
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0451527149992986e-05,
+      "loss": 0.7211,
+      "step": 13100
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0444256754429683e-05,
+      "loss": 0.7433,
+      "step": 13101
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.043698684698219e-05,
+      "loss": 17.2996,
+      "step": 13102
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0429717427831972e-05,
+      "loss": 18.5026,
+      "step": 13103
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0422448497160493e-05,
+      "loss": 20.1731,
+      "step": 13104
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.041518005514915e-05,
+      "loss": 0.7679,
+      "step": 13105
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0407912101979396e-05,
+      "loss": 0.7643,
+      "step": 13106
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.040064463783262e-05,
+      "loss": 17.0845,
+      "step": 13107
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0393377662890227e-05,
+      "loss": 17.8749,
+      "step": 13108
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0386111177333615e-05,
+      "loss": 0.6911,
+      "step": 13109
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.037884518134414e-05,
+      "loss": 0.7767,
+      "step": 13110
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0371579675103184e-05,
+      "loss": 19.8018,
+      "step": 13111
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.036431465879206e-05,
+      "loss": 19.6567,
+      "step": 13112
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.035705013259213e-05,
+      "loss": 0.7782,
+      "step": 13113
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.034978609668471e-05,
+      "loss": 0.7543,
+      "step": 13114
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.034252255125113e-05,
+      "loss": 0.7238,
+      "step": 13115
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0335259496472667e-05,
+      "loss": 0.7534,
+      "step": 13116
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.032799693253061e-05,
+      "loss": 16.3174,
+      "step": 13117
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.032073485960627e-05,
+      "loss": 17.9873,
+      "step": 13118
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0313473277880854e-05,
+      "loss": 0.7661,
+      "step": 13119
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0306212187535653e-05,
+      "loss": 17.072,
+      "step": 13120
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0298951588751878e-05,
+      "loss": 16.6607,
+      "step": 13121
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.029169148171077e-05,
+      "loss": 0.7405,
+      "step": 13122
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.028443186659356e-05,
+      "loss": 0.782,
+      "step": 13123
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0277172743581417e-05,
+      "loss": 0.7587,
+      "step": 13124
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0269914112855568e-05,
+      "loss": 0.7372,
+      "step": 13125
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0262655974597144e-05,
+      "loss": 0.7192,
+      "step": 13126
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.025539832898734e-05,
+      "loss": 0.7204,
+      "step": 13127
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0248141176207296e-05,
+      "loss": 17.7517,
+      "step": 13128
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0240884516438173e-05,
+      "loss": 0.7221,
+      "step": 13129
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0233628349861065e-05,
+      "loss": 18.3127,
+      "step": 13130
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0226372676657118e-05,
+      "loss": 16.3853,
+      "step": 13131
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0219117497007437e-05,
+      "loss": 19.0265,
+      "step": 13132
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0211862811093082e-05,
+      "loss": 0.7279,
+      "step": 13133
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.020460861909516e-05,
+      "loss": 17.6627,
+      "step": 13134
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0197354921194714e-05,
+      "loss": 0.7747,
+      "step": 13135
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0190101717572815e-05,
+      "loss": 19.7947,
+      "step": 13136
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0182849008410508e-05,
+      "loss": 0.7788,
+      "step": 13137
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0175596793888804e-05,
+      "loss": 0.7484,
+      "step": 13138
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0168345074188753e-05,
+      "loss": 0.7667,
+      "step": 13139
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.016109384949131e-05,
+      "loss": 0.7313,
+      "step": 13140
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0153843119977497e-05,
+      "loss": 0.7546,
+      "step": 13141
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0146592885828295e-05,
+      "loss": 0.7417,
+      "step": 13142
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0139343147224676e-05,
+      "loss": 16.6875,
+      "step": 13143
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.013209390434758e-05,
+      "loss": 0.7418,
+      "step": 13144
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0124845157377952e-05,
+      "loss": 16.6261,
+      "step": 13145
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0117596906496747e-05,
+      "loss": 15.2025,
+      "step": 13146
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.011034915188484e-05,
+      "loss": 17.1571,
+      "step": 13147
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0103101893723185e-05,
+      "loss": 18.8083,
+      "step": 13148
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0095855132192625e-05,
+      "loss": 18.9709,
+      "step": 13149
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0088608867474065e-05,
+      "loss": 0.765,
+      "step": 13150
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0081363099748394e-05,
+      "loss": 0.7004,
+      "step": 13151
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.007411782919644e-05,
+      "loss": 0.7658,
+      "step": 13152
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0066873055999057e-05,
+      "loss": 0.7991,
+      "step": 13153
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0059628780337067e-05,
+      "loss": 18.7359,
+      "step": 13154
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0052385002391292e-05,
+      "loss": 18.82,
+      "step": 13155
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0045141722342542e-05,
+      "loss": 0.7786,
+      "step": 13156
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.003789894037162e-05,
+      "loss": 0.711,
+      "step": 13157
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.003065665665929e-05,
+      "loss": 0.7238,
+      "step": 13158
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.002341487138635e-05,
+      "loss": 0.789,
+      "step": 13159
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0016173584733513e-05,
+      "loss": 16.3972,
+      "step": 13160
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.000893279688155e-05,
+      "loss": 0.7392,
+      "step": 13161
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.0001692508011198e-05,
+      "loss": 0.7602,
+      "step": 13162
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9994452718303158e-05,
+      "loss": 0.8123,
+      "step": 13163
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9987213427938144e-05,
+      "loss": 0.7806,
+      "step": 13164
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9979974637096864e-05,
+      "loss": 0.761,
+      "step": 13165
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9972736345959982e-05,
+      "loss": 0.7975,
+      "step": 13166
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9965498554708183e-05,
+      "loss": 0.7312,
+      "step": 13167
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.99582612635221e-05,
+      "loss": 0.7551,
+      "step": 13168
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9951024472582396e-05,
+      "loss": 17.6198,
+      "step": 13169
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.994378818206969e-05,
+      "loss": 15.4795,
+      "step": 13170
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.993655239216463e-05,
+      "loss": 0.7827,
+      "step": 13171
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9929317103047787e-05,
+      "loss": 0.7267,
+      "step": 13172
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9922082314899796e-05,
+      "loss": 17.1686,
+      "step": 13173
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9914848027901192e-05,
+      "loss": 0.8049,
+      "step": 13174
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9907614242232568e-05,
+      "loss": 0.739,
+      "step": 13175
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9900380958074492e-05,
+      "loss": 0.7643,
+      "step": 13176
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.989314817560749e-05,
+      "loss": 0.7498,
+      "step": 13177
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.988591589501209e-05,
+      "loss": 0.7765,
+      "step": 13178
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9878684116468835e-05,
+      "loss": 18.1366,
+      "step": 13179
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.987145284015822e-05,
+      "loss": 0.8289,
+      "step": 13180
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.986422206626074e-05,
+      "loss": 0.7419,
+      "step": 13181
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9856991794956868e-05,
+      "loss": 0.7659,
+      "step": 13182
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9849762026427075e-05,
+      "loss": 0.7866,
+      "step": 13183
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9842532760851826e-05,
+      "loss": 18.4389,
+      "step": 13184
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.983530399841157e-05,
+      "loss": 18.1645,
+      "step": 13185
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.982807573928672e-05,
+      "loss": 0.7304,
+      "step": 13186
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9820847983657725e-05,
+      "loss": 0.7397,
+      "step": 13187
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.981362073170495e-05,
+      "loss": 0.7366,
+      "step": 13188
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9806393983608815e-05,
+      "loss": 18.0783,
+      "step": 13189
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9799167739549705e-05,
+      "loss": 19.2118,
+      "step": 13190
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9791941999707967e-05,
+      "loss": 17.6952,
+      "step": 13191
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9784716764263974e-05,
+      "loss": 0.7715,
+      "step": 13192
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.977749203339808e-05,
+      "loss": 19.3535,
+      "step": 13193
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9770267807290597e-05,
+      "loss": 0.7436,
+      "step": 13194
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.976304408612185e-05,
+      "loss": 0.7782,
+      "step": 13195
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9755820870072137e-05,
+      "loss": 0.7492,
+      "step": 13196
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.974859815932175e-05,
+      "loss": 17.4705,
+      "step": 13197
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.974137595405098e-05,
+      "loss": 0.7332,
+      "step": 13198
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9734154254440107e-05,
+      "loss": 0.7357,
+      "step": 13199
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.972693306066936e-05,
+      "loss": 17.4302,
+      "step": 13200
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9719712372919007e-05,
+      "loss": 18.2456,
+      "step": 13201
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9712492191369244e-05,
+      "loss": 0.737,
+      "step": 13202
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.970527251620031e-05,
+      "loss": 18.7022,
+      "step": 13203
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.969805334759242e-05,
+      "loss": 0.8023,
+      "step": 13204
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9690834685725745e-05,
+      "loss": 15.4425,
+      "step": 13205
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9683616530780466e-05,
+      "loss": 19.7143,
+      "step": 13206
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9676398882936774e-05,
+      "loss": 17.5593,
+      "step": 13207
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.966918174237481e-05,
+      "loss": 0.7235,
+      "step": 13208
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9661965109274703e-05,
+      "loss": 0.7658,
+      "step": 13209
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.965474898381658e-05,
+      "loss": 16.6479,
+      "step": 13210
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9647533366180573e-05,
+      "loss": 16.8434,
+      "step": 13211
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9640318256546785e-05,
+      "loss": 0.7628,
+      "step": 13212
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9633103655095306e-05,
+      "loss": 0.7966,
+      "step": 13213
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9625889562006198e-05,
+      "loss": 0.7267,
+      "step": 13214
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9618675977459558e-05,
+      "loss": 17.1421,
+      "step": 13215
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.96114629016354e-05,
+      "loss": 19.9019,
+      "step": 13216
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.960425033471379e-05,
+      "loss": 17.7649,
+      "step": 13217
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9597038276874757e-05,
+      "loss": 18.322,
+      "step": 13218
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.958982672829829e-05,
+      "loss": 19.1112,
+      "step": 13219
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.958261568916442e-05,
+      "loss": 0.742,
+      "step": 13220
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.957540515965312e-05,
+      "loss": 17.2114,
+      "step": 13221
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9568195139944377e-05,
+      "loss": 20.2206,
+      "step": 13222
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9560985630218153e-05,
+      "loss": 0.7713,
+      "step": 13223
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9553776630654372e-05,
+      "loss": 17.7582,
+      "step": 13224
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9546568141433006e-05,
+      "loss": 20.8914,
+      "step": 13225
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9539360162733963e-05,
+      "loss": 0.8072,
+      "step": 13226
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.953215269473718e-05,
+      "loss": 0.7474,
+      "step": 13227
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9524945737622544e-05,
+      "loss": 0.8072,
+      "step": 13228
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9517739291569923e-05,
+      "loss": 18.1644,
+      "step": 13229
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9510533356759197e-05,
+      "loss": 0.7608,
+      "step": 13230
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.950332793337024e-05,
+      "loss": 0.7517,
+      "step": 13231
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.949612302158291e-05,
+      "loss": 0.8014,
+      "step": 13232
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9488918621577023e-05,
+      "loss": 17.3105,
+      "step": 13233
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9481714733532407e-05,
+      "loss": 17.5188,
+      "step": 13234
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9474511357628898e-05,
+      "loss": 0.721,
+      "step": 13235
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9467308494046248e-05,
+      "loss": 17.7045,
+      "step": 13236
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9460106142964277e-05,
+      "loss": 18.1658,
+      "step": 13237
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.945290430456274e-05,
+      "loss": 0.7788,
+      "step": 13238
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.94457029790214e-05,
+      "loss": 0.8316,
+      "step": 13239
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.943850216652001e-05,
+      "loss": 17.6056,
+      "step": 13240
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9431301867238305e-05,
+      "loss": 0.7001,
+      "step": 13241
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.942410208135602e-05,
+      "loss": 17.101,
+      "step": 13242
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9416902809052815e-05,
+      "loss": 0.7345,
+      "step": 13243
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9409704050508424e-05,
+      "loss": 19.4251,
+      "step": 13244
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.940250580590252e-05,
+      "loss": 0.7721,
+      "step": 13245
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9395308075414784e-05,
+      "loss": 0.7062,
+      "step": 13246
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.938811085922485e-05,
+      "loss": 0.7378,
+      "step": 13247
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9380914157512373e-05,
+      "loss": 0.7439,
+      "step": 13248
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9373717970457014e-05,
+      "loss": 0.7652,
+      "step": 13249
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.936652229823834e-05,
+      "loss": 18.8845,
+      "step": 13250
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9359327141035996e-05,
+      "loss": 0.7769,
+      "step": 13251
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.935213249902954e-05,
+      "loss": 0.7563,
+      "step": 13252
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.934493837239858e-05,
+      "loss": 19.024,
+      "step": 13253
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.933774476132267e-05,
+      "loss": 0.8175,
+      "step": 13254
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9330551665981375e-05,
+      "loss": 19.331,
+      "step": 13255
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9323359086554247e-05,
+      "loss": 17.9299,
+      "step": 13256
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9316167023220777e-05,
+      "loss": 0.8559,
+      "step": 13257
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9308975476160495e-05,
+      "loss": 0.8169,
+      "step": 13258
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9301784445552914e-05,
+      "loss": 0.746,
+      "step": 13259
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9294593931577528e-05,
+      "loss": 18.1742,
+      "step": 13260
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9287403934413793e-05,
+      "loss": 0.7697,
+      "step": 13261
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.928021445424119e-05,
+      "loss": 17.9886,
+      "step": 13262
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.927302549123918e-05,
+      "loss": 16.7428,
+      "step": 13263
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.926583704558717e-05,
+      "loss": 18.8501,
+      "step": 13264
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.925864911746461e-05,
+      "loss": 16.9128,
+      "step": 13265
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.92514617070509e-05,
+      "loss": 15.9724,
+      "step": 13266
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9244274814525428e-05,
+      "loss": 0.7252,
+      "step": 13267
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9237088440067605e-05,
+      "loss": 19.6301,
+      "step": 13268
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9229902583856807e-05,
+      "loss": 0.7696,
+      "step": 13269
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9222717246072396e-05,
+      "loss": 18.8945,
+      "step": 13270
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9215532426893678e-05,
+      "loss": 0.7901,
+      "step": 13271
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9208348126500025e-05,
+      "loss": 18.4926,
+      "step": 13272
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9201164345070742e-05,
+      "loss": 20.133,
+      "step": 13273
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.919398108278516e-05,
+      "loss": 0.7837,
+      "step": 13274
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.918679833982254e-05,
+      "loss": 17.2384,
+      "step": 13275
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9179616116362206e-05,
+      "loss": 20.3027,
+      "step": 13276
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9172434412583428e-05,
+      "loss": 0.7425,
+      "step": 13277
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9165253228665402e-05,
+      "loss": 0.7187,
+      "step": 13278
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9158072564787432e-05,
+      "loss": 0.7586,
+      "step": 13279
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9150892421128712e-05,
+      "loss": 0.7646,
+      "step": 13280
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9143712797868495e-05,
+      "loss": 0.7611,
+      "step": 13281
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.913653369518597e-05,
+      "loss": 0.7371,
+      "step": 13282
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9129355113260305e-05,
+      "loss": 0.7118,
+      "step": 13283
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.912217705227075e-05,
+      "loss": 16.6826,
+      "step": 13284
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9114999512396375e-05,
+      "loss": 18.3383,
+      "step": 13285
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9107822493816406e-05,
+      "loss": 18.217,
+      "step": 13286
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.910064599670995e-05,
+      "loss": 18.086,
+      "step": 13287
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.909347002125613e-05,
+      "loss": 18.4298,
+      "step": 13288
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9086294567634088e-05,
+      "loss": 16.5034,
+      "step": 13289
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9079119636022904e-05,
+      "loss": 18.3498,
+      "step": 13290
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.907194522660168e-05,
+      "loss": 0.7169,
+      "step": 13291
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9064771339549478e-05,
+      "loss": 0.7573,
+      "step": 13292
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.905759797504534e-05,
+      "loss": 16.0765,
+      "step": 13293
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9050425133268366e-05,
+      "loss": 0.7632,
+      "step": 13294
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.904325281439756e-05,
+      "loss": 0.7546,
+      "step": 13295
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9036081018611925e-05,
+      "loss": 0.7681,
+      "step": 13296
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9028909746090517e-05,
+      "loss": 16.0327,
+      "step": 13297
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.902173899701231e-05,
+      "loss": 0.7569,
+      "step": 13298
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.901456877155629e-05,
+      "loss": 16.5298,
+      "step": 13299
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.900739906990142e-05,
+      "loss": 19.0371,
+      "step": 13300
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.9000229892226648e-05,
+      "loss": 20.2669,
+      "step": 13301
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.899306123871095e-05,
+      "loss": 20.1773,
+      "step": 13302
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8985893109533223e-05,
+      "loss": 0.746,
+      "step": 13303
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.897872550487244e-05,
+      "loss": 18.7337,
+      "step": 13304
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8971558424907448e-05,
+      "loss": 17.4681,
+      "step": 13305
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8964391869817143e-05,
+      "loss": 13.3945,
+      "step": 13306
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8957225839780434e-05,
+      "loss": 18.5329,
+      "step": 13307
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8950060334976153e-05,
+      "loss": 17.8827,
+      "step": 13308
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8942895355583195e-05,
+      "loss": 0.6968,
+      "step": 13309
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8935730901780377e-05,
+      "loss": 18.3831,
+      "step": 13310
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8928566973746518e-05,
+      "loss": 16.8569,
+      "step": 13311
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8921403571660448e-05,
+      "loss": 18.0019,
+      "step": 13312
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8914240695700934e-05,
+      "loss": 16.5823,
+      "step": 13313
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8907078346046813e-05,
+      "loss": 15.4058,
+      "step": 13314
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8899916522876824e-05,
+      "loss": 17.2791,
+      "step": 13315
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8892755226369717e-05,
+      "loss": 17.8469,
+      "step": 13316
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.888559445670428e-05,
+      "loss": 0.7786,
+      "step": 13317
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.887843421405923e-05,
+      "loss": 0.7671,
+      "step": 13318
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.887127449861328e-05,
+      "loss": 0.7819,
+      "step": 13319
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8864115310545146e-05,
+      "loss": 17.5202,
+      "step": 13320
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8856956650033502e-05,
+      "loss": 0.7571,
+      "step": 13321
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8849798517257065e-05,
+      "loss": 16.7699,
+      "step": 13322
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8842640912394482e-05,
+      "loss": 0.7718,
+      "step": 13323
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8835483835624398e-05,
+      "loss": 19.8444,
+      "step": 13324
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.882832728712551e-05,
+      "loss": 0.7403,
+      "step": 13325
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8821171267076365e-05,
+      "loss": 0.7547,
+      "step": 13326
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8814015775655646e-05,
+      "loss": 17.3355,
+      "step": 13327
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8806860813041926e-05,
+      "loss": 0.7528,
+      "step": 13328
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.879970637941377e-05,
+      "loss": 0.7488,
+      "step": 13329
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.879255247494982e-05,
+      "loss": 17.1859,
+      "step": 13330
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.878539909982857e-05,
+      "loss": 0.8143,
+      "step": 13331
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8778246254228646e-05,
+      "loss": 18.4259,
+      "step": 13332
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.877109393832853e-05,
+      "loss": 0.8305,
+      "step": 13333
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8763942152306728e-05,
+      "loss": 0.7786,
+      "step": 13334
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.87567908963418e-05,
+      "loss": 16.9922,
+      "step": 13335
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.87496401706122e-05,
+      "loss": 18.0672,
+      "step": 13336
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8742489975296454e-05,
+      "loss": 0.7495,
+      "step": 13337
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.873534031057301e-05,
+      "loss": 0.7575,
+      "step": 13338
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8728191176620335e-05,
+      "loss": 16.8384,
+      "step": 13339
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8721042573616858e-05,
+      "loss": 0.7808,
+      "step": 13340
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8713894501741e-05,
+      "loss": 0.8097,
+      "step": 13341
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8706746961171216e-05,
+      "loss": 0.7635,
+      "step": 13342
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8699599952085886e-05,
+      "loss": 0.7683,
+      "step": 13343
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8692453474663387e-05,
+      "loss": 0.7725,
+      "step": 13344
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.868530752908214e-05,
+      "loss": 18.5365,
+      "step": 13345
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8678162115520473e-05,
+      "loss": 17.8875,
+      "step": 13346
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8671017234156756e-05,
+      "loss": 17.3666,
+      "step": 13347
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8663872885169317e-05,
+      "loss": 0.7197,
+      "step": 13348
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8656729068736458e-05,
+      "loss": 0.7142,
+      "step": 13349
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8649585785036536e-05,
+      "loss": 16.7126,
+      "step": 13350
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8642443034247836e-05,
+      "loss": 18.4308,
+      "step": 13351
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8635300816548604e-05,
+      "loss": 16.4968,
+      "step": 13352
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8628159132117193e-05,
+      "loss": 17.342,
+      "step": 13353
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.862101798113176e-05,
+      "loss": 0.7391,
+      "step": 13354
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8613877363770626e-05,
+      "loss": 0.7839,
+      "step": 13355
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8606737280211993e-05,
+      "loss": 0.7549,
+      "step": 13356
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.859959773063406e-05,
+      "loss": 18.0632,
+      "step": 13357
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8592458715215077e-05,
+      "loss": 0.8123,
+      "step": 13358
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.858532023413319e-05,
+      "loss": 16.1042,
+      "step": 13359
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8578182287566646e-05,
+      "loss": 18.7521,
+      "step": 13360
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8571044875693532e-05,
+      "loss": 17.7131,
+      "step": 13361
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.856390799869202e-05,
+      "loss": 0.7898,
+      "step": 13362
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.855677165674028e-05,
+      "loss": 0.7775,
+      "step": 13363
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8549635850016388e-05,
+      "loss": 0.7585,
+      "step": 13364
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8542500578698507e-05,
+      "loss": 18.424,
+      "step": 13365
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.853536584296471e-05,
+      "loss": 16.6425,
+      "step": 13366
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8528231642993085e-05,
+      "loss": 18.103,
+      "step": 13367
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.85210979789617e-05,
+      "loss": 0.7765,
+      "step": 13368
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.851396485104859e-05,
+      "loss": 0.7097,
+      "step": 13369
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8506832259431847e-05,
+      "loss": 0.8006,
+      "step": 13370
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.849970020428948e-05,
+      "loss": 0.7866,
+      "step": 13371
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8492568685799482e-05,
+      "loss": 16.5749,
+      "step": 13372
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.84854377041399e-05,
+      "loss": 17.4237,
+      "step": 13373
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.84783072594887e-05,
+      "loss": 17.9124,
+      "step": 13374
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8471177352023877e-05,
+      "loss": 0.7594,
+      "step": 13375
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8464047981923374e-05,
+      "loss": 0.7726,
+      "step": 13376
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8456919149365137e-05,
+      "loss": 0.7159,
+      "step": 13377
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.844979085452713e-05,
+      "loss": 0.7329,
+      "step": 13378
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.844266309758727e-05,
+      "loss": 19.8529,
+      "step": 13379
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8435535878723457e-05,
+      "loss": 16.6976,
+      "step": 13380
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.842840919811359e-05,
+      "loss": 0.7487,
+      "step": 13381
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8421283055935544e-05,
+      "loss": 16.8408,
+      "step": 13382
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8414157452367218e-05,
+      "loss": 0.7295,
+      "step": 13383
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8407032387586442e-05,
+      "loss": 0.7275,
+      "step": 13384
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8399907861771058e-05,
+      "loss": 17.8657,
+      "step": 13385
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8392783875098917e-05,
+      "loss": 0.7846,
+      "step": 13386
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.838566042774783e-05,
+      "loss": 18.7753,
+      "step": 13387
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8378537519895593e-05,
+      "loss": 17.6094,
+      "step": 13388
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8371415151719994e-05,
+      "loss": 17.5815,
+      "step": 13389
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8364293323398795e-05,
+      "loss": 0.781,
+      "step": 13390
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.835717203510979e-05,
+      "loss": 0.8025,
+      "step": 13391
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8350051287030698e-05,
+      "loss": 0.7286,
+      "step": 13392
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8342931079339287e-05,
+      "loss": 0.7869,
+      "step": 13393
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8335811412213253e-05,
+      "loss": 16.1471,
+      "step": 13394
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8328692285830326e-05,
+      "loss": 0.7617,
+      "step": 13395
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8321573700368175e-05,
+      "loss": 0.7981,
+      "step": 13396
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8314455656004485e-05,
+      "loss": 0.7567,
+      "step": 13397
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.830733815291695e-05,
+      "loss": 0.7956,
+      "step": 13398
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8300221191283205e-05,
+      "loss": 18.9101,
+      "step": 13399
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8293104771280876e-05,
+      "loss": 0.7277,
+      "step": 13400
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.828598889308765e-05,
+      "loss": 16.3759,
+      "step": 13401
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.827887355688106e-05,
+      "loss": 0.7378,
+      "step": 13402
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.827175876283876e-05,
+      "loss": 19.8221,
+      "step": 13403
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8264644511138326e-05,
+      "loss": 17.9433,
+      "step": 13404
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.825753080195731e-05,
+      "loss": 17.3329,
+      "step": 13405
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8250417635473302e-05,
+      "loss": 0.7584,
+      "step": 13406
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.824330501186384e-05,
+      "loss": 0.7405,
+      "step": 13407
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.823619293130646e-05,
+      "loss": 0.7343,
+      "step": 13408
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.822908139397866e-05,
+      "loss": 0.7774,
+      "step": 13409
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.822197040005795e-05,
+      "loss": 15.1997,
+      "step": 13410
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.821485994972185e-05,
+      "loss": 0.7592,
+      "step": 13411
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.820775004314782e-05,
+      "loss": 0.7526,
+      "step": 13412
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8200640680513302e-05,
+      "loss": 0.7753,
+      "step": 13413
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8193531861995793e-05,
+      "loss": 0.8323,
+      "step": 13414
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8186423587772704e-05,
+      "loss": 0.7671,
+      "step": 13415
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.817931585802146e-05,
+      "loss": 16.0975,
+      "step": 13416
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.817220867291948e-05,
+      "loss": 18.7812,
+      "step": 13417
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.816510203264413e-05,
+      "loss": 0.7487,
+      "step": 13418
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.815799593737285e-05,
+      "loss": 17.6778,
+      "step": 13419
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8150890387282946e-05,
+      "loss": 18.923,
+      "step": 13420
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8143785382551835e-05,
+      "loss": 0.7866,
+      "step": 13421
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.813668092335683e-05,
+      "loss": 0.7383,
+      "step": 13422
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8129577009875262e-05,
+      "loss": 0.7414,
+      "step": 13423
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.812247364228444e-05,
+      "loss": 18.4142,
+      "step": 13424
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8115370820761654e-05,
+      "loss": 18.0972,
+      "step": 13425
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.810826854548423e-05,
+      "loss": 18.4348,
+      "step": 13426
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.810116681662942e-05,
+      "loss": 17.1609,
+      "step": 13427
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8094065634374467e-05,
+      "loss": 0.7289,
+      "step": 13428
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.808696499889668e-05,
+      "loss": 0.7661,
+      "step": 13429
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.807986491037321e-05,
+      "loss": 0.7245,
+      "step": 13430
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8072765368981336e-05,
+      "loss": 0.6799,
+      "step": 13431
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8065666374898248e-05,
+      "loss": 16.9378,
+      "step": 13432
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8058567928301106e-05,
+      "loss": 17.4351,
+      "step": 13433
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8051470029367142e-05,
+      "loss": 0.7228,
+      "step": 13434
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.80443726782735e-05,
+      "loss": 16.0416,
+      "step": 13435
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8037275875197333e-05,
+      "loss": 17.7886,
+      "step": 13436
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8030179620315767e-05,
+      "loss": 0.747,
+      "step": 13437
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.802308391380592e-05,
+      "loss": 0.7326,
+      "step": 13438
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8015988755844936e-05,
+      "loss": 0.7386,
+      "step": 13439
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8008894146609886e-05,
+      "loss": 18.9081,
+      "step": 13440
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.8001800086277848e-05,
+      "loss": 17.8743,
+      "step": 13441
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.799470657502592e-05,
+      "loss": 0.7521,
+      "step": 13442
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.798761361303114e-05,
+      "loss": 0.7259,
+      "step": 13443
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7980521200470545e-05,
+      "loss": 17.5845,
+      "step": 13444
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7973429337521174e-05,
+      "loss": 21.2799,
+      "step": 13445
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7966338024360018e-05,
+      "loss": 18.7376,
+      "step": 13446
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7959247261164114e-05,
+      "loss": 16.6209,
+      "step": 13447
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7952157048110406e-05,
+      "loss": 0.7166,
+      "step": 13448
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7945067385375912e-05,
+      "loss": 0.7624,
+      "step": 13449
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7937978273137577e-05,
+      "loss": 0.8008,
+      "step": 13450
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.793088971157234e-05,
+      "loss": 17.3656,
+      "step": 13451
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7923801700857132e-05,
+      "loss": 17.9128,
+      "step": 13452
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7916714241168862e-05,
+      "loss": 0.7466,
+      "step": 13453
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7909627332684464e-05,
+      "loss": 16.8102,
+      "step": 13454
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7902540975580804e-05,
+      "loss": 0.7379,
+      "step": 13455
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7895455170034777e-05,
+      "loss": 17.6039,
+      "step": 13456
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7888369916223234e-05,
+      "loss": 16.4091,
+      "step": 13457
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7881285214323004e-05,
+      "loss": 0.7482,
+      "step": 13458
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7874201064510962e-05,
+      "loss": 17.5916,
+      "step": 13459
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.786711746696392e-05,
+      "loss": 0.703,
+      "step": 13460
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7860034421858655e-05,
+      "loss": 19.5457,
+      "step": 13461
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7852951929372005e-05,
+      "loss": 0.7694,
+      "step": 13462
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7845869989680738e-05,
+      "loss": 0.7751,
+      "step": 13463
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.783878860296161e-05,
+      "loss": 17.1033,
+      "step": 13464
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.783170776939138e-05,
+      "loss": 0.8573,
+      "step": 13465
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7824627489146767e-05,
+      "loss": 15.3311,
+      "step": 13466
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7817547762404535e-05,
+      "loss": 16.688,
+      "step": 13467
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.781046858934137e-05,
+      "loss": 0.8277,
+      "step": 13468
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7803389970133965e-05,
+      "loss": 0.7521,
+      "step": 13469
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.779631190495905e-05,
+      "loss": 0.7581,
+      "step": 13470
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7789234393993213e-05,
+      "loss": 0.6982,
+      "step": 13471
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.778215743741318e-05,
+      "loss": 0.7351,
+      "step": 13472
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7775081035395554e-05,
+      "loss": 0.6949,
+      "step": 13473
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7768005188117e-05,
+      "loss": 0.841,
+      "step": 13474
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7760929895754116e-05,
+      "loss": 16.1711,
+      "step": 13475
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.775385515848348e-05,
+      "loss": 0.7617,
+      "step": 13476
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.774678097648172e-05,
+      "loss": 17.0072,
+      "step": 13477
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.773970734992537e-05,
+      "loss": 0.7534,
+      "step": 13478
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7732634278991022e-05,
+      "loss": 0.7473,
+      "step": 13479
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.77255617638552e-05,
+      "loss": 18.8734,
+      "step": 13480
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7718489804694424e-05,
+      "loss": 0.706,
+      "step": 13481
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7711418401685253e-05,
+      "loss": 0.7517,
+      "step": 13482
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7704347555004173e-05,
+      "loss": 18.7614,
+      "step": 13483
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.769727726482766e-05,
+      "loss": 0.7521,
+      "step": 13484
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.76902075313322e-05,
+      "loss": 0.7317,
+      "step": 13485
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7683138354694237e-05,
+      "loss": 0.6976,
+      "step": 13486
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7676069735090255e-05,
+      "loss": 17.7389,
+      "step": 13487
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7669001672696672e-05,
+      "loss": 17.843,
+      "step": 13488
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7661934167689884e-05,
+      "loss": 15.9388,
+      "step": 13489
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.765486722024634e-05,
+      "loss": 17.4302,
+      "step": 13490
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7647800830542408e-05,
+      "loss": 0.7304,
+      "step": 13491
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7640734998754476e-05,
+      "loss": 0.7589,
+      "step": 13492
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7633669725058896e-05,
+      "loss": 0.7013,
+      "step": 13493
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7626605009632016e-05,
+      "loss": 0.823,
+      "step": 13494
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7619540852650204e-05,
+      "loss": 18.4914,
+      "step": 13495
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.761247725428976e-05,
+      "loss": 0.8019,
+      "step": 13496
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7605414214726972e-05,
+      "loss": 0.7679,
+      "step": 13497
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7598351734138196e-05,
+      "loss": 0.7724,
+      "step": 13498
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7591289812699644e-05,
+      "loss": 16.4974,
+      "step": 13499
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.758422845058763e-05,
+      "loss": 0.7236,
+      "step": 13500
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7577167647978376e-05,
+      "loss": 18.215,
+      "step": 13501
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.757010740504815e-05,
+      "loss": 17.7185,
+      "step": 13502
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7563047721973166e-05,
+      "loss": 20.5245,
+      "step": 13503
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7555988598929623e-05,
+      "loss": 0.7437,
+      "step": 13504
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7548930036093763e-05,
+      "loss": 0.7642,
+      "step": 13505
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.754187203364169e-05,
+      "loss": 0.7444,
+      "step": 13506
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7534814591749646e-05,
+      "loss": 16.7205,
+      "step": 13507
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.752775771059376e-05,
+      "loss": 0.6985,
+      "step": 13508
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7520701390350155e-05,
+      "loss": 18.355,
+      "step": 13509
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.751364563119499e-05,
+      "loss": 0.8414,
+      "step": 13510
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7506590433304367e-05,
+      "loss": 17.4402,
+      "step": 13511
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7499535796854382e-05,
+      "loss": 0.7536,
+      "step": 13512
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.749248172202112e-05,
+      "loss": 18.4553,
+      "step": 13513
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7485428208980636e-05,
+      "loss": 18.5527,
+      "step": 13514
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7478375257909016e-05,
+      "loss": 18.1398,
+      "step": 13515
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.74713228689823e-05,
+      "loss": 15.7543,
+      "step": 13516
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7464271042376487e-05,
+      "loss": 17.4155,
+      "step": 13517
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.745721977826764e-05,
+      "loss": 0.7092,
+      "step": 13518
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7450169076831728e-05,
+      "loss": 0.8592,
+      "step": 13519
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7443118938244746e-05,
+      "loss": 0.7833,
+      "step": 13520
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7436069362682664e-05,
+      "loss": 0.7554,
+      "step": 13521
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7429020350321422e-05,
+      "loss": 0.7197,
+      "step": 13522
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7421971901337012e-05,
+      "loss": 0.8459,
+      "step": 13523
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7414924015905337e-05,
+      "loss": 18.677,
+      "step": 13524
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7407876694202294e-05,
+      "loss": 0.694,
+      "step": 13525
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7400829936403844e-05,
+      "loss": 16.9944,
+      "step": 13526
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.73937837426858e-05,
+      "loss": 18.7037,
+      "step": 13527
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7386738113224097e-05,
+      "loss": 20.8238,
+      "step": 13528
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7379693048194554e-05,
+      "loss": 0.7368,
+      "step": 13529
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.737264854777306e-05,
+      "loss": 16.5844,
+      "step": 13530
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.736560461213542e-05,
+      "loss": 17.8958,
+      "step": 13531
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7358561241457458e-05,
+      "loss": 19.023,
+      "step": 13532
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.735151843591498e-05,
+      "loss": 17.2541,
+      "step": 13533
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.734447619568375e-05,
+      "loss": 17.6869,
+      "step": 13534
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7337434520939587e-05,
+      "loss": 19.1719,
+      "step": 13535
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7330393411858237e-05,
+      "loss": 0.8158,
+      "step": 13536
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7323352868615417e-05,
+      "loss": 0.8098,
+      "step": 13537
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7316312891386907e-05,
+      "loss": 0.7574,
+      "step": 13538
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.730927348034842e-05,
+      "loss": 18.4702,
+      "step": 13539
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7302234635675632e-05,
+      "loss": 16.1099,
+      "step": 13540
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.729519635754426e-05,
+      "loss": 0.7188,
+      "step": 13541
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7288158646129945e-05,
+      "loss": 18.2063,
+      "step": 13542
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7281121501608398e-05,
+      "loss": 0.7859,
+      "step": 13543
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7274084924155246e-05,
+      "loss": 17.1671,
+      "step": 13544
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7267048913946103e-05,
+      "loss": 18.2972,
+      "step": 13545
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7260013471156655e-05,
+      "loss": 0.7737,
+      "step": 13546
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7252978595962415e-05,
+      "loss": 17.2986,
+      "step": 13547
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.724594428853905e-05,
+      "loss": 0.7724,
+      "step": 13548
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7238910549062103e-05,
+      "loss": 17.843,
+      "step": 13549
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7231877377707126e-05,
+      "loss": 17.3413,
+      "step": 13550
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7224844774649717e-05,
+      "loss": 0.737,
+      "step": 13551
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7217812740065373e-05,
+      "loss": 0.7454,
+      "step": 13552
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.721078127412964e-05,
+      "loss": 18.3012,
+      "step": 13553
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7203750377017995e-05,
+      "loss": 0.7073,
+      "step": 13554
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7196720048905932e-05,
+      "loss": 15.0541,
+      "step": 13555
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.718969028996896e-05,
+      "loss": 0.7268,
+      "step": 13556
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7182661100382505e-05,
+      "loss": 17.214,
+      "step": 13557
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7175632480322066e-05,
+      "loss": 0.7457,
+      "step": 13558
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7168604429963045e-05,
+      "loss": 17.8244,
+      "step": 13559
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7161576949480873e-05,
+      "loss": 17.9434,
+      "step": 13560
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7154550039050952e-05,
+      "loss": 0.8008,
+      "step": 13561
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7147523698848663e-05,
+      "loss": 17.3859,
+      "step": 13562
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7140497929049424e-05,
+      "loss": 20.5254,
+      "step": 13563
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.713347272982857e-05,
+      "loss": 0.7609,
+      "step": 13564
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7126448101361434e-05,
+      "loss": 0.7453,
+      "step": 13565
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7119424043823403e-05,
+      "loss": 0.7391,
+      "step": 13566
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.711240055738977e-05,
+      "loss": 18.2756,
+      "step": 13567
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7105377642235848e-05,
+      "loss": 18.713,
+      "step": 13568
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7098355298536926e-05,
+      "loss": 0.6951,
+      "step": 13569
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7091333526468266e-05,
+      "loss": 0.8068,
+      "step": 13570
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7084312326205165e-05,
+      "loss": 17.005,
+      "step": 13571
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7077291697922868e-05,
+      "loss": 0.7415,
+      "step": 13572
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.707027164179658e-05,
+      "loss": 15.6309,
+      "step": 13573
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7063252158001583e-05,
+      "loss": 0.7229,
+      "step": 13574
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.705623324671302e-05,
+      "loss": 18.585,
+      "step": 13575
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7049214908106124e-05,
+      "loss": 0.7624,
+      "step": 13576
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7042197142356072e-05,
+      "loss": 0.7207,
+      "step": 13577
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7035179949638e-05,
+      "loss": 0.7884,
+      "step": 13578
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.70281633301271e-05,
+      "loss": 0.7079,
+      "step": 13579
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.702114728399848e-05,
+      "loss": 18.0412,
+      "step": 13580
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7014131811427278e-05,
+      "loss": 0.7717,
+      "step": 13581
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.70071169125886e-05,
+      "loss": 0.7569,
+      "step": 13582
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7000102587657504e-05,
+      "loss": 18.5331,
+      "step": 13583
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6993088836809122e-05,
+      "loss": 0.7898,
+      "step": 13584
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.698607566021848e-05,
+      "loss": 18.6592,
+      "step": 13585
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6979063058060662e-05,
+      "loss": 16.3619,
+      "step": 13586
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.697205103051068e-05,
+      "loss": 17.5844,
+      "step": 13587
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.696503957774357e-05,
+      "loss": 18.124,
+      "step": 13588
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6958028699934327e-05,
+      "loss": 0.7798,
+      "step": 13589
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6951018397257933e-05,
+      "loss": 15.5576,
+      "step": 13590
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6944008669889396e-05,
+      "loss": 17.0788,
+      "step": 13591
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6936999518003674e-05,
+      "loss": 0.7815,
+      "step": 13592
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6929990941775684e-05,
+      "loss": 0.7416,
+      "step": 13593
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6922982941380414e-05,
+      "loss": 0.7585,
+      "step": 13594
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.691597551699276e-05,
+      "loss": 16.262,
+      "step": 13595
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.690896866878762e-05,
+      "loss": 17.6581,
+      "step": 13596
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.69019623969399e-05,
+      "loss": 0.7924,
+      "step": 13597
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6894956701624453e-05,
+      "loss": 0.7648,
+      "step": 13598
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.688795158301618e-05,
+      "loss": 20.9772,
+      "step": 13599
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6880947041289906e-05,
+      "loss": 0.7927,
+      "step": 13600
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6873943076620467e-05,
+      "loss": 0.7605,
+      "step": 13601
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6866939689182718e-05,
+      "loss": 0.7539,
+      "step": 13602
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.68599368791514e-05,
+      "loss": 0.7651,
+      "step": 13603
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6852934646701362e-05,
+      "loss": 0.743,
+      "step": 13604
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.684593299200736e-05,
+      "loss": 0.7532,
+      "step": 13605
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6838931915244137e-05,
+      "loss": 16.798,
+      "step": 13606
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6831931416586475e-05,
+      "loss": 17.4223,
+      "step": 13607
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6824931496209105e-05,
+      "loss": 16.5972,
+      "step": 13608
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.681793215428673e-05,
+      "loss": 17.5562,
+      "step": 13609
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6810933390994066e-05,
+      "loss": 16.5069,
+      "step": 13610
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.680393520650578e-05,
+      "loss": 19.1046,
+      "step": 13611
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6796937600996587e-05,
+      "loss": 0.7211,
+      "step": 13612
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6789940574641102e-05,
+      "loss": 18.0601,
+      "step": 13613
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.678294412761403e-05,
+      "loss": 0.7638,
+      "step": 13614
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6775948260089984e-05,
+      "loss": 16.8423,
+      "step": 13615
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.676895297224354e-05,
+      "loss": 0.7846,
+      "step": 13616
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6761958264249347e-05,
+      "loss": 17.2008,
+      "step": 13617
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.675496413628197e-05,
+      "loss": 16.7682,
+      "step": 13618
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6747970588516008e-05,
+      "loss": 18.9397,
+      "step": 13619
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.674097762112602e-05,
+      "loss": 0.749,
+      "step": 13620
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.673398523428651e-05,
+      "loss": 19.4014,
+      "step": 13621
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6726993428172086e-05,
+      "loss": 19.7448,
+      "step": 13622
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6720002202957172e-05,
+      "loss": 18.291,
+      "step": 13623
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6713011558816346e-05,
+      "loss": 15.7704,
+      "step": 13624
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.670602149592407e-05,
+      "loss": 18.8463,
+      "step": 13625
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6699032014454794e-05,
+      "loss": 0.7584,
+      "step": 13626
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.669204311458302e-05,
+      "loss": 0.7474,
+      "step": 13627
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6685054796483173e-05,
+      "loss": 0.7261,
+      "step": 13628
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6678067060329682e-05,
+      "loss": 19.4041,
+      "step": 13629
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6671079906296958e-05,
+      "loss": 17.4005,
+      "step": 13630
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.666409333455939e-05,
+      "loss": 15.8223,
+      "step": 13631
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.66571073452914e-05,
+      "loss": 0.7749,
+      "step": 13632
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6650121938667338e-05,
+      "loss": 0.7685,
+      "step": 13633
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6643137114861544e-05,
+      "loss": 0.7348,
+      "step": 13634
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6636152874048402e-05,
+      "loss": 17.798,
+      "step": 13635
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6629169216402218e-05,
+      "loss": 17.3536,
+      "step": 13636
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.662218614209731e-05,
+      "loss": 0.7656,
+      "step": 13637
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6615203651307974e-05,
+      "loss": 0.7592,
+      "step": 13638
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.660822174420847e-05,
+      "loss": 18.245,
+      "step": 13639
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6601240420973116e-05,
+      "loss": 0.7471,
+      "step": 13640
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.659425968177612e-05,
+      "loss": 0.7404,
+      "step": 13641
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6587279526791763e-05,
+      "loss": 17.4592,
+      "step": 13642
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.658029995619429e-05,
+      "loss": 0.7913,
+      "step": 13643
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6573320970157822e-05,
+      "loss": 18.5648,
+      "step": 13644
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.656634256885664e-05,
+      "loss": 0.6948,
+      "step": 13645
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6559364752464876e-05,
+      "loss": 18.3973,
+      "step": 13646
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6552387521156734e-05,
+      "loss": 0.7099,
+      "step": 13647
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.654541087510636e-05,
+      "loss": 0.757,
+      "step": 13648
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6538434814487867e-05,
+      "loss": 16.9049,
+      "step": 13649
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6531459339475433e-05,
+      "loss": 0.7821,
+      "step": 13650
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6524484450243093e-05,
+      "loss": 0.8042,
+      "step": 13651
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.651751014696501e-05,
+      "loss": 20.8168,
+      "step": 13652
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6510536429815226e-05,
+      "loss": 18.401,
+      "step": 13653
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6503563298967797e-05,
+      "loss": 15.625,
+      "step": 13654
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6496590754596817e-05,
+      "loss": 0.7607,
+      "step": 13655
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.648961879687629e-05,
+      "loss": 16.8711,
+      "step": 13656
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6482647425980257e-05,
+      "loss": 17.1257,
+      "step": 13657
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.647567664208271e-05,
+      "loss": 0.8193,
+      "step": 13658
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6468706445357627e-05,
+      "loss": 18.0845,
+      "step": 13659
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6461736835979018e-05,
+      "loss": 0.8025,
+      "step": 13660
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6454767814120834e-05,
+      "loss": 0.7915,
+      "step": 13661
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6447799379957006e-05,
+      "loss": 17.9921,
+      "step": 13662
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6440831533661498e-05,
+      "loss": 15.9316,
+      "step": 13663
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6433864275408215e-05,
+      "loss": 0.7014,
+      "step": 13664
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.642689760537106e-05,
+      "loss": 19.6846,
+      "step": 13665
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.641993152372393e-05,
+      "loss": 16.1546,
+      "step": 13666
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.641296603064066e-05,
+      "loss": 19.4213,
+      "step": 13667
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6406001126295178e-05,
+      "loss": 0.7629,
+      "step": 13668
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6399036810861267e-05,
+      "loss": 18.8491,
+      "step": 13669
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6392073084512808e-05,
+      "loss": 16.3397,
+      "step": 13670
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6385109947423615e-05,
+      "loss": 0.7728,
+      "step": 13671
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6378147399767428e-05,
+      "loss": 0.7388,
+      "step": 13672
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.63711854417181e-05,
+      "loss": 0.779,
+      "step": 13673
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.636422407344935e-05,
+      "loss": 16.7517,
+      "step": 13674
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6357263295134995e-05,
+      "loss": 0.7459,
+      "step": 13675
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6350303106948737e-05,
+      "loss": 18.0278,
+      "step": 13676
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.63433435090643e-05,
+      "loss": 19.9251,
+      "step": 13677
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6336384501655446e-05,
+      "loss": 0.7836,
+      "step": 13678
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6329426084895802e-05,
+      "loss": 0.7073,
+      "step": 13679
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6322468258959104e-05,
+      "loss": 0.7787,
+      "step": 13680
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6315511024019007e-05,
+      "loss": 19.9212,
+      "step": 13681
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.630855438024914e-05,
+      "loss": 16.9617,
+      "step": 13682
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6301598327823184e-05,
+      "loss": 17.7095,
+      "step": 13683
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6294642866914754e-05,
+      "loss": 16.7081,
+      "step": 13684
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6287687997697442e-05,
+      "loss": 17.5479,
+      "step": 13685
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6280733720344853e-05,
+      "loss": 15.277,
+      "step": 13686
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.627378003503055e-05,
+      "loss": 0.7279,
+      "step": 13687
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.626682694192813e-05,
+      "loss": 0.7802,
+      "step": 13688
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.625987444121113e-05,
+      "loss": 0.7396,
+      "step": 13689
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6252922533053064e-05,
+      "loss": 15.7904,
+      "step": 13690
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6245971217627518e-05,
+      "loss": 18.2724,
+      "step": 13691
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6239020495107914e-05,
+      "loss": 17.183,
+      "step": 13692
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6232070365667793e-05,
+      "loss": 18.7607,
+      "step": 13693
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.622512082948063e-05,
+      "loss": 0.7988,
+      "step": 13694
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6218171886719866e-05,
+      "loss": 18.407,
+      "step": 13695
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.621122353755897e-05,
+      "loss": 18.1508,
+      "step": 13696
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6204275782171356e-05,
+      "loss": 18.0515,
+      "step": 13697
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6197328620730488e-05,
+      "loss": 17.1418,
+      "step": 13698
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6190382053409713e-05,
+      "loss": 17.4235,
+      "step": 13699
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6183436080382416e-05,
+      "loss": 16.2103,
+      "step": 13700
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6176490701822008e-05,
+      "loss": 17.2806,
+      "step": 13701
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6169545917901823e-05,
+      "loss": 14.4002,
+      "step": 13702
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6162601728795227e-05,
+      "loss": 0.7525,
+      "step": 13703
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.615565813467553e-05,
+      "loss": 0.7472,
+      "step": 13704
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.614871513571606e-05,
+      "loss": 17.4084,
+      "step": 13705
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6141772732090103e-05,
+      "loss": 18.5032,
+      "step": 13706
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.613483092397092e-05,
+      "loss": 17.2246,
+      "step": 13707
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6127889711531837e-05,
+      "loss": 18.5993,
+      "step": 13708
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6120949094946068e-05,
+      "loss": 17.548,
+      "step": 13709
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6114009074386846e-05,
+      "loss": 0.7387,
+      "step": 13710
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6107069650027434e-05,
+      "loss": 0.7211,
+      "step": 13711
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6100130822041013e-05,
+      "loss": 18.954,
+      "step": 13712
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.609319259060079e-05,
+      "loss": 0.7809,
+      "step": 13713
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.608625495587993e-05,
+      "loss": 0.7297,
+      "step": 13714
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6079317918051594e-05,
+      "loss": 0.7161,
+      "step": 13715
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6072381477288964e-05,
+      "loss": 0.7736,
+      "step": 13716
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.606544563376515e-05,
+      "loss": 20.2279,
+      "step": 13717
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6058510387653267e-05,
+      "loss": 17.7395,
+      "step": 13718
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6051575739126467e-05,
+      "loss": 16.8007,
+      "step": 13719
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6044641688357764e-05,
+      "loss": 0.7939,
+      "step": 13720
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.603770823552029e-05,
+      "loss": 18.3563,
+      "step": 13721
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6030775380787097e-05,
+      "loss": 0.7444,
+      "step": 13722
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6023843124331203e-05,
+      "loss": 0.7436,
+      "step": 13723
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6016911466325676e-05,
+      "loss": 0.7929,
+      "step": 13724
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.600998040694349e-05,
+      "loss": 0.7409,
+      "step": 13725
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.6003049946357717e-05,
+      "loss": 18.7222,
+      "step": 13726
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5996120084741277e-05,
+      "loss": 0.7256,
+      "step": 13727
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5989190822267138e-05,
+      "loss": 18.2145,
+      "step": 13728
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.59822621591083e-05,
+      "loss": 0.7742,
+      "step": 13729
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5975334095437663e-05,
+      "loss": 0.7419,
+      "step": 13730
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5968406631428188e-05,
+      "loss": 17.2699,
+      "step": 13731
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.596147976725278e-05,
+      "loss": 21.7535,
+      "step": 13732
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5954553503084318e-05,
+      "loss": 17.0227,
+      "step": 13733
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.594762783909569e-05,
+      "loss": 0.7703,
+      "step": 13734
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5940702775459747e-05,
+      "loss": 15.4751,
+      "step": 13735
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5933778312349378e-05,
+      "loss": 0.7225,
+      "step": 13736
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5926854449937388e-05,
+      "loss": 0.7394,
+      "step": 13737
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5919931188396594e-05,
+      "loss": 0.7637,
+      "step": 13738
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.591300852789984e-05,
+      "loss": 0.7594,
+      "step": 13739
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5906086468619885e-05,
+      "loss": 18.4009,
+      "step": 13740
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.589916501072952e-05,
+      "loss": 16.9595,
+      "step": 13741
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5892244154401492e-05,
+      "loss": 0.7978,
+      "step": 13742
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.588532389980854e-05,
+      "loss": 18.2857,
+      "step": 13743
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5878404247123428e-05,
+      "loss": 18.7815,
+      "step": 13744
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.587148519651885e-05,
+      "loss": 0.7303,
+      "step": 13745
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5864566748167495e-05,
+      "loss": 19.7681,
+      "step": 13746
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5857648902242103e-05,
+      "loss": 16.3993,
+      "step": 13747
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5850731658915272e-05,
+      "loss": 0.7874,
+      "step": 13748
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5843815018359706e-05,
+      "loss": 0.6995,
+      "step": 13749
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5836898980748036e-05,
+      "loss": 16.9816,
+      "step": 13750
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.582998354625286e-05,
+      "loss": 0.7835,
+      "step": 13751
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.582306871504684e-05,
+      "loss": 16.8632,
+      "step": 13752
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5816154487302523e-05,
+      "loss": 15.1389,
+      "step": 13753
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5809240863192553e-05,
+      "loss": 17.3247,
+      "step": 13754
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5802327842889427e-05,
+      "loss": 18.3192,
+      "step": 13755
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5795415426565712e-05,
+      "loss": 17.6281,
+      "step": 13756
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5788503614393973e-05,
+      "loss": 17.4293,
+      "step": 13757
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.578159240654669e-05,
+      "loss": 19.8399,
+      "step": 13758
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.577468180319641e-05,
+      "loss": 0.7253,
+      "step": 13759
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5767771804515605e-05,
+      "loss": 17.8654,
+      "step": 13760
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5760862410676744e-05,
+      "loss": 0.7136,
+      "step": 13761
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.57539536218523e-05,
+      "loss": 0.7876,
+      "step": 13762
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5747045438214683e-05,
+      "loss": 0.7443,
+      "step": 13763
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5740137859936375e-05,
+      "loss": 0.7263,
+      "step": 13764
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5733230887189764e-05,
+      "loss": 18.0002,
+      "step": 13765
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5726324520147226e-05,
+      "loss": 18.7115,
+      "step": 13766
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5719418758981212e-05,
+      "loss": 0.74,
+      "step": 13767
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5712513603864012e-05,
+      "loss": 18.1717,
+      "step": 13768
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.570560905496804e-05,
+      "loss": 17.3837,
+      "step": 13769
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5698705112465605e-05,
+      "loss": 19.2125,
+      "step": 13770
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5691801776529028e-05,
+      "loss": 0.7645,
+      "step": 13771
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5684899047330645e-05,
+      "loss": 0.7757,
+      "step": 13772
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5677996925042736e-05,
+      "loss": 0.6984,
+      "step": 13773
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5671095409837574e-05,
+      "loss": 16.7849,
+      "step": 13774
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.566419450188743e-05,
+      "loss": 18.5809,
+      "step": 13775
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5657294201364522e-05,
+      "loss": 0.7371,
+      "step": 13776
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5650394508441133e-05,
+      "loss": 0.6875,
+      "step": 13777
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5643495423289454e-05,
+      "loss": 0.7491,
+      "step": 13778
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.563659694608167e-05,
+      "loss": 0.7143,
+      "step": 13779
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.562969907699001e-05,
+      "loss": 0.7588,
+      "step": 13780
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5622801816186627e-05,
+      "loss": 0.761,
+      "step": 13781
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5615905163843677e-05,
+      "loss": 0.7437,
+      "step": 13782
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5609009120133297e-05,
+      "loss": 17.6299,
+      "step": 13783
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5602113685227603e-05,
+      "loss": 17.2285,
+      "step": 13784
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.559521885929874e-05,
+      "loss": 0.7803,
+      "step": 13785
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5588324642518773e-05,
+      "loss": 0.7324,
+      "step": 13786
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5581431035059812e-05,
+      "loss": 0.8066,
+      "step": 13787
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.557453803709391e-05,
+      "loss": 0.7251,
+      "step": 13788
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5567645648793114e-05,
+      "loss": 0.7458,
+      "step": 13789
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5560753870329456e-05,
+      "loss": 17.3087,
+      "step": 13790
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5553862701874948e-05,
+      "loss": 17.813,
+      "step": 13791
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5546972143601627e-05,
+      "loss": 15.8861,
+      "step": 13792
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5540082195681464e-05,
+      "loss": 0.7702,
+      "step": 13793
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5533192858286413e-05,
+      "loss": 0.736,
+      "step": 13794
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5526304131588497e-05,
+      "loss": 0.749,
+      "step": 13795
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.551941601575958e-05,
+      "loss": 0.6969,
+      "step": 13796
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5512528510971646e-05,
+      "loss": 19.2008,
+      "step": 13797
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5505641617396593e-05,
+      "loss": 0.7768,
+      "step": 13798
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5498755335206297e-05,
+      "loss": 0.7706,
+      "step": 13799
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5491869664572686e-05,
+      "loss": 17.2963,
+      "step": 13800
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5484984605667596e-05,
+      "loss": 16.2019,
+      "step": 13801
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5478100158662894e-05,
+      "loss": 0.7945,
+      "step": 13802
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5471216323730407e-05,
+      "loss": 16.1402,
+      "step": 13803
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5464333101041947e-05,
+      "loss": 19.3536,
+      "step": 13804
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5457450490769353e-05,
+      "loss": 17.9978,
+      "step": 13805
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5450568493084393e-05,
+      "loss": 0.7064,
+      "step": 13806
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5443687108158836e-05,
+      "loss": 18.5289,
+      "step": 13807
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5436806336164476e-05,
+      "loss": 0.767,
+      "step": 13808
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5429926177273035e-05,
+      "loss": 19.2618,
+      "step": 13809
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5423046631656243e-05,
+      "loss": 0.7099,
+      "step": 13810
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.541616769948582e-05,
+      "loss": 0.7721,
+      "step": 13811
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5409289380933448e-05,
+      "loss": 18.3911,
+      "step": 13812
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5402411676170846e-05,
+      "loss": 17.5132,
+      "step": 13813
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.539553458536965e-05,
+      "loss": 16.9209,
+      "step": 13814
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5388658108701545e-05,
+      "loss": 0.7412,
+      "step": 13815
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.538178224633815e-05,
+      "loss": 0.7869,
+      "step": 13816
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.537490699845109e-05,
+      "loss": 0.7588,
+      "step": 13817
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.536803236521198e-05,
+      "loss": 0.7612,
+      "step": 13818
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5361158346792384e-05,
+      "loss": 0.6849,
+      "step": 13819
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5354284943363926e-05,
+      "loss": 18.3804,
+      "step": 13820
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.534741215509814e-05,
+      "loss": 17.1199,
+      "step": 13821
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.534053998216655e-05,
+      "loss": 19.6608,
+      "step": 13822
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.533366842474075e-05,
+      "loss": 17.7216,
+      "step": 13823
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5326797482992194e-05,
+      "loss": 0.7355,
+      "step": 13824
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5319927157092416e-05,
+      "loss": 0.6668,
+      "step": 13825
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.531305744721289e-05,
+      "loss": 0.7175,
+      "step": 13826
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5306188353525073e-05,
+      "loss": 18.9957,
+      "step": 13827
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5299319876200444e-05,
+      "loss": 17.833,
+      "step": 13828
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.529245201541044e-05,
+      "loss": 0.7559,
+      "step": 13829
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.528558477132647e-05,
+      "loss": 16.2435,
+      "step": 13830
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.527871814411995e-05,
+      "loss": 16.7081,
+      "step": 13831
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5271852133962248e-05,
+      "loss": 0.7968,
+      "step": 13832
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5264986741024783e-05,
+      "loss": 17.761,
+      "step": 13833
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.52581219654789e-05,
+      "loss": 0.7655,
+      "step": 13834
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5251257807495915e-05,
+      "loss": 16.8401,
+      "step": 13835
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5244394267247228e-05,
+      "loss": 0.7473,
+      "step": 13836
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5237531344904076e-05,
+      "loss": 15.8634,
+      "step": 13837
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5230669040637812e-05,
+      "loss": 18.5286,
+      "step": 13838
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.522380735461971e-05,
+      "loss": 17.1693,
+      "step": 13839
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.521694628702101e-05,
+      "loss": 17.3902,
+      "step": 13840
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.521008583801301e-05,
+      "loss": 16.2418,
+      "step": 13841
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5203226007766907e-05,
+      "loss": 18.2121,
+      "step": 13842
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5196366796453986e-05,
+      "loss": 18.5623,
+      "step": 13843
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.518950820424537e-05,
+      "loss": 16.4272,
+      "step": 13844
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5182650231312316e-05,
+      "loss": 0.748,
+      "step": 13845
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5175792877825978e-05,
+      "loss": 0.764,
+      "step": 13846
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.51689361439575e-05,
+      "loss": 16.7374,
+      "step": 13847
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5162080029878067e-05,
+      "loss": 18.6857,
+      "step": 13848
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.515522453575878e-05,
+      "loss": 0.7747,
+      "step": 13849
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5148369661770765e-05,
+      "loss": 17.3864,
+      "step": 13850
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5141515408085127e-05,
+      "loss": 0.6827,
+      "step": 13851
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5134661774872908e-05,
+      "loss": 0.6587,
+      "step": 13852
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5127808762305233e-05,
+      "loss": 0.7138,
+      "step": 13853
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5120956370553128e-05,
+      "loss": 0.7412,
+      "step": 13854
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5114104599787613e-05,
+      "loss": 15.7118,
+      "step": 13855
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5107253450179757e-05,
+      "loss": 17.0259,
+      "step": 13856
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5100402921900536e-05,
+      "loss": 0.7731,
+      "step": 13857
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5093553015120935e-05,
+      "loss": 17.6887,
+      "step": 13858
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.508670373001194e-05,
+      "loss": 17.2652,
+      "step": 13859
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5079855066744496e-05,
+      "loss": 17.1426,
+      "step": 13860
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.507300702548957e-05,
+      "loss": 17.5845,
+      "step": 13861
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.506615960641809e-05,
+      "loss": 18.7839,
+      "step": 13862
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.505931280970093e-05,
+      "loss": 0.7504,
+      "step": 13863
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5052466635509066e-05,
+      "loss": 0.761,
+      "step": 13864
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.504562108401328e-05,
+      "loss": 0.7377,
+      "step": 13865
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.503877615538452e-05,
+      "loss": 0.7392,
+      "step": 13866
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.50319318497936e-05,
+      "loss": 18.8526,
+      "step": 13867
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5025088167411348e-05,
+      "loss": 19.7894,
+      "step": 13868
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.501824510840861e-05,
+      "loss": 18.6053,
+      "step": 13869
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5011402672956165e-05,
+      "loss": 18.4661,
+      "step": 13870
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.5004560861224857e-05,
+      "loss": 15.1871,
+      "step": 13871
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4997719673385374e-05,
+      "loss": 19.2915,
+      "step": 13872
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4990879109608527e-05,
+      "loss": 17.456,
+      "step": 13873
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.498403917006506e-05,
+      "loss": 16.8988,
+      "step": 13874
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4977199854925664e-05,
+      "loss": 17.7367,
+      "step": 13875
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.497036116436109e-05,
+      "loss": 16.3936,
+      "step": 13876
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.496352309854202e-05,
+      "loss": 17.1075,
+      "step": 13877
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4956685657639128e-05,
+      "loss": 17.3089,
+      "step": 13878
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4949848841823082e-05,
+      "loss": 18.268,
+      "step": 13879
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4943012651264508e-05,
+      "loss": 0.7729,
+      "step": 13880
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4936177086134076e-05,
+      "loss": 19.9825,
+      "step": 13881
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4929342146602393e-05,
+      "loss": 16.5131,
+      "step": 13882
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.492250783284003e-05,
+      "loss": 0.7759,
+      "step": 13883
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.491567414501762e-05,
+      "loss": 19.3519,
+      "step": 13884
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4908841083305707e-05,
+      "loss": 17.3215,
+      "step": 13885
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4902008647874852e-05,
+      "loss": 0.7128,
+      "step": 13886
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.489517683889559e-05,
+      "loss": 16.2514,
+      "step": 13887
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4888345656538432e-05,
+      "loss": 0.783,
+      "step": 13888
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4881515100973913e-05,
+      "loss": 18.4982,
+      "step": 13889
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4874685172372512e-05,
+      "loss": 18.9181,
+      "step": 13890
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4867855870904694e-05,
+      "loss": 0.7595,
+      "step": 13891
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4861027196740965e-05,
+      "loss": 16.9161,
+      "step": 13892
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4854199150051697e-05,
+      "loss": 0.8225,
+      "step": 13893
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4847371731007386e-05,
+      "loss": 0.7614,
+      "step": 13894
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4840544939778397e-05,
+      "loss": 0.7454,
+      "step": 13895
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.483371877653517e-05,
+      "loss": 0.7813,
+      "step": 13896
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.482689324144807e-05,
+      "loss": 17.1874,
+      "step": 13897
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.482006833468744e-05,
+      "loss": 0.7428,
+      "step": 13898
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.481324405642369e-05,
+      "loss": 0.7691,
+      "step": 13899
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4806420406827087e-05,
+      "loss": 18.8422,
+      "step": 13900
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4799597386067996e-05,
+      "loss": 0.6925,
+      "step": 13901
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4792774994316713e-05,
+      "loss": 17.7507,
+      "step": 13902
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4785953231743502e-05,
+      "loss": 0.7286,
+      "step": 13903
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4779132098518675e-05,
+      "loss": 0.7792,
+      "step": 13904
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.477231159481247e-05,
+      "loss": 18.5125,
+      "step": 13905
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4765491720795125e-05,
+      "loss": 0.7715,
+      "step": 13906
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.475867247663687e-05,
+      "loss": 0.7246,
+      "step": 13907
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4751853862507894e-05,
+      "loss": 16.9165,
+      "step": 13908
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4745035878578426e-05,
+      "loss": 0.7301,
+      "step": 13909
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4738218525018636e-05,
+      "loss": 0.7406,
+      "step": 13910
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4731401801998656e-05,
+      "loss": 0.7259,
+      "step": 13911
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4724585709688696e-05,
+      "loss": 0.7631,
+      "step": 13912
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4717770248258808e-05,
+      "loss": 18.326,
+      "step": 13913
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4710955417879166e-05,
+      "loss": 17.7494,
+      "step": 13914
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4704141218719857e-05,
+      "loss": 0.7408,
+      "step": 13915
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4697327650950934e-05,
+      "loss": 17.4171,
+      "step": 13916
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4690514714742513e-05,
+      "loss": 18.223,
+      "step": 13917
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.468370241026462e-05,
+      "loss": 16.2638,
+      "step": 13918
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4676890737687296e-05,
+      "loss": 17.7037,
+      "step": 13919
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4670079697180566e-05,
+      "loss": 17.0573,
+      "step": 13920
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4663269288914404e-05,
+      "loss": 0.7343,
+      "step": 13921
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.465645951305885e-05,
+      "loss": 0.772,
+      "step": 13922
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4649650369783833e-05,
+      "loss": 0.8561,
+      "step": 13923
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4642841859259345e-05,
+      "loss": 0.763,
+      "step": 13924
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4636033981655322e-05,
+      "loss": 17.6277,
+      "step": 13925
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4629226737141677e-05,
+      "loss": 17.1519,
+      "step": 13926
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.462242012588833e-05,
+      "loss": 0.7507,
+      "step": 13927
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4615614148065147e-05,
+      "loss": 0.7397,
+      "step": 13928
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.460880880384205e-05,
+      "loss": 0.769,
+      "step": 13929
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4602004093388886e-05,
+      "loss": 0.7645,
+      "step": 13930
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.459520001687548e-05,
+      "loss": 0.7336,
+      "step": 13931
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.45883965744717e-05,
+      "loss": 0.8128,
+      "step": 13932
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4581593766347343e-05,
+      "loss": 0.7658,
+      "step": 13933
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.457479159267221e-05,
+      "loss": 0.7384,
+      "step": 13934
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4567990053616086e-05,
+      "loss": 0.7706,
+      "step": 13935
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.456118914934871e-05,
+      "loss": 18.6016,
+      "step": 13936
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.455438888003989e-05,
+      "loss": 15.1188,
+      "step": 13937
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.454758924585933e-05,
+      "loss": 18.1263,
+      "step": 13938
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4540790246976735e-05,
+      "loss": 19.6407,
+      "step": 13939
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4533991883561868e-05,
+      "loss": 17.6867,
+      "step": 13940
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.452719415578434e-05,
+      "loss": 16.9227,
+      "step": 13941
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.452039706381388e-05,
+      "loss": 17.5129,
+      "step": 13942
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4513600607820125e-05,
+      "loss": 18.8284,
+      "step": 13943
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4506804787972702e-05,
+      "loss": 0.7855,
+      "step": 13944
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4500009604441272e-05,
+      "loss": 17.2828,
+      "step": 13945
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.449321505739542e-05,
+      "loss": 16.0315,
+      "step": 13946
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.448642114700475e-05,
+      "loss": 18.291,
+      "step": 13947
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4479627873438833e-05,
+      "loss": 16.7008,
+      "step": 13948
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.447283523686721e-05,
+      "loss": 17.3864,
+      "step": 13949
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4466043237459463e-05,
+      "loss": 18.7063,
+      "step": 13950
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4459251875385092e-05,
+      "loss": 18.511,
+      "step": 13951
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.445246115081365e-05,
+      "loss": 0.7653,
+      "step": 13952
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4445671063914605e-05,
+      "loss": 0.7702,
+      "step": 13953
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4438881614857444e-05,
+      "loss": 0.7637,
+      "step": 13954
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4432092803811635e-05,
+      "loss": 0.7603,
+      "step": 13955
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.442530463094661e-05,
+      "loss": 16.4832,
+      "step": 13956
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.441851709643184e-05,
+      "loss": 18.4431,
+      "step": 13957
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4411730200436727e-05,
+      "loss": 0.7078,
+      "step": 13958
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.440494394313065e-05,
+      "loss": 18.6989,
+      "step": 13959
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4398158324683034e-05,
+      "loss": 18.3023,
+      "step": 13960
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.439137334526323e-05,
+      "loss": 0.7885,
+      "step": 13961
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.43845890050406e-05,
+      "loss": 19.1129,
+      "step": 13962
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.437780530418447e-05,
+      "loss": 0.7897,
+      "step": 13963
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4371022242864154e-05,
+      "loss": 16.6469,
+      "step": 13964
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4364239821248996e-05,
+      "loss": 19.2063,
+      "step": 13965
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4357458039508263e-05,
+      "loss": 19.1376,
+      "step": 13966
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4350676897811215e-05,
+      "loss": 0.7796,
+      "step": 13967
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4343896396327165e-05,
+      "loss": 15.8304,
+      "step": 13968
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.433711653522528e-05,
+      "loss": 16.7063,
+      "step": 13969
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4330337314674852e-05,
+      "loss": 17.1202,
+      "step": 13970
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4323558734845053e-05,
+      "loss": 15.2355,
+      "step": 13971
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4316780795905082e-05,
+      "loss": 17.0461,
+      "step": 13972
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4310003498024142e-05,
+      "loss": 17.6701,
+      "step": 13973
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4303226841371378e-05,
+      "loss": 0.7894,
+      "step": 13974
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4296450826115946e-05,
+      "loss": 15.6859,
+      "step": 13975
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4289675452426962e-05,
+      "loss": 0.7826,
+      "step": 13976
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.428290072047354e-05,
+      "loss": 0.7405,
+      "step": 13977
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.42761266304248e-05,
+      "loss": 0.7718,
+      "step": 13978
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4269353182449796e-05,
+      "loss": 0.7814,
+      "step": 13979
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4262580376717624e-05,
+      "loss": 17.632,
+      "step": 13980
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.425580821339733e-05,
+      "loss": 0.7379,
+      "step": 13981
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.424903669265794e-05,
+      "loss": 0.7706,
+      "step": 13982
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4242265814668473e-05,
+      "loss": 0.7819,
+      "step": 13983
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4235495579597915e-05,
+      "loss": 16.3958,
+      "step": 13984
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4228725987615286e-05,
+      "loss": 0.8092,
+      "step": 13985
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4221957038889548e-05,
+      "loss": 15.277,
+      "step": 13986
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4215188733589623e-05,
+      "loss": 18.6105,
+      "step": 13987
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4208421071884513e-05,
+      "loss": 0.7203,
+      "step": 13988
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4201654053943063e-05,
+      "loss": 0.6847,
+      "step": 13989
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.419488767993424e-05,
+      "loss": 16.2305,
+      "step": 13990
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.418812195002691e-05,
+      "loss": 17.1633,
+      "step": 13991
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.418135686438993e-05,
+      "loss": 18.6881,
+      "step": 13992
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.41745924231922e-05,
+      "loss": 0.7633,
+      "step": 13993
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.416782862660254e-05,
+      "loss": 16.8905,
+      "step": 13994
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4161065474789772e-05,
+      "loss": 0.771,
+      "step": 13995
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4154302967922715e-05,
+      "loss": 0.7443,
+      "step": 13996
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4147541106170136e-05,
+      "loss": 17.0563,
+      "step": 13997
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4140779889700853e-05,
+      "loss": 18.1447,
+      "step": 13998
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.413401931868361e-05,
+      "loss": 0.7575,
+      "step": 13999
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4127259393287128e-05,
+      "loss": 18.4429,
+      "step": 14000
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4120500113680177e-05,
+      "loss": 17.835,
+      "step": 14001
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4113741480031455e-05,
+      "loss": 0.7402,
+      "step": 14002
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4106983492509653e-05,
+      "loss": 0.7339,
+      "step": 14003
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.410022615128346e-05,
+      "loss": 0.765,
+      "step": 14004
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4093469456521516e-05,
+      "loss": 0.7423,
+      "step": 14005
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4086713408392507e-05,
+      "loss": 0.7258,
+      "step": 14006
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.407995800706503e-05,
+      "loss": 0.781,
+      "step": 14007
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4073203252707736e-05,
+      "loss": 17.5729,
+      "step": 14008
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4066449145489233e-05,
+      "loss": 0.7621,
+      "step": 14009
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4059695685578038e-05,
+      "loss": 0.787,
+      "step": 14010
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4052942873142785e-05,
+      "loss": 18.3762,
+      "step": 14011
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.404619070835198e-05,
+      "loss": 0.8038,
+      "step": 14012
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.40394391913742e-05,
+      "loss": 18.2135,
+      "step": 14013
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4032688322377943e-05,
+      "loss": 0.7596,
+      "step": 14014
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4025938101531702e-05,
+      "loss": 0.7571,
+      "step": 14015
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.401918852900401e-05,
+      "loss": 0.7551,
+      "step": 14016
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4012439604963264e-05,
+      "loss": 18.4911,
+      "step": 14017
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4005691329577984e-05,
+      "loss": 0.7547,
+      "step": 14018
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3998943703016586e-05,
+      "loss": 0.7813,
+      "step": 14019
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3992196725447462e-05,
+      "loss": 0.721,
+      "step": 14020
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3985450397039073e-05,
+      "loss": 0.801,
+      "step": 14021
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3978704717959775e-05,
+      "loss": 17.7748,
+      "step": 14022
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3971959688377955e-05,
+      "loss": 0.7103,
+      "step": 14023
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3965215308461962e-05,
+      "loss": 0.7639,
+      "step": 14024
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.395847157838012e-05,
+      "loss": 0.777,
+      "step": 14025
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3951728498300792e-05,
+      "loss": 0.7386,
+      "step": 14026
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3944986068392266e-05,
+      "loss": 0.7676,
+      "step": 14027
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.393824428882282e-05,
+      "loss": 0.7248,
+      "step": 14028
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3931503159760772e-05,
+      "loss": 17.1893,
+      "step": 14029
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3924762681374348e-05,
+      "loss": 15.9434,
+      "step": 14030
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3918022853831807e-05,
+      "loss": 16.9048,
+      "step": 14031
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3911283677301368e-05,
+      "loss": 17.8124,
+      "step": 14032
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3904545151951223e-05,
+      "loss": 17.714,
+      "step": 14033
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.389780727794962e-05,
+      "loss": 0.7594,
+      "step": 14034
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3891070055464677e-05,
+      "loss": 0.7865,
+      "step": 14035
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.388433348466461e-05,
+      "loss": 18.2268,
+      "step": 14036
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3877597565717563e-05,
+      "loss": 16.8148,
+      "step": 14037
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3870862298791602e-05,
+      "loss": 16.6737,
+      "step": 14038
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.386412768405491e-05,
+      "loss": 0.7738,
+      "step": 14039
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3857393721675538e-05,
+      "loss": 18.1192,
+      "step": 14040
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.38506604118216e-05,
+      "loss": 16.7388,
+      "step": 14041
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3843927754661156e-05,
+      "loss": 0.7889,
+      "step": 14042
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3837195750362225e-05,
+      "loss": 16.8179,
+      "step": 14043
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.38304643990929e-05,
+      "loss": 17.5359,
+      "step": 14044
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3823733701021124e-05,
+      "loss": 0.7716,
+      "step": 14045
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3817003656314945e-05,
+      "loss": 0.7684,
+      "step": 14046
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3810274265142336e-05,
+      "loss": 0.7367,
+      "step": 14047
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3803545527671246e-05,
+      "loss": 0.7218,
+      "step": 14048
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3796817444069652e-05,
+      "loss": 0.8047,
+      "step": 14049
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3790090014505484e-05,
+      "loss": 16.4854,
+      "step": 14050
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3783363239146656e-05,
+      "loss": 18.282,
+      "step": 14051
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.377663711816106e-05,
+      "loss": 17.2688,
+      "step": 14052
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3769911651716576e-05,
+      "loss": 0.7855,
+      "step": 14053
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.37631868399811e-05,
+      "loss": 16.5458,
+      "step": 14054
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3756462683122478e-05,
+      "loss": 19.2601,
+      "step": 14055
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.374973918130851e-05,
+      "loss": 0.7855,
+      "step": 14056
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3743016334707075e-05,
+      "loss": 16.5355,
+      "step": 14057
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3736294143485943e-05,
+      "loss": 17.2245,
+      "step": 14058
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3729572607812907e-05,
+      "loss": 17.2023,
+      "step": 14059
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.372285172785574e-05,
+      "loss": 0.7275,
+      "step": 14060
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.371613150378217e-05,
+      "loss": 15.9464,
+      "step": 14061
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3709411935759978e-05,
+      "loss": 17.7512,
+      "step": 14062
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.370269302395685e-05,
+      "loss": 17.7337,
+      "step": 14063
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3695974768540546e-05,
+      "loss": 0.7587,
+      "step": 14064
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3689257169678703e-05,
+      "loss": 0.7697,
+      "step": 14065
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3682540227538984e-05,
+      "loss": 0.7188,
+      "step": 14066
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3675823942289087e-05,
+      "loss": 19.4957,
+      "step": 14067
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.366910831409662e-05,
+      "loss": 0.7308,
+      "step": 14068
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3662393343129242e-05,
+      "loss": 17.6216,
+      "step": 14069
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.365567902955454e-05,
+      "loss": 0.8056,
+      "step": 14070
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.36489653735401e-05,
+      "loss": 17.0018,
+      "step": 14071
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3642252375253503e-05,
+      "loss": 15.6061,
+      "step": 14072
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.363554003486229e-05,
+      "loss": 19.4571,
+      "step": 14073
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.362882835253403e-05,
+      "loss": 0.7777,
+      "step": 14074
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3622117328436238e-05,
+      "loss": 0.7541,
+      "step": 14075
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3615406962736402e-05,
+      "loss": 0.7463,
+      "step": 14076
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3608697255602053e-05,
+      "loss": 0.7563,
+      "step": 14077
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.360198820720065e-05,
+      "loss": 19.1392,
+      "step": 14078
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.359527981769964e-05,
+      "loss": 0.7717,
+      "step": 14079
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.358857208726648e-05,
+      "loss": 0.7317,
+      "step": 14080
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.358186501606857e-05,
+      "loss": 0.7169,
+      "step": 14081
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.357515860427336e-05,
+      "loss": 0.7238,
+      "step": 14082
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3568452852048228e-05,
+      "loss": 0.7447,
+      "step": 14083
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3561747759560526e-05,
+      "loss": 0.7194,
+      "step": 14084
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3555043326977677e-05,
+      "loss": 16.203,
+      "step": 14085
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.354833955446694e-05,
+      "loss": 17.7479,
+      "step": 14086
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3541636442195712e-05,
+      "loss": 0.7655,
+      "step": 14087
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3534933990331282e-05,
+      "loss": 18.256,
+      "step": 14088
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3528232199040922e-05,
+      "loss": 0.8216,
+      "step": 14089
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3521531068491948e-05,
+      "loss": 17.3142,
+      "step": 14090
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3514830598851585e-05,
+      "loss": 0.683,
+      "step": 14091
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3508130790287135e-05,
+      "loss": 16.7422,
+      "step": 14092
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.350143164296577e-05,
+      "loss": 0.7584,
+      "step": 14093
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3494733157054705e-05,
+      "loss": 0.7495,
+      "step": 14094
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.348803533272117e-05,
+      "loss": 0.7309,
+      "step": 14095
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3481338170132313e-05,
+      "loss": 0.7474,
+      "step": 14096
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.347464166945532e-05,
+      "loss": 16.133,
+      "step": 14097
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.346794583085733e-05,
+      "loss": 0.7161,
+      "step": 14098
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3461250654505474e-05,
+      "loss": 0.7443,
+      "step": 14099
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.345455614056686e-05,
+      "loss": 18.3627,
+      "step": 14100
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3447862289208567e-05,
+      "loss": 0.7753,
+      "step": 14101
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.344116910059771e-05,
+      "loss": 15.5216,
+      "step": 14102
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3434476574901338e-05,
+      "loss": 0.78,
+      "step": 14103
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3427784712286478e-05,
+      "loss": 17.5534,
+      "step": 14104
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3421093512920194e-05,
+      "loss": 0.6838,
+      "step": 14105
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.341440297696949e-05,
+      "loss": 16.1055,
+      "step": 14106
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3407713104601357e-05,
+      "loss": 17.4347,
+      "step": 14107
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3401023895982776e-05,
+      "loss": 19.1133,
+      "step": 14108
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3394335351280694e-05,
+      "loss": 18.7761,
+      "step": 14109
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3387647470662095e-05,
+      "loss": 0.7915,
+      "step": 14110
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.338096025429389e-05,
+      "loss": 18.2355,
+      "step": 14111
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3374273702342976e-05,
+      "loss": 18.7452,
+      "step": 14112
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3367587814976305e-05,
+      "loss": 0.7931,
+      "step": 14113
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3360902592360688e-05,
+      "loss": 0.7475,
+      "step": 14114
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3354218034663045e-05,
+      "loss": 0.7694,
+      "step": 14115
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3347534142050203e-05,
+      "loss": 17.5882,
+      "step": 14116
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3340850914688977e-05,
+      "loss": 0.7968,
+      "step": 14117
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3334168352746217e-05,
+      "loss": 0.7607,
+      "step": 14118
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3327486456388685e-05,
+      "loss": 15.6698,
+      "step": 14119
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.332080522578322e-05,
+      "loss": 18.3397,
+      "step": 14120
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3314124661096527e-05,
+      "loss": 0.7355,
+      "step": 14121
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3307444762495356e-05,
+      "loss": 0.7431,
+      "step": 14122
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.330076553014648e-05,
+      "loss": 0.8338,
+      "step": 14123
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.329408696421657e-05,
+      "loss": 16.4854,
+      "step": 14124
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3287409064872374e-05,
+      "loss": 17.5868,
+      "step": 14125
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.328073183228054e-05,
+      "loss": 18.9426,
+      "step": 14126
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3274055266607742e-05,
+      "loss": 0.7636,
+      "step": 14127
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3267379368020626e-05,
+      "loss": 16.0722,
+      "step": 14128
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.326070413668581e-05,
+      "loss": 18.7262,
+      "step": 14129
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3254029572769937e-05,
+      "loss": 15.6045,
+      "step": 14130
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.324735567643959e-05,
+      "loss": 15.2784,
+      "step": 14131
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3240682447861346e-05,
+      "loss": 18.2398,
+      "step": 14132
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.323400988720179e-05,
+      "loss": 0.7308,
+      "step": 14133
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3227337994627457e-05,
+      "loss": 0.7214,
+      "step": 14134
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3220666770304884e-05,
+      "loss": 0.7765,
+      "step": 14135
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3213996214400584e-05,
+      "loss": 0.7337,
+      "step": 14136
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.320732632708103e-05,
+      "loss": 18.5185,
+      "step": 14137
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3200657108512753e-05,
+      "loss": 0.7282,
+      "step": 14138
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3193988558862194e-05,
+      "loss": 16.8734,
+      "step": 14139
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3187320678295797e-05,
+      "loss": 17.549,
+      "step": 14140
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3180653466980002e-05,
+      "loss": 17.0249,
+      "step": 14141
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.31739869250812e-05,
+      "loss": 0.7901,
+      "step": 14142
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3167321052765828e-05,
+      "loss": 0.7945,
+      "step": 14143
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3160655850200248e-05,
+      "loss": 0.7517,
+      "step": 14144
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.315399131755081e-05,
+      "loss": 0.7395,
+      "step": 14145
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.314732745498389e-05,
+      "loss": 0.7728,
+      "step": 14146
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3140664262665817e-05,
+      "loss": 17.2587,
+      "step": 14147
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3134001740762895e-05,
+      "loss": 0.8007,
+      "step": 14148
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3127339889441425e-05,
+      "loss": 0.7541,
+      "step": 14149
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.312067870886766e-05,
+      "loss": 0.7708,
+      "step": 14150
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.311401819920792e-05,
+      "loss": 17.7671,
+      "step": 14151
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.31073583606284e-05,
+      "loss": 0.7453,
+      "step": 14152
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3100699193295376e-05,
+      "loss": 0.7509,
+      "step": 14153
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.309404069737504e-05,
+      "loss": 17.1347,
+      "step": 14154
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.30873828730336e-05,
+      "loss": 0.7865,
+      "step": 14155
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3080725720437218e-05,
+      "loss": 0.7255,
+      "step": 14156
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3074069239752054e-05,
+      "loss": 18.2127,
+      "step": 14157
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.306741343114429e-05,
+      "loss": 0.7503,
+      "step": 14158
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.306075829478004e-05,
+      "loss": 15.8427,
+      "step": 14159
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3054103830825387e-05,
+      "loss": 17.4142,
+      "step": 14160
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3047450039446493e-05,
+      "loss": 0.7588,
+      "step": 14161
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3040796920809372e-05,
+      "loss": 0.7144,
+      "step": 14162
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3034144475080127e-05,
+      "loss": 0.7602,
+      "step": 14163
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.302749270242479e-05,
+      "loss": 18.4483,
+      "step": 14164
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.302084160300938e-05,
+      "loss": 19.5702,
+      "step": 14165
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3014191176999938e-05,
+      "loss": 20.3629,
+      "step": 14166
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3007541424562453e-05,
+      "loss": 18.4014,
+      "step": 14167
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.300089234586289e-05,
+      "loss": 18.7594,
+      "step": 14168
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2994243941067223e-05,
+      "loss": 16.8925,
+      "step": 14169
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2987596210341377e-05,
+      "loss": 0.6808,
+      "step": 14170
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.298094915385131e-05,
+      "loss": 0.799,
+      "step": 14171
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2974302771762923e-05,
+      "loss": 0.7649,
+      "step": 14172
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2967657064242092e-05,
+      "loss": 0.7594,
+      "step": 14173
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2961012031454733e-05,
+      "loss": 0.7133,
+      "step": 14174
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2954367673566686e-05,
+      "loss": 0.7751,
+      "step": 14175
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2947723990743796e-05,
+      "loss": 18.2567,
+      "step": 14176
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2941080983151896e-05,
+      "loss": 0.7524,
+      "step": 14177
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.293443865095677e-05,
+      "loss": 15.8024,
+      "step": 14178
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.292779699432426e-05,
+      "loss": 0.7521,
+      "step": 14179
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.292115601342009e-05,
+      "loss": 18.7814,
+      "step": 14180
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2914515708410073e-05,
+      "loss": 0.7487,
+      "step": 14181
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.290787607945993e-05,
+      "loss": 17.5062,
+      "step": 14182
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2901237126735385e-05,
+      "loss": 17.5736,
+      "step": 14183
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.289459885040215e-05,
+      "loss": 0.762,
+      "step": 14184
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2887961250625894e-05,
+      "loss": 16.3348,
+      "step": 14185
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2881324327572333e-05,
+      "loss": 0.7478,
+      "step": 14186
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2874688081407114e-05,
+      "loss": 18.4374,
+      "step": 14187
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.286805251229585e-05,
+      "loss": 18.5869,
+      "step": 14188
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2861417620404235e-05,
+      "loss": 0.7397,
+      "step": 14189
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.285478340589779e-05,
+      "loss": 20.6858,
+      "step": 14190
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.284814986894217e-05,
+      "loss": 17.9,
+      "step": 14191
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.284151700970293e-05,
+      "loss": 0.7867,
+      "step": 14192
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2834884828345605e-05,
+      "loss": 16.121,
+      "step": 14193
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2828253325035776e-05,
+      "loss": 0.7549,
+      "step": 14194
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.282162249993895e-05,
+      "loss": 0.754,
+      "step": 14195
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2814992353220632e-05,
+      "loss": 19.4419,
+      "step": 14196
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2808362885046308e-05,
+      "loss": 0.739,
+      "step": 14197
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.280173409558144e-05,
+      "loss": 0.804,
+      "step": 14198
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2795105984991516e-05,
+      "loss": 17.3341,
+      "step": 14199
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.278847855344196e-05,
+      "loss": 0.7299,
+      "step": 14200
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2781851801098177e-05,
+      "loss": 0.8151,
+      "step": 14201
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2775225728125598e-05,
+      "loss": 0.7981,
+      "step": 14202
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.276860033468961e-05,
+      "loss": 17.1675,
+      "step": 14203
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2761975620955566e-05,
+      "loss": 0.7686,
+      "step": 14204
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2755351587088834e-05,
+      "loss": 0.7771,
+      "step": 14205
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2748728233254735e-05,
+      "loss": 16.8112,
+      "step": 14206
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2742105559618616e-05,
+      "loss": 17.0376,
+      "step": 14207
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2735483566345744e-05,
+      "loss": 18.5585,
+      "step": 14208
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2728862253601445e-05,
+      "loss": 0.7411,
+      "step": 14209
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2722241621550977e-05,
+      "loss": 17.6082,
+      "step": 14210
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.271562167035958e-05,
+      "loss": 0.7572,
+      "step": 14211
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2709002400192504e-05,
+      "loss": 17.0088,
+      "step": 14212
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2702383811214934e-05,
+      "loss": 17.7704,
+      "step": 14213
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.269576590359212e-05,
+      "loss": 16.2713,
+      "step": 14214
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.268914867748922e-05,
+      "loss": 18.0591,
+      "step": 14215
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2682532133071406e-05,
+      "loss": 0.7491,
+      "step": 14216
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.267591627050383e-05,
+      "loss": 15.7453,
+      "step": 14217
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.26693010899516e-05,
+      "loss": 0.7093,
+      "step": 14218
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2662686591579873e-05,
+      "loss": 17.0016,
+      "step": 14219
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.265607277555374e-05,
+      "loss": 0.7454,
+      "step": 14220
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2649459642038247e-05,
+      "loss": 0.7653,
+      "step": 14221
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2642847191198502e-05,
+      "loss": 16.9968,
+      "step": 14222
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2636235423199542e-05,
+      "loss": 18.2575,
+      "step": 14223
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2629624338206395e-05,
+      "loss": 18.5687,
+      "step": 14224
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2623013936384073e-05,
+      "loss": 16.0589,
+      "step": 14225
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2616404217897552e-05,
+      "loss": 16.9307,
+      "step": 14226
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2609795182911857e-05,
+      "loss": 0.7776,
+      "step": 14227
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.260318683159193e-05,
+      "loss": 0.7274,
+      "step": 14228
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2596579164102695e-05,
+      "loss": 0.7848,
+      "step": 14229
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2589972180609138e-05,
+      "loss": 0.7714,
+      "step": 14230
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2583365881276093e-05,
+      "loss": 16.4512,
+      "step": 14231
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2576760266268515e-05,
+      "loss": 18.8326,
+      "step": 14232
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2570155335751263e-05,
+      "loss": 17.4123,
+      "step": 14233
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2563551089889184e-05,
+      "loss": 18.6333,
+      "step": 14234
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.255694752884715e-05,
+      "loss": 17.4595,
+      "step": 14235
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.255034465278995e-05,
+      "loss": 20.6826,
+      "step": 14236
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2543742461882455e-05,
+      "loss": 0.7061,
+      "step": 14237
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2537140956289377e-05,
+      "loss": 17.0197,
+      "step": 14238
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2530540136175556e-05,
+      "loss": 0.7253,
+      "step": 14239
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2523940001705717e-05,
+      "loss": 18.7686,
+      "step": 14240
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2517340553044598e-05,
+      "loss": 19.0459,
+      "step": 14241
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.251074179035695e-05,
+      "loss": 17.3496,
+      "step": 14242
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2504143713807456e-05,
+      "loss": 0.7482,
+      "step": 14243
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2497546323560825e-05,
+      "loss": 16.9657,
+      "step": 14244
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2490949619781708e-05,
+      "loss": 16.6849,
+      "step": 14245
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2484353602634756e-05,
+      "loss": 0.7276,
+      "step": 14246
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2477758272284638e-05,
+      "loss": 17.8572,
+      "step": 14247
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2471163628895957e-05,
+      "loss": 0.7299,
+      "step": 14248
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2464569672633302e-05,
+      "loss": 18.9652,
+      "step": 14249
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.245797640366129e-05,
+      "loss": 0.7562,
+      "step": 14250
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.245138382214448e-05,
+      "loss": 17.3895,
+      "step": 14251
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2444791928247416e-05,
+      "loss": 17.2058,
+      "step": 14252
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.243820072213464e-05,
+      "loss": 19.1428,
+      "step": 14253
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2431610203970648e-05,
+      "loss": 0.7959,
+      "step": 14254
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2425020373919975e-05,
+      "loss": 17.7934,
+      "step": 14255
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2418431232147093e-05,
+      "loss": 0.7227,
+      "step": 14256
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2411842778816443e-05,
+      "loss": 0.7198,
+      "step": 14257
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2405255014092542e-05,
+      "loss": 17.4407,
+      "step": 14258
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2398667938139732e-05,
+      "loss": 0.7323,
+      "step": 14259
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.239208155112249e-05,
+      "loss": 16.0335,
+      "step": 14260
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2385495853205203e-05,
+      "loss": 17.6107,
+      "step": 14261
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.237891084455222e-05,
+      "loss": 0.7087,
+      "step": 14262
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2372326525327954e-05,
+      "loss": 0.7541,
+      "step": 14263
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.23657428956967e-05,
+      "loss": 0.7317,
+      "step": 14264
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2359159955822856e-05,
+      "loss": 17.818,
+      "step": 14265
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2352577705870648e-05,
+      "loss": 17.9497,
+      "step": 14266
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2345996146004433e-05,
+      "loss": 17.3922,
+      "step": 14267
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2339415276388476e-05,
+      "loss": 0.7683,
+      "step": 14268
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2332835097187005e-05,
+      "loss": 19.1226,
+      "step": 14269
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2326255608564306e-05,
+      "loss": 0.794,
+      "step": 14270
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.231967681068458e-05,
+      "loss": 19.1995,
+      "step": 14271
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2313098703712048e-05,
+      "loss": 18.0558,
+      "step": 14272
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2306521287810894e-05,
+      "loss": 19.2024,
+      "step": 14273
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2299944563145274e-05,
+      "loss": 0.741,
+      "step": 14274
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2293368529879378e-05,
+      "loss": 0.763,
+      "step": 14275
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.228679318817733e-05,
+      "loss": 0.7784,
+      "step": 14276
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2280218538203234e-05,
+      "loss": 0.7882,
+      "step": 14277
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2273644580121234e-05,
+      "loss": 17.8247,
+      "step": 14278
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.226707131409539e-05,
+      "loss": 17.7469,
+      "step": 14279
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.226049874028978e-05,
+      "loss": 17.6428,
+      "step": 14280
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2253926858868462e-05,
+      "loss": 17.6807,
+      "step": 14281
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.224735566999544e-05,
+      "loss": 18.1042,
+      "step": 14282
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2240785173834777e-05,
+      "loss": 16.7916,
+      "step": 14283
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2234215370550456e-05,
+      "loss": 0.7518,
+      "step": 14284
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2227646260306435e-05,
+      "loss": 0.7278,
+      "step": 14285
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2221077843266748e-05,
+      "loss": 17.8985,
+      "step": 14286
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.221451011959526e-05,
+      "loss": 15.224,
+      "step": 14287
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.220794308945596e-05,
+      "loss": 17.4812,
+      "step": 14288
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2201376753012732e-05,
+      "loss": 0.7018,
+      "step": 14289
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2194811110429508e-05,
+      "loss": 0.7714,
+      "step": 14290
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2188246161870142e-05,
+      "loss": 0.7856,
+      "step": 14291
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2181681907498502e-05,
+      "loss": 16.8674,
+      "step": 14292
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.217511834747844e-05,
+      "loss": 0.7706,
+      "step": 14293
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.216855548197375e-05,
+      "loss": 17.421,
+      "step": 14294
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.21619933111483e-05,
+      "loss": 16.6884,
+      "step": 14295
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2155431835165842e-05,
+      "loss": 18.3025,
+      "step": 14296
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.214887105419015e-05,
+      "loss": 16.5943,
+      "step": 14297
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.214231096838501e-05,
+      "loss": 0.7636,
+      "step": 14298
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2135751577914156e-05,
+      "loss": 0.79,
+      "step": 14299
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2129192882941303e-05,
+      "loss": 16.4734,
+      "step": 14300
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2122634883630155e-05,
+      "loss": 0.7697,
+      "step": 14301
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.211607758014439e-05,
+      "loss": 16.3634,
+      "step": 14302
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2109520972647713e-05,
+      "loss": 16.9441,
+      "step": 14303
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.210296506130376e-05,
+      "loss": 0.7387,
+      "step": 14304
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.209640984627615e-05,
+      "loss": 0.7743,
+      "step": 14305
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.208985532772856e-05,
+      "loss": 0.7972,
+      "step": 14306
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2083301505824517e-05,
+      "loss": 0.7843,
+      "step": 14307
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.207674838072766e-05,
+      "loss": 0.7771,
+      "step": 14308
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.207019595260154e-05,
+      "loss": 0.7299,
+      "step": 14309
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2063644221609688e-05,
+      "loss": 0.7677,
+      "step": 14310
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2057093187915672e-05,
+      "loss": 16.9756,
+      "step": 14311
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2050542851682992e-05,
+      "loss": 0.7509,
+      "step": 14312
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.204399321307514e-05,
+      "loss": 17.6865,
+      "step": 14313
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2037444272255604e-05,
+      "loss": 0.7406,
+      "step": 14314
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2030896029387827e-05,
+      "loss": 19.896,
+      "step": 14315
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2024348484635288e-05,
+      "loss": 0.7678,
+      "step": 14316
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.201780163816138e-05,
+      "loss": 17.4527,
+      "step": 14317
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.2011255490129544e-05,
+      "loss": 16.8148,
+      "step": 14318
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.200471004070317e-05,
+      "loss": 0.8062,
+      "step": 14319
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1998165290045626e-05,
+      "loss": 0.7919,
+      "step": 14320
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1991621238320264e-05,
+      "loss": 17.3415,
+      "step": 14321
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.198507788569042e-05,
+      "loss": 18.807,
+      "step": 14322
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1978535232319442e-05,
+      "loss": 18.0578,
+      "step": 14323
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1971993278370622e-05,
+      "loss": 0.7458,
+      "step": 14324
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.196545202400723e-05,
+      "loss": 17.7616,
+      "step": 14325
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1958911469392585e-05,
+      "loss": 18.8935,
+      "step": 14326
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1952371614689903e-05,
+      "loss": 18.5324,
+      "step": 14327
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1945832460062432e-05,
+      "loss": 18.7976,
+      "step": 14328
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1939294005673388e-05,
+      "loss": 0.7333,
+      "step": 14329
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1932756251685954e-05,
+      "loss": 17.0624,
+      "step": 14330
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1926219198263354e-05,
+      "loss": 0.7651,
+      "step": 14331
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.191968284556873e-05,
+      "loss": 0.777,
+      "step": 14332
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1913147193765215e-05,
+      "loss": 18.5863,
+      "step": 14333
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1906612243015996e-05,
+      "loss": 16.8851,
+      "step": 14334
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1900077993484115e-05,
+      "loss": 18.2661,
+      "step": 14335
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.189354444533272e-05,
+      "loss": 0.7491,
+      "step": 14336
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1887011598724872e-05,
+      "loss": 0.7306,
+      "step": 14337
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1880479453823615e-05,
+      "loss": 19.1124,
+      "step": 14338
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1873948010792032e-05,
+      "loss": 17.1503,
+      "step": 14339
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.186741726979313e-05,
+      "loss": 17.9702,
+      "step": 14340
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1860887230989914e-05,
+      "loss": 16.3869,
+      "step": 14341
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.185435789454538e-05,
+      "loss": 0.7986,
+      "step": 14342
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1847829260622477e-05,
+      "loss": 17.5101,
+      "step": 14343
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1841301329384204e-05,
+      "loss": 20.1289,
+      "step": 14344
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1834774100993465e-05,
+      "loss": 18.1518,
+      "step": 14345
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.182824757561321e-05,
+      "loss": 16.8879,
+      "step": 14346
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1821721753406325e-05,
+      "loss": 17.4301,
+      "step": 14347
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.18151966345357e-05,
+      "loss": 17.9192,
+      "step": 14348
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1808672219164205e-05,
+      "loss": 18.1761,
+      "step": 14349
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.180214850745467e-05,
+      "loss": 15.7804,
+      "step": 14350
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1795625499569966e-05,
+      "loss": 0.7881,
+      "step": 14351
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1789103195672887e-05,
+      "loss": 0.742,
+      "step": 14352
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1782581595926217e-05,
+      "loss": 0.8052,
+      "step": 14353
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.177606070049277e-05,
+      "loss": 0.759,
+      "step": 14354
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1769540509535297e-05,
+      "loss": 0.7251,
+      "step": 14355
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1763021023216535e-05,
+      "loss": 17.8281,
+      "step": 14356
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.175650224169921e-05,
+      "loss": 18.3602,
+      "step": 14357
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1749984165146027e-05,
+      "loss": 15.717,
+      "step": 14358
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1743466793719707e-05,
+      "loss": 0.7738,
+      "step": 14359
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1736950127582907e-05,
+      "loss": 0.8499,
+      "step": 14360
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1730434166898268e-05,
+      "loss": 0.7604,
+      "step": 14361
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1723918911828477e-05,
+      "loss": 17.9002,
+      "step": 14362
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.17174043625361e-05,
+      "loss": 17.916,
+      "step": 14363
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1710890519183775e-05,
+      "loss": 0.7297,
+      "step": 14364
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1704377381934087e-05,
+      "loss": 18.5933,
+      "step": 14365
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1697864950949588e-05,
+      "loss": 0.7538,
+      "step": 14366
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.169135322639285e-05,
+      "loss": 18.4928,
+      "step": 14367
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1684842208426403e-05,
+      "loss": 16.6152,
+      "step": 14368
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.167833189721276e-05,
+      "loss": 0.7065,
+      "step": 14369
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1671822292914423e-05,
+      "loss": 17.8605,
+      "step": 14370
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1665313395693847e-05,
+      "loss": 16.894,
+      "step": 14371
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.165880520571354e-05,
+      "loss": 20.1172,
+      "step": 14372
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1652297723135904e-05,
+      "loss": 0.7737,
+      "step": 14373
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1645790948123414e-05,
+      "loss": 16.3046,
+      "step": 14374
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1639284880838467e-05,
+      "loss": 0.7518,
+      "step": 14375
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1632779521443413e-05,
+      "loss": 15.8842,
+      "step": 14376
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1626274870100675e-05,
+      "loss": 0.7817,
+      "step": 14377
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1619770926972577e-05,
+      "loss": 18.2198,
+      "step": 14378
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1613267692221496e-05,
+      "loss": 17.6403,
+      "step": 14379
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1606765166009735e-05,
+      "loss": 17.415,
+      "step": 14380
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1600263348499576e-05,
+      "loss": 0.7244,
+      "step": 14381
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.159376223985337e-05,
+      "loss": 0.7442,
+      "step": 14382
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1587261840233308e-05,
+      "loss": 17.4238,
+      "step": 14383
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.158076214980169e-05,
+      "loss": 14.8836,
+      "step": 14384
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.157426316872075e-05,
+      "loss": 0.755,
+      "step": 14385
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1567764897152665e-05,
+      "loss": 18.4685,
+      "step": 14386
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.156126733525968e-05,
+      "loss": 18.3184,
+      "step": 14387
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.155477048320395e-05,
+      "loss": 18.1424,
+      "step": 14388
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.154827434114765e-05,
+      "loss": 0.7912,
+      "step": 14389
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.154177890925292e-05,
+      "loss": 17.8498,
+      "step": 14390
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1535284187681864e-05,
+      "loss": 0.7733,
+      "step": 14391
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.152879017659663e-05,
+      "loss": 18.0647,
+      "step": 14392
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1522296876159298e-05,
+      "loss": 17.8923,
+      "step": 14393
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.151580428653192e-05,
+      "loss": 19.2499,
+      "step": 14394
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1509312407876593e-05,
+      "loss": 18.3662,
+      "step": 14395
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1502821240355324e-05,
+      "loss": 0.746,
+      "step": 14396
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1496330784130154e-05,
+      "loss": 16.3127,
+      "step": 14397
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.148984103936307e-05,
+      "loss": 0.7673,
+      "step": 14398
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1483352006216047e-05,
+      "loss": 0.7204,
+      "step": 14399
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1476863684851084e-05,
+      "loss": 0.7641,
+      "step": 14400
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1470376075430098e-05,
+      "loss": 16.6963,
+      "step": 14401
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.146388917811506e-05,
+      "loss": 17.0758,
+      "step": 14402
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1457402993067883e-05,
+      "loss": 15.1195,
+      "step": 14403
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1450917520450407e-05,
+      "loss": 0.7436,
+      "step": 14404
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1444432760424565e-05,
+      "loss": 18.4109,
+      "step": 14405
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1437948713152185e-05,
+      "loss": 0.7867,
+      "step": 14406
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1431465378795147e-05,
+      "loss": 16.1852,
+      "step": 14407
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.142498275751525e-05,
+      "loss": 17.5829,
+      "step": 14408
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1418500849474294e-05,
+      "loss": 0.7599,
+      "step": 14409
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.141201965483412e-05,
+      "loss": 17.4243,
+      "step": 14410
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1405539173756422e-05,
+      "loss": 0.7342,
+      "step": 14411
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1399059406403014e-05,
+      "loss": 0.7492,
+      "step": 14412
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.139258035293562e-05,
+      "loss": 17.5792,
+      "step": 14413
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.138610201351593e-05,
+      "loss": 17.4931,
+      "step": 14414
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1379624388305687e-05,
+      "loss": 19.3544,
+      "step": 14415
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.137314747746655e-05,
+      "loss": 17.0229,
+      "step": 14416
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.136667128116019e-05,
+      "loss": 0.7561,
+      "step": 14417
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1360195799548255e-05,
+      "loss": 16.5887,
+      "step": 14418
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.135372103279235e-05,
+      "loss": 0.7382,
+      "step": 14419
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.134724698105413e-05,
+      "loss": 20.0532,
+      "step": 14420
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1340773644495166e-05,
+      "loss": 19.8018,
+      "step": 14421
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.133430102327702e-05,
+      "loss": 15.3111,
+      "step": 14422
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1327829117561284e-05,
+      "loss": 18.5868,
+      "step": 14423
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1321357927509478e-05,
+      "loss": 16.9677,
+      "step": 14424
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.131488745328313e-05,
+      "loss": 0.7623,
+      "step": 14425
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.130841769504373e-05,
+      "loss": 0.7755,
+      "step": 14426
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.130194865295277e-05,
+      "loss": 17.1441,
+      "step": 14427
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1295480327171736e-05,
+      "loss": 0.748,
+      "step": 14428
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.128901271786205e-05,
+      "loss": 18.3736,
+      "step": 14429
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.128254582518518e-05,
+      "loss": 0.7828,
+      "step": 14430
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.127607964930254e-05,
+      "loss": 16.2404,
+      "step": 14431
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1269614190375475e-05,
+      "loss": 17.5451,
+      "step": 14432
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1263149448565424e-05,
+      "loss": 18.4335,
+      "step": 14433
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1256685424033694e-05,
+      "loss": 16.879,
+      "step": 14434
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1250222116941686e-05,
+      "loss": 0.8241,
+      "step": 14435
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1243759527450695e-05,
+      "loss": 0.7503,
+      "step": 14436
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1237297655722017e-05,
+      "loss": 18.1381,
+      "step": 14437
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.123083650191699e-05,
+      "loss": 17.2647,
+      "step": 14438
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.122437606619682e-05,
+      "loss": 0.7364,
+      "step": 14439
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.121791634872281e-05,
+      "loss": 0.7379,
+      "step": 14440
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1211457349656177e-05,
+      "loss": 0.7947,
+      "step": 14441
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1204999069158128e-05,
+      "loss": 16.2428,
+      "step": 14442
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1198541507389892e-05,
+      "loss": 15.3278,
+      "step": 14443
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.119208466451264e-05,
+      "loss": 18.732,
+      "step": 14444
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.118562854068753e-05,
+      "loss": 20.5333,
+      "step": 14445
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.117917313607571e-05,
+      "loss": 21.5856,
+      "step": 14446
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.117271845083829e-05,
+      "loss": 15.7939,
+      "step": 14447
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1166264485136422e-05,
+      "loss": 16.907,
+      "step": 14448
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.115981123913117e-05,
+      "loss": 0.7519,
+      "step": 14449
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.11533587129836e-05,
+      "loss": 18.8853,
+      "step": 14450
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1146906906854817e-05,
+      "loss": 0.7693,
+      "step": 14451
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1140455820905798e-05,
+      "loss": 19.036,
+      "step": 14452
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1134005455297602e-05,
+      "loss": 17.2019,
+      "step": 14453
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.112755581019122e-05,
+      "loss": 0.705,
+      "step": 14454
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.112110688574762e-05,
+      "loss": 19.2313,
+      "step": 14455
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1114658682127807e-05,
+      "loss": 17.9432,
+      "step": 14456
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.110821119949269e-05,
+      "loss": 17.0009,
+      "step": 14457
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.110176443800325e-05,
+      "loss": 18.2276,
+      "step": 14458
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1095318397820357e-05,
+      "loss": 18.1176,
+      "step": 14459
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.108887307910489e-05,
+      "loss": 0.7402,
+      "step": 14460
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.108242848201778e-05,
+      "loss": 0.7901,
+      "step": 14461
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1075984606719833e-05,
+      "loss": 0.7021,
+      "step": 14462
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1069541453371933e-05,
+      "loss": 17.2835,
+      "step": 14463
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1063099022134884e-05,
+      "loss": 0.731,
+      "step": 14464
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1056657313169494e-05,
+      "loss": 17.1917,
+      "step": 14465
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1050216326636546e-05,
+      "loss": 17.7353,
+      "step": 14466
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1043776062696783e-05,
+      "loss": 18.3825,
+      "step": 14467
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1037336521511007e-05,
+      "loss": 0.724,
+      "step": 14468
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.103089770323992e-05,
+      "loss": 0.7804,
+      "step": 14469
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1024459608044227e-05,
+      "loss": 18.7721,
+      "step": 14470
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1018022236084655e-05,
+      "loss": 18.9744,
+      "step": 14471
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1011585587521866e-05,
+      "loss": 16.4884,
+      "step": 14472
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.1005149662516517e-05,
+      "loss": 0.7919,
+      "step": 14473
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.099871446122925e-05,
+      "loss": 0.786,
+      "step": 14474
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.099227998382068e-05,
+      "loss": 16.5774,
+      "step": 14475
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0985846230451438e-05,
+      "loss": 0.7665,
+      "step": 14476
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0979413201282106e-05,
+      "loss": 16.6686,
+      "step": 14477
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.097298089647322e-05,
+      "loss": 0.7788,
+      "step": 14478
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0966549316185402e-05,
+      "loss": 18.0473,
+      "step": 14479
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0960118460579104e-05,
+      "loss": 0.7605,
+      "step": 14480
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0953688329814897e-05,
+      "loss": 0.7851,
+      "step": 14481
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.094725892405326e-05,
+      "loss": 0.8168,
+      "step": 14482
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0940830243454657e-05,
+      "loss": 0.7794,
+      "step": 14483
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.093440228817958e-05,
+      "loss": 20.1574,
+      "step": 14484
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.092797505838845e-05,
+      "loss": 16.5652,
+      "step": 14485
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0921548554241732e-05,
+      "loss": 0.7459,
+      "step": 14486
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.091512277589978e-05,
+      "loss": 18.0619,
+      "step": 14487
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.090869772352299e-05,
+      "loss": 0.7229,
+      "step": 14488
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0902273397271765e-05,
+      "loss": 0.7826,
+      "step": 14489
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0895849797306423e-05,
+      "loss": 0.7466,
+      "step": 14490
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.088942692378733e-05,
+      "loss": 0.7679,
+      "step": 14491
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0883004776874788e-05,
+      "loss": 0.8204,
+      "step": 14492
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.08765833567291e-05,
+      "loss": 0.7912,
+      "step": 14493
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0870162663510535e-05,
+      "loss": 15.3522,
+      "step": 14494
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0863742697379345e-05,
+      "loss": 0.7476,
+      "step": 14495
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0857323458495808e-05,
+      "loss": 17.0317,
+      "step": 14496
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0850904947020133e-05,
+      "loss": 18.8501,
+      "step": 14497
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.084448716311251e-05,
+      "loss": 17.6969,
+      "step": 14498
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.083807010693316e-05,
+      "loss": 16.5656,
+      "step": 14499
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.083165377864224e-05,
+      "loss": 0.7499,
+      "step": 14500
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0825238178399903e-05,
+      "loss": 18.0409,
+      "step": 14501
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0818823306366275e-05,
+      "loss": 17.723,
+      "step": 14502
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.081240916270147e-05,
+      "loss": 0.7333,
+      "step": 14503
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.080599574756561e-05,
+      "loss": 16.5686,
+      "step": 14504
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0799583061118767e-05,
+      "loss": 0.8129,
+      "step": 14505
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0793171103520975e-05,
+      "loss": 16.8375,
+      "step": 14506
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.078675987493234e-05,
+      "loss": 0.7798,
+      "step": 14507
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0780349375512815e-05,
+      "loss": 16.3438,
+      "step": 14508
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0773939605422455e-05,
+      "loss": 18.3593,
+      "step": 14509
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0767530564821237e-05,
+      "loss": 0.7435,
+      "step": 14510
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0761122253869107e-05,
+      "loss": 18.9078,
+      "step": 14511
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0754714672726067e-05,
+      "loss": 16.1574,
+      "step": 14512
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.074830782155201e-05,
+      "loss": 17.8331,
+      "step": 14513
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0741901700506898e-05,
+      "loss": 0.7343,
+      "step": 14514
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0735496309750585e-05,
+      "loss": 0.7526,
+      "step": 14515
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.072909164944295e-05,
+      "loss": 0.6494,
+      "step": 14516
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0722687719743887e-05,
+      "loss": 18.189,
+      "step": 14517
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0716284520813207e-05,
+      "loss": 0.7591,
+      "step": 14518
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.070988205281077e-05,
+      "loss": 0.7445,
+      "step": 14519
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.070348031589636e-05,
+      "loss": 0.7792,
+      "step": 14520
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0697079310229772e-05,
+      "loss": 16.6512,
+      "step": 14521
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.069067903597078e-05,
+      "loss": 17.5568,
+      "step": 14522
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0684279493279107e-05,
+      "loss": 0.7402,
+      "step": 14523
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.067788068231454e-05,
+      "loss": 15.5607,
+      "step": 14524
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0671482603236754e-05,
+      "loss": 0.7572,
+      "step": 14525
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0665085256205446e-05,
+      "loss": 0.7912,
+      "step": 14526
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.065868864138034e-05,
+      "loss": 0.7309,
+      "step": 14527
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0652292758921036e-05,
+      "loss": 0.776,
+      "step": 14528
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.064589760898722e-05,
+      "loss": 0.7303,
+      "step": 14529
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.06395031917385e-05,
+      "loss": 17.5152,
+      "step": 14530
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0633109507334475e-05,
+      "loss": 17.1526,
+      "step": 14531
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.062671655593475e-05,
+      "loss": 16.1824,
+      "step": 14532
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0620324337698892e-05,
+      "loss": 17.6889,
+      "step": 14533
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0613932852786442e-05,
+      "loss": 15.7735,
+      "step": 14534
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0607542101356945e-05,
+      "loss": 17.4747,
+      "step": 14535
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0601152083569887e-05,
+      "loss": 0.7523,
+      "step": 14536
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0594762799584805e-05,
+      "loss": 18.7936,
+      "step": 14537
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0588374249561154e-05,
+      "loss": 0.7295,
+      "step": 14538
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0581986433658377e-05,
+      "loss": 19.6492,
+      "step": 14539
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0575599352035964e-05,
+      "loss": 0.753,
+      "step": 14540
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.05692130048533e-05,
+      "loss": 17.4718,
+      "step": 14541
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.05628273922698e-05,
+      "loss": 0.7797,
+      "step": 14542
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.055644251444485e-05,
+      "loss": 17.3621,
+      "step": 14543
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0550058371537806e-05,
+      "loss": 16.8375,
+      "step": 14544
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0543674963708043e-05,
+      "loss": 0.7619,
+      "step": 14545
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0537292291114862e-05,
+      "loss": 0.7954,
+      "step": 14546
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.053091035391761e-05,
+      "loss": 18.1895,
+      "step": 14547
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.052452915227556e-05,
+      "loss": 21.0089,
+      "step": 14548
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0518148686348e-05,
+      "loss": 0.7406,
+      "step": 14549
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0511768956294175e-05,
+      "loss": 0.7629,
+      "step": 14550
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0505389962273313e-05,
+      "loss": 17.4381,
+      "step": 14551
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.049901170444467e-05,
+      "loss": 16.5268,
+      "step": 14552
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.049263418296743e-05,
+      "loss": 17.5316,
+      "step": 14553
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0486257398000765e-05,
+      "loss": 0.7346,
+      "step": 14554
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0479881349703883e-05,
+      "loss": 16.9516,
+      "step": 14555
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0473506038235866e-05,
+      "loss": 0.7764,
+      "step": 14556
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0467131463755906e-05,
+      "loss": 0.7088,
+      "step": 14557
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0460757626423082e-05,
+      "loss": 17.3672,
+      "step": 14558
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.045438452639648e-05,
+      "loss": 0.7632,
+      "step": 14559
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.04480121638352e-05,
+      "loss": 17.5634,
+      "step": 14560
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.044164053889829e-05,
+      "loss": 0.7642,
+      "step": 14561
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.043526965174477e-05,
+      "loss": 0.7379,
+      "step": 14562
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0428899502533678e-05,
+      "loss": 0.7082,
+      "step": 14563
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0422530091423985e-05,
+      "loss": 0.7427,
+      "step": 14564
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0416161418574713e-05,
+      "loss": 0.763,
+      "step": 14565
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.040979348414481e-05,
+      "loss": 18.3103,
+      "step": 14566
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.04034262882932e-05,
+      "loss": 0.6986,
+      "step": 14567
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0397059831178845e-05,
+      "loss": 17.5577,
+      "step": 14568
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.039069411296063e-05,
+      "loss": 0.6942,
+      "step": 14569
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0384329133797453e-05,
+      "loss": 17.5321,
+      "step": 14570
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.037796489384819e-05,
+      "loss": 0.7263,
+      "step": 14571
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.037160139327166e-05,
+      "loss": 0.6891,
+      "step": 14572
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0365238632226747e-05,
+      "loss": 17.7064,
+      "step": 14573
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0358876610872223e-05,
+      "loss": 0.7592,
+      "step": 14574
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0352515329366928e-05,
+      "loss": 0.7431,
+      "step": 14575
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.034615478786962e-05,
+      "loss": 0.7879,
+      "step": 14576
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.033979498653906e-05,
+      "loss": 17.17,
+      "step": 14577
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0333435925533982e-05,
+      "loss": 18.6566,
+      "step": 14578
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0327077605013108e-05,
+      "loss": 15.7647,
+      "step": 14579
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0320720025135166e-05,
+      "loss": 0.7107,
+      "step": 14580
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0314363186058832e-05,
+      "loss": 0.7184,
+      "step": 14581
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0308007087942747e-05,
+      "loss": 18.1634,
+      "step": 14582
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0301651730945627e-05,
+      "loss": 0.7836,
+      "step": 14583
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.029529711522602e-05,
+      "loss": 0.7729,
+      "step": 14584
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0288943240942597e-05,
+      "loss": 0.7761,
+      "step": 14585
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0282590108253942e-05,
+      "loss": 17.6396,
+      "step": 14586
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0276237717318598e-05,
+      "loss": 0.778,
+      "step": 14587
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0269886068295168e-05,
+      "loss": 0.7567,
+      "step": 14588
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0263535161342167e-05,
+      "loss": 17.9649,
+      "step": 14589
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.025718499661812e-05,
+      "loss": 0.7803,
+      "step": 14590
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.025083557428153e-05,
+      "loss": 18.8191,
+      "step": 14591
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0244486894490856e-05,
+      "loss": 0.7422,
+      "step": 14592
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0238138957404602e-05,
+      "loss": 0.7227,
+      "step": 14593
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0231791763181195e-05,
+      "loss": 0.72,
+      "step": 14594
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.022544531197904e-05,
+      "loss": 17.4329,
+      "step": 14595
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.021909960395661e-05,
+      "loss": 17.0533,
+      "step": 14596
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.021275463927222e-05,
+      "loss": 17.7678,
+      "step": 14597
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0206410418084286e-05,
+      "loss": 0.7214,
+      "step": 14598
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0200066940551155e-05,
+      "loss": 17.8688,
+      "step": 14599
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.019372420683114e-05,
+      "loss": 0.7153,
+      "step": 14600
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.018738221708259e-05,
+      "loss": 0.7357,
+      "step": 14601
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0181040971463773e-05,
+      "loss": 0.7686,
+      "step": 14602
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.017470047013301e-05,
+      "loss": 17.8,
+      "step": 14603
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0168360713248507e-05,
+      "loss": 0.7518,
+      "step": 14604
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.016202170096854e-05,
+      "loss": 0.7811,
+      "step": 14605
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0155683433451334e-05,
+      "loss": 0.7646,
+      "step": 14606
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.014934591085506e-05,
+      "loss": 18.2803,
+      "step": 14607
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.014300913333795e-05,
+      "loss": 0.762,
+      "step": 14608
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0136673101058145e-05,
+      "loss": 17.0863,
+      "step": 14609
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.01303378141738e-05,
+      "loss": 0.7636,
+      "step": 14610
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0124003272843045e-05,
+      "loss": 16.6249,
+      "step": 14611
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0117669477223967e-05,
+      "loss": 0.782,
+      "step": 14612
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0111336427474703e-05,
+      "loss": 0.7745,
+      "step": 14613
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.010500412375331e-05,
+      "loss": 16.2943,
+      "step": 14614
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0098672566217813e-05,
+      "loss": 0.7805,
+      "step": 14615
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.00923417550263e-05,
+      "loss": 17.4917,
+      "step": 14616
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0086011690336765e-05,
+      "loss": 0.7887,
+      "step": 14617
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0079682372307207e-05,
+      "loss": 0.7808,
+      "step": 14618
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.007335380109561e-05,
+      "loss": 0.7678,
+      "step": 14619
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0067025976859914e-05,
+      "loss": 18.0902,
+      "step": 14620
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0060698899758106e-05,
+      "loss": 0.7124,
+      "step": 14621
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0054372569948093e-05,
+      "loss": 15.1589,
+      "step": 14622
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0048046987587756e-05,
+      "loss": 19.9861,
+      "step": 14623
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0041722152835042e-05,
+      "loss": 0.7646,
+      "step": 14624
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0035398065847744e-05,
+      "loss": 15.8989,
+      "step": 14625
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0029074726783776e-05,
+      "loss": 18.0115,
+      "step": 14626
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0022752135800936e-05,
+      "loss": 0.7768,
+      "step": 14627
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0016430293057037e-05,
+      "loss": 18.2996,
+      "step": 14628
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0010109198709902e-05,
+      "loss": 18.3957,
+      "step": 14629
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0003788852917272e-05,
+      "loss": 19.4471,
+      "step": 14630
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9997469255836955e-05,
+      "loss": 0.723,
+      "step": 14631
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.999115040762662e-05,
+      "loss": 16.569,
+      "step": 14632
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.998483230844404e-05,
+      "loss": 0.7461,
+      "step": 14633
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9978514958446898e-05,
+      "loss": 17.1281,
+      "step": 14634
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9972198357792866e-05,
+      "loss": 17.7789,
+      "step": 14635
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9965882506639627e-05,
+      "loss": 18.2174,
+      "step": 14636
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9959567405144825e-05,
+      "loss": 15.1974,
+      "step": 14637
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.995325305346608e-05,
+      "loss": 18.1602,
+      "step": 14638
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9946939451760998e-05,
+      "loss": 0.7439,
+      "step": 14639
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.994062660018715e-05,
+      "loss": 18.3594,
+      "step": 14640
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9934314498902147e-05,
+      "loss": 0.7145,
+      "step": 14641
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9928003148063512e-05,
+      "loss": 16.3414,
+      "step": 14642
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9921692547828775e-05,
+      "loss": 17.2837,
+      "step": 14643
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9915382698355473e-05,
+      "loss": 0.7497,
+      "step": 14644
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9909073599801087e-05,
+      "loss": 0.7484,
+      "step": 14645
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9902765252323098e-05,
+      "loss": 17.6132,
+      "step": 14646
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.989645765607896e-05,
+      "loss": 0.7163,
+      "step": 14647
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9890150811226094e-05,
+      "loss": 17.7785,
+      "step": 14648
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9883844717921956e-05,
+      "loss": 18.2016,
+      "step": 14649
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.987753937632393e-05,
+      "loss": 0.8406,
+      "step": 14650
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.987123478658938e-05,
+      "loss": 0.7599,
+      "step": 14651
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9864930948875726e-05,
+      "loss": 16.6689,
+      "step": 14652
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9858627863340246e-05,
+      "loss": 19.3303,
+      "step": 14653
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9852325530140308e-05,
+      "loss": 19.1745,
+      "step": 14654
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9846023949433205e-05,
+      "loss": 0.824,
+      "step": 14655
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.983972312137622e-05,
+      "loss": 0.7738,
+      "step": 14656
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9833423046126648e-05,
+      "loss": 0.7277,
+      "step": 14657
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9827123723841707e-05,
+      "loss": 16.9419,
+      "step": 14658
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.982082515467868e-05,
+      "loss": 17.5218,
+      "step": 14659
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9814527338794716e-05,
+      "loss": 0.8044,
+      "step": 14660
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.980823027634705e-05,
+      "loss": 0.7895,
+      "step": 14661
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9801933967492858e-05,
+      "loss": 0.8229,
+      "step": 14662
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.979563841238927e-05,
+      "loss": 19.0205,
+      "step": 14663
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9789343611193467e-05,
+      "loss": 0.7448,
+      "step": 14664
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.978304956406254e-05,
+      "loss": 0.7599,
+      "step": 14665
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9776756271153596e-05,
+      "loss": 17.3252,
+      "step": 14666
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.977046373262371e-05,
+      "loss": 16.9268,
+      "step": 14667
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9764171948629945e-05,
+      "loss": 16.9738,
+      "step": 14668
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9757880919329362e-05,
+      "loss": 17.8701,
+      "step": 14669
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.975159064487898e-05,
+      "loss": 0.7616,
+      "step": 14670
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9745301125435783e-05,
+      "loss": 0.7405,
+      "step": 14671
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9739012361156815e-05,
+      "loss": 18.6193,
+      "step": 14672
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9732724352198973e-05,
+      "loss": 0.7288,
+      "step": 14673
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9726437098719257e-05,
+      "loss": 0.7454,
+      "step": 14674
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9720150600874586e-05,
+      "loss": 18.542,
+      "step": 14675
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.971386485882185e-05,
+      "loss": 0.7544,
+      "step": 14676
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9707579872717978e-05,
+      "loss": 0.7472,
+      "step": 14677
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9701295642719836e-05,
+      "loss": 0.791,
+      "step": 14678
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9695012168984265e-05,
+      "loss": 16.639,
+      "step": 14679
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9688729451668114e-05,
+      "loss": 16.4307,
+      "step": 14680
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.968244749092818e-05,
+      "loss": 0.7572,
+      "step": 14681
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9676166286921295e-05,
+      "loss": 16.3969,
+      "step": 14682
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9669885839804207e-05,
+      "loss": 0.7649,
+      "step": 14683
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.966360614973371e-05,
+      "loss": 0.719,
+      "step": 14684
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.965732721686653e-05,
+      "loss": 17.2002,
+      "step": 14685
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9651049041359392e-05,
+      "loss": 0.7288,
+      "step": 14686
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9644771623368998e-05,
+      "loss": 18.5786,
+      "step": 14687
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9638494963052018e-05,
+      "loss": 16.6868,
+      "step": 14688
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9632219060565153e-05,
+      "loss": 0.787,
+      "step": 14689
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9625943916065037e-05,
+      "loss": 0.7261,
+      "step": 14690
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.961966952970828e-05,
+      "loss": 0.7396,
+      "step": 14691
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.961339590165152e-05,
+      "loss": 17.2007,
+      "step": 14692
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9607123032051337e-05,
+      "loss": 0.7856,
+      "step": 14693
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9600850921064305e-05,
+      "loss": 0.7291,
+      "step": 14694
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9594579568846978e-05,
+      "loss": 18.6867,
+      "step": 14695
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9588308975555864e-05,
+      "loss": 0.7532,
+      "step": 14696
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9582039141347525e-05,
+      "loss": 17.5275,
+      "step": 14697
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.957577006637843e-05,
+      "loss": 0.7769,
+      "step": 14698
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9569501750805047e-05,
+      "loss": 17.8647,
+      "step": 14699
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9563234194783882e-05,
+      "loss": 17.2043,
+      "step": 14700
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9556967398471304e-05,
+      "loss": 0.7339,
+      "step": 14701
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9550701362023788e-05,
+      "loss": 17.8121,
+      "step": 14702
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.954443608559772e-05,
+      "loss": 15.8909,
+      "step": 14703
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9538171569349468e-05,
+      "loss": 0.718,
+      "step": 14704
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9531907813435417e-05,
+      "loss": 0.7337,
+      "step": 14705
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9525644818011913e-05,
+      "loss": 18.9033,
+      "step": 14706
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.951938258323527e-05,
+      "loss": 0.7786,
+      "step": 14707
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9513121109261794e-05,
+      "loss": 16.7695,
+      "step": 14708
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.950686039624776e-05,
+      "loss": 0.7614,
+      "step": 14709
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.950060044434947e-05,
+      "loss": 16.7359,
+      "step": 14710
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9494341253723142e-05,
+      "loss": 17.9457,
+      "step": 14711
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.948808282452504e-05,
+      "loss": 18.2588,
+      "step": 14712
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9481825156911355e-05,
+      "loss": 18.3775,
+      "step": 14713
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9475568251038285e-05,
+      "loss": 0.7303,
+      "step": 14714
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9469312107061993e-05,
+      "loss": 18.6949,
+      "step": 14715
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9463056725138634e-05,
+      "loss": 0.7268,
+      "step": 14716
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9456802105424365e-05,
+      "loss": 16.7542,
+      "step": 14717
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.945054824807529e-05,
+      "loss": 16.6761,
+      "step": 14718
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9444295153247487e-05,
+      "loss": 16.5221,
+      "step": 14719
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.943804282109707e-05,
+      "loss": 0.736,
+      "step": 14720
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9431791251780084e-05,
+      "loss": 16.6535,
+      "step": 14721
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9425540445452572e-05,
+      "loss": 17.1284,
+      "step": 14722
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9419290402270544e-05,
+      "loss": 0.7448,
+      "step": 14723
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9413041122390004e-05,
+      "loss": 17.5843,
+      "step": 14724
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9406792605966944e-05,
+      "loss": 20.084,
+      "step": 14725
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9400544853157336e-05,
+      "loss": 0.7842,
+      "step": 14726
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9394297864117094e-05,
+      "loss": 0.755,
+      "step": 14727
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9388051639002202e-05,
+      "loss": 16.3167,
+      "step": 14728
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9381806177968498e-05,
+      "loss": 18.8625,
+      "step": 14729
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9375561481171917e-05,
+      "loss": 18.0224,
+      "step": 14730
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9369317548768317e-05,
+      "loss": 0.7555,
+      "step": 14731
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9363074380913525e-05,
+      "loss": 17.798,
+      "step": 14732
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9356831977763408e-05,
+      "loss": 16.1878,
+      "step": 14733
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9350590339473763e-05,
+      "loss": 0.7214,
+      "step": 14734
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.934434946620038e-05,
+      "loss": 16.266,
+      "step": 14735
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9338109358099037e-05,
+      "loss": 0.7062,
+      "step": 14736
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9331870015325464e-05,
+      "loss": 0.7363,
+      "step": 14737
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.932563143803544e-05,
+      "loss": 0.7448,
+      "step": 14738
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.931939362638464e-05,
+      "loss": 17.6117,
+      "step": 14739
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9313156580528796e-05,
+      "loss": 0.7869,
+      "step": 14740
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.930692030062357e-05,
+      "loss": 18.437,
+      "step": 14741
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9300684786824625e-05,
+      "loss": 18.9351,
+      "step": 14742
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9294450039287593e-05,
+      "loss": 19.3077,
+      "step": 14743
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9288216058168083e-05,
+      "loss": 16.5642,
+      "step": 14744
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9281982843621732e-05,
+      "loss": 19.3662,
+      "step": 14745
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9275750395804103e-05,
+      "loss": 18.5886,
+      "step": 14746
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9269518714870744e-05,
+      "loss": 18.7829,
+      "step": 14747
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9263287800977242e-05,
+      "loss": 0.7884,
+      "step": 14748
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9257057654279065e-05,
+      "loss": 15.766,
+      "step": 14749
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9250828274931765e-05,
+      "loss": 17.2551,
+      "step": 14750
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.924459966309081e-05,
+      "loss": 15.6103,
+      "step": 14751
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9238371818911656e-05,
+      "loss": 18.1363,
+      "step": 14752
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9232144742549785e-05,
+      "loss": 0.7414,
+      "step": 14753
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.92259184341606e-05,
+      "loss": 0.7686,
+      "step": 14754
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.921969289389952e-05,
+      "loss": 15.7295,
+      "step": 14755
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9213468121921933e-05,
+      "loss": 0.7428,
+      "step": 14756
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9207244118383193e-05,
+      "loss": 17.2308,
+      "step": 14757
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.920102088343869e-05,
+      "loss": 0.7992,
+      "step": 14758
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9194798417243736e-05,
+      "loss": 0.7794,
+      "step": 14759
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9188576719953633e-05,
+      "loss": 0.789,
+      "step": 14760
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9182355791723706e-05,
+      "loss": 0.7095,
+      "step": 14761
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9176135632709214e-05,
+      "loss": 18.0511,
+      "step": 14762
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9169916243065416e-05,
+      "loss": 17.563,
+      "step": 14763
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9163697622947548e-05,
+      "loss": 0.7258,
+      "step": 14764
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9157479772510812e-05,
+      "loss": 0.7485,
+      "step": 14765
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9151262691910437e-05,
+      "loss": 0.7728,
+      "step": 14766
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9145046381301574e-05,
+      "loss": 0.7726,
+      "step": 14767
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9138830840839418e-05,
+      "loss": 0.8062,
+      "step": 14768
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9132616070679103e-05,
+      "loss": 17.8577,
+      "step": 14769
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.912640207097571e-05,
+      "loss": 19.1084,
+      "step": 14770
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.912018884188438e-05,
+      "loss": 0.7802,
+      "step": 14771
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9113976383560172e-05,
+      "loss": 16.8824,
+      "step": 14772
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9107764696158187e-05,
+      "loss": 21.4135,
+      "step": 14773
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9101553779833448e-05,
+      "loss": 17.1844,
+      "step": 14774
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9095343634740965e-05,
+      "loss": 19.7645,
+      "step": 14775
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.908913426103579e-05,
+      "loss": 18.7079,
+      "step": 14776
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9082925658872853e-05,
+      "loss": 17.8698,
+      "step": 14777
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.907671782840717e-05,
+      "loss": 16.3757,
+      "step": 14778
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9070510769793663e-05,
+      "loss": 16.6305,
+      "step": 14779
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9064304483187255e-05,
+      "loss": 0.7434,
+      "step": 14780
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9058098968742892e-05,
+      "loss": 16.7587,
+      "step": 14781
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9051894226615442e-05,
+      "loss": 15.4236,
+      "step": 14782
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9045690256959776e-05,
+      "loss": 18.891,
+      "step": 14783
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.903948705993075e-05,
+      "loss": 15.9642,
+      "step": 14784
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.903328463568318e-05,
+      "loss": 17.4566,
+      "step": 14785
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9027082984371913e-05,
+      "loss": 0.7768,
+      "step": 14786
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9020882106151728e-05,
+      "loss": 0.7959,
+      "step": 14787
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.901468200117738e-05,
+      "loss": 0.7666,
+      "step": 14788
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.900848266960366e-05,
+      "loss": 18.6747,
+      "step": 14789
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.900228411158529e-05,
+      "loss": 17.3935,
+      "step": 14790
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.899608632727699e-05,
+      "loss": 16.6454,
+      "step": 14791
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.898988931683345e-05,
+      "loss": 0.7448,
+      "step": 14792
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8983693080409336e-05,
+      "loss": 17.3053,
+      "step": 14793
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8977497618159347e-05,
+      "loss": 17.6949,
+      "step": 14794
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8971302930238076e-05,
+      "loss": 0.7695,
+      "step": 14795
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.896510901680019e-05,
+      "loss": 17.9464,
+      "step": 14796
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8958915878000282e-05,
+      "loss": 17.0508,
+      "step": 14797
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8952723513992893e-05,
+      "loss": 17.3113,
+      "step": 14798
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.894653192493263e-05,
+      "loss": 0.7801,
+      "step": 14799
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8940341110974002e-05,
+      "loss": 17.3552,
+      "step": 14800
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8934151072271573e-05,
+      "loss": 19.919,
+      "step": 14801
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8927961808979822e-05,
+      "loss": 0.7694,
+      "step": 14802
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8921773321253222e-05,
+      "loss": 16.5333,
+      "step": 14803
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.891558560924629e-05,
+      "loss": 16.8284,
+      "step": 14804
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8909398673113404e-05,
+      "loss": 19.216,
+      "step": 14805
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8903212513009043e-05,
+      "loss": 0.72,
+      "step": 14806
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8897027129087604e-05,
+      "loss": 0.7452,
+      "step": 14807
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8890842521503444e-05,
+      "loss": 0.7576,
+      "step": 14808
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.888465869041098e-05,
+      "loss": 17.7881,
+      "step": 14809
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8878475635964538e-05,
+      "loss": 0.7318,
+      "step": 14810
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.887229335831845e-05,
+      "loss": 17.6295,
+      "step": 14811
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8866111857627028e-05,
+      "loss": 17.6161,
+      "step": 14812
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8859931134044546e-05,
+      "loss": 16.6892,
+      "step": 14813
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.885375118772531e-05,
+      "loss": 0.6884,
+      "step": 14814
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8847572018823555e-05,
+      "loss": 0.7493,
+      "step": 14815
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8841393627493496e-05,
+      "loss": 0.7708,
+      "step": 14816
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8835216013889388e-05,
+      "loss": 0.7221,
+      "step": 14817
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8829039178165404e-05,
+      "loss": 0.7441,
+      "step": 14818
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8822863120475713e-05,
+      "loss": 0.804,
+      "step": 14819
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8816687840974483e-05,
+      "loss": 0.716,
+      "step": 14820
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8810513339815826e-05,
+      "loss": 15.6225,
+      "step": 14821
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.88043396171539e-05,
+      "loss": 18.6775,
+      "step": 14822
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8798166673142754e-05,
+      "loss": 21.2187,
+      "step": 14823
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8791994507936516e-05,
+      "loss": 15.3126,
+      "step": 14824
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8785823121689234e-05,
+      "loss": 0.7102,
+      "step": 14825
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8779652514554902e-05,
+      "loss": 16.422,
+      "step": 14826
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8773482686687587e-05,
+      "loss": 18.3012,
+      "step": 14827
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8767313638241258e-05,
+      "loss": 16.8415,
+      "step": 14828
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8761145369369932e-05,
+      "loss": 0.7919,
+      "step": 14829
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.875497788022754e-05,
+      "loss": 0.7885,
+      "step": 14830
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8748811170968043e-05,
+      "loss": 17.445,
+      "step": 14831
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8742645241745353e-05,
+      "loss": 0.7356,
+      "step": 14832
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.873648009271336e-05,
+      "loss": 0.7662,
+      "step": 14833
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8730315724025976e-05,
+      "loss": 0.7375,
+      "step": 14834
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8724152135837052e-05,
+      "loss": 0.6802,
+      "step": 14835
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.871798932830042e-05,
+      "loss": 0.7521,
+      "step": 14836
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.871182730156993e-05,
+      "loss": 0.7572,
+      "step": 14837
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.870566605579937e-05,
+      "loss": 0.7046,
+      "step": 14838
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8699505591142535e-05,
+      "loss": 0.7837,
+      "step": 14839
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.869334590775319e-05,
+      "loss": 19.7961,
+      "step": 14840
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8687187005785056e-05,
+      "loss": 16.3949,
+      "step": 14841
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8681028885391904e-05,
+      "loss": 0.7712,
+      "step": 14842
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8674871546727425e-05,
+      "loss": 19.2489,
+      "step": 14843
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8668714989945286e-05,
+      "loss": 16.4499,
+      "step": 14844
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8662559215199204e-05,
+      "loss": 0.7486,
+      "step": 14845
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.865640422264277e-05,
+      "loss": 0.7586,
+      "step": 14846
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.865025001242966e-05,
+      "loss": 15.9297,
+      "step": 14847
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8644096584713466e-05,
+      "loss": 18.2703,
+      "step": 14848
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8637943939647768e-05,
+      "loss": 0.7835,
+      "step": 14849
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.863179207738616e-05,
+      "loss": 0.7561,
+      "step": 14850
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.862564099808217e-05,
+      "loss": 16.569,
+      "step": 14851
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8619490701889386e-05,
+      "loss": 16.0238,
+      "step": 14852
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8613341188961252e-05,
+      "loss": 0.7705,
+      "step": 14853
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.860719245945128e-05,
+      "loss": 19.3672,
+      "step": 14854
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.860104451351296e-05,
+      "loss": 18.5223,
+      "step": 14855
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8594897351299728e-05,
+      "loss": 15.5186,
+      "step": 14856
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8588750972965043e-05,
+      "loss": 17.824,
+      "step": 14857
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8582605378662306e-05,
+      "loss": 16.1757,
+      "step": 14858
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.857646056854491e-05,
+      "loss": 0.8004,
+      "step": 14859
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8570316542766235e-05,
+      "loss": 0.7568,
+      "step": 14860
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.856417330147962e-05,
+      "loss": 17.689,
+      "step": 14861
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.855803084483843e-05,
+      "loss": 0.7535,
+      "step": 14862
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8551889172995974e-05,
+      "loss": 17.7613,
+      "step": 14863
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8545748286105523e-05,
+      "loss": 18.2076,
+      "step": 14864
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.85396081843204e-05,
+      "loss": 0.7697,
+      "step": 14865
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8533468867793834e-05,
+      "loss": 0.8159,
+      "step": 14866
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8527330336679076e-05,
+      "loss": 0.7668,
+      "step": 14867
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8521192591129338e-05,
+      "loss": 17.9455,
+      "step": 14868
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.85150556312978e-05,
+      "loss": 18.2634,
+      "step": 14869
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.850891945733768e-05,
+      "loss": 18.7101,
+      "step": 14870
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8502784069402117e-05,
+      "loss": 16.8847,
+      "step": 14871
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8496649467644243e-05,
+      "loss": 0.8504,
+      "step": 14872
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8490515652217226e-05,
+      "loss": 18.7721,
+      "step": 14873
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.84843826232741e-05,
+      "loss": 18.0695,
+      "step": 14874
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.847825038096799e-05,
+      "loss": 16.61,
+      "step": 14875
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8472118925451954e-05,
+      "loss": 0.743,
+      "step": 14876
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8465988256879014e-05,
+      "loss": 0.805,
+      "step": 14877
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8459858375402218e-05,
+      "loss": 15.6159,
+      "step": 14878
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8453729281174544e-05,
+      "loss": 0.7935,
+      "step": 14879
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8447600974349022e-05,
+      "loss": 17.0037,
+      "step": 14880
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8441473455078567e-05,
+      "loss": 0.8168,
+      "step": 14881
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8435346723516128e-05,
+      "loss": 18.1526,
+      "step": 14882
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8429220779814653e-05,
+      "loss": 0.751,
+      "step": 14883
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8423095624127025e-05,
+      "loss": 0.7298,
+      "step": 14884
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8416971256606152e-05,
+      "loss": 0.7314,
+      "step": 14885
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.841084767740489e-05,
+      "loss": 18.0552,
+      "step": 14886
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8404724886676084e-05,
+      "loss": 16.9808,
+      "step": 14887
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.839860288457256e-05,
+      "loss": 0.7437,
+      "step": 14888
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8392481671247097e-05,
+      "loss": 0.7438,
+      "step": 14889
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8386361246852534e-05,
+      "loss": 0.7549,
+      "step": 14890
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.838024161154161e-05,
+      "loss": 0.7567,
+      "step": 14891
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.837412276546706e-05,
+      "loss": 0.7769,
+      "step": 14892
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8368004708781644e-05,
+      "loss": 0.6957,
+      "step": 14893
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8361887441638048e-05,
+      "loss": 17.2776,
+      "step": 14894
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8355770964188974e-05,
+      "loss": 0.7558,
+      "step": 14895
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8349655276587074e-05,
+      "loss": 0.6852,
+      "step": 14896
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.834354037898499e-05,
+      "loss": 18.8091,
+      "step": 14897
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8337426271535384e-05,
+      "loss": 16.2215,
+      "step": 14898
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.833131295439085e-05,
+      "loss": 0.7655,
+      "step": 14899
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.832520042770396e-05,
+      "loss": 0.689,
+      "step": 14900
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8319088691627327e-05,
+      "loss": 0.729,
+      "step": 14901
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.831297774631344e-05,
+      "loss": 17.3288,
+      "step": 14902
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.830686759191489e-05,
+      "loss": 16.8637,
+      "step": 14903
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8300758228584154e-05,
+      "loss": 0.7321,
+      "step": 14904
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8294649656473716e-05,
+      "loss": 0.8038,
+      "step": 14905
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8288541875736077e-05,
+      "loss": 16.409,
+      "step": 14906
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8282434886523676e-05,
+      "loss": 0.7202,
+      "step": 14907
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8276328688988935e-05,
+      "loss": 16.8575,
+      "step": 14908
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8270223283284284e-05,
+      "loss": 17.4503,
+      "step": 14909
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8264118669562082e-05,
+      "loss": 0.7888,
+      "step": 14910
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.825801484797474e-05,
+      "loss": 0.7417,
+      "step": 14911
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8251911818674576e-05,
+      "loss": 19.1764,
+      "step": 14912
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8245809581813967e-05,
+      "loss": 0.738,
+      "step": 14913
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8239708137545197e-05,
+      "loss": 0.7767,
+      "step": 14914
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8233607486020566e-05,
+      "loss": 0.7412,
+      "step": 14915
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8227507627392344e-05,
+      "loss": 0.7049,
+      "step": 14916
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8221408561812763e-05,
+      "loss": 17.8818,
+      "step": 14917
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8215310289434108e-05,
+      "loss": 17.6827,
+      "step": 14918
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8209212810408555e-05,
+      "loss": 0.7605,
+      "step": 14919
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.82031161248883e-05,
+      "loss": 0.7597,
+      "step": 14920
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8197020233025557e-05,
+      "loss": 18.5471,
+      "step": 14921
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.819092513497242e-05,
+      "loss": 18.297,
+      "step": 14922
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8184830830881067e-05,
+      "loss": 16.0769,
+      "step": 14923
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.81787373209036e-05,
+      "loss": 15.9526,
+      "step": 14924
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.81726446051921e-05,
+      "loss": 19.8763,
+      "step": 14925
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.816655268389868e-05,
+      "loss": 18.2643,
+      "step": 14926
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.816046155717537e-05,
+      "loss": 0.8006,
+      "step": 14927
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8154371225174206e-05,
+      "loss": 17.0866,
+      "step": 14928
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8148281688047213e-05,
+      "loss": 0.7504,
+      "step": 14929
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8142192945946363e-05,
+      "loss": 0.7932,
+      "step": 14930
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8136104999023663e-05,
+      "loss": 18.7127,
+      "step": 14931
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8130017847431064e-05,
+      "loss": 0.776,
+      "step": 14932
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8123931491320478e-05,
+      "loss": 0.7555,
+      "step": 14933
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8117845930843857e-05,
+      "loss": 14.8027,
+      "step": 14934
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.811176116615308e-05,
+      "loss": 16.4648,
+      "step": 14935
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8105677197400024e-05,
+      "loss": 16.876,
+      "step": 14936
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.809959402473655e-05,
+      "loss": 0.7761,
+      "step": 14937
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8093511648314477e-05,
+      "loss": 0.7533,
+      "step": 14938
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8087430068285656e-05,
+      "loss": 15.7339,
+      "step": 14939
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8081349284801845e-05,
+      "loss": 0.7382,
+      "step": 14940
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8075269298014863e-05,
+      "loss": 0.812,
+      "step": 14941
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.806919010807645e-05,
+      "loss": 0.7322,
+      "step": 14942
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8063111715138344e-05,
+      "loss": 0.7254,
+      "step": 14943
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.805703411935226e-05,
+      "loss": 16.5345,
+      "step": 14944
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8050957320869883e-05,
+      "loss": 16.6483,
+      "step": 14945
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8044881319842923e-05,
+      "loss": 17.7776,
+      "step": 14946
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.803880611642302e-05,
+      "loss": 0.7909,
+      "step": 14947
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.803273171076179e-05,
+      "loss": 18.3406,
+      "step": 14948
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.802665810301092e-05,
+      "loss": 15.3783,
+      "step": 14949
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8020585293321923e-05,
+      "loss": 16.373,
+      "step": 14950
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.801451328184644e-05,
+      "loss": 16.5449,
+      "step": 14951
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.800844206873601e-05,
+      "loss": 0.7499,
+      "step": 14952
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.8002371654142152e-05,
+      "loss": 0.7698,
+      "step": 14953
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7996302038216417e-05,
+      "loss": 16.7943,
+      "step": 14954
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7990233221110297e-05,
+      "loss": 0.7528,
+      "step": 14955
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.798416520297526e-05,
+      "loss": 18.4217,
+      "step": 14956
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.797809798396277e-05,
+      "loss": 17.6522,
+      "step": 14957
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.797203156422425e-05,
+      "loss": 18.5117,
+      "step": 14958
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7965965943911156e-05,
+      "loss": 16.962,
+      "step": 14959
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.795990112317486e-05,
+      "loss": 17.4079,
+      "step": 14960
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.795383710216673e-05,
+      "loss": 18.0306,
+      "step": 14961
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7947773881038166e-05,
+      "loss": 0.7153,
+      "step": 14962
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.794171145994048e-05,
+      "loss": 0.7458,
+      "step": 14963
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7935649839025e-05,
+      "loss": 0.7387,
+      "step": 14964
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7929589018443016e-05,
+      "loss": 18.3877,
+      "step": 14965
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.79235289983458e-05,
+      "loss": 17.552,
+      "step": 14966
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7917469778884638e-05,
+      "loss": 0.7723,
+      "step": 14967
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7911411360210738e-05,
+      "loss": 0.7952,
+      "step": 14968
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.790535374247536e-05,
+      "loss": 17.0313,
+      "step": 14969
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7899296925829677e-05,
+      "loss": 16.0552,
+      "step": 14970
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7893240910424876e-05,
+      "loss": 15.7029,
+      "step": 14971
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7887185696412108e-05,
+      "loss": 16.6231,
+      "step": 14972
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.78811312839425e-05,
+      "loss": 17.5988,
+      "step": 14973
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7875077673167205e-05,
+      "loss": 17.542,
+      "step": 14974
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.786902486423731e-05,
+      "loss": 19.8305,
+      "step": 14975
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7862972857303874e-05,
+      "loss": 15.1327,
+      "step": 14976
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7856921652518007e-05,
+      "loss": 0.7526,
+      "step": 14977
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7850871250030675e-05,
+      "loss": 0.7126,
+      "step": 14978
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.784482164999296e-05,
+      "loss": 0.7076,
+      "step": 14979
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7838772852555842e-05,
+      "loss": 0.7048,
+      "step": 14980
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.783272485787027e-05,
+      "loss": 0.765,
+      "step": 14981
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7826677666087255e-05,
+      "loss": 16.961,
+      "step": 14982
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7820631277357708e-05,
+      "loss": 17.9223,
+      "step": 14983
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7814585691832553e-05,
+      "loss": 16.3901,
+      "step": 14984
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7808540909662692e-05,
+      "loss": 0.728,
+      "step": 14985
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7802496930998974e-05,
+      "loss": 0.7475,
+      "step": 14986
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.779645375599231e-05,
+      "loss": 15.4645,
+      "step": 14987
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.779041138479351e-05,
+      "loss": 0.7989,
+      "step": 14988
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7784369817553376e-05,
+      "loss": 17.8599,
+      "step": 14989
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7778329054422766e-05,
+      "loss": 16.2023,
+      "step": 14990
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7772289095552387e-05,
+      "loss": 0.7569,
+      "step": 14991
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7766249941093042e-05,
+      "loss": 17.2467,
+      "step": 14992
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7760211591195465e-05,
+      "loss": 17.0156,
+      "step": 14993
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7754174046010348e-05,
+      "loss": 15.7172,
+      "step": 14994
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7748137305688422e-05,
+      "loss": 18.4865,
+      "step": 14995
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.774210137038034e-05,
+      "loss": 14.7908,
+      "step": 14996
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7736066240236804e-05,
+      "loss": 0.765,
+      "step": 14997
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.773003191540839e-05,
+      "loss": 20.5708,
+      "step": 14998
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7723998396045765e-05,
+      "loss": 0.7784,
+      "step": 14999
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7717965682299513e-05,
+      "loss": 0.7904,
+      "step": 15000
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.771193377432019e-05,
+      "loss": 0.728,
+      "step": 15001
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7705902672258383e-05,
+      "loss": 0.7527,
+      "step": 15002
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7699872376264625e-05,
+      "loss": 0.7819,
+      "step": 15003
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7693842886489427e-05,
+      "loss": 18.1453,
+      "step": 15004
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7687814203083286e-05,
+      "loss": 0.7948,
+      "step": 15005
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7681786326196663e-05,
+      "loss": 0.7329,
+      "step": 15006
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.767575925598005e-05,
+      "loss": 17.0622,
+      "step": 15007
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7669732992583866e-05,
+      "loss": 17.0227,
+      "step": 15008
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7663707536158507e-05,
+      "loss": 18.0084,
+      "step": 15009
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7657682886854415e-05,
+      "loss": 16.4188,
+      "step": 15010
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7651659044821945e-05,
+      "loss": 18.765,
+      "step": 15011
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7645636010211447e-05,
+      "loss": 17.2558,
+      "step": 15012
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7639613783173258e-05,
+      "loss": 19.6357,
+      "step": 15013
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7633592363857688e-05,
+      "loss": 0.8394,
+      "step": 15014
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7627571752415055e-05,
+      "loss": 18.3095,
+      "step": 15015
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7621551948995623e-05,
+      "loss": 18.1511,
+      "step": 15016
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.761553295374963e-05,
+      "loss": 0.747,
+      "step": 15017
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7609514766827367e-05,
+      "loss": 19.0898,
+      "step": 15018
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7603497388378976e-05,
+      "loss": 18.2298,
+      "step": 15019
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.75974808185547e-05,
+      "loss": 0.7723,
+      "step": 15020
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.75914650575047e-05,
+      "loss": 0.7387,
+      "step": 15021
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7585450105379125e-05,
+      "loss": 0.7439,
+      "step": 15022
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7579435962328127e-05,
+      "loss": 16.3113,
+      "step": 15023
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.757342262850179e-05,
+      "loss": 16.8533,
+      "step": 15024
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.756741010405027e-05,
+      "loss": 19.5258,
+      "step": 15025
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7561398389123557e-05,
+      "loss": 0.791,
+      "step": 15026
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7555387483871765e-05,
+      "loss": 16.4621,
+      "step": 15027
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7549377388444914e-05,
+      "loss": 17.4829,
+      "step": 15028
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7543368102992997e-05,
+      "loss": 0.7205,
+      "step": 15029
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7537359627666034e-05,
+      "loss": 20.1287,
+      "step": 15030
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.753135196261399e-05,
+      "loss": 17.1238,
+      "step": 15031
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7525345107986814e-05,
+      "loss": 19.1306,
+      "step": 15032
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7519339063934443e-05,
+      "loss": 18.2612,
+      "step": 15033
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.751333383060676e-05,
+      "loss": 0.7876,
+      "step": 15034
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7507329408153705e-05,
+      "loss": 18.5282,
+      "step": 15035
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7501325796725126e-05,
+      "loss": 0.791,
+      "step": 15036
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.749532299647086e-05,
+      "loss": 0.7182,
+      "step": 15037
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7489321007540765e-05,
+      "loss": 18.4257,
+      "step": 15038
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.748331983008465e-05,
+      "loss": 20.4744,
+      "step": 15039
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.747731946425229e-05,
+      "loss": 17.3346,
+      "step": 15040
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7471319910193462e-05,
+      "loss": 18.1239,
+      "step": 15041
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.74653211680579e-05,
+      "loss": 0.7151,
+      "step": 15042
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7459323237995367e-05,
+      "loss": 17.2844,
+      "step": 15043
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.745332612015556e-05,
+      "loss": 18.276,
+      "step": 15044
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7447329814688145e-05,
+      "loss": 0.7653,
+      "step": 15045
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.744133432174284e-05,
+      "loss": 16.7403,
+      "step": 15046
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.743533964146924e-05,
+      "loss": 16.4434,
+      "step": 15047
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.742934577401701e-05,
+      "loss": 0.7241,
+      "step": 15048
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.742335271953574e-05,
+      "loss": 14.7975,
+      "step": 15049
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7417360478175016e-05,
+      "loss": 15.3232,
+      "step": 15050
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7411369050084424e-05,
+      "loss": 17.769,
+      "step": 15051
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7405378435413493e-05,
+      "loss": 0.7472,
+      "step": 15052
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7399388634311785e-05,
+      "loss": 0.755,
+      "step": 15053
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7393399646928755e-05,
+      "loss": 0.7681,
+      "step": 15054
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7387411473413923e-05,
+      "loss": 0.7075,
+      "step": 15055
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.738142411391675e-05,
+      "loss": 15.8433,
+      "step": 15056
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7375437568586666e-05,
+      "loss": 18.0413,
+      "step": 15057
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.736945183757312e-05,
+      "loss": 0.7088,
+      "step": 15058
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.736346692102551e-05,
+      "loss": 18.1915,
+      "step": 15059
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.735748281909322e-05,
+      "loss": 0.7287,
+      "step": 15060
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7351499531925607e-05,
+      "loss": 0.732,
+      "step": 15061
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7345517059672006e-05,
+      "loss": 0.7557,
+      "step": 15062
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.733953540248177e-05,
+      "loss": 0.7496,
+      "step": 15063
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.733355456050419e-05,
+      "loss": 19.3749,
+      "step": 15064
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.732757453388853e-05,
+      "loss": 18.4794,
+      "step": 15065
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.73215953227841e-05,
+      "loss": 20.0187,
+      "step": 15066
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7315616927340085e-05,
+      "loss": 0.7587,
+      "step": 15067
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.730963934770575e-05,
+      "loss": 19.7151,
+      "step": 15068
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7303662584030272e-05,
+      "loss": 0.7108,
+      "step": 15069
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7297686636462834e-05,
+      "loss": 0.7588,
+      "step": 15070
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.729171150515262e-05,
+      "loss": 17.0621,
+      "step": 15071
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.728573719024875e-05,
+      "loss": 17.6058,
+      "step": 15072
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7279763691900353e-05,
+      "loss": 0.77,
+      "step": 15073
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7273791010256525e-05,
+      "loss": 18.6982,
+      "step": 15074
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.726781914546633e-05,
+      "loss": 0.7897,
+      "step": 15075
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7261848097678856e-05,
+      "loss": 0.7547,
+      "step": 15076
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7255877867043117e-05,
+      "loss": 18.7607,
+      "step": 15077
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7249908453708158e-05,
+      "loss": 0.7389,
+      "step": 15078
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.724393985782296e-05,
+      "loss": 17.5211,
+      "step": 15079
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7237972079536502e-05,
+      "loss": 18.6185,
+      "step": 15080
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7232005118997745e-05,
+      "loss": 16.7164,
+      "step": 15081
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.72260389763556e-05,
+      "loss": 15.8865,
+      "step": 15082
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7220073651759026e-05,
+      "loss": 0.7547,
+      "step": 15083
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7214109145356898e-05,
+      "loss": 17.383,
+      "step": 15084
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7208145457298074e-05,
+      "loss": 18.7651,
+      "step": 15085
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7202182587731446e-05,
+      "loss": 18.9607,
+      "step": 15086
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.719622053680583e-05,
+      "loss": 18.844,
+      "step": 15087
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7190259304670038e-05,
+      "loss": 17.4954,
+      "step": 15088
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.718429889147286e-05,
+      "loss": 15.5123,
+      "step": 15089
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.717833929736306e-05,
+      "loss": 0.7924,
+      "step": 15090
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7172380522489428e-05,
+      "loss": 0.7581,
+      "step": 15091
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7166422567000666e-05,
+      "loss": 0.7118,
+      "step": 15092
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.716046543104548e-05,
+      "loss": 15.2022,
+      "step": 15093
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.715450911477261e-05,
+      "loss": 0.7087,
+      "step": 15094
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.714855361833066e-05,
+      "loss": 17.8129,
+      "step": 15095
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7142598941868327e-05,
+      "loss": 0.75,
+      "step": 15096
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7136645085534225e-05,
+      "loss": 0.8116,
+      "step": 15097
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.713069204947695e-05,
+      "loss": 0.7498,
+      "step": 15098
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.712473983384512e-05,
+      "loss": 0.7198,
+      "step": 15099
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.71187884387873e-05,
+      "loss": 0.7888,
+      "step": 15100
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7112837864452018e-05,
+      "loss": 0.7501,
+      "step": 15101
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.710688811098782e-05,
+      "loss": 0.7432,
+      "step": 15102
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.710093917854318e-05,
+      "loss": 17.2131,
+      "step": 15103
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.709499106726663e-05,
+      "loss": 19.5021,
+      "step": 15104
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.70890437773066e-05,
+      "loss": 0.7662,
+      "step": 15105
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7083097308811574e-05,
+      "loss": 0.7953,
+      "step": 15106
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7077151661929945e-05,
+      "loss": 0.7551,
+      "step": 15107
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7071206836810134e-05,
+      "loss": 19.2335,
+      "step": 15108
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.706526283360052e-05,
+      "loss": 0.7168,
+      "step": 15109
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7059319652449445e-05,
+      "loss": 0.8479,
+      "step": 15110
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.705337729350529e-05,
+      "loss": 17.7088,
+      "step": 15111
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7047435756916364e-05,
+      "loss": 0.7099,
+      "step": 15112
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7041495042830945e-05,
+      "loss": 0.7838,
+      "step": 15113
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7035555151397355e-05,
+      "loss": 0.7723,
+      "step": 15114
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7029616082763834e-05,
+      "loss": 17.204,
+      "step": 15115
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.702367783707862e-05,
+      "loss": 16.0327,
+      "step": 15116
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7017740414489946e-05,
+      "loss": 20.0066,
+      "step": 15117
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.7011803815145975e-05,
+      "loss": 0.7559,
+      "step": 15118
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.700586803919494e-05,
+      "loss": 17.3742,
+      "step": 15119
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6999933086784963e-05,
+      "loss": 19.0484,
+      "step": 15120
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.699399895806418e-05,
+      "loss": 0.6915,
+      "step": 15121
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6988065653180745e-05,
+      "loss": 0.8233,
+      "step": 15122
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6982133172282705e-05,
+      "loss": 17.4338,
+      "step": 15123
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6976201515518168e-05,
+      "loss": 0.7567,
+      "step": 15124
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.697027068303519e-05,
+      "loss": 0.7725,
+      "step": 15125
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6964340674981778e-05,
+      "loss": 16.2041,
+      "step": 15126
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6958411491505976e-05,
+      "loss": 18.1023,
+      "step": 15127
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6952483132755776e-05,
+      "loss": 0.7451,
+      "step": 15128
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.694655559887914e-05,
+      "loss": 0.7736,
+      "step": 15129
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6940628890024025e-05,
+      "loss": 18.1746,
+      "step": 15130
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6934703006338338e-05,
+      "loss": 16.0041,
+      "step": 15131
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.692877794797004e-05,
+      "loss": 0.7776,
+      "step": 15132
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.692285371506697e-05,
+      "loss": 16.7191,
+      "step": 15133
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6916930307777047e-05,
+      "loss": 0.7925,
+      "step": 15134
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6911007726248106e-05,
+      "loss": 0.7156,
+      "step": 15135
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6905085970627944e-05,
+      "loss": 0.8075,
+      "step": 15136
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.68991650410644e-05,
+      "loss": 17.2595,
+      "step": 15137
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6893244937705242e-05,
+      "loss": 0.7254,
+      "step": 15138
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.688732566069827e-05,
+      "loss": 18.2763,
+      "step": 15139
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.688140721019121e-05,
+      "loss": 16.7965,
+      "step": 15140
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6875489586331766e-05,
+      "loss": 17.7794,
+      "step": 15141
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.68695727892677e-05,
+      "loss": 0.7387,
+      "step": 15142
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.686365681914663e-05,
+      "loss": 0.7879,
+      "step": 15143
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.685774167611627e-05,
+      "loss": 17.8014,
+      "step": 15144
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6851827360324242e-05,
+      "loss": 19.0301,
+      "step": 15145
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6845913871918155e-05,
+      "loss": 0.7516,
+      "step": 15146
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6840001211045642e-05,
+      "loss": 18.2072,
+      "step": 15147
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6834089377854278e-05,
+      "loss": 0.7445,
+      "step": 15148
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.682817837249161e-05,
+      "loss": 17.2012,
+      "step": 15149
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6822268195105185e-05,
+      "loss": 0.7451,
+      "step": 15150
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6816358845842506e-05,
+      "loss": 17.1314,
+      "step": 15151
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6810450324851102e-05,
+      "loss": 19.9022,
+      "step": 15152
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6804542632278436e-05,
+      "loss": 18.4399,
+      "step": 15153
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.679863576827195e-05,
+      "loss": 17.8943,
+      "step": 15154
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6792729732979113e-05,
+      "loss": 0.7701,
+      "step": 15155
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6786824526547324e-05,
+      "loss": 20.1411,
+      "step": 15156
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6780920149123974e-05,
+      "loss": 0.7781,
+      "step": 15157
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6775016600856446e-05,
+      "loss": 17.2084,
+      "step": 15158
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6769113881892074e-05,
+      "loss": 17.448,
+      "step": 15159
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6763211992378223e-05,
+      "loss": 0.8277,
+      "step": 15160
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6757310932462173e-05,
+      "loss": 17.3455,
+      "step": 15161
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6751410702291247e-05,
+      "loss": 0.7574,
+      "step": 15162
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6745511302012722e-05,
+      "loss": 15.6575,
+      "step": 15163
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.673961273177379e-05,
+      "loss": 0.7673,
+      "step": 15164
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.673371499172174e-05,
+      "loss": 0.7613,
+      "step": 15165
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.672781808200374e-05,
+      "loss": 18.6752,
+      "step": 15166
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6721922002767016e-05,
+      "loss": 0.7655,
+      "step": 15167
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6716026754158714e-05,
+      "loss": 19.3448,
+      "step": 15168
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6710132336325974e-05,
+      "loss": 0.7093,
+      "step": 15169
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6704238749415957e-05,
+      "loss": 18.161,
+      "step": 15170
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6698345993575714e-05,
+      "loss": 0.7996,
+      "step": 15171
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6692454068952372e-05,
+      "loss": 0.8064,
+      "step": 15172
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6686562975692986e-05,
+      "loss": 21.4437,
+      "step": 15173
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6680672713944577e-05,
+      "loss": 17.6712,
+      "step": 15174
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.66747832838542e-05,
+      "loss": 18.0349,
+      "step": 15175
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6668894685568843e-05,
+      "loss": 0.7195,
+      "step": 15176
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6663006919235485e-05,
+      "loss": 16.2176,
+      "step": 15177
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6657119985001084e-05,
+      "loss": 17.3255,
+      "step": 15178
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6651233883012567e-05,
+      "loss": 0.7533,
+      "step": 15179
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6645348613416882e-05,
+      "loss": 17.1349,
+      "step": 15180
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6639464176360913e-05,
+      "loss": 15.5157,
+      "step": 15181
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.663358057199152e-05,
+      "loss": 17.7378,
+      "step": 15182
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.662769780045559e-05,
+      "loss": 0.8026,
+      "step": 15183
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6621815861899943e-05,
+      "loss": 18.9438,
+      "step": 15184
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.661593475647139e-05,
+      "loss": 0.7493,
+      "step": 15185
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.661005448431673e-05,
+      "loss": 18.2736,
+      "step": 15186
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6604175045582717e-05,
+      "loss": 19.8022,
+      "step": 15187
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.659829644041614e-05,
+      "loss": 0.8018,
+      "step": 15188
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6592418668963687e-05,
+      "loss": 17.3952,
+      "step": 15189
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6586541731372113e-05,
+      "loss": 0.7828,
+      "step": 15190
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6580665627788104e-05,
+      "loss": 18.5357,
+      "step": 15191
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6574790358358276e-05,
+      "loss": 17.7104,
+      "step": 15192
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6568915923229332e-05,
+      "loss": 0.7407,
+      "step": 15193
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.656304232254786e-05,
+      "loss": 0.7857,
+      "step": 15194
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6557169556460507e-05,
+      "loss": 20.2894,
+      "step": 15195
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6551297625113832e-05,
+      "loss": 15.5233,
+      "step": 15196
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6545426528654394e-05,
+      "loss": 19.0138,
+      "step": 15197
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6539556267228778e-05,
+      "loss": 19.6582,
+      "step": 15198
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6533686840983447e-05,
+      "loss": 0.7668,
+      "step": 15199
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6527818250064953e-05,
+      "loss": 16.5338,
+      "step": 15200
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6521950494619755e-05,
+      "loss": 17.1189,
+      "step": 15201
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.65160835747943e-05,
+      "loss": 17.916,
+      "step": 15202
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6510217490735057e-05,
+      "loss": 17.0824,
+      "step": 15203
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.650435224258844e-05,
+      "loss": 0.8141,
+      "step": 15204
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6498487830500835e-05,
+      "loss": 16.5361,
+      "step": 15205
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6492624254618623e-05,
+      "loss": 0.7667,
+      "step": 15206
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6486761515088144e-05,
+      "loss": 0.8132,
+      "step": 15207
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6480899612055766e-05,
+      "loss": 19.5597,
+      "step": 15208
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6475038545667787e-05,
+      "loss": 16.8576,
+      "step": 15209
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6469178316070488e-05,
+      "loss": 19.1317,
+      "step": 15210
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.646331892341018e-05,
+      "loss": 0.7396,
+      "step": 15211
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.645746036783306e-05,
+      "loss": 0.7735,
+      "step": 15212
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6451602649485408e-05,
+      "loss": 0.7588,
+      "step": 15213
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.644574576851341e-05,
+      "loss": 17.8535,
+      "step": 15214
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6439889725063247e-05,
+      "loss": 0.7446,
+      "step": 15215
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6434034519281106e-05,
+      "loss": 17.6773,
+      "step": 15216
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6428180151313117e-05,
+      "loss": 0.7838,
+      "step": 15217
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.642232662130545e-05,
+      "loss": 18.1965,
+      "step": 15218
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6416473929404164e-05,
+      "loss": 0.7672,
+      "step": 15219
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6410622075755332e-05,
+      "loss": 0.7611,
+      "step": 15220
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6404771060505064e-05,
+      "loss": 17.017,
+      "step": 15221
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.639892088379937e-05,
+      "loss": 19.8182,
+      "step": 15222
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.639307154578429e-05,
+      "loss": 17.1846,
+      "step": 15223
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.638722304660582e-05,
+      "loss": 0.7277,
+      "step": 15224
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6381375386409943e-05,
+      "loss": 0.8418,
+      "step": 15225
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6375528565342606e-05,
+      "loss": 16.3005,
+      "step": 15226
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.636968258354974e-05,
+      "loss": 17.9346,
+      "step": 15227
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6363837441177292e-05,
+      "loss": 0.7099,
+      "step": 15228
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6357993138371142e-05,
+      "loss": 17.6814,
+      "step": 15229
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6352149675277145e-05,
+      "loss": 15.6244,
+      "step": 15230
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6346307052041195e-05,
+      "loss": 17.4827,
+      "step": 15231
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6340465268809102e-05,
+      "loss": 17.7894,
+      "step": 15232
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6334624325726683e-05,
+      "loss": 17.1478,
+      "step": 15233
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.632878422293973e-05,
+      "loss": 0.7792,
+      "step": 15234
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.632294496059399e-05,
+      "loss": 0.774,
+      "step": 15235
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6317106538835252e-05,
+      "loss": 17.4773,
+      "step": 15236
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6311268957809227e-05,
+      "loss": 19.8692,
+      "step": 15237
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6305432217661605e-05,
+      "loss": 0.7689,
+      "step": 15238
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.629959631853812e-05,
+      "loss": 18.0137,
+      "step": 15239
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6293761260584367e-05,
+      "loss": 0.7714,
+      "step": 15240
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6287927043946056e-05,
+      "loss": 17.1227,
+      "step": 15241
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.628209366876877e-05,
+      "loss": 0.7782,
+      "step": 15242
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6276261135198122e-05,
+      "loss": 16.827,
+      "step": 15243
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.627042944337971e-05,
+      "loss": 0.7703,
+      "step": 15244
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6264598593459057e-05,
+      "loss": 0.7608,
+      "step": 15245
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6258768585581764e-05,
+      "loss": 17.6838,
+      "step": 15246
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6252939419893292e-05,
+      "loss": 19.0269,
+      "step": 15247
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6247111096539143e-05,
+      "loss": 17.6847,
+      "step": 15248
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6241283615664827e-05,
+      "loss": 16.8649,
+      "step": 15249
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6235456977415757e-05,
+      "loss": 0.7856,
+      "step": 15250
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.622963118193741e-05,
+      "loss": 17.8688,
+      "step": 15251
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6223806229375182e-05,
+      "loss": 0.7738,
+      "step": 15252
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6217982119874458e-05,
+      "loss": 0.768,
+      "step": 15253
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6212158853580616e-05,
+      "loss": 0.7237,
+      "step": 15254
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.620633643063899e-05,
+      "loss": 0.7899,
+      "step": 15255
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6200514851194935e-05,
+      "loss": 16.624,
+      "step": 15256
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6194694115393748e-05,
+      "loss": 15.6828,
+      "step": 15257
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.61888742233807e-05,
+      "loss": 0.7541,
+      "step": 15258
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6183055175301086e-05,
+      "loss": 0.7552,
+      "step": 15259
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.617723697130014e-05,
+      "loss": 16.2336,
+      "step": 15260
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6171419611523074e-05,
+      "loss": 16.7213,
+      "step": 15261
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6165603096115107e-05,
+      "loss": 18.5129,
+      "step": 15262
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.615978742522139e-05,
+      "loss": 0.742,
+      "step": 15263
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6153972598987126e-05,
+      "loss": 17.7063,
+      "step": 15264
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6148158617557435e-05,
+      "loss": 15.6335,
+      "step": 15265
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6142345481077413e-05,
+      "loss": 0.7102,
+      "step": 15266
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.613653318969221e-05,
+      "loss": 18.1794,
+      "step": 15267
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6130721743546845e-05,
+      "loss": 0.7041,
+      "step": 15268
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.612491114278641e-05,
+      "loss": 0.7706,
+      "step": 15269
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.611910138755593e-05,
+      "loss": 0.7443,
+      "step": 15270
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6113292478000396e-05,
+      "loss": 0.7175,
+      "step": 15271
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.610748441426484e-05,
+      "loss": 0.7671,
+      "step": 15272
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.610167719649419e-05,
+      "loss": 0.7535,
+      "step": 15273
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6095870824833455e-05,
+      "loss": 0.7934,
+      "step": 15274
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6090065299427503e-05,
+      "loss": 19.2035,
+      "step": 15275
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6084260620421254e-05,
+      "loss": 16.2433,
+      "step": 15276
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6078456787959618e-05,
+      "loss": 0.7649,
+      "step": 15277
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6072653802187426e-05,
+      "loss": 18.2032,
+      "step": 15278
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.606685166324956e-05,
+      "loss": 0.7901,
+      "step": 15279
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6061050371290825e-05,
+      "loss": 18.0823,
+      "step": 15280
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6055249926456024e-05,
+      "loss": 16.4261,
+      "step": 15281
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.604945032888993e-05,
+      "loss": 0.7249,
+      "step": 15282
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6043651578737297e-05,
+      "loss": 0.7944,
+      "step": 15283
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6037853676142887e-05,
+      "loss": 0.7134,
+      "step": 15284
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6032056621251407e-05,
+      "loss": 0.7499,
+      "step": 15285
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6026260414207527e-05,
+      "loss": 17.6588,
+      "step": 15286
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6020465055155988e-05,
+      "loss": 18.4893,
+      "step": 15287
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6014670544241354e-05,
+      "loss": 0.7699,
+      "step": 15288
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.600887688160833e-05,
+      "loss": 18.1805,
+      "step": 15289
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.6003084067401497e-05,
+      "loss": 0.6707,
+      "step": 15290
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5997292101765427e-05,
+      "loss": 15.0991,
+      "step": 15291
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.599150098484473e-05,
+      "loss": 18.7858,
+      "step": 15292
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5985710716783935e-05,
+      "loss": 0.748,
+      "step": 15293
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5979921297727568e-05,
+      "loss": 17.8989,
+      "step": 15294
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.597413272782013e-05,
+      "loss": 0.7387,
+      "step": 15295
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5968345007206093e-05,
+      "loss": 15.7361,
+      "step": 15296
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5962558136029948e-05,
+      "loss": 0.7481,
+      "step": 15297
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.595677211443613e-05,
+      "loss": 15.4983,
+      "step": 15298
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.595098694256903e-05,
+      "loss": 0.7628,
+      "step": 15299
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.594520262057309e-05,
+      "loss": 15.8254,
+      "step": 15300
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5939419148592673e-05,
+      "loss": 0.7361,
+      "step": 15301
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5933636526772122e-05,
+      "loss": 0.7668,
+      "step": 15302
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.592785475525579e-05,
+      "loss": 0.7657,
+      "step": 15303
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.592207383418796e-05,
+      "loss": 19.3437,
+      "step": 15304
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.591629376371297e-05,
+      "loss": 0.7095,
+      "step": 15305
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5910514543975046e-05,
+      "loss": 19.4099,
+      "step": 15306
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.590473617511849e-05,
+      "loss": 0.7152,
+      "step": 15307
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5898958657287494e-05,
+      "loss": 18.1853,
+      "step": 15308
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5893181990626282e-05,
+      "loss": 0.7587,
+      "step": 15309
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5887406175279034e-05,
+      "loss": 17.7465,
+      "step": 15310
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5881631211389898e-05,
+      "loss": 0.7843,
+      "step": 15311
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5875857099103052e-05,
+      "loss": 0.7619,
+      "step": 15312
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.587008383856261e-05,
+      "loss": 0.7863,
+      "step": 15313
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.586431142991265e-05,
+      "loss": 17.6843,
+      "step": 15314
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5858539873297302e-05,
+      "loss": 0.7179,
+      "step": 15315
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5852769168860565e-05,
+      "loss": 0.7667,
+      "step": 15316
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.584699931674652e-05,
+      "loss": 0.8051,
+      "step": 15317
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.584123031709917e-05,
+      "loss": 16.8359,
+      "step": 15318
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.583546217006249e-05,
+      "loss": 0.75,
+      "step": 15319
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5829694875780506e-05,
+      "loss": 0.7193,
+      "step": 15320
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.582392843439713e-05,
+      "loss": 16.2971,
+      "step": 15321
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.58181628460563e-05,
+      "loss": 17.6936,
+      "step": 15322
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.581239811090193e-05,
+      "loss": 18.2728,
+      "step": 15323
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5806634229077897e-05,
+      "loss": 0.7345,
+      "step": 15324
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.58008712007281e-05,
+      "loss": 20.1203,
+      "step": 15325
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5795109025996365e-05,
+      "loss": 0.7422,
+      "step": 15326
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.57893477050265e-05,
+      "loss": 0.7613,
+      "step": 15327
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5783587237962343e-05,
+      "loss": 17.1185,
+      "step": 15328
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5777827624947665e-05,
+      "loss": 18.3312,
+      "step": 15329
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5772068866126222e-05,
+      "loss": 0.7769,
+      "step": 15330
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5766310961641756e-05,
+      "loss": 0.7548,
+      "step": 15331
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.576055391163797e-05,
+      "loss": 0.7682,
+      "step": 15332
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.575479771625859e-05,
+      "loss": 0.7868,
+      "step": 15333
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.574904237564726e-05,
+      "loss": 18.2337,
+      "step": 15334
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.574328788994767e-05,
+      "loss": 0.7036,
+      "step": 15335
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.573753425930344e-05,
+      "loss": 18.9461,
+      "step": 15336
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5731781483858178e-05,
+      "loss": 0.7544,
+      "step": 15337
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5726029563755467e-05,
+      "loss": 0.7568,
+      "step": 15338
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5720278499138873e-05,
+      "loss": 16.845,
+      "step": 15339
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.571452829015197e-05,
+      "loss": 17.6778,
+      "step": 15340
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5708778936938272e-05,
+      "loss": 17.6404,
+      "step": 15341
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5703030439641264e-05,
+      "loss": 18.6719,
+      "step": 15342
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5697282798404478e-05,
+      "loss": 15.2359,
+      "step": 15343
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5691536013371312e-05,
+      "loss": 0.7647,
+      "step": 15344
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.568579008468526e-05,
+      "loss": 0.7618,
+      "step": 15345
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.568004501248972e-05,
+      "loss": 17.9325,
+      "step": 15346
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5674300796928075e-05,
+      "loss": 0.7474,
+      "step": 15347
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5668557438143743e-05,
+      "loss": 17.1165,
+      "step": 15348
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5662814936280052e-05,
+      "loss": 0.7905,
+      "step": 15349
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.565707329148034e-05,
+      "loss": 14.8337,
+      "step": 15350
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.565133250388792e-05,
+      "loss": 0.7006,
+      "step": 15351
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5645592573646067e-05,
+      "loss": 17.5585,
+      "step": 15352
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5639853500898087e-05,
+      "loss": 17.9399,
+      "step": 15353
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5634115285787203e-05,
+      "loss": 16.421,
+      "step": 15354
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.562837792845664e-05,
+      "loss": 0.6839,
+      "step": 15355
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5622641429049622e-05,
+      "loss": 19.2455,
+      "step": 15356
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5616905787709324e-05,
+      "loss": 17.5031,
+      "step": 15357
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5611171004578912e-05,
+      "loss": 0.7283,
+      "step": 15358
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.560543707980152e-05,
+      "loss": 0.7275,
+      "step": 15359
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5599704013520255e-05,
+      "loss": 15.2717,
+      "step": 15360
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.559397180587825e-05,
+      "loss": 0.8312,
+      "step": 15361
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5588240457018548e-05,
+      "loss": 16.5803,
+      "step": 15362
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.558250996708426e-05,
+      "loss": 0.7107,
+      "step": 15363
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.557678033621834e-05,
+      "loss": 18.8589,
+      "step": 15364
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5571051564563865e-05,
+      "loss": 17.6741,
+      "step": 15365
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5565323652263804e-05,
+      "loss": 17.557,
+      "step": 15366
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5559596599461105e-05,
+      "loss": 0.7571,
+      "step": 15367
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.555387040629876e-05,
+      "loss": 0.7924,
+      "step": 15368
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5548145072919675e-05,
+      "loss": 0.7443,
+      "step": 15369
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5542420599466752e-05,
+      "loss": 18.2198,
+      "step": 15370
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5536696986082878e-05,
+      "loss": 0.7642,
+      "step": 15371
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5530974232910906e-05,
+      "loss": 18.0379,
+      "step": 15372
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.55252523400937e-05,
+      "loss": 15.8937,
+      "step": 15373
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5519531307774064e-05,
+      "loss": 0.772,
+      "step": 15374
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5513811136094787e-05,
+      "loss": 0.7674,
+      "step": 15375
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5508091825198674e-05,
+      "loss": 17.7174,
+      "step": 15376
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5502373375228462e-05,
+      "loss": 0.7643,
+      "step": 15377
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5496655786326893e-05,
+      "loss": 0.6932,
+      "step": 15378
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5490939058636667e-05,
+      "loss": 17.5487,
+      "step": 15379
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.548522319230047e-05,
+      "loss": 18.0098,
+      "step": 15380
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5479508187461e-05,
+      "loss": 0.8138,
+      "step": 15381
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.547379404426088e-05,
+      "loss": 0.7921,
+      "step": 15382
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5468080762842735e-05,
+      "loss": 18.9837,
+      "step": 15383
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.546236834334921e-05,
+      "loss": 18.5055,
+      "step": 15384
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5456656785922824e-05,
+      "loss": 0.8411,
+      "step": 15385
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.545094609070619e-05,
+      "loss": 0.7476,
+      "step": 15386
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5445236257841828e-05,
+      "loss": 18.0402,
+      "step": 15387
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5439527287472243e-05,
+      "loss": 17.6693,
+      "step": 15388
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5433819179739968e-05,
+      "loss": 17.4516,
+      "step": 15389
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5428111934787437e-05,
+      "loss": 18.3276,
+      "step": 15390
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5422405552757157e-05,
+      "loss": 0.7512,
+      "step": 15391
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5416700033791503e-05,
+      "loss": 18.6465,
+      "step": 15392
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5410995378032927e-05,
+      "loss": 16.3947,
+      "step": 15393
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5405291585623806e-05,
+      "loss": 16.9806,
+      "step": 15394
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.539958865670648e-05,
+      "loss": 0.8035,
+      "step": 15395
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.539388659142335e-05,
+      "loss": 0.7703,
+      "step": 15396
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5388185389916705e-05,
+      "loss": 0.7252,
+      "step": 15397
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.538248505232886e-05,
+      "loss": 0.7155,
+      "step": 15398
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.537678557880209e-05,
+      "loss": 0.7367,
+      "step": 15399
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.537108696947864e-05,
+      "loss": 0.7881,
+      "step": 15400
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.536538922450078e-05,
+      "loss": 0.7725,
+      "step": 15401
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5359692344010718e-05,
+      "loss": 0.7477,
+      "step": 15402
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5353996328150628e-05,
+      "loss": 19.4052,
+      "step": 15403
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5348301177062713e-05,
+      "loss": 17.7039,
+      "step": 15404
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5342606890889116e-05,
+      "loss": 16.2019,
+      "step": 15405
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.533691346977196e-05,
+      "loss": 16.1011,
+      "step": 15406
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.533122091385337e-05,
+      "loss": 0.7866,
+      "step": 15407
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.532552922327539e-05,
+      "loss": 19.334,
+      "step": 15408
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5319838398180142e-05,
+      "loss": 18.1286,
+      "step": 15409
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5314148438709647e-05,
+      "loss": 0.7763,
+      "step": 15410
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.530845934500591e-05,
+      "loss": 16.8125,
+      "step": 15411
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.530277111721098e-05,
+      "loss": 17.0497,
+      "step": 15412
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5297083755466775e-05,
+      "loss": 17.6506,
+      "step": 15413
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5291397259915298e-05,
+      "loss": 0.7702,
+      "step": 15414
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5285711630698468e-05,
+      "loss": 0.8222,
+      "step": 15415
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5280026867958185e-05,
+      "loss": 17.4474,
+      "step": 15416
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.527434297183638e-05,
+      "loss": 0.7855,
+      "step": 15417
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5268659942474885e-05,
+      "loss": 17.0126,
+      "step": 15418
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5262977780015602e-05,
+      "loss": 0.7863,
+      "step": 15419
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5257296484600291e-05,
+      "loss": 0.7812,
+      "step": 15420
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.525161605637081e-05,
+      "loss": 15.3859,
+      "step": 15421
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5245936495468922e-05,
+      "loss": 16.3614,
+      "step": 15422
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5240257802036385e-05,
+      "loss": 16.222,
+      "step": 15423
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5234579976214957e-05,
+      "loss": 17.4784,
+      "step": 15424
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5228903018146357e-05,
+      "loss": 17.0778,
+      "step": 15425
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5223226927972278e-05,
+      "loss": 16.7206,
+      "step": 15426
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5217551705834394e-05,
+      "loss": 0.7506,
+      "step": 15427
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5211877351874344e-05,
+      "loss": 17.5797,
+      "step": 15428
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5206203866233798e-05,
+      "loss": 0.7179,
+      "step": 15429
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5200531249054345e-05,
+      "loss": 0.8479,
+      "step": 15430
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5194859500477559e-05,
+      "loss": 17.5652,
+      "step": 15431
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5189188620645051e-05,
+      "loss": 18.2928,
+      "step": 15432
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5183518609698338e-05,
+      "loss": 0.7324,
+      "step": 15433
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5177849467778949e-05,
+      "loss": 0.7043,
+      "step": 15434
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.517218119502839e-05,
+      "loss": 0.7638,
+      "step": 15435
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5166513791588122e-05,
+      "loss": 17.0688,
+      "step": 15436
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5160847257599642e-05,
+      "loss": 0.7977,
+      "step": 15437
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5155181593204371e-05,
+      "loss": 0.7297,
+      "step": 15438
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5149516798543723e-05,
+      "loss": 18.0403,
+      "step": 15439
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.514385287375909e-05,
+      "loss": 0.7562,
+      "step": 15440
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5138189818991832e-05,
+      "loss": 0.7431,
+      "step": 15441
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5132527634383337e-05,
+      "loss": 16.9327,
+      "step": 15442
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5126866320074905e-05,
+      "loss": 18.348,
+      "step": 15443
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5121205876207844e-05,
+      "loss": 18.9309,
+      "step": 15444
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5115546302923462e-05,
+      "loss": 16.7286,
+      "step": 15445
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.510988760036301e-05,
+      "loss": 17.9564,
+      "step": 15446
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5104229768667727e-05,
+      "loss": 18.3849,
+      "step": 15447
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5098572807978823e-05,
+      "loss": 19.3009,
+      "step": 15448
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5092916718437528e-05,
+      "loss": 16.65,
+      "step": 15449
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5087261500185002e-05,
+      "loss": 19.7165,
+      "step": 15450
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5081607153362382e-05,
+      "loss": 15.7132,
+      "step": 15451
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5075953678110838e-05,
+      "loss": 17.4053,
+      "step": 15452
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5070301074571464e-05,
+      "loss": 0.8109,
+      "step": 15453
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5064649342885344e-05,
+      "loss": 0.708,
+      "step": 15454
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5058998483193566e-05,
+      "loss": 15.9717,
+      "step": 15455
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5053348495637138e-05,
+      "loss": 17.8422,
+      "step": 15456
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5047699380357133e-05,
+      "loss": 0.778,
+      "step": 15457
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5042051137494529e-05,
+      "loss": 18.7171,
+      "step": 15458
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5036403767190294e-05,
+      "loss": 0.7387,
+      "step": 15459
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5030757269585438e-05,
+      "loss": 15.4287,
+      "step": 15460
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5025111644820832e-05,
+      "loss": 0.7681,
+      "step": 15461
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5019466893037442e-05,
+      "loss": 0.7795,
+      "step": 15462
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5013823014376144e-05,
+      "loss": 17.5216,
+      "step": 15463
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5008180008977795e-05,
+      "loss": 18.8966,
+      "step": 15464
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.500253787698328e-05,
+      "loss": 0.7773,
+      "step": 15465
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4996896618533412e-05,
+      "loss": 15.9746,
+      "step": 15466
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4991256233768997e-05,
+      "loss": 0.8286,
+      "step": 15467
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.498561672283082e-05,
+      "loss": 17.0319,
+      "step": 15468
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4979978085859626e-05,
+      "loss": 18.0997,
+      "step": 15469
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4974340322996195e-05,
+      "loss": 0.6965,
+      "step": 15470
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4968703434381204e-05,
+      "loss": 15.7183,
+      "step": 15471
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4963067420155397e-05,
+      "loss": 17.0878,
+      "step": 15472
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4957432280459422e-05,
+      "loss": 17.5305,
+      "step": 15473
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4951798015433942e-05,
+      "loss": 17.6104,
+      "step": 15474
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4946164625219583e-05,
+      "loss": 16.1929,
+      "step": 15475
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4940532109956945e-05,
+      "loss": 0.7914,
+      "step": 15476
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.493490046978665e-05,
+      "loss": 18.2637,
+      "step": 15477
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4929269704849241e-05,
+      "loss": 18.0989,
+      "step": 15478
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.492363981528525e-05,
+      "loss": 0.7654,
+      "step": 15479
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4918010801235239e-05,
+      "loss": 17.6307,
+      "step": 15480
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4912382662839686e-05,
+      "loss": 19.6752,
+      "step": 15481
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4906755400239076e-05,
+      "loss": 20.0425,
+      "step": 15482
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4901129013573862e-05,
+      "loss": 17.4822,
+      "step": 15483
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4895503502984464e-05,
+      "loss": 18.1828,
+      "step": 15484
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.488987886861133e-05,
+      "loss": 0.7927,
+      "step": 15485
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4884255110594842e-05,
+      "loss": 0.7515,
+      "step": 15486
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.487863222907534e-05,
+      "loss": 0.7075,
+      "step": 15487
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4873010224193235e-05,
+      "loss": 16.6655,
+      "step": 15488
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4867389096088779e-05,
+      "loss": 0.8427,
+      "step": 15489
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4861768844902336e-05,
+      "loss": 0.7555,
+      "step": 15490
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4856149470774162e-05,
+      "loss": 0.7649,
+      "step": 15491
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4850530973844501e-05,
+      "loss": 0.748,
+      "step": 15492
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.484491335425363e-05,
+      "loss": 16.555,
+      "step": 15493
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4839296612141757e-05,
+      "loss": 0.7202,
+      "step": 15494
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4833680747649059e-05,
+      "loss": 0.824,
+      "step": 15495
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4828065760915726e-05,
+      "loss": 19.5071,
+      "step": 15496
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.482245165208188e-05,
+      "loss": 0.7423,
+      "step": 15497
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4816838421287694e-05,
+      "loss": 19.6587,
+      "step": 15498
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4811226068673234e-05,
+      "loss": 0.7592,
+      "step": 15499
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4805614594378626e-05,
+      "loss": 17.32,
+      "step": 15500
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4800003998543915e-05,
+      "loss": 17.0602,
+      "step": 15501
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4794394281309137e-05,
+      "loss": 17.1849,
+      "step": 15502
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4788785442814317e-05,
+      "loss": 0.7114,
+      "step": 15503
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.478317748319944e-05,
+      "loss": 0.811,
+      "step": 15504
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4777570402604501e-05,
+      "loss": 0.7529,
+      "step": 15505
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4771964201169458e-05,
+      "loss": 18.5453,
+      "step": 15506
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.476635887903421e-05,
+      "loss": 17.4362,
+      "step": 15507
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4760754436338703e-05,
+      "loss": 19.1961,
+      "step": 15508
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4755150873222817e-05,
+      "loss": 17.5807,
+      "step": 15509
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4749548189826412e-05,
+      "loss": 0.7495,
+      "step": 15510
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4743946386289332e-05,
+      "loss": 16.9992,
+      "step": 15511
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4738345462751379e-05,
+      "loss": 0.7971,
+      "step": 15512
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4732745419352395e-05,
+      "loss": 17.1056,
+      "step": 15513
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4727146256232138e-05,
+      "loss": 0.8406,
+      "step": 15514
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.472154797353037e-05,
+      "loss": 18.9621,
+      "step": 15515
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4715950571386811e-05,
+      "loss": 0.7637,
+      "step": 15516
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4710354049941172e-05,
+      "loss": 17.5099,
+      "step": 15517
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4704758409333163e-05,
+      "loss": 0.7559,
+      "step": 15518
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4699163649702447e-05,
+      "loss": 0.7014,
+      "step": 15519
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4693569771188653e-05,
+      "loss": 19.0266,
+      "step": 15520
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4687976773931427e-05,
+      "loss": 0.7312,
+      "step": 15521
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.468238465807037e-05,
+      "loss": 0.6915,
+      "step": 15522
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4676793423745056e-05,
+      "loss": 0.7718,
+      "step": 15523
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.467120307109504e-05,
+      "loss": 17.0863,
+      "step": 15524
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4665613600259854e-05,
+      "loss": 16.6884,
+      "step": 15525
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.466002501137903e-05,
+      "loss": 16.9805,
+      "step": 15526
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4654437304592034e-05,
+      "loss": 0.7546,
+      "step": 15527
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4648850480038374e-05,
+      "loss": 0.7811,
+      "step": 15528
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4643264537857488e-05,
+      "loss": 15.6948,
+      "step": 15529
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4637679478188759e-05,
+      "loss": 0.7778,
+      "step": 15530
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4632095301171644e-05,
+      "loss": 0.7629,
+      "step": 15531
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4626512006945491e-05,
+      "loss": 0.7371,
+      "step": 15532
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4620929595649691e-05,
+      "loss": 0.7589,
+      "step": 15533
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4615348067423567e-05,
+      "loss": 0.7124,
+      "step": 15534
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4609767422406417e-05,
+      "loss": 0.7502,
+      "step": 15535
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4604187660737584e-05,
+      "loss": 0.6892,
+      "step": 15536
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4598608782556277e-05,
+      "loss": 17.5402,
+      "step": 15537
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4593030788001794e-05,
+      "loss": 18.7025,
+      "step": 15538
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4587453677213348e-05,
+      "loss": 0.7613,
+      "step": 15539
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4581877450330129e-05,
+      "loss": 0.837,
+      "step": 15540
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4576302107491353e-05,
+      "loss": 15.5461,
+      "step": 15541
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4570727648836158e-05,
+      "loss": 16.7585,
+      "step": 15542
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4565154074503695e-05,
+      "loss": 0.7342,
+      "step": 15543
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.455958138463308e-05,
+      "loss": 0.7371,
+      "step": 15544
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4554009579363386e-05,
+      "loss": 0.7452,
+      "step": 15545
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4548438658833724e-05,
+      "loss": 0.7214,
+      "step": 15546
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4542868623183125e-05,
+      "loss": 0.7697,
+      "step": 15547
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4537299472550609e-05,
+      "loss": 0.7723,
+      "step": 15548
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4531731207075205e-05,
+      "loss": 16.1113,
+      "step": 15549
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4526163826895895e-05,
+      "loss": 16.6974,
+      "step": 15550
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.452059733215163e-05,
+      "loss": 0.7722,
+      "step": 15551
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.451503172298136e-05,
+      "loss": 0.8011,
+      "step": 15552
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4509466999523985e-05,
+      "loss": 0.7464,
+      "step": 15553
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.450390316191843e-05,
+      "loss": 15.726,
+      "step": 15554
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4498340210303547e-05,
+      "loss": 16.3954,
+      "step": 15555
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4492778144818208e-05,
+      "loss": 0.7502,
+      "step": 15556
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4487216965601252e-05,
+      "loss": 0.7323,
+      "step": 15557
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4481656672791438e-05,
+      "loss": 0.7346,
+      "step": 15558
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.44760972665276e-05,
+      "loss": 17.7046,
+      "step": 15559
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.447053874694847e-05,
+      "loss": 0.7449,
+      "step": 15560
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4464981114192822e-05,
+      "loss": 0.7423,
+      "step": 15561
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4459424368399355e-05,
+      "loss": 17.0839,
+      "step": 15562
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4453868509706758e-05,
+      "loss": 17.3789,
+      "step": 15563
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.444831353825375e-05,
+      "loss": 16.4702,
+      "step": 15564
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4442759454178917e-05,
+      "loss": 0.7299,
+      "step": 15565
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.443720625762095e-05,
+      "loss": 18.5564,
+      "step": 15566
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4431653948718432e-05,
+      "loss": 18.2567,
+      "step": 15567
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4426102527609936e-05,
+      "loss": 0.7877,
+      "step": 15568
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4420551994434061e-05,
+      "loss": 0.7818,
+      "step": 15569
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4415002349329332e-05,
+      "loss": 0.7578,
+      "step": 15570
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4409453592434269e-05,
+      "loss": 16.1321,
+      "step": 15571
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4403905723887367e-05,
+      "loss": 0.7357,
+      "step": 15572
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4398358743827095e-05,
+      "loss": 17.2835,
+      "step": 15573
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4392812652391935e-05,
+      "loss": 17.92,
+      "step": 15574
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4387267449720293e-05,
+      "loss": 17.2384,
+      "step": 15575
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4381723135950582e-05,
+      "loss": 0.746,
+      "step": 15576
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.43761797112212e-05,
+      "loss": 0.7377,
+      "step": 15577
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4370637175670515e-05,
+      "loss": 17.0434,
+      "step": 15578
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4365095529436856e-05,
+      "loss": 17.3994,
+      "step": 15579
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4359554772658552e-05,
+      "loss": 0.827,
+      "step": 15580
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4354014905473879e-05,
+      "loss": 16.6739,
+      "step": 15581
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4348475928021149e-05,
+      "loss": 17.6392,
+      "step": 15582
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4342937840438586e-05,
+      "loss": 0.6997,
+      "step": 15583
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4337400642864452e-05,
+      "loss": 0.7326,
+      "step": 15584
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4331864335436957e-05,
+      "loss": 0.797,
+      "step": 15585
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4326328918294241e-05,
+      "loss": 0.8139,
+      "step": 15586
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4320794391574516e-05,
+      "loss": 0.7499,
+      "step": 15587
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4315260755415894e-05,
+      "loss": 17.9181,
+      "step": 15588
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4309728009956524e-05,
+      "loss": 18.3306,
+      "step": 15589
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4304196155334499e-05,
+      "loss": 0.8261,
+      "step": 15590
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4298665191687883e-05,
+      "loss": 0.7571,
+      "step": 15591
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4293135119154733e-05,
+      "loss": 16.5279,
+      "step": 15592
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4287605937873072e-05,
+      "loss": 0.7491,
+      "step": 15593
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4282077647980935e-05,
+      "loss": 0.7732,
+      "step": 15594
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.42765502496163e-05,
+      "loss": 17.3904,
+      "step": 15595
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4271023742917106e-05,
+      "loss": 17.4557,
+      "step": 15596
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4265498128021338e-05,
+      "loss": 19.0318,
+      "step": 15597
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.425997340506689e-05,
+      "loss": 0.824,
+      "step": 15598
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4254449574191669e-05,
+      "loss": 0.7019,
+      "step": 15599
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.424892663553355e-05,
+      "loss": 0.7425,
+      "step": 15600
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.424340458923037e-05,
+      "loss": 0.7825,
+      "step": 15601
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.423788343541999e-05,
+      "loss": 0.7305,
+      "step": 15602
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4232363174240205e-05,
+      "loss": 18.7197,
+      "step": 15603
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4226843805828782e-05,
+      "loss": 17.6632,
+      "step": 15604
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4221325330323543e-05,
+      "loss": 0.7649,
+      "step": 15605
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4215807747862158e-05,
+      "loss": 0.7573,
+      "step": 15606
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4210291058582403e-05,
+      "loss": 16.3143,
+      "step": 15607
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.420477526262195e-05,
+      "loss": 20.5469,
+      "step": 15608
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4199260360118465e-05,
+      "loss": 0.7511,
+      "step": 15609
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4193746351209635e-05,
+      "loss": 0.7603,
+      "step": 15610
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.418823323603305e-05,
+      "loss": 0.7784,
+      "step": 15611
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4182721014726374e-05,
+      "loss": 16.4528,
+      "step": 15612
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4177209687427146e-05,
+      "loss": 0.8111,
+      "step": 15613
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.417169925427293e-05,
+      "loss": 19.7582,
+      "step": 15614
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.416618971540129e-05,
+      "loss": 0.7307,
+      "step": 15615
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4160681070949727e-05,
+      "loss": 0.6963,
+      "step": 15616
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4155173321055759e-05,
+      "loss": 18.499,
+      "step": 15617
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4149666465856853e-05,
+      "loss": 17.5064,
+      "step": 15618
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4144160505490455e-05,
+      "loss": 17.3387,
+      "step": 15619
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4138655440094e-05,
+      "loss": 0.729,
+      "step": 15620
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4133151269804873e-05,
+      "loss": 0.7131,
+      "step": 15621
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4127647994760501e-05,
+      "loss": 17.6975,
+      "step": 15622
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4122145615098226e-05,
+      "loss": 18.5218,
+      "step": 15623
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.411664413095537e-05,
+      "loss": 0.7351,
+      "step": 15624
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4111143542469285e-05,
+      "loss": 0.7553,
+      "step": 15625
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.410564384977725e-05,
+      "loss": 0.7668,
+      "step": 15626
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4100145053016545e-05,
+      "loss": 16.77,
+      "step": 15627
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4094647152324414e-05,
+      "loss": 0.7335,
+      "step": 15628
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.408915014783807e-05,
+      "loss": 17.515,
+      "step": 15629
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.408365403969476e-05,
+      "loss": 0.737,
+      "step": 15630
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.407815882803164e-05,
+      "loss": 16.6674,
+      "step": 15631
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4072664512985867e-05,
+      "loss": 17.8945,
+      "step": 15632
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4067171094694615e-05,
+      "loss": 0.7938,
+      "step": 15633
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4061678573294951e-05,
+      "loss": 19.278,
+      "step": 15634
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.405618694892401e-05,
+      "loss": 0.7422,
+      "step": 15635
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4050696221718851e-05,
+      "loss": 0.7092,
+      "step": 15636
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4045206391816507e-05,
+      "loss": 18.7612,
+      "step": 15637
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4039717459354035e-05,
+      "loss": 18.8073,
+      "step": 15638
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4034229424468415e-05,
+      "loss": 17.7022,
+      "step": 15639
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4028742287296664e-05,
+      "loss": 16.7361,
+      "step": 15640
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4023256047975709e-05,
+      "loss": 18.6568,
+      "step": 15641
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.401777070664248e-05,
+      "loss": 0.7259,
+      "step": 15642
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4012286263433928e-05,
+      "loss": 0.7075,
+      "step": 15643
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.400680271848691e-05,
+      "loss": 16.657,
+      "step": 15644
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.400132007193833e-05,
+      "loss": 0.7411,
+      "step": 15645
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.399583832392502e-05,
+      "loss": 0.7271,
+      "step": 15646
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.399035747458381e-05,
+      "loss": 0.742,
+      "step": 15647
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3984877524051493e-05,
+      "loss": 0.7747,
+      "step": 15648
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3979398472464844e-05,
+      "loss": 17.9219,
+      "step": 15649
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3973920319960655e-05,
+      "loss": 17.3032,
+      "step": 15650
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3968443066675634e-05,
+      "loss": 0.7676,
+      "step": 15651
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3962966712746484e-05,
+      "loss": 18.1556,
+      "step": 15652
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3957491258309934e-05,
+      "loss": 16.2503,
+      "step": 15653
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.395201670350263e-05,
+      "loss": 0.7992,
+      "step": 15654
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3946543048461218e-05,
+      "loss": 0.7141,
+      "step": 15655
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3941070293322323e-05,
+      "loss": 0.7745,
+      "step": 15656
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3935598438222536e-05,
+      "loss": 15.0025,
+      "step": 15657
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3930127483298456e-05,
+      "loss": 0.7382,
+      "step": 15658
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3924657428686632e-05,
+      "loss": 0.7539,
+      "step": 15659
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3919188274523588e-05,
+      "loss": 0.7248,
+      "step": 15660
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3913720020945864e-05,
+      "loss": 17.2957,
+      "step": 15661
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3908252668089899e-05,
+      "loss": 18.5472,
+      "step": 15662
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3902786216092206e-05,
+      "loss": 0.7286,
+      "step": 15663
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.389732066508921e-05,
+      "loss": 17.7965,
+      "step": 15664
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3891856015217319e-05,
+      "loss": 16.4099,
+      "step": 15665
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3886392266612957e-05,
+      "loss": 0.7504,
+      "step": 15666
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3880929419412498e-05,
+      "loss": 0.7638,
+      "step": 15667
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.387546747375228e-05,
+      "loss": 18.3715,
+      "step": 15668
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3870006429768645e-05,
+      "loss": 16.9399,
+      "step": 15669
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3864546287597886e-05,
+      "loss": 0.7789,
+      "step": 15670
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3859087047376318e-05,
+      "loss": 16.5835,
+      "step": 15671
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3853628709240169e-05,
+      "loss": 0.7433,
+      "step": 15672
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3848171273325722e-05,
+      "loss": 18.9554,
+      "step": 15673
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3842714739769186e-05,
+      "loss": 17.3528,
+      "step": 15674
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3837259108706713e-05,
+      "loss": 19.8813,
+      "step": 15675
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3831804380274532e-05,
+      "loss": 0.74,
+      "step": 15676
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3826350554608753e-05,
+      "loss": 0.7716,
+      "step": 15677
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3820897631845537e-05,
+      "loss": 17.8921,
+      "step": 15678
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3815445612120981e-05,
+      "loss": 17.3955,
+      "step": 15679
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3809994495571144e-05,
+      "loss": 15.6304,
+      "step": 15680
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3804544282332143e-05,
+      "loss": 17.2294,
+      "step": 15681
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3799094972539945e-05,
+      "loss": 17.8116,
+      "step": 15682
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.379364656633062e-05,
+      "loss": 0.788,
+      "step": 15683
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3788199063840147e-05,
+      "loss": 0.6919,
+      "step": 15684
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3782752465204479e-05,
+      "loss": 0.7775,
+      "step": 15685
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.377730677055959e-05,
+      "loss": 0.7595,
+      "step": 15686
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3771861980041395e-05,
+      "loss": 0.7078,
+      "step": 15687
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3766418093785799e-05,
+      "loss": 15.3945,
+      "step": 15688
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3760975111928681e-05,
+      "loss": 19.4306,
+      "step": 15689
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3755533034605883e-05,
+      "loss": 0.7359,
+      "step": 15690
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3750091861953268e-05,
+      "loss": 0.77,
+      "step": 15691
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.374465159410664e-05,
+      "loss": 18.531,
+      "step": 15692
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.373921223120177e-05,
+      "loss": 0.7565,
+      "step": 15693
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3733773773374459e-05,
+      "loss": 16.6611,
+      "step": 15694
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3728336220760436e-05,
+      "loss": 0.7632,
+      "step": 15695
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3722899573495418e-05,
+      "loss": 18.5006,
+      "step": 15696
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3717463831715116e-05,
+      "loss": 0.7466,
+      "step": 15697
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3712028995555182e-05,
+      "loss": 0.8293,
+      "step": 15698
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3706595065151306e-05,
+      "loss": 0.7474,
+      "step": 15699
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3701162040639087e-05,
+      "loss": 16.8216,
+      "step": 15700
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3695729922154171e-05,
+      "loss": 15.7954,
+      "step": 15701
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3690298709832134e-05,
+      "loss": 15.9903,
+      "step": 15702
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.368486840380851e-05,
+      "loss": 0.7529,
+      "step": 15703
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.367943900421887e-05,
+      "loss": 0.7442,
+      "step": 15704
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3674010511198714e-05,
+      "loss": 18.3877,
+      "step": 15705
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3668582924883567e-05,
+      "loss": 0.7133,
+      "step": 15706
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3663156245408882e-05,
+      "loss": 16.4679,
+      "step": 15707
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3657730472910102e-05,
+      "loss": 0.7586,
+      "step": 15708
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3652305607522692e-05,
+      "loss": 0.7924,
+      "step": 15709
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.364688164938201e-05,
+      "loss": 18.0809,
+      "step": 15710
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3641458598623474e-05,
+      "loss": 0.7424,
+      "step": 15711
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3636036455382434e-05,
+      "loss": 16.4856,
+      "step": 15712
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3630615219794213e-05,
+      "loss": 20.5174,
+      "step": 15713
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3625194891994152e-05,
+      "loss": 15.2189,
+      "step": 15714
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3619775472117535e-05,
+      "loss": 17.9819,
+      "step": 15715
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3614356960299629e-05,
+      "loss": 19.5491,
+      "step": 15716
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3608939356675677e-05,
+      "loss": 18.3368,
+      "step": 15717
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3603522661380896e-05,
+      "loss": 0.7752,
+      "step": 15718
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3598106874550514e-05,
+      "loss": 0.7137,
+      "step": 15719
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3592691996319695e-05,
+      "loss": 0.751,
+      "step": 15720
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3587278026823586e-05,
+      "loss": 0.713,
+      "step": 15721
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3581864966197345e-05,
+      "loss": 17.0486,
+      "step": 15722
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3576452814576068e-05,
+      "loss": 0.7343,
+      "step": 15723
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3571041572094845e-05,
+      "loss": 0.7619,
+      "step": 15724
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3565631238888744e-05,
+      "loss": 17.3253,
+      "step": 15725
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3560221815092788e-05,
+      "loss": 19.4328,
+      "step": 15726
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3554813300842034e-05,
+      "loss": 0.7233,
+      "step": 15727
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3549405696271445e-05,
+      "loss": 0.7366,
+      "step": 15728
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3543999001516028e-05,
+      "loss": 0.7462,
+      "step": 15729
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.353859321671072e-05,
+      "loss": 0.7619,
+      "step": 15730
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3533188341990455e-05,
+      "loss": 18.244,
+      "step": 15731
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3527784377490132e-05,
+      "loss": 18.7495,
+      "step": 15732
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3522381323344618e-05,
+      "loss": 0.7598,
+      "step": 15733
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3516979179688816e-05,
+      "loss": 18.9738,
+      "step": 15734
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3511577946657544e-05,
+      "loss": 0.7873,
+      "step": 15735
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3506177624385602e-05,
+      "loss": 16.4947,
+      "step": 15736
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.350077821300783e-05,
+      "loss": 18.7388,
+      "step": 15737
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3495379712658935e-05,
+      "loss": 0.7763,
+      "step": 15738
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3489982123473716e-05,
+      "loss": 17.323,
+      "step": 15739
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3484585445586873e-05,
+      "loss": 0.8017,
+      "step": 15740
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3479189679133109e-05,
+      "loss": 0.7644,
+      "step": 15741
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3473794824247115e-05,
+      "loss": 0.7541,
+      "step": 15742
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3468400881063548e-05,
+      "loss": 0.7478,
+      "step": 15743
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3463007849717036e-05,
+      "loss": 14.9177,
+      "step": 15744
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3457615730342188e-05,
+      "loss": 16.4626,
+      "step": 15745
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.345222452307358e-05,
+      "loss": 0.8258,
+      "step": 15746
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3446834228045812e-05,
+      "loss": 18.446,
+      "step": 15747
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3441444845393408e-05,
+      "loss": 0.742,
+      "step": 15748
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3436056375250866e-05,
+      "loss": 18.8454,
+      "step": 15749
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.343066881775274e-05,
+      "loss": 16.6957,
+      "step": 15750
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3425282173033443e-05,
+      "loss": 16.2945,
+      "step": 15751
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3419896441227464e-05,
+      "loss": 16.7954,
+      "step": 15752
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3414511622469228e-05,
+      "loss": 0.7401,
+      "step": 15753
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3409127716893117e-05,
+      "loss": 0.6717,
+      "step": 15754
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3403744724633543e-05,
+      "loss": 15.8456,
+      "step": 15755
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3398362645824846e-05,
+      "loss": 0.7411,
+      "step": 15756
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3392981480601407e-05,
+      "loss": 14.8143,
+      "step": 15757
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3387601229097479e-05,
+      "loss": 0.7449,
+      "step": 15758
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3382221891447394e-05,
+      "loss": 17.0794,
+      "step": 15759
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3376843467785415e-05,
+      "loss": 0.7644,
+      "step": 15760
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3371465958245771e-05,
+      "loss": 15.7593,
+      "step": 15761
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3366089362962714e-05,
+      "loss": 17.0502,
+      "step": 15762
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3360713682070431e-05,
+      "loss": 0.773,
+      "step": 15763
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3355338915703097e-05,
+      "loss": 16.8695,
+      "step": 15764
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3349965063994874e-05,
+      "loss": 18.4861,
+      "step": 15765
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3344592127079874e-05,
+      "loss": 0.7635,
+      "step": 15766
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3339220105092242e-05,
+      "loss": 17.8517,
+      "step": 15767
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3333848998166048e-05,
+      "loss": 0.7149,
+      "step": 15768
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3328478806435335e-05,
+      "loss": 18.3657,
+      "step": 15769
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3323109530034183e-05,
+      "loss": 16.7108,
+      "step": 15770
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3317741169096592e-05,
+      "loss": 15.619,
+      "step": 15771
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3312373723756556e-05,
+      "loss": 17.2586,
+      "step": 15772
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3307007194148046e-05,
+      "loss": 0.7623,
+      "step": 15773
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3301641580405006e-05,
+      "loss": 17.4927,
+      "step": 15774
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3296276882661378e-05,
+      "loss": 17.5328,
+      "step": 15775
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3290913101051067e-05,
+      "loss": 0.7333,
+      "step": 15776
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3285550235707928e-05,
+      "loss": 17.4217,
+      "step": 15777
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3280188286765872e-05,
+      "loss": 0.8247,
+      "step": 15778
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3274827254358662e-05,
+      "loss": 0.7185,
+      "step": 15779
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3269467138620173e-05,
+      "loss": 15.7406,
+      "step": 15780
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3264107939684167e-05,
+      "loss": 17.7686,
+      "step": 15781
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.32587496576844e-05,
+      "loss": 18.4362,
+      "step": 15782
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3253392292754647e-05,
+      "loss": 17.3609,
+      "step": 15783
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3248035845028595e-05,
+      "loss": 0.7586,
+      "step": 15784
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3242680314639993e-05,
+      "loss": 17.7157,
+      "step": 15785
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3237325701722452e-05,
+      "loss": 17.7791,
+      "step": 15786
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3231972006409671e-05,
+      "loss": 16.3816,
+      "step": 15787
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3226619228835258e-05,
+      "loss": 18.3135,
+      "step": 15788
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3221267369132818e-05,
+      "loss": 18.7322,
+      "step": 15789
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3215916427435953e-05,
+      "loss": 0.7534,
+      "step": 15790
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3210566403878217e-05,
+      "loss": 17.0978,
+      "step": 15791
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3205217298593143e-05,
+      "loss": 0.7658,
+      "step": 15792
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3199869111714241e-05,
+      "loss": 0.7386,
+      "step": 15793
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3194521843374996e-05,
+      "loss": 17.5343,
+      "step": 15794
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.318917549370891e-05,
+      "loss": 0.7482,
+      "step": 15795
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.31838300628494e-05,
+      "loss": 0.8021,
+      "step": 15796
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3178485550929892e-05,
+      "loss": 0.7369,
+      "step": 15797
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3173141958083801e-05,
+      "loss": 16.6457,
+      "step": 15798
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3167799284444499e-05,
+      "loss": 0.7674,
+      "step": 15799
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3162457530145334e-05,
+      "loss": 0.7723,
+      "step": 15800
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3157116695319638e-05,
+      "loss": 14.7072,
+      "step": 15801
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3151776780100711e-05,
+      "loss": 0.7907,
+      "step": 15802
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3146437784621862e-05,
+      "loss": 19.9155,
+      "step": 15803
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3141099709016341e-05,
+      "loss": 16.86,
+      "step": 15804
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3135762553417375e-05,
+      "loss": 0.7569,
+      "step": 15805
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3130426317958216e-05,
+      "loss": 17.7922,
+      "step": 15806
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3125091002772011e-05,
+      "loss": 0.7656,
+      "step": 15807
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3119756607991968e-05,
+      "loss": 19.0815,
+      "step": 15808
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.311442313375122e-05,
+      "loss": 16.3106,
+      "step": 15809
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3109090580182875e-05,
+      "loss": 16.7677,
+      "step": 15810
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.310375894742007e-05,
+      "loss": 0.8134,
+      "step": 15811
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.309842823559585e-05,
+      "loss": 18.0294,
+      "step": 15812
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3093098444843321e-05,
+      "loss": 16.2213,
+      "step": 15813
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3087769575295449e-05,
+      "loss": 0.7012,
+      "step": 15814
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.308244162708529e-05,
+      "loss": 0.7532,
+      "step": 15815
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3077114600345818e-05,
+      "loss": 15.9876,
+      "step": 15816
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3071788495209986e-05,
+      "loss": 0.7949,
+      "step": 15817
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3066463311810756e-05,
+      "loss": 0.7531,
+      "step": 15818
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3061139050281036e-05,
+      "loss": 0.7355,
+      "step": 15819
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3055815710753715e-05,
+      "loss": 0.7478,
+      "step": 15820
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3050493293361671e-05,
+      "loss": 0.7338,
+      "step": 15821
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3045171798237738e-05,
+      "loss": 16.7424,
+      "step": 15822
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3039851225514771e-05,
+      "loss": 16.8314,
+      "step": 15823
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3034531575325554e-05,
+      "loss": 0.8078,
+      "step": 15824
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3029212847802858e-05,
+      "loss": 19.5357,
+      "step": 15825
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3023895043079476e-05,
+      "loss": 16.5716,
+      "step": 15826
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3018578161288086e-05,
+      "loss": 0.7303,
+      "step": 15827
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3013262202561448e-05,
+      "loss": 17.3032,
+      "step": 15828
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.300794716703223e-05,
+      "loss": 0.7522,
+      "step": 15829
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.300263305483308e-05,
+      "loss": 17.6642,
+      "step": 15830
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2997319866096674e-05,
+      "loss": 0.7921,
+      "step": 15831
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2992007600955613e-05,
+      "loss": 0.7647,
+      "step": 15832
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.298669625954249e-05,
+      "loss": 17.3277,
+      "step": 15833
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2981385841989879e-05,
+      "loss": 0.7624,
+      "step": 15834
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2976076348430317e-05,
+      "loss": 16.3001,
+      "step": 15835
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2970767778996357e-05,
+      "loss": 15.5347,
+      "step": 15836
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.296546013382049e-05,
+      "loss": 0.6906,
+      "step": 15837
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2960153413035176e-05,
+      "loss": 16.2717,
+      "step": 15838
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2954847616772909e-05,
+      "loss": 17.0691,
+      "step": 15839
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2949542745166099e-05,
+      "loss": 0.778,
+      "step": 15840
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2944238798347163e-05,
+      "loss": 0.7256,
+      "step": 15841
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2938935776448474e-05,
+      "loss": 17.6674,
+      "step": 15842
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2933633679602424e-05,
+      "loss": 0.7862,
+      "step": 15843
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2928332507941342e-05,
+      "loss": 0.7944,
+      "step": 15844
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2923032261597534e-05,
+      "loss": 0.7956,
+      "step": 15845
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.291773294070332e-05,
+      "loss": 16.9289,
+      "step": 15846
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2912434545390955e-05,
+      "loss": 19.0678,
+      "step": 15847
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2907137075792691e-05,
+      "loss": 18.1432,
+      "step": 15848
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.290184053204076e-05,
+      "loss": 19.0123,
+      "step": 15849
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2896544914267339e-05,
+      "loss": 17.0851,
+      "step": 15850
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2891250222604651e-05,
+      "loss": 0.8221,
+      "step": 15851
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2885956457184822e-05,
+      "loss": 15.8575,
+      "step": 15852
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2880663618139987e-05,
+      "loss": 16.7556,
+      "step": 15853
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.287537170560229e-05,
+      "loss": 16.2194,
+      "step": 15854
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2870080719703759e-05,
+      "loss": 0.7124,
+      "step": 15855
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.286479066057651e-05,
+      "loss": 17.0374,
+      "step": 15856
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2859501528352563e-05,
+      "loss": 0.6876,
+      "step": 15857
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2854213323163927e-05,
+      "loss": 0.6831,
+      "step": 15858
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2848926045142617e-05,
+      "loss": 0.7557,
+      "step": 15859
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.28436396944206e-05,
+      "loss": 0.713,
+      "step": 15860
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2838354271129822e-05,
+      "loss": 0.7264,
+      "step": 15861
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2833069775402201e-05,
+      "loss": 0.7284,
+      "step": 15862
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2827786207369635e-05,
+      "loss": 17.7136,
+      "step": 15863
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2822503567164023e-05,
+      "loss": 16.5054,
+      "step": 15864
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.281722185491721e-05,
+      "loss": 0.7778,
+      "step": 15865
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.281194107076102e-05,
+      "loss": 16.1203,
+      "step": 15866
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2806661214827287e-05,
+      "loss": 0.7491,
+      "step": 15867
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2801382287247776e-05,
+      "loss": 0.7752,
+      "step": 15868
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.279610428815426e-05,
+      "loss": 16.8952,
+      "step": 15869
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2790827217678464e-05,
+      "loss": 17.9049,
+      "step": 15870
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2785551075952129e-05,
+      "loss": 0.7287,
+      "step": 15871
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2780275863106932e-05,
+      "loss": 0.8027,
+      "step": 15872
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2775001579274536e-05,
+      "loss": 19.2021,
+      "step": 15873
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.276972822458662e-05,
+      "loss": 18.7112,
+      "step": 15874
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2764455799174784e-05,
+      "loss": 16.751,
+      "step": 15875
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.275918430317063e-05,
+      "loss": 0.7306,
+      "step": 15876
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2753913736705741e-05,
+      "loss": 16.6202,
+      "step": 15877
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.274864409991166e-05,
+      "loss": 18.4664,
+      "step": 15878
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2743375392919937e-05,
+      "loss": 17.6583,
+      "step": 15879
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2738107615862072e-05,
+      "loss": 0.7703,
+      "step": 15880
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2732840768869537e-05,
+      "loss": 0.7471,
+      "step": 15881
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2727574852073832e-05,
+      "loss": 0.7681,
+      "step": 15882
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2722309865606342e-05,
+      "loss": 16.6529,
+      "step": 15883
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2717045809598521e-05,
+      "loss": 0.8011,
+      "step": 15884
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.271178268418175e-05,
+      "loss": 0.7611,
+      "step": 15885
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2706520489487383e-05,
+      "loss": 18.3305,
+      "step": 15886
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2701259225646795e-05,
+      "loss": 17.3929,
+      "step": 15887
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2695998892791295e-05,
+      "loss": 0.6987,
+      "step": 15888
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2690739491052178e-05,
+      "loss": 17.3874,
+      "step": 15889
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2685481020560724e-05,
+      "loss": 16.2868,
+      "step": 15890
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2680223481448166e-05,
+      "loss": 0.7884,
+      "step": 15891
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2674966873845767e-05,
+      "loss": 0.6747,
+      "step": 15892
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2669711197884705e-05,
+      "loss": 18.0693,
+      "step": 15893
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2664456453696183e-05,
+      "loss": 0.7548,
+      "step": 15894
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2659202641411377e-05,
+      "loss": 0.7513,
+      "step": 15895
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.265394976116136e-05,
+      "loss": 0.7113,
+      "step": 15896
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2648697813077304e-05,
+      "loss": 15.6877,
+      "step": 15897
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2643446797290266e-05,
+      "loss": 15.7202,
+      "step": 15898
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2638196713931338e-05,
+      "loss": 0.7491,
+      "step": 15899
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2632947563131553e-05,
+      "loss": 16.4535,
+      "step": 15900
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.262769934502191e-05,
+      "loss": 19.9453,
+      "step": 15901
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2622452059733458e-05,
+      "loss": 17.5948,
+      "step": 15902
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2617205707397107e-05,
+      "loss": 18.4222,
+      "step": 15903
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2611960288143849e-05,
+      "loss": 16.5495,
+      "step": 15904
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2606715802104602e-05,
+      "loss": 17.1563,
+      "step": 15905
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2601472249410245e-05,
+      "loss": 0.7859,
+      "step": 15906
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2596229630191702e-05,
+      "loss": 17.6125,
+      "step": 15907
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2590987944579807e-05,
+      "loss": 0.746,
+      "step": 15908
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2585747192705394e-05,
+      "loss": 14.9746,
+      "step": 15909
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2580507374699275e-05,
+      "loss": 16.7864,
+      "step": 15910
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2575268490692216e-05,
+      "loss": 17.8271,
+      "step": 15911
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.257003054081502e-05,
+      "loss": 0.7309,
+      "step": 15912
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2564793525198404e-05,
+      "loss": 0.793,
+      "step": 15913
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.255955744397308e-05,
+      "loss": 17.2326,
+      "step": 15914
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.255432229726976e-05,
+      "loss": 17.1285,
+      "step": 15915
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2549088085219113e-05,
+      "loss": 17.2535,
+      "step": 15916
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2543854807951771e-05,
+      "loss": 18.9467,
+      "step": 15917
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.253862246559837e-05,
+      "loss": 0.7329,
+      "step": 15918
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.253339105828949e-05,
+      "loss": 0.8313,
+      "step": 15919
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2528160586155734e-05,
+      "loss": 0.7837,
+      "step": 15920
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2522931049327636e-05,
+      "loss": 0.7223,
+      "step": 15921
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2517702447935747e-05,
+      "loss": 0.7578,
+      "step": 15922
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2512474782110584e-05,
+      "loss": 0.7675,
+      "step": 15923
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2507248051982579e-05,
+      "loss": 0.7402,
+      "step": 15924
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2502022257682234e-05,
+      "loss": 0.7841,
+      "step": 15925
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2496797399339965e-05,
+      "loss": 17.129,
+      "step": 15926
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2491573477086205e-05,
+      "loss": 17.01,
+      "step": 15927
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.248635049105134e-05,
+      "loss": 15.7566,
+      "step": 15928
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2481128441365719e-05,
+      "loss": 0.7605,
+      "step": 15929
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2475907328159725e-05,
+      "loss": 18.4051,
+      "step": 15930
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.247068715156362e-05,
+      "loss": 0.7778,
+      "step": 15931
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.246546791170775e-05,
+      "loss": 19.184,
+      "step": 15932
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2460249608722374e-05,
+      "loss": 0.7522,
+      "step": 15933
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2455032242737724e-05,
+      "loss": 18.1822,
+      "step": 15934
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2449815813884052e-05,
+      "loss": 0.7454,
+      "step": 15935
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2444600322291555e-05,
+      "loss": 0.6761,
+      "step": 15936
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2439385768090405e-05,
+      "loss": 15.0496,
+      "step": 15937
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2434172151410767e-05,
+      "loss": 0.7632,
+      "step": 15938
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2428959472382757e-05,
+      "loss": 17.593,
+      "step": 15939
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2423747731136503e-05,
+      "loss": 0.7046,
+      "step": 15940
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2418536927802094e-05,
+      "loss": 17.8994,
+      "step": 15941
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2413327062509567e-05,
+      "loss": 0.7621,
+      "step": 15942
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.240811813538899e-05,
+      "loss": 0.7891,
+      "step": 15943
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2402910146570368e-05,
+      "loss": 17.4818,
+      "step": 15944
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.23977030961837e-05,
+      "loss": 17.8182,
+      "step": 15945
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2392496984358942e-05,
+      "loss": 0.751,
+      "step": 15946
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2387291811226032e-05,
+      "loss": 0.792,
+      "step": 15947
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2382087576914925e-05,
+      "loss": 0.7597,
+      "step": 15948
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2376884281555484e-05,
+      "loss": 0.7363,
+      "step": 15949
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2371681925277618e-05,
+      "loss": 0.7006,
+      "step": 15950
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.236648050821117e-05,
+      "loss": 0.6701,
+      "step": 15951
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2361280030485939e-05,
+      "loss": 0.7935,
+      "step": 15952
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.235608049223177e-05,
+      "loss": 17.8909,
+      "step": 15953
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.235088189357841e-05,
+      "loss": 0.7786,
+      "step": 15954
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2345684234655647e-05,
+      "loss": 0.7606,
+      "step": 15955
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2340487515593201e-05,
+      "loss": 18.3546,
+      "step": 15956
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2335291736520771e-05,
+      "loss": 0.7967,
+      "step": 15957
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2330096897568095e-05,
+      "loss": 0.7003,
+      "step": 15958
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2324902998864769e-05,
+      "loss": 0.7778,
+      "step": 15959
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2319710040540478e-05,
+      "loss": 16.1799,
+      "step": 15960
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.231451802272483e-05,
+      "loss": 19.3672,
+      "step": 15961
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.23093269455474e-05,
+      "loss": 0.7997,
+      "step": 15962
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2304136809137796e-05,
+      "loss": 0.7703,
+      "step": 15963
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2298947613625538e-05,
+      "loss": 0.7375,
+      "step": 15964
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.229375935914016e-05,
+      "loss": 0.8306,
+      "step": 15965
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2288572045811158e-05,
+      "loss": 17.9577,
+      "step": 15966
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2283385673767995e-05,
+      "loss": 16.2024,
+      "step": 15967
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2278200243140147e-05,
+      "loss": 16.99,
+      "step": 15968
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2273015754057038e-05,
+      "loss": 0.7893,
+      "step": 15969
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.226783220664805e-05,
+      "loss": 0.822,
+      "step": 15970
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.226264960104262e-05,
+      "loss": 0.7145,
+      "step": 15971
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2257467937370043e-05,
+      "loss": 0.7701,
+      "step": 15972
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2252287215759694e-05,
+      "loss": 17.2603,
+      "step": 15973
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2247107436340882e-05,
+      "loss": 0.791,
+      "step": 15974
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.224192859924287e-05,
+      "loss": 0.6931,
+      "step": 15975
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2236750704594951e-05,
+      "loss": 16.3192,
+      "step": 15976
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2231573752526348e-05,
+      "loss": 17.4782,
+      "step": 15977
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2226397743166319e-05,
+      "loss": 0.7389,
+      "step": 15978
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2221222676644012e-05,
+      "loss": 14.8403,
+      "step": 15979
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2216048553088594e-05,
+      "loss": 0.7783,
+      "step": 15980
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2210875372629248e-05,
+      "loss": 0.6912,
+      "step": 15981
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.220570313539507e-05,
+      "loss": 17.8572,
+      "step": 15982
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2200531841515177e-05,
+      "loss": 0.7343,
+      "step": 15983
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2195361491118651e-05,
+      "loss": 17.2687,
+      "step": 15984
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.219019208433453e-05,
+      "loss": 17.8787,
+      "step": 15985
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.218502362129184e-05,
+      "loss": 16.8792,
+      "step": 15986
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2179856102119591e-05,
+      "loss": 16.5153,
+      "step": 15987
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2174689526946781e-05,
+      "loss": 18.2468,
+      "step": 15988
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2169523895902351e-05,
+      "loss": 17.8629,
+      "step": 15989
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2164359209115234e-05,
+      "loss": 17.5403,
+      "step": 15990
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2159195466714363e-05,
+      "loss": 18.6381,
+      "step": 15991
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2154032668828613e-05,
+      "loss": 15.6318,
+      "step": 15992
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2148870815586849e-05,
+      "loss": 0.8524,
+      "step": 15993
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2143709907117912e-05,
+      "loss": 0.7537,
+      "step": 15994
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2138549943550603e-05,
+      "loss": 18.2102,
+      "step": 15995
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2133390925013749e-05,
+      "loss": 17.0711,
+      "step": 15996
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2128232851636107e-05,
+      "loss": 0.7957,
+      "step": 15997
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2123075723546402e-05,
+      "loss": 0.7432,
+      "step": 15998
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2117919540873402e-05,
+      "loss": 15.8814,
+      "step": 15999
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2112764303745754e-05,
+      "loss": 20.2477,
+      "step": 16000
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2107610012292176e-05,
+      "loss": 17.5567,
+      "step": 16001
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2102456666641299e-05,
+      "loss": 18.714,
+      "step": 16002
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2097304266921743e-05,
+      "loss": 0.7224,
+      "step": 16003
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2092152813262148e-05,
+      "loss": 15.7089,
+      "step": 16004
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2087002305791056e-05,
+      "loss": 0.7488,
+      "step": 16005
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2081852744637068e-05,
+      "loss": 15.3025,
+      "step": 16006
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2076704129928684e-05,
+      "loss": 0.7292,
+      "step": 16007
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2071556461794409e-05,
+      "loss": 0.7523,
+      "step": 16008
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2066409740362756e-05,
+      "loss": 17.6888,
+      "step": 16009
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2061263965762165e-05,
+      "loss": 16.3134,
+      "step": 16010
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2056119138121102e-05,
+      "loss": 0.7924,
+      "step": 16011
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2050975257567971e-05,
+      "loss": 17.2396,
+      "step": 16012
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.204583232423116e-05,
+      "loss": 17.8547,
+      "step": 16013
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2040690338239035e-05,
+      "loss": 16.9251,
+      "step": 16014
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.203554929971994e-05,
+      "loss": 0.7782,
+      "step": 16015
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2030409208802218e-05,
+      "loss": 0.7592,
+      "step": 16016
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2025270065614146e-05,
+      "loss": 18.8753,
+      "step": 16017
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.202013187028399e-05,
+      "loss": 0.7802,
+      "step": 16018
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2014994622940035e-05,
+      "loss": 17.2239,
+      "step": 16019
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2009858323710482e-05,
+      "loss": 0.78,
+      "step": 16020
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2004722972723542e-05,
+      "loss": 17.3759,
+      "step": 16021
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.199958857010739e-05,
+      "loss": 17.8701,
+      "step": 16022
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1994455115990171e-05,
+      "loss": 0.7415,
+      "step": 16023
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1989322610500043e-05,
+      "loss": 0.8166,
+      "step": 16024
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1984191053765103e-05,
+      "loss": 0.8337,
+      "step": 16025
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.197906044591342e-05,
+      "loss": 18.2834,
+      "step": 16026
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.19739307870731e-05,
+      "loss": 0.7174,
+      "step": 16027
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1968802077372116e-05,
+      "loss": 0.7439,
+      "step": 16028
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1963674316938534e-05,
+      "loss": 17.2708,
+      "step": 16029
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1958547505900326e-05,
+      "loss": 16.159,
+      "step": 16030
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1953421644385443e-05,
+      "loss": 0.6973,
+      "step": 16031
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1948296732521858e-05,
+      "loss": 16.537,
+      "step": 16032
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1943172770437462e-05,
+      "loss": 16.0317,
+      "step": 16033
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1938049758260188e-05,
+      "loss": 0.7525,
+      "step": 16034
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1932927696117867e-05,
+      "loss": 0.803,
+      "step": 16035
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1927806584138357e-05,
+      "loss": 15.7443,
+      "step": 16036
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1922686422449492e-05,
+      "loss": 0.7845,
+      "step": 16037
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.191756721117906e-05,
+      "loss": 16.1592,
+      "step": 16038
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1912448950454858e-05,
+      "loss": 0.7833,
+      "step": 16039
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1907331640404628e-05,
+      "loss": 17.8653,
+      "step": 16040
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1902215281156099e-05,
+      "loss": 17.4674,
+      "step": 16041
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.189709987283697e-05,
+      "loss": 16.6954,
+      "step": 16042
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.189198541557492e-05,
+      "loss": 18.9758,
+      "step": 16043
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1886871909497626e-05,
+      "loss": 21.1557,
+      "step": 16044
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1881759354732714e-05,
+      "loss": 0.7297,
+      "step": 16045
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1876647751407772e-05,
+      "loss": 0.7896,
+      "step": 16046
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1871537099650442e-05,
+      "loss": 0.7178,
+      "step": 16047
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.186642739958822e-05,
+      "loss": 17.7694,
+      "step": 16048
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.186131865134869e-05,
+      "loss": 18.5781,
+      "step": 16049
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1856210855059358e-05,
+      "loss": 17.6152,
+      "step": 16050
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1851104010847697e-05,
+      "loss": 0.8129,
+      "step": 16051
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1845998118841207e-05,
+      "loss": 16.9998,
+      "step": 16052
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1840893179167312e-05,
+      "loss": 0.7798,
+      "step": 16053
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1835789191953434e-05,
+      "loss": 19.9545,
+      "step": 16054
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1830686157326975e-05,
+      "loss": 0.7695,
+      "step": 16055
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.182558407541529e-05,
+      "loss": 17.299,
+      "step": 16056
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1820482946345762e-05,
+      "loss": 0.7948,
+      "step": 16057
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1815382770245692e-05,
+      "loss": 0.7075,
+      "step": 16058
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1810283547242374e-05,
+      "loss": 0.765,
+      "step": 16059
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.180518527746311e-05,
+      "loss": 14.8822,
+      "step": 16060
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1800087961035144e-05,
+      "loss": 0.7625,
+      "step": 16061
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1794991598085704e-05,
+      "loss": 0.7345,
+      "step": 16062
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1789896188742e-05,
+      "loss": 18.2304,
+      "step": 16063
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1784801733131196e-05,
+      "loss": 0.699,
+      "step": 16064
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1779708231380482e-05,
+      "loss": 18.0302,
+      "step": 16065
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1774615683616958e-05,
+      "loss": 0.7751,
+      "step": 16066
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1769524089967771e-05,
+      "loss": 0.7482,
+      "step": 16067
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1764433450560009e-05,
+      "loss": 17.3794,
+      "step": 16068
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1759343765520687e-05,
+      "loss": 16.4726,
+      "step": 16069
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1754255034976891e-05,
+      "loss": 0.7837,
+      "step": 16070
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.174916725905561e-05,
+      "loss": 0.7739,
+      "step": 16071
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1744080437883859e-05,
+      "loss": 17.2131,
+      "step": 16072
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1738994571588596e-05,
+      "loss": 0.8063,
+      "step": 16073
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1733909660296754e-05,
+      "loss": 17.143,
+      "step": 16074
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1728825704135288e-05,
+      "loss": 0.787,
+      "step": 16075
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1723742703231045e-05,
+      "loss": 0.7524,
+      "step": 16076
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1718660657710939e-05,
+      "loss": 0.8102,
+      "step": 16077
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1713579567701798e-05,
+      "loss": 0.8003,
+      "step": 16078
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1708499433330445e-05,
+      "loss": 0.7861,
+      "step": 16079
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1703420254723702e-05,
+      "loss": 0.7199,
+      "step": 16080
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1698342032008331e-05,
+      "loss": 0.7558,
+      "step": 16081
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.169326476531109e-05,
+      "loss": 0.7917,
+      "step": 16082
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1688188454758708e-05,
+      "loss": 18.8935,
+      "step": 16083
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1683113100477871e-05,
+      "loss": 15.6515,
+      "step": 16084
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1678038702595301e-05,
+      "loss": 17.9714,
+      "step": 16085
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1672965261237633e-05,
+      "loss": 0.7351,
+      "step": 16086
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1667892776531487e-05,
+      "loss": 0.766,
+      "step": 16087
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1662821248603506e-05,
+      "loss": 17.7491,
+      "step": 16088
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.165775067758026e-05,
+      "loss": 0.7681,
+      "step": 16089
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1652681063588306e-05,
+      "loss": 15.9745,
+      "step": 16090
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1647612406754194e-05,
+      "loss": 17.2102,
+      "step": 16091
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1642544707204417e-05,
+      "loss": 0.733,
+      "step": 16092
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1637477965065496e-05,
+      "loss": 16.9372,
+      "step": 16093
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1632412180463875e-05,
+      "loss": 0.7728,
+      "step": 16094
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1627347353526019e-05,
+      "loss": 0.7748,
+      "step": 16095
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1622283484378343e-05,
+      "loss": 18.2372,
+      "step": 16096
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1617220573147215e-05,
+      "loss": 0.7805,
+      "step": 16097
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1612158619959029e-05,
+      "loss": 0.7447,
+      "step": 16098
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1607097624940122e-05,
+      "loss": 0.7467,
+      "step": 16099
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1602037588216836e-05,
+      "loss": 17.0643,
+      "step": 16100
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1596978509915463e-05,
+      "loss": 17.5186,
+      "step": 16101
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1591920390162259e-05,
+      "loss": 0.7752,
+      "step": 16102
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1586863229083516e-05,
+      "loss": 0.7533,
+      "step": 16103
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1581807026805414e-05,
+      "loss": 18.6345,
+      "step": 16104
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1576751783454193e-05,
+      "loss": 16.3811,
+      "step": 16105
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.157169749915602e-05,
+      "loss": 16.7797,
+      "step": 16106
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.156664417403704e-05,
+      "loss": 19.1806,
+      "step": 16107
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.156159180822341e-05,
+      "loss": 0.8288,
+      "step": 16108
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.155654040184122e-05,
+      "loss": 18.2153,
+      "step": 16109
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1551489955016564e-05,
+      "loss": 0.7573,
+      "step": 16110
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1546440467875497e-05,
+      "loss": 16.5258,
+      "step": 16111
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1541391940544044e-05,
+      "loss": 16.9533,
+      "step": 16112
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1536344373148244e-05,
+      "loss": 0.7861,
+      "step": 16113
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.153129776581407e-05,
+      "loss": 17.2956,
+      "step": 16114
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1526252118667468e-05,
+      "loss": 16.9074,
+      "step": 16115
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1521207431834418e-05,
+      "loss": 17.957,
+      "step": 16116
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1516163705440813e-05,
+      "loss": 0.7273,
+      "step": 16117
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1511120939612553e-05,
+      "loss": 17.7802,
+      "step": 16118
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1506079134475495e-05,
+      "loss": 0.7736,
+      "step": 16119
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1501038290155486e-05,
+      "loss": 0.7898,
+      "step": 16120
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1495998406778358e-05,
+      "loss": 15.8948,
+      "step": 16121
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1490959484469893e-05,
+      "loss": 0.7457,
+      "step": 16122
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1485921523355897e-05,
+      "loss": 0.7533,
+      "step": 16123
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1480884523562063e-05,
+      "loss": 17.0765,
+      "step": 16124
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.147584848521417e-05,
+      "loss": 16.1148,
+      "step": 16125
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1470813408437885e-05,
+      "loss": 0.7888,
+      "step": 16126
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1465779293358886e-05,
+      "loss": 18.6346,
+      "step": 16127
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1460746140102845e-05,
+      "loss": 0.7202,
+      "step": 16128
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1455713948795382e-05,
+      "loss": 18.283,
+      "step": 16129
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.14506827195621e-05,
+      "loss": 16.5604,
+      "step": 16130
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1445652452528577e-05,
+      "loss": 0.772,
+      "step": 16131
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1440623147820361e-05,
+      "loss": 15.5961,
+      "step": 16132
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.143559480556301e-05,
+      "loss": 0.7814,
+      "step": 16133
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1430567425882016e-05,
+      "loss": 0.7852,
+      "step": 16134
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1425541008902851e-05,
+      "loss": 17.4889,
+      "step": 16135
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1420515554751004e-05,
+      "loss": 17.505,
+      "step": 16136
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1415491063551898e-05,
+      "loss": 0.7259,
+      "step": 16137
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1410467535430946e-05,
+      "loss": 18.2705,
+      "step": 16138
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1405444970513535e-05,
+      "loss": 0.744,
+      "step": 16139
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1400423368925017e-05,
+      "loss": 0.7489,
+      "step": 16140
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1395402730790755e-05,
+      "loss": 0.7293,
+      "step": 16141
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.139038305623606e-05,
+      "loss": 17.4086,
+      "step": 16142
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1385364345386206e-05,
+      "loss": 18.9307,
+      "step": 16143
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1380346598366498e-05,
+      "loss": 0.7934,
+      "step": 16144
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1375329815302133e-05,
+      "loss": 0.7539,
+      "step": 16145
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1370313996318366e-05,
+      "loss": 0.7562,
+      "step": 16146
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1365299141540386e-05,
+      "loss": 0.7305,
+      "step": 16147
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1360285251093345e-05,
+      "loss": 0.7904,
+      "step": 16148
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1355272325102417e-05,
+      "loss": 0.7691,
+      "step": 16149
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1350260363692706e-05,
+      "loss": 0.7666,
+      "step": 16150
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1345249366989347e-05,
+      "loss": 0.7774,
+      "step": 16151
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1340239335117364e-05,
+      "loss": 16.9625,
+      "step": 16152
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1335230268201846e-05,
+      "loss": 16.7499,
+      "step": 16153
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.133022216636781e-05,
+      "loss": 0.7527,
+      "step": 16154
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1325215029740244e-05,
+      "loss": 17.4714,
+      "step": 16155
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1320208858444159e-05,
+      "loss": 16.12,
+      "step": 16156
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1315203652604489e-05,
+      "loss": 18.5682,
+      "step": 16157
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1310199412346167e-05,
+      "loss": 16.9418,
+      "step": 16158
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1305196137794106e-05,
+      "loss": 0.7248,
+      "step": 16159
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1300193829073164e-05,
+      "loss": 18.0281,
+      "step": 16160
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1295192486308242e-05,
+      "loss": 0.7567,
+      "step": 16161
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1290192109624148e-05,
+      "loss": 0.7167,
+      "step": 16162
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1285192699145685e-05,
+      "loss": 0.7985,
+      "step": 16163
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1280194254997667e-05,
+      "loss": 0.7868,
+      "step": 16164
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1275196777304842e-05,
+      "loss": 18.9409,
+      "step": 16165
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1270200266191944e-05,
+      "loss": 0.7718,
+      "step": 16166
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1265204721783689e-05,
+      "loss": 18.2144,
+      "step": 16167
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1260210144204757e-05,
+      "loss": 16.1401,
+      "step": 16168
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1255216533579844e-05,
+      "loss": 17.7419,
+      "step": 16169
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1250223890033568e-05,
+      "loss": 16.6855,
+      "step": 16170
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1245232213690537e-05,
+      "loss": 0.6677,
+      "step": 16171
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1240241504675391e-05,
+      "loss": 0.7062,
+      "step": 16172
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1235251763112636e-05,
+      "loss": 0.781,
+      "step": 16173
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.123026298912686e-05,
+      "loss": 16.5892,
+      "step": 16174
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1225275182842576e-05,
+      "loss": 14.9204,
+      "step": 16175
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.122028834438426e-05,
+      "loss": 0.7752,
+      "step": 16176
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1215302473876416e-05,
+      "loss": 16.7744,
+      "step": 16177
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1210317571443468e-05,
+      "loss": 0.7541,
+      "step": 16178
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1205333637209875e-05,
+      "loss": 18.3253,
+      "step": 16179
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.120035067129998e-05,
+      "loss": 0.8007,
+      "step": 16180
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1195368673838213e-05,
+      "loss": 15.6556,
+      "step": 16181
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1190387644948902e-05,
+      "loss": 18.2435,
+      "step": 16182
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1185407584756358e-05,
+      "loss": 17.5051,
+      "step": 16183
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.118042849338492e-05,
+      "loss": 18.2944,
+      "step": 16184
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1175450370958852e-05,
+      "loss": 0.7556,
+      "step": 16185
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1170473217602407e-05,
+      "loss": 18.244,
+      "step": 16186
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.116549703343982e-05,
+      "loss": 0.756,
+      "step": 16187
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1160521818595276e-05,
+      "loss": 17.1042,
+      "step": 16188
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1155547573192988e-05,
+      "loss": 17.1519,
+      "step": 16189
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1150574297357103e-05,
+      "loss": 0.6864,
+      "step": 16190
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1145601991211741e-05,
+      "loss": 18.9089,
+      "step": 16191
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.114063065488104e-05,
+      "loss": 18.357,
+      "step": 16192
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1135660288489069e-05,
+      "loss": 18.2908,
+      "step": 16193
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1130690892159889e-05,
+      "loss": 0.767,
+      "step": 16194
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1125722466017547e-05,
+      "loss": 18.1656,
+      "step": 16195
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1120755010186024e-05,
+      "loss": 18.9731,
+      "step": 16196
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1115788524789356e-05,
+      "loss": 19.2957,
+      "step": 16197
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1110823009951482e-05,
+      "loss": 0.7675,
+      "step": 16198
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1105858465796343e-05,
+      "loss": 16.6989,
+      "step": 16199
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1100894892447861e-05,
+      "loss": 0.7799,
+      "step": 16200
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1095932290029903e-05,
+      "loss": 16.271,
+      "step": 16201
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1090970658666378e-05,
+      "loss": 0.7341,
+      "step": 16202
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1086009998481112e-05,
+      "loss": 0.7836,
+      "step": 16203
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.10810503095979e-05,
+      "loss": 0.7937,
+      "step": 16204
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.107609159214058e-05,
+      "loss": 0.7416,
+      "step": 16205
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1071133846232896e-05,
+      "loss": 16.3199,
+      "step": 16206
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1066177071998601e-05,
+      "loss": 17.6186,
+      "step": 16207
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1061221269561406e-05,
+      "loss": 18.5893,
+      "step": 16208
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1056266439045032e-05,
+      "loss": 19.5021,
+      "step": 16209
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1051312580573136e-05,
+      "loss": 0.7806,
+      "step": 16210
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1046359694269359e-05,
+      "loss": 0.7962,
+      "step": 16211
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1041407780257352e-05,
+      "loss": 0.7789,
+      "step": 16212
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1036456838660702e-05,
+      "loss": 19.2658,
+      "step": 16213
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1031506869602986e-05,
+      "loss": 0.7265,
+      "step": 16214
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1026557873207754e-05,
+      "loss": 0.7476,
+      "step": 16215
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1021609849598525e-05,
+      "loss": 17.2062,
+      "step": 16216
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.101666279889883e-05,
+      "loss": 14.3126,
+      "step": 16217
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.101171672123213e-05,
+      "loss": 16.8913,
+      "step": 16218
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1006771616721867e-05,
+      "loss": 0.7822,
+      "step": 16219
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1001827485491522e-05,
+      "loss": 0.8193,
+      "step": 16220
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0996884327664436e-05,
+      "loss": 0.7622,
+      "step": 16221
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0991942143364032e-05,
+      "loss": 0.765,
+      "step": 16222
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0987000932713665e-05,
+      "loss": 19.5672,
+      "step": 16223
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0982060695836643e-05,
+      "loss": 18.0206,
+      "step": 16224
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0977121432856314e-05,
+      "loss": 19.0675,
+      "step": 16225
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0972183143895942e-05,
+      "loss": 18.3581,
+      "step": 16226
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0967245829078793e-05,
+      "loss": 15.8016,
+      "step": 16227
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.09623094885281e-05,
+      "loss": 0.7603,
+      "step": 16228
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0957374122367059e-05,
+      "loss": 0.7341,
+      "step": 16229
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0952439730718888e-05,
+      "loss": 0.8333,
+      "step": 16230
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0947506313706734e-05,
+      "loss": 0.7748,
+      "step": 16231
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0942573871453733e-05,
+      "loss": 0.7718,
+      "step": 16232
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0937642404083015e-05,
+      "loss": 17.7533,
+      "step": 16233
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0932711911717663e-05,
+      "loss": 0.7526,
+      "step": 16234
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.092778239448074e-05,
+      "loss": 0.7691,
+      "step": 16235
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.092285385249528e-05,
+      "loss": 0.739,
+      "step": 16236
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0917926285884322e-05,
+      "loss": 17.0561,
+      "step": 16237
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0912999694770848e-05,
+      "loss": 18.2018,
+      "step": 16238
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0908074079277814e-05,
+      "loss": 18.8454,
+      "step": 16239
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0903149439528188e-05,
+      "loss": 19.0677,
+      "step": 16240
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0898225775644877e-05,
+      "loss": 0.7232,
+      "step": 16241
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0893303087750783e-05,
+      "loss": 0.7459,
+      "step": 16242
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.088838137596877e-05,
+      "loss": 14.7406,
+      "step": 16243
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0883460640421678e-05,
+      "loss": 0.7381,
+      "step": 16244
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0878540881232352e-05,
+      "loss": 0.7298,
+      "step": 16245
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0873622098523578e-05,
+      "loss": 0.7313,
+      "step": 16246
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0868704292418114e-05,
+      "loss": 18.7347,
+      "step": 16247
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0863787463038755e-05,
+      "loss": 0.7372,
+      "step": 16248
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.085887161050817e-05,
+      "loss": 0.7457,
+      "step": 16249
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.08539567349491e-05,
+      "loss": 18.4879,
+      "step": 16250
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0849042836484208e-05,
+      "loss": 17.7393,
+      "step": 16251
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0844129915236134e-05,
+      "loss": 18.2502,
+      "step": 16252
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.083921797132753e-05,
+      "loss": 0.7533,
+      "step": 16253
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.083430700488099e-05,
+      "loss": 0.7463,
+      "step": 16254
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0829397016019089e-05,
+      "loss": 17.627,
+      "step": 16255
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0824488004864375e-05,
+      "loss": 17.7471,
+      "step": 16256
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0819579971539379e-05,
+      "loss": 19.697,
+      "step": 16257
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0814672916166624e-05,
+      "loss": 0.7769,
+      "step": 16258
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.080976683886858e-05,
+      "loss": 0.7744,
+      "step": 16259
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.080486173976769e-05,
+      "loss": 16.8871,
+      "step": 16260
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0799957618986417e-05,
+      "loss": 0.7468,
+      "step": 16261
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0795054476647154e-05,
+      "loss": 0.7316,
+      "step": 16262
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0790152312872276e-05,
+      "loss": 0.7771,
+      "step": 16263
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.078525112778414e-05,
+      "loss": 15.5551,
+      "step": 16264
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.07803509215051e-05,
+      "loss": 17.1389,
+      "step": 16265
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0775451694157462e-05,
+      "loss": 16.1687,
+      "step": 16266
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0770553445863484e-05,
+      "loss": 15.9077,
+      "step": 16267
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.076565617674547e-05,
+      "loss": 0.758,
+      "step": 16268
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.076075988692563e-05,
+      "loss": 0.7282,
+      "step": 16269
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0755864576526192e-05,
+      "loss": 0.8017,
+      "step": 16270
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0750970245669328e-05,
+      "loss": 16.6213,
+      "step": 16271
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0746076894477192e-05,
+      "loss": 17.6983,
+      "step": 16272
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0741184523071957e-05,
+      "loss": 18.4432,
+      "step": 16273
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.073629313157572e-05,
+      "loss": 16.1276,
+      "step": 16274
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0731402720110573e-05,
+      "loss": 0.7379,
+      "step": 16275
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0726513288798579e-05,
+      "loss": 16.2843,
+      "step": 16276
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0721624837761768e-05,
+      "loss": 0.712,
+      "step": 16277
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0716737367122182e-05,
+      "loss": 0.71,
+      "step": 16278
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.07118508770018e-05,
+      "loss": 0.7504,
+      "step": 16279
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0706965367522582e-05,
+      "loss": 0.7331,
+      "step": 16280
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.070208083880649e-05,
+      "loss": 19.8892,
+      "step": 16281
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.069719729097543e-05,
+      "loss": 17.9718,
+      "step": 16282
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0692314724151304e-05,
+      "loss": 18.051,
+      "step": 16283
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0687433138455977e-05,
+      "loss": 16.1494,
+      "step": 16284
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0682552534011282e-05,
+      "loss": 0.7438,
+      "step": 16285
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0677672910939064e-05,
+      "loss": 0.7332,
+      "step": 16286
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0672794269361091e-05,
+      "loss": 0.8199,
+      "step": 16287
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0667916609399176e-05,
+      "loss": 0.7094,
+      "step": 16288
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.066303993117505e-05,
+      "loss": 17.8023,
+      "step": 16289
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0658164234810403e-05,
+      "loss": 0.7992,
+      "step": 16290
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0653289520426979e-05,
+      "loss": 18.3302,
+      "step": 16291
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.064841578814641e-05,
+      "loss": 0.7817,
+      "step": 16292
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0643543038090387e-05,
+      "loss": 17.7677,
+      "step": 16293
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0638671270380517e-05,
+      "loss": 0.7284,
+      "step": 16294
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0633800485138385e-05,
+      "loss": 0.7443,
+      "step": 16295
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0628930682485606e-05,
+      "loss": 17.1582,
+      "step": 16296
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0624061862543683e-05,
+      "loss": 18.2518,
+      "step": 16297
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0619194025434175e-05,
+      "loss": 17.7249,
+      "step": 16298
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0614327171278581e-05,
+      "loss": 17.414,
+      "step": 16299
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0609461300198359e-05,
+      "loss": 0.7478,
+      "step": 16300
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0604596412314987e-05,
+      "loss": 0.7774,
+      "step": 16301
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0599732507749887e-05,
+      "loss": 0.727,
+      "step": 16302
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0594869586624461e-05,
+      "loss": 18.8831,
+      "step": 16303
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0590007649060085e-05,
+      "loss": 17.8512,
+      "step": 16304
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0585146695178106e-05,
+      "loss": 0.7233,
+      "step": 16305
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0580286725099876e-05,
+      "loss": 0.7601,
+      "step": 16306
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0575427738946692e-05,
+      "loss": 19.0066,
+      "step": 16307
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0570569736839814e-05,
+      "loss": 17.9632,
+      "step": 16308
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.056571271890054e-05,
+      "loss": 16.8185,
+      "step": 16309
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0560856685250075e-05,
+      "loss": 0.7368,
+      "step": 16310
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0556001636009633e-05,
+      "loss": 0.7929,
+      "step": 16311
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0551147571300396e-05,
+      "loss": 0.7705,
+      "step": 16312
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0546294491243513e-05,
+      "loss": 0.7079,
+      "step": 16313
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0541442395960139e-05,
+      "loss": 18.6505,
+      "step": 16314
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0536591285571356e-05,
+      "loss": 19.2559,
+      "step": 16315
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0531741160198282e-05,
+      "loss": 15.424,
+      "step": 16316
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0526892019961975e-05,
+      "loss": 19.619,
+      "step": 16317
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0522043864983428e-05,
+      "loss": 0.7841,
+      "step": 16318
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0517196695383696e-05,
+      "loss": 15.7703,
+      "step": 16319
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.051235051128373e-05,
+      "loss": 0.7842,
+      "step": 16320
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0507505312804533e-05,
+      "loss": 0.742,
+      "step": 16321
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0502661100067012e-05,
+      "loss": 0.7983,
+      "step": 16322
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0497817873192084e-05,
+      "loss": 0.8329,
+      "step": 16323
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0492975632300662e-05,
+      "loss": 0.7898,
+      "step": 16324
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0488134377513564e-05,
+      "loss": 0.7943,
+      "step": 16325
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0483294108951663e-05,
+      "loss": 0.7732,
+      "step": 16326
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0478454826735767e-05,
+      "loss": 18.1388,
+      "step": 16327
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0473616530986652e-05,
+      "loss": 16.193,
+      "step": 16328
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0468779221825103e-05,
+      "loss": 17.9985,
+      "step": 16329
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0463942899371853e-05,
+      "loss": 0.7786,
+      "step": 16330
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0459107563747617e-05,
+      "loss": 0.7772,
+      "step": 16331
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0454273215073079e-05,
+      "loss": 15.8881,
+      "step": 16332
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.04494398534689e-05,
+      "loss": 0.7242,
+      "step": 16333
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0444607479055745e-05,
+      "loss": 17.3488,
+      "step": 16334
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0439776091954223e-05,
+      "loss": 15.3829,
+      "step": 16335
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0434945692284909e-05,
+      "loss": 0.766,
+      "step": 16336
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0430116280168394e-05,
+      "loss": 13.9428,
+      "step": 16337
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0425287855725214e-05,
+      "loss": 0.8,
+      "step": 16338
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0420460419075883e-05,
+      "loss": 0.7627,
+      "step": 16339
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.04156339703409e-05,
+      "loss": 17.4997,
+      "step": 16340
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0410808509640718e-05,
+      "loss": 17.6196,
+      "step": 16341
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0405984037095806e-05,
+      "loss": 17.75,
+      "step": 16342
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0401160552826562e-05,
+      "loss": 0.7946,
+      "step": 16343
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0396338056953402e-05,
+      "loss": 18.0654,
+      "step": 16344
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0391516549596702e-05,
+      "loss": 0.7242,
+      "step": 16345
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0386696030876769e-05,
+      "loss": 0.7401,
+      "step": 16346
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0381876500913956e-05,
+      "loss": 0.7606,
+      "step": 16347
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.037705795982854e-05,
+      "loss": 15.3383,
+      "step": 16348
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0372240407740819e-05,
+      "loss": 17.85,
+      "step": 16349
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0367423844771023e-05,
+      "loss": 17.5311,
+      "step": 16350
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.036260827103938e-05,
+      "loss": 0.7456,
+      "step": 16351
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.035779368666609e-05,
+      "loss": 0.7398,
+      "step": 16352
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0352980091771298e-05,
+      "loss": 0.8484,
+      "step": 16353
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0348167486475196e-05,
+      "loss": 17.476,
+      "step": 16354
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.034335587089788e-05,
+      "loss": 0.7603,
+      "step": 16355
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0338545245159447e-05,
+      "loss": 0.7806,
+      "step": 16356
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0333735609379997e-05,
+      "loss": 16.4894,
+      "step": 16357
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0328926963679558e-05,
+      "loss": 17.2547,
+      "step": 16358
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0324119308178166e-05,
+      "loss": 0.7975,
+      "step": 16359
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.031931264299581e-05,
+      "loss": 0.7222,
+      "step": 16360
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0314506968252458e-05,
+      "loss": 0.7309,
+      "step": 16361
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0309702284068085e-05,
+      "loss": 16.0105,
+      "step": 16362
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0304898590562612e-05,
+      "loss": 19.4444,
+      "step": 16363
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0300095887855915e-05,
+      "loss": 18.2392,
+      "step": 16364
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0295294176067915e-05,
+      "loss": 18.2723,
+      "step": 16365
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0290493455318417e-05,
+      "loss": 18.2743,
+      "step": 16366
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.028569372572728e-05,
+      "loss": 17.1209,
+      "step": 16367
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0280894987414291e-05,
+      "loss": 0.7805,
+      "step": 16368
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0276097240499222e-05,
+      "loss": 16.5312,
+      "step": 16369
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0271300485101853e-05,
+      "loss": 0.7596,
+      "step": 16370
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0266504721341879e-05,
+      "loss": 0.7406,
+      "step": 16371
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0261709949339043e-05,
+      "loss": 0.742,
+      "step": 16372
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0256916169212994e-05,
+      "loss": 15.9974,
+      "step": 16373
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0252123381083373e-05,
+      "loss": 0.7558,
+      "step": 16374
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0247331585069842e-05,
+      "loss": 0.7527,
+      "step": 16375
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0242540781291981e-05,
+      "loss": 18.1552,
+      "step": 16376
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0237750969869391e-05,
+      "loss": 17.1211,
+      "step": 16377
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0232962150921616e-05,
+      "loss": 0.6933,
+      "step": 16378
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0228174324568186e-05,
+      "loss": 0.7505,
+      "step": 16379
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0223387490928604e-05,
+      "loss": 0.7523,
+      "step": 16380
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.021860165012234e-05,
+      "loss": 0.7558,
+      "step": 16381
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0213816802268878e-05,
+      "loss": 18.2041,
+      "step": 16382
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0209032947487629e-05,
+      "loss": 17.1613,
+      "step": 16383
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0204250085897992e-05,
+      "loss": 15.3999,
+      "step": 16384
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0199468217619373e-05,
+      "loss": 17.4084,
+      "step": 16385
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0194687342771115e-05,
+      "loss": 15.9505,
+      "step": 16386
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0189907461472548e-05,
+      "loss": 18.3676,
+      "step": 16387
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0185128573842979e-05,
+      "loss": 0.713,
+      "step": 16388
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0180350680001682e-05,
+      "loss": 17.0099,
+      "step": 16389
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0175573780067938e-05,
+      "loss": 16.5212,
+      "step": 16390
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0170797874160964e-05,
+      "loss": 18.3993,
+      "step": 16391
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0166022962399952e-05,
+      "loss": 0.7355,
+      "step": 16392
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0161249044904136e-05,
+      "loss": 17.0005,
+      "step": 16393
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0156476121792607e-05,
+      "loss": 16.0045,
+      "step": 16394
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0151704193184553e-05,
+      "loss": 19.006,
+      "step": 16395
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0146933259199054e-05,
+      "loss": 16.6991,
+      "step": 16396
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.014216331995519e-05,
+      "loss": 0.7362,
+      "step": 16397
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.013739437557204e-05,
+      "loss": 17.2628,
+      "step": 16398
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0132626426168618e-05,
+      "loss": 0.7977,
+      "step": 16399
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.012785947186397e-05,
+      "loss": 17.6954,
+      "step": 16400
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0123093512777038e-05,
+      "loss": 0.7136,
+      "step": 16401
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0118328549026784e-05,
+      "loss": 0.7878,
+      "step": 16402
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0113564580732171e-05,
+      "loss": 17.0291,
+      "step": 16403
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.010880160801208e-05,
+      "loss": 16.8963,
+      "step": 16404
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0104039630985424e-05,
+      "loss": 16.6611,
+      "step": 16405
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0099278649771044e-05,
+      "loss": 0.74,
+      "step": 16406
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0094518664487778e-05,
+      "loss": 15.8109,
+      "step": 16407
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0089759675254446e-05,
+      "loss": 0.8082,
+      "step": 16408
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0085001682189804e-05,
+      "loss": 0.7599,
+      "step": 16409
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0080244685412648e-05,
+      "loss": 17.6993,
+      "step": 16410
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0075488685041707e-05,
+      "loss": 18.4743,
+      "step": 16411
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0070733681195665e-05,
+      "loss": 16.9501,
+      "step": 16412
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0065979673993243e-05,
+      "loss": 0.7674,
+      "step": 16413
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.006122666355308e-05,
+      "loss": 14.6688,
+      "step": 16414
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0056474649993824e-05,
+      "loss": 0.7304,
+      "step": 16415
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.005172363343408e-05,
+      "loss": 0.7559,
+      "step": 16416
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0046973613992421e-05,
+      "loss": 18.118,
+      "step": 16417
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0042224591787442e-05,
+      "loss": 0.7124,
+      "step": 16418
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0037476566937654e-05,
+      "loss": 0.761,
+      "step": 16419
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0032729539561569e-05,
+      "loss": 18.6701,
+      "step": 16420
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.00279835097777e-05,
+      "loss": 18.2824,
+      "step": 16421
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0023238477704466e-05,
+      "loss": 16.5437,
+      "step": 16422
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0018494443460341e-05,
+      "loss": 18.4138,
+      "step": 16423
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0013751407163723e-05,
+      "loss": 18.481,
+      "step": 16424
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0009009368932986e-05,
+      "loss": 0.783,
+      "step": 16425
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0004268328886523e-05,
+      "loss": 0.7724,
+      "step": 16426
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.999528287142651e-06,
+      "loss": 0.7674,
+      "step": 16427
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.994789243819685e-06,
+      "loss": 17.3181,
+      "step": 16428
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.990051199035921e-06,
+      "loss": 17.626,
+      "step": 16429
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.985314152909592e-06,
+      "loss": 16.5203,
+      "step": 16430
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.98057810555898e-06,
+      "loss": 17.1926,
+      "step": 16431
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.975843057102258e-06,
+      "loss": 17.5694,
+      "step": 16432
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.971109007657647e-06,
+      "loss": 17.5425,
+      "step": 16433
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.966375957343305e-06,
+      "loss": 0.7771,
+      "step": 16434
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.961643906277334e-06,
+      "loss": 17.8535,
+      "step": 16435
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.956912854577887e-06,
+      "loss": 17.4632,
+      "step": 16436
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.95218280236303e-06,
+      "loss": 15.8614,
+      "step": 16437
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.947453749750846e-06,
+      "loss": 18.2581,
+      "step": 16438
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.942725696859356e-06,
+      "loss": 0.7968,
+      "step": 16439
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.937998643806568e-06,
+      "loss": 0.7617,
+      "step": 16440
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.933272590710507e-06,
+      "loss": 0.7483,
+      "step": 16441
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.928547537689087e-06,
+      "loss": 16.3487,
+      "step": 16442
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.923823484860285e-06,
+      "loss": 16.8658,
+      "step": 16443
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.919100432341994e-06,
+      "loss": 20.3612,
+      "step": 16444
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.914378380252098e-06,
+      "loss": 18.6167,
+      "step": 16445
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.909657328708482e-06,
+      "loss": 17.8246,
+      "step": 16446
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.904937277828974e-06,
+      "loss": 0.7751,
+      "step": 16447
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.900218227731384e-06,
+      "loss": 0.7949,
+      "step": 16448
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.895500178533507e-06,
+      "loss": 17.256,
+      "step": 16449
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.890783130353087e-06,
+      "loss": 18.539,
+      "step": 16450
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.886067083307892e-06,
+      "loss": 0.692,
+      "step": 16451
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.881352037515618e-06,
+      "loss": 0.6805,
+      "step": 16452
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.876637993093951e-06,
+      "loss": 0.7556,
+      "step": 16453
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.871924950160566e-06,
+      "loss": 16.1825,
+      "step": 16454
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.867212908833107e-06,
+      "loss": 15.6622,
+      "step": 16455
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.862501869229168e-06,
+      "loss": 17.7297,
+      "step": 16456
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.857791831466351e-06,
+      "loss": 0.768,
+      "step": 16457
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.853082795662205e-06,
+      "loss": 0.7796,
+      "step": 16458
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.848374761934287e-06,
+      "loss": 17.6399,
+      "step": 16459
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.843667730400092e-06,
+      "loss": 17.6492,
+      "step": 16460
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.838961701177135e-06,
+      "loss": 16.7407,
+      "step": 16461
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.834256674382875e-06,
+      "loss": 18.376,
+      "step": 16462
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.82955265013471e-06,
+      "loss": 19.7188,
+      "step": 16463
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.824849628550097e-06,
+      "loss": 0.7532,
+      "step": 16464
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.820147609746399e-06,
+      "loss": 0.7273,
+      "step": 16465
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.815446593841005e-06,
+      "loss": 0.7837,
+      "step": 16466
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.810746580951242e-06,
+      "loss": 0.755,
+      "step": 16467
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.806047571194399e-06,
+      "loss": 0.7365,
+      "step": 16468
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.801349564687817e-06,
+      "loss": 16.7673,
+      "step": 16469
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.796652561548703e-06,
+      "loss": 0.8269,
+      "step": 16470
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.791956561894334e-06,
+      "loss": 0.7914,
+      "step": 16471
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.78726156584191e-06,
+      "loss": 15.7826,
+      "step": 16472
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.782567573508605e-06,
+      "loss": 17.2917,
+      "step": 16473
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.77787458501161e-06,
+      "loss": 0.6868,
+      "step": 16474
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.77318260046805e-06,
+      "loss": 16.9564,
+      "step": 16475
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.768491619995034e-06,
+      "loss": 0.7516,
+      "step": 16476
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.763801643709658e-06,
+      "loss": 16.3009,
+      "step": 16477
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.759112671728964e-06,
+      "loss": 18.5439,
+      "step": 16478
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.754424704170017e-06,
+      "loss": 0.7022,
+      "step": 16479
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.749737741149823e-06,
+      "loss": 16.8771,
+      "step": 16480
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.745051782785348e-06,
+      "loss": 0.734,
+      "step": 16481
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.740366829193587e-06,
+      "loss": 0.7985,
+      "step": 16482
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.735682880491465e-06,
+      "loss": 16.8591,
+      "step": 16483
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.730999936795888e-06,
+      "loss": 0.7789,
+      "step": 16484
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.72631799822375e-06,
+      "loss": 0.7524,
+      "step": 16485
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.721637064891892e-06,
+      "loss": 0.7336,
+      "step": 16486
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.716957136917188e-06,
+      "loss": 15.9938,
+      "step": 16487
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.712278214416421e-06,
+      "loss": 0.7713,
+      "step": 16488
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.707600297506398e-06,
+      "loss": 16.951,
+      "step": 16489
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.70292338630388e-06,
+      "loss": 17.4957,
+      "step": 16490
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.698247480925577e-06,
+      "loss": 0.7369,
+      "step": 16491
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.693572581488225e-06,
+      "loss": 17.6188,
+      "step": 16492
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.688898688108495e-06,
+      "loss": 18.5206,
+      "step": 16493
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.684225800903073e-06,
+      "loss": 17.6143,
+      "step": 16494
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.679553919988577e-06,
+      "loss": 16.5777,
+      "step": 16495
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.674883045481615e-06,
+      "loss": 18.6635,
+      "step": 16496
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.6702131774988e-06,
+      "loss": 16.1279,
+      "step": 16497
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.665544316156645e-06,
+      "loss": 0.8006,
+      "step": 16498
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.660876461571728e-06,
+      "loss": 0.7689,
+      "step": 16499
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.656209613860546e-06,
+      "loss": 0.755,
+      "step": 16500
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.65154377313957e-06,
+      "loss": 0.7297,
+      "step": 16501
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.646878939525288e-06,
+      "loss": 17.5366,
+      "step": 16502
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.642215113134123e-06,
+      "loss": 0.7805,
+      "step": 16503
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.63755229408248e-06,
+      "loss": 0.7464,
+      "step": 16504
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.63289048248675e-06,
+      "loss": 0.7722,
+      "step": 16505
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.628229678463274e-06,
+      "loss": 0.7797,
+      "step": 16506
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.62356988212842e-06,
+      "loss": 0.7577,
+      "step": 16507
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.618911093598481e-06,
+      "loss": 0.7769,
+      "step": 16508
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.61425331298973e-06,
+      "loss": 0.7293,
+      "step": 16509
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.609596540418458e-06,
+      "loss": 0.787,
+      "step": 16510
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.604940776000854e-06,
+      "loss": 17.8283,
+      "step": 16511
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.600286019853167e-06,
+      "loss": 0.7883,
+      "step": 16512
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.595632272091565e-06,
+      "loss": 0.7357,
+      "step": 16513
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.590979532832195e-06,
+      "loss": 0.7218,
+      "step": 16514
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.586327802191213e-06,
+      "loss": 0.7373,
+      "step": 16515
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.581677080284707e-06,
+      "loss": 0.7659,
+      "step": 16516
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.577027367228792e-06,
+      "loss": 17.1639,
+      "step": 16517
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.57237866313948e-06,
+      "loss": 14.9951,
+      "step": 16518
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.567730968132838e-06,
+      "loss": 18.5062,
+      "step": 16519
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.56308428232487e-06,
+      "loss": 18.8196,
+      "step": 16520
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.558438605831527e-06,
+      "loss": 0.7877,
+      "step": 16521
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.553793938768812e-06,
+      "loss": 0.7367,
+      "step": 16522
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.549150281252633e-06,
+      "loss": 18.8449,
+      "step": 16523
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.544507633398897e-06,
+      "loss": 18.3418,
+      "step": 16524
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.539865995323487e-06,
+      "loss": 16.59,
+      "step": 16525
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.535225367142253e-06,
+      "loss": 0.7959,
+      "step": 16526
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.530585748971044e-06,
+      "loss": 18.3155,
+      "step": 16527
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.52594714092565e-06,
+      "loss": 16.7995,
+      "step": 16528
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.521309543121848e-06,
+      "loss": 16.5402,
+      "step": 16529
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.516672955675416e-06,
+      "loss": 0.7481,
+      "step": 16530
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.51203737870207e-06,
+      "loss": 17.7752,
+      "step": 16531
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.507402812317512e-06,
+      "loss": 0.7763,
+      "step": 16532
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.502769256637429e-06,
+      "loss": 17.6152,
+      "step": 16533
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.49813671177746e-06,
+      "loss": 0.7614,
+      "step": 16534
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.493505177853256e-06,
+      "loss": 0.7608,
+      "step": 16535
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.488874654980417e-06,
+      "loss": 19.8277,
+      "step": 16536
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.484245143274501e-06,
+      "loss": 20.1078,
+      "step": 16537
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.479616642851097e-06,
+      "loss": 0.7546,
+      "step": 16538
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.474989153825698e-06,
+      "loss": 0.7568,
+      "step": 16539
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.470362676313827e-06,
+      "loss": 0.7745,
+      "step": 16540
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.465737210430964e-06,
+      "loss": 0.7654,
+      "step": 16541
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.46111275629254e-06,
+      "loss": 16.8028,
+      "step": 16542
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.45648931401401e-06,
+      "loss": 18.3899,
+      "step": 16543
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.451866883710747e-06,
+      "loss": 0.7682,
+      "step": 16544
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.447245465498177e-06,
+      "loss": 16.2342,
+      "step": 16545
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.442625059491595e-06,
+      "loss": 0.7851,
+      "step": 16546
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.438005665806359e-06,
+      "loss": 17.7267,
+      "step": 16547
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.433387284557765e-06,
+      "loss": 15.1448,
+      "step": 16548
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.42876991586108e-06,
+      "loss": 0.7276,
+      "step": 16549
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.424153559831566e-06,
+      "loss": 17.7608,
+      "step": 16550
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.41953821658445e-06,
+      "loss": 0.7964,
+      "step": 16551
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.414923886234923e-06,
+      "loss": 16.4926,
+      "step": 16552
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.410310568898162e-06,
+      "loss": 18.6797,
+      "step": 16553
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.405698264689305e-06,
+      "loss": 0.8144,
+      "step": 16554
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.4010869737235e-06,
+      "loss": 0.7693,
+      "step": 16555
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.396476696115835e-06,
+      "loss": 17.808,
+      "step": 16556
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.391867431981372e-06,
+      "loss": 15.8889,
+      "step": 16557
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.387259181435176e-06,
+      "loss": 16.9426,
+      "step": 16558
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.38265194459227e-06,
+      "loss": 0.699,
+      "step": 16559
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.378045721567646e-06,
+      "loss": 17.1169,
+      "step": 16560
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.373440512476272e-06,
+      "loss": 0.7795,
+      "step": 16561
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.368836317433089e-06,
+      "loss": 0.739,
+      "step": 16562
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.36423313655304e-06,
+      "loss": 17.2371,
+      "step": 16563
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.35963096995101e-06,
+      "loss": 18.1476,
+      "step": 16564
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.355029817741862e-06,
+      "loss": 0.7474,
+      "step": 16565
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.350429680040473e-06,
+      "loss": 18.3821,
+      "step": 16566
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.345830556961615e-06,
+      "loss": 0.7858,
+      "step": 16567
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.341232448620118e-06,
+      "loss": 0.7547,
+      "step": 16568
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.336635355130746e-06,
+      "loss": 19.4199,
+      "step": 16569
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.33203927660823e-06,
+      "loss": 0.7669,
+      "step": 16570
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.327444213167309e-06,
+      "loss": 0.761,
+      "step": 16571
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.322850164922653e-06,
+      "loss": 15.3329,
+      "step": 16572
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.318257131988973e-06,
+      "loss": 0.7623,
+      "step": 16573
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.313665114480858e-06,
+      "loss": 16.7733,
+      "step": 16574
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.309074112512966e-06,
+      "loss": 15.5376,
+      "step": 16575
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.304484126199869e-06,
+      "loss": 16.9214,
+      "step": 16576
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.299895155656135e-06,
+      "loss": 16.9239,
+      "step": 16577
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.295307200996317e-06,
+      "loss": 0.7557,
+      "step": 16578
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.290720262334929e-06,
+      "loss": 0.7524,
+      "step": 16579
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.286134339786463e-06,
+      "loss": 0.7875,
+      "step": 16580
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.28154943346537e-06,
+      "loss": 16.8891,
+      "step": 16581
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.276965543486094e-06,
+      "loss": 15.6455,
+      "step": 16582
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.27238266996307e-06,
+      "loss": 0.7496,
+      "step": 16583
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.267800813010669e-06,
+      "loss": 0.7319,
+      "step": 16584
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.263219972743248e-06,
+      "loss": 18.5117,
+      "step": 16585
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.258640149275189e-06,
+      "loss": 0.8072,
+      "step": 16586
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.254061342720744e-06,
+      "loss": 0.7734,
+      "step": 16587
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.249483553194243e-06,
+      "loss": 0.7101,
+      "step": 16588
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.244906780809943e-06,
+      "loss": 0.727,
+      "step": 16589
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.24033102568206e-06,
+      "loss": 0.7687,
+      "step": 16590
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.235756287924829e-06,
+      "loss": 0.7351,
+      "step": 16591
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.231182567652436e-06,
+      "loss": 0.7232,
+      "step": 16592
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.22660986497903e-06,
+      "loss": 0.7698,
+      "step": 16593
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.22203818001875e-06,
+      "loss": 0.7438,
+      "step": 16594
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.217467512885703e-06,
+      "loss": 0.7645,
+      "step": 16595
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.212897863693982e-06,
+      "loss": 0.785,
+      "step": 16596
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.20832923255765e-06,
+      "loss": 0.8,
+      "step": 16597
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.203761619590718e-06,
+      "loss": 0.7627,
+      "step": 16598
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.199195024907225e-06,
+      "loss": 0.7969,
+      "step": 16599
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.194629448621139e-06,
+      "loss": 15.8488,
+      "step": 16600
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.190064890846423e-06,
+      "loss": 0.7705,
+      "step": 16601
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.18550135169699e-06,
+      "loss": 0.7869,
+      "step": 16602
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.180938831286774e-06,
+      "loss": 0.7998,
+      "step": 16603
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.176377329729646e-06,
+      "loss": 0.7462,
+      "step": 16604
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.171816847139448e-06,
+      "loss": 0.7384,
+      "step": 16605
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.16725738363004e-06,
+      "loss": 0.7397,
+      "step": 16606
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.162698939315206e-06,
+      "loss": 0.7733,
+      "step": 16607
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.158141514308733e-06,
+      "loss": 0.7286,
+      "step": 16608
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.153585108724372e-06,
+      "loss": 0.7401,
+      "step": 16609
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.149029722675845e-06,
+      "loss": 0.7512,
+      "step": 16610
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.14447535627687e-06,
+      "loss": 18.6627,
+      "step": 16611
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.139922009641123e-06,
+      "loss": 17.0058,
+      "step": 16612
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.135369682882239e-06,
+      "loss": 17.4506,
+      "step": 16613
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.13081837611388e-06,
+      "loss": 0.7238,
+      "step": 16614
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.126268089449602e-06,
+      "loss": 0.7513,
+      "step": 16615
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.121718823003023e-06,
+      "loss": 16.5981,
+      "step": 16616
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.117170576887668e-06,
+      "loss": 16.644,
+      "step": 16617
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.112623351217064e-06,
+      "loss": 16.9893,
+      "step": 16618
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.108077146104727e-06,
+      "loss": 0.7143,
+      "step": 16619
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.103531961664118e-06,
+      "loss": 18.9468,
+      "step": 16620
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.098987798008696e-06,
+      "loss": 0.7551,
+      "step": 16621
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.094444655251877e-06,
+      "loss": 19.2976,
+      "step": 16622
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.089902533507039e-06,
+      "loss": 0.7114,
+      "step": 16623
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.085361432887595e-06,
+      "loss": 0.6987,
+      "step": 16624
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.080821353506868e-06,
+      "loss": 17.1707,
+      "step": 16625
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.076282295478173e-06,
+      "loss": 16.2761,
+      "step": 16626
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.071744258914828e-06,
+      "loss": 17.0587,
+      "step": 16627
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.067207243930087e-06,
+      "loss": 18.4821,
+      "step": 16628
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.062671250637201e-06,
+      "loss": 16.6255,
+      "step": 16629
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.058136279149377e-06,
+      "loss": 20.2492,
+      "step": 16630
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.053602329579835e-06,
+      "loss": 0.7103,
+      "step": 16631
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.049069402041727e-06,
+      "loss": 0.7725,
+      "step": 16632
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.044537496648181e-06,
+      "loss": 16.212,
+      "step": 16633
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.040006613512341e-06,
+      "loss": 16.791,
+      "step": 16634
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.035476752747295e-06,
+      "loss": 16.8186,
+      "step": 16635
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.030947914466098e-06,
+      "loss": 16.5013,
+      "step": 16636
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.026420098781796e-06,
+      "loss": 18.8543,
+      "step": 16637
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.021893305807393e-06,
+      "loss": 0.8138,
+      "step": 16638
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.017367535655902e-06,
+      "loss": 17.7971,
+      "step": 16639
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.01284278844028e-06,
+      "loss": 18.2316,
+      "step": 16640
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.008319064273441e-06,
+      "loss": 0.7633,
+      "step": 16641
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.00379636326834e-06,
+      "loss": 0.8263,
+      "step": 16642
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.999274685537823e-06,
+      "loss": 0.7471,
+      "step": 16643
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.994754031194785e-06,
+      "loss": 0.7901,
+      "step": 16644
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.990234400352044e-06,
+      "loss": 17.7896,
+      "step": 16645
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.985715793122407e-06,
+      "loss": 18.0293,
+      "step": 16646
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.98119820961868e-06,
+      "loss": 0.7308,
+      "step": 16647
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.976681649953611e-06,
+      "loss": 15.9502,
+      "step": 16648
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.972166114239939e-06,
+      "loss": 0.7215,
+      "step": 16649
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.967651602590365e-06,
+      "loss": 0.7692,
+      "step": 16650
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.963138115117564e-06,
+      "loss": 0.7175,
+      "step": 16651
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.958625651934216e-06,
+      "loss": 17.3251,
+      "step": 16652
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.954114213152947e-06,
+      "loss": 16.73,
+      "step": 16653
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.949603798886347e-06,
+      "loss": 0.6863,
+      "step": 16654
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.945094409247028e-06,
+      "loss": 16.7378,
+      "step": 16655
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.940586044347505e-06,
+      "loss": 17.4704,
+      "step": 16656
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.936078704300344e-06,
+      "loss": 0.7758,
+      "step": 16657
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.931572389218019e-06,
+      "loss": 15.3938,
+      "step": 16658
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.927067099213038e-06,
+      "loss": 18.2268,
+      "step": 16659
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.92256283439784e-06,
+      "loss": 0.7408,
+      "step": 16660
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.918059594884842e-06,
+      "loss": 16.6524,
+      "step": 16661
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.91355738078648e-06,
+      "loss": 0.7227,
+      "step": 16662
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.909056192215081e-06,
+      "loss": 0.7261,
+      "step": 16663
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.90455602928304e-06,
+      "loss": 20.1319,
+      "step": 16664
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.90005689210266e-06,
+      "loss": 0.7044,
+      "step": 16665
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.895558780786228e-06,
+      "loss": 16.6549,
+      "step": 16666
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.891061695446051e-06,
+      "loss": 17.8336,
+      "step": 16667
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.886565636194361e-06,
+      "loss": 17.684,
+      "step": 16668
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.882070603143378e-06,
+      "loss": 0.7696,
+      "step": 16669
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.8775765964053e-06,
+      "loss": 15.0076,
+      "step": 16670
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.873083616092292e-06,
+      "loss": 0.7146,
+      "step": 16671
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.868591662316517e-06,
+      "loss": 16.7576,
+      "step": 16672
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.864100735190084e-06,
+      "loss": 0.7588,
+      "step": 16673
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.859610834825078e-06,
+      "loss": 0.7561,
+      "step": 16674
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.855121961333596e-06,
+      "loss": 18.2262,
+      "step": 16675
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.85063411482766e-06,
+      "loss": 17.2421,
+      "step": 16676
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.846147295419293e-06,
+      "loss": 0.7171,
+      "step": 16677
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.841661503220484e-06,
+      "loss": 0.7906,
+      "step": 16678
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.837176738343189e-06,
+      "loss": 19.7218,
+      "step": 16679
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.832693000899378e-06,
+      "loss": 17.9303,
+      "step": 16680
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.828210291000932e-06,
+      "loss": 0.7515,
+      "step": 16681
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.823728608759774e-06,
+      "loss": 16.2743,
+      "step": 16682
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.819247954287763e-06,
+      "loss": 16.2417,
+      "step": 16683
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.814768327696698e-06,
+      "loss": 17.9573,
+      "step": 16684
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.810289729098437e-06,
+      "loss": 0.7611,
+      "step": 16685
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.805812158604737e-06,
+      "loss": 0.8255,
+      "step": 16686
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.801335616327377e-06,
+      "loss": 16.5255,
+      "step": 16687
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.796860102378097e-06,
+      "loss": 0.7434,
+      "step": 16688
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.792385616868576e-06,
+      "loss": 18.8191,
+      "step": 16689
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.787912159910544e-06,
+      "loss": 17.6212,
+      "step": 16690
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.78343973161561e-06,
+      "loss": 0.7874,
+      "step": 16691
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.77896833209545e-06,
+      "loss": 0.7627,
+      "step": 16692
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.774497961461647e-06,
+      "loss": 0.7936,
+      "step": 16693
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.770028619825777e-06,
+      "loss": 0.7756,
+      "step": 16694
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.76556030729942e-06,
+      "loss": 0.7543,
+      "step": 16695
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.761093023994088e-06,
+      "loss": 0.7377,
+      "step": 16696
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.756626770021292e-06,
+      "loss": 0.7487,
+      "step": 16697
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.752161545492515e-06,
+      "loss": 17.5593,
+      "step": 16698
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.747697350519185e-06,
+      "loss": 16.4347,
+      "step": 16699
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.743234185212762e-06,
+      "loss": 19.6826,
+      "step": 16700
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.738772049684635e-06,
+      "loss": 0.7575,
+      "step": 16701
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.734310944046164e-06,
+      "loss": 0.7792,
+      "step": 16702
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.729850868408729e-06,
+      "loss": 0.8052,
+      "step": 16703
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.725391822883638e-06,
+      "loss": 15.3335,
+      "step": 16704
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.720933807582193e-06,
+      "loss": 0.7552,
+      "step": 16705
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.716476822615665e-06,
+      "loss": 0.7388,
+      "step": 16706
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.71202086809529e-06,
+      "loss": 16.2487,
+      "step": 16707
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.707565944132312e-06,
+      "loss": 17.2869,
+      "step": 16708
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.703112050837902e-06,
+      "loss": 0.7661,
+      "step": 16709
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.69865918832326e-06,
+      "loss": 0.7611,
+      "step": 16710
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.69420735669953e-06,
+      "loss": 0.7533,
+      "step": 16711
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.689756556077788e-06,
+      "loss": 17.265,
+      "step": 16712
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.685306786569163e-06,
+      "loss": 14.7145,
+      "step": 16713
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.680858048284707e-06,
+      "loss": 0.7501,
+      "step": 16714
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.676410341335478e-06,
+      "loss": 18.2947,
+      "step": 16715
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.671963665832483e-06,
+      "loss": 18.519,
+      "step": 16716
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.6675180218867e-06,
+      "loss": 0.6999,
+      "step": 16717
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.663073409609129e-06,
+      "loss": 16.9721,
+      "step": 16718
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.65862982911066e-06,
+      "loss": 0.7467,
+      "step": 16719
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.65418728050224e-06,
+      "loss": 17.319,
+      "step": 16720
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.64974576389475e-06,
+      "loss": 17.5665,
+      "step": 16721
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.645305279399036e-06,
+      "loss": 19.6814,
+      "step": 16722
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.640865827125948e-06,
+      "loss": 0.8138,
+      "step": 16723
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.636427407186298e-06,
+      "loss": 20.5575,
+      "step": 16724
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.631990019690867e-06,
+      "loss": 18.7317,
+      "step": 16725
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.62755366475041e-06,
+      "loss": 0.709,
+      "step": 16726
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.623118342475644e-06,
+      "loss": 18.1292,
+      "step": 16727
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.618684052977305e-06,
+      "loss": 16.9678,
+      "step": 16728
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.614250796366064e-06,
+      "loss": 17.1744,
+      "step": 16729
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.60981857275256e-06,
+      "loss": 0.7416,
+      "step": 16730
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.605387382247455e-06,
+      "loss": 0.7636,
+      "step": 16731
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.600957224961309e-06,
+      "loss": 0.8469,
+      "step": 16732
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.596528101004737e-06,
+      "loss": 16.32,
+      "step": 16733
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.592100010488279e-06,
+      "loss": 16.2957,
+      "step": 16734
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.587672953522447e-06,
+      "loss": 0.7292,
+      "step": 16735
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.583246930217765e-06,
+      "loss": 18.5806,
+      "step": 16736
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.578821940684678e-06,
+      "loss": 16.9558,
+      "step": 16737
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.574397985033683e-06,
+      "loss": 0.7457,
+      "step": 16738
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.569975063375164e-06,
+      "loss": 18.2943,
+      "step": 16739
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.565553175819513e-06,
+      "loss": 0.7213,
+      "step": 16740
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.561132322477123e-06,
+      "loss": 0.7755,
+      "step": 16741
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.556712503458325e-06,
+      "loss": 0.8019,
+      "step": 16742
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.552293718873456e-06,
+      "loss": 20.246,
+      "step": 16743
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.547875968832797e-06,
+      "loss": 0.7761,
+      "step": 16744
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.543459253446623e-06,
+      "loss": 0.7785,
+      "step": 16745
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.539043572825174e-06,
+      "loss": 15.2523,
+      "step": 16746
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.534628927078641e-06,
+      "loss": 16.8288,
+      "step": 16747
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.530215316317259e-06,
+      "loss": 17.0347,
+      "step": 16748
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.525802740651168e-06,
+      "loss": 0.7751,
+      "step": 16749
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.521391200190504e-06,
+      "loss": 0.7596,
+      "step": 16750
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.51698069504539e-06,
+      "loss": 18.3051,
+      "step": 16751
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.512571225325917e-06,
+      "loss": 16.5086,
+      "step": 16752
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.508162791142132e-06,
+      "loss": 0.7794,
+      "step": 16753
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.503755392604084e-06,
+      "loss": 0.7748,
+      "step": 16754
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.499349029821763e-06,
+      "loss": 18.2719,
+      "step": 16755
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.494943702905173e-06,
+      "loss": 18.0444,
+      "step": 16756
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.490539411964272e-06,
+      "loss": 17.3229,
+      "step": 16757
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.486136157108965e-06,
+      "loss": 0.7394,
+      "step": 16758
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.48173393844921e-06,
+      "loss": 17.0653,
+      "step": 16759
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.47733275609483e-06,
+      "loss": 16.1388,
+      "step": 16760
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.472932610155715e-06,
+      "loss": 0.7238,
+      "step": 16761
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.468533500741683e-06,
+      "loss": 0.7269,
+      "step": 16762
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.464135427962527e-06,
+      "loss": 17.6094,
+      "step": 16763
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.459738391928052e-06,
+      "loss": 0.7273,
+      "step": 16764
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.455342392747973e-06,
+      "loss": 0.768,
+      "step": 16765
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.450947430532057e-06,
+      "loss": 0.7685,
+      "step": 16766
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.44655350538997e-06,
+      "loss": 0.76,
+      "step": 16767
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.442160617431388e-06,
+      "loss": 0.7329,
+      "step": 16768
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.437768766765974e-06,
+      "loss": 0.7349,
+      "step": 16769
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.433377953503323e-06,
+      "loss": 15.3016,
+      "step": 16770
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.428988177753071e-06,
+      "loss": 0.739,
+      "step": 16771
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.424599439624765e-06,
+      "loss": 0.7721,
+      "step": 16772
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.420211739227945e-06,
+      "loss": 0.7039,
+      "step": 16773
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.415825076672135e-06,
+      "loss": 0.759,
+      "step": 16774
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.411439452066816e-06,
+      "loss": 0.7583,
+      "step": 16775
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.407054865521475e-06,
+      "loss": 0.7416,
+      "step": 16776
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.402671317145543e-06,
+      "loss": 0.7919,
+      "step": 16777
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.398288807048421e-06,
+      "loss": 0.821,
+      "step": 16778
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.393907335339517e-06,
+      "loss": 0.7143,
+      "step": 16779
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.389526902128191e-06,
+      "loss": 0.7202,
+      "step": 16780
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.385147507523772e-06,
+      "loss": 0.7693,
+      "step": 16781
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.380769151635575e-06,
+      "loss": 0.7574,
+      "step": 16782
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.376391834572866e-06,
+      "loss": 0.7179,
+      "step": 16783
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.37201555644494e-06,
+      "loss": 0.7621,
+      "step": 16784
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.367640317361003e-06,
+      "loss": 0.7774,
+      "step": 16785
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.363266117430263e-06,
+      "loss": 0.7697,
+      "step": 16786
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.358892956761927e-06,
+      "loss": 0.8134,
+      "step": 16787
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.354520835465112e-06,
+      "loss": 0.8016,
+      "step": 16788
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.350149753648973e-06,
+      "loss": 0.6988,
+      "step": 16789
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.345779711422608e-06,
+      "loss": 0.8164,
+      "step": 16790
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.34141070889508e-06,
+      "loss": 0.7661,
+      "step": 16791
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.337042746175467e-06,
+      "loss": 0.7379,
+      "step": 16792
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.332675823372765e-06,
+      "loss": 16.7431,
+      "step": 16793
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.328309940596013e-06,
+      "loss": 19.2217,
+      "step": 16794
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.32394509795415e-06,
+      "loss": 0.7849,
+      "step": 16795
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.319581295556123e-06,
+      "loss": 17.6408,
+      "step": 16796
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.315218533510872e-06,
+      "loss": 16.9486,
+      "step": 16797
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.31085681192727e-06,
+      "loss": 0.7356,
+      "step": 16798
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.306496130914216e-06,
+      "loss": 17.2349,
+      "step": 16799
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.302136490580536e-06,
+      "loss": 0.764,
+      "step": 16800
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.297777891035053e-06,
+      "loss": 15.7657,
+      "step": 16801
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.293420332386553e-06,
+      "loss": 0.81,
+      "step": 16802
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.289063814743798e-06,
+      "loss": 17.1589,
+      "step": 16803
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.28470833821554e-06,
+      "loss": 17.4341,
+      "step": 16804
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.280353902910487e-06,
+      "loss": 0.7833,
+      "step": 16805
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.276000508937315e-06,
+      "loss": 16.9939,
+      "step": 16806
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.271648156404727e-06,
+      "loss": 15.3953,
+      "step": 16807
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.267296845421296e-06,
+      "loss": 0.7599,
+      "step": 16808
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.262946576095682e-06,
+      "loss": 18.0489,
+      "step": 16809
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.25859734853645e-06,
+      "loss": 17.4815,
+      "step": 16810
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.254249162852146e-06,
+      "loss": 16.777,
+      "step": 16811
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.24990201915133e-06,
+      "loss": 16.6784,
+      "step": 16812
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.245555917542485e-06,
+      "loss": 16.5024,
+      "step": 16813
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.241210858134108e-06,
+      "loss": 17.1011,
+      "step": 16814
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.236866841034635e-06,
+      "loss": 16.5703,
+      "step": 16815
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.232523866352492e-06,
+      "loss": 16.9207,
+      "step": 16816
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.2281819341961e-06,
+      "loss": 19.6245,
+      "step": 16817
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.22384104467383e-06,
+      "loss": 0.7784,
+      "step": 16818
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.21950119789401e-06,
+      "loss": 16.9951,
+      "step": 16819
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.215162393964992e-06,
+      "loss": 17.5768,
+      "step": 16820
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.210824632995062e-06,
+      "loss": 0.7653,
+      "step": 16821
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.206487915092493e-06,
+      "loss": 19.742,
+      "step": 16822
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.202152240365524e-06,
+      "loss": 0.7548,
+      "step": 16823
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.19781760892237e-06,
+      "loss": 19.9114,
+      "step": 16824
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.193484020871244e-06,
+      "loss": 0.7599,
+      "step": 16825
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.189151476320295e-06,
+      "loss": 0.7762,
+      "step": 16826
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.184819975377679e-06,
+      "loss": 18.5609,
+      "step": 16827
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.180489518151513e-06,
+      "loss": 17.1744,
+      "step": 16828
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.176160104749858e-06,
+      "loss": 17.1314,
+      "step": 16829
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.17183173528081e-06,
+      "loss": 16.683,
+      "step": 16830
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.167504409852377e-06,
+      "loss": 0.7508,
+      "step": 16831
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.163178128572596e-06,
+      "loss": 0.7615,
+      "step": 16832
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.158852891549446e-06,
+      "loss": 0.7134,
+      "step": 16833
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.154528698890867e-06,
+      "loss": 0.7647,
+      "step": 16834
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.150205550704831e-06,
+      "loss": 15.9279,
+      "step": 16835
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.145883447099195e-06,
+      "loss": 18.6146,
+      "step": 16836
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.141562388181879e-06,
+      "loss": 0.7381,
+      "step": 16837
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.137242374060722e-06,
+      "loss": 16.9068,
+      "step": 16838
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.13292340484354e-06,
+      "loss": 0.7799,
+      "step": 16839
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.128605480638162e-06,
+      "loss": 0.7823,
+      "step": 16840
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.124288601552355e-06,
+      "loss": 0.7606,
+      "step": 16841
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.119972767693862e-06,
+      "loss": 17.2688,
+      "step": 16842
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.11565797917041e-06,
+      "loss": 18.0886,
+      "step": 16843
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.11134423608969e-06,
+      "loss": 18.9064,
+      "step": 16844
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.107031538559396e-06,
+      "loss": 0.8166,
+      "step": 16845
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.102719886687155e-06,
+      "loss": 0.7228,
+      "step": 16846
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.098409280580582e-06,
+      "loss": 17.4465,
+      "step": 16847
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.094099720347293e-06,
+      "loss": 0.7338,
+      "step": 16848
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.089791206094843e-06,
+      "loss": 0.7703,
+      "step": 16849
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.085483737930776e-06,
+      "loss": 0.7608,
+      "step": 16850
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.081177315962602e-06,
+      "loss": 16.8075,
+      "step": 16851
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.076871940297803e-06,
+      "loss": 17.9121,
+      "step": 16852
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.072567611043869e-06,
+      "loss": 0.7551,
+      "step": 16853
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.0682643283082e-06,
+      "loss": 16.382,
+      "step": 16854
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.063962092198246e-06,
+      "loss": 19.4648,
+      "step": 16855
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.05966090282138e-06,
+      "loss": 16.0834,
+      "step": 16856
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.055360760284935e-06,
+      "loss": 18.1514,
+      "step": 16857
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.051061664696274e-06,
+      "loss": 16.4396,
+      "step": 16858
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.046763616162679e-06,
+      "loss": 14.9838,
+      "step": 16859
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.042466614791455e-06,
+      "loss": 0.7618,
+      "step": 16860
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.038170660689848e-06,
+      "loss": 0.7555,
+      "step": 16861
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.033875753965064e-06,
+      "loss": 0.7444,
+      "step": 16862
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.029581894724353e-06,
+      "loss": 19.2191,
+      "step": 16863
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.025289083074832e-06,
+      "loss": 15.8187,
+      "step": 16864
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.020997319123697e-06,
+      "loss": 0.8038,
+      "step": 16865
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.016706602978053e-06,
+      "loss": 0.784,
+      "step": 16866
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.012416934744981e-06,
+      "loss": 0.749,
+      "step": 16867
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.008128314531587e-06,
+      "loss": 16.3234,
+      "step": 16868
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.003840742444896e-06,
+      "loss": 0.7633,
+      "step": 16869
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.999554218591925e-06,
+      "loss": 19.3557,
+      "step": 16870
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.995268743079675e-06,
+      "loss": 0.8018,
+      "step": 16871
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.990984316015099e-06,
+      "loss": 16.4232,
+      "step": 16872
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.986700937505154e-06,
+      "loss": 15.5357,
+      "step": 16873
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.98241860765675e-06,
+      "loss": 0.7876,
+      "step": 16874
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.978137326576757e-06,
+      "loss": 0.723,
+      "step": 16875
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.973857094372061e-06,
+      "loss": 16.5329,
+      "step": 16876
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.96957791114949e-06,
+      "loss": 0.7435,
+      "step": 16877
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.96529977701585e-06,
+      "loss": 17.0616,
+      "step": 16878
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.96102269207793e-06,
+      "loss": 0.7368,
+      "step": 16879
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.956746656442466e-06,
+      "loss": 17.1233,
+      "step": 16880
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.952471670216217e-06,
+      "loss": 0.7192,
+      "step": 16881
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.948197733505863e-06,
+      "loss": 18.0923,
+      "step": 16882
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.943924846418121e-06,
+      "loss": 0.7054,
+      "step": 16883
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.939653009059606e-06,
+      "loss": 0.7794,
+      "step": 16884
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.935382221536942e-06,
+      "loss": 0.7713,
+      "step": 16885
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.931112483956754e-06,
+      "loss": 20.4267,
+      "step": 16886
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.926843796425598e-06,
+      "loss": 0.7862,
+      "step": 16887
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.922576159050033e-06,
+      "loss": 0.7376,
+      "step": 16888
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.918309571936578e-06,
+      "loss": 15.8243,
+      "step": 16889
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.91404403519172e-06,
+      "loss": 19.7631,
+      "step": 16890
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.909779548921937e-06,
+      "loss": 18.9988,
+      "step": 16891
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.905516113233653e-06,
+      "loss": 19.6323,
+      "step": 16892
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.90125372823331e-06,
+      "loss": 0.7341,
+      "step": 16893
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.896992394027286e-06,
+      "loss": 14.6188,
+      "step": 16894
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.892732110721935e-06,
+      "loss": 0.7609,
+      "step": 16895
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.888472878423615e-06,
+      "loss": 0.7139,
+      "step": 16896
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.88421469723863e-06,
+      "loss": 15.16,
+      "step": 16897
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.87995756727326e-06,
+      "loss": 0.7712,
+      "step": 16898
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.875701488633764e-06,
+      "loss": 13.5417,
+      "step": 16899
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.871446461426368e-06,
+      "loss": 0.7004,
+      "step": 16900
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.867192485757301e-06,
+      "loss": 16.581,
+      "step": 16901
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.862939561732729e-06,
+      "loss": 16.9773,
+      "step": 16902
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.858687689458794e-06,
+      "loss": 0.7679,
+      "step": 16903
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.854436869041659e-06,
+      "loss": 17.3487,
+      "step": 16904
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.850187100587381e-06,
+      "loss": 0.731,
+      "step": 16905
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.845938384202067e-06,
+      "loss": 0.754,
+      "step": 16906
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.841690719991757e-06,
+      "loss": 16.5576,
+      "step": 16907
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.837444108062459e-06,
+      "loss": 0.7428,
+      "step": 16908
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.8331985485202e-06,
+      "loss": 0.7213,
+      "step": 16909
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.828954041470914e-06,
+      "loss": 19.1396,
+      "step": 16910
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.824710587020596e-06,
+      "loss": 0.7584,
+      "step": 16911
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.820468185275104e-06,
+      "loss": 17.5189,
+      "step": 16912
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.816226836340368e-06,
+      "loss": 17.8371,
+      "step": 16913
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.811986540322242e-06,
+      "loss": 0.7523,
+      "step": 16914
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.80774729732655e-06,
+      "loss": 0.78,
+      "step": 16915
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.803509107459133e-06,
+      "loss": 0.7128,
+      "step": 16916
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.799271970825766e-06,
+      "loss": 15.6665,
+      "step": 16917
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.795035887532205e-06,
+      "loss": 18.5269,
+      "step": 16918
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.790800857684188e-06,
+      "loss": 16.7652,
+      "step": 16919
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.786566881387402e-06,
+      "loss": 17.2507,
+      "step": 16920
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.782333958747562e-06,
+      "loss": 17.0851,
+      "step": 16921
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.778102089870299e-06,
+      "loss": 16.801,
+      "step": 16922
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.773871274861244e-06,
+      "loss": 0.7908,
+      "step": 16923
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.76964151382601e-06,
+      "loss": 17.9695,
+      "step": 16924
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.76541280687017e-06,
+      "loss": 17.4857,
+      "step": 16925
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.761185154099266e-06,
+      "loss": 0.7117,
+      "step": 16926
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.756958555618826e-06,
+      "loss": 17.3382,
+      "step": 16927
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.752733011534335e-06,
+      "loss": 0.7719,
+      "step": 16928
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.748508521951286e-06,
+      "loss": 18.5772,
+      "step": 16929
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.744285086975112e-06,
+      "loss": 16.3367,
+      "step": 16930
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.740062706711221e-06,
+      "loss": 17.5376,
+      "step": 16931
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.735841381265036e-06,
+      "loss": 18.2526,
+      "step": 16932
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.731621110741872e-06,
+      "loss": 0.7676,
+      "step": 16933
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.72740189524711e-06,
+      "loss": 19.8115,
+      "step": 16934
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.723183734886052e-06,
+      "loss": 16.8214,
+      "step": 16935
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.718966629763969e-06,
+      "loss": 0.7386,
+      "step": 16936
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.71475057998614e-06,
+      "loss": 0.7288,
+      "step": 16937
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.710535585657786e-06,
+      "loss": 0.7393,
+      "step": 16938
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.70632164688414e-06,
+      "loss": 18.7152,
+      "step": 16939
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.702108763770338e-06,
+      "loss": 16.1903,
+      "step": 16940
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.697896936421573e-06,
+      "loss": 0.779,
+      "step": 16941
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.693686164942958e-06,
+      "loss": 18.1466,
+      "step": 16942
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.689476449439586e-06,
+      "loss": 0.7567,
+      "step": 16943
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.685267790016553e-06,
+      "loss": 17.0887,
+      "step": 16944
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.681060186778899e-06,
+      "loss": 0.7936,
+      "step": 16945
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.676853639831643e-06,
+      "loss": 0.7431,
+      "step": 16946
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.672648149279787e-06,
+      "loss": 0.7143,
+      "step": 16947
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.668443715228285e-06,
+      "loss": 17.162,
+      "step": 16948
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.664240337782102e-06,
+      "loss": 0.7685,
+      "step": 16949
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.66003801704615e-06,
+      "loss": 18.7369,
+      "step": 16950
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.6558367531253e-06,
+      "loss": 17.9986,
+      "step": 16951
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.651636546124452e-06,
+      "loss": 16.7886,
+      "step": 16952
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.647437396148421e-06,
+      "loss": 0.7923,
+      "step": 16953
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.643239303302024e-06,
+      "loss": 18.004,
+      "step": 16954
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.639042267690044e-06,
+      "loss": 0.7478,
+      "step": 16955
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.63484628941723e-06,
+      "loss": 19.622,
+      "step": 16956
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.630651368588337e-06,
+      "loss": 17.5762,
+      "step": 16957
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.626457505308066e-06,
+      "loss": 0.7162,
+      "step": 16958
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.622264699681081e-06,
+      "loss": 16.8615,
+      "step": 16959
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.618072951812055e-06,
+      "loss": 14.7944,
+      "step": 16960
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.6138822618055885e-06,
+      "loss": 17.3389,
+      "step": 16961
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.609692629766308e-06,
+      "loss": 17.58,
+      "step": 16962
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.605504055798779e-06,
+      "loss": 17.1693,
+      "step": 16963
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.601316540007542e-06,
+      "loss": 0.7684,
+      "step": 16964
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.597130082497134e-06,
+      "loss": 0.7337,
+      "step": 16965
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.592944683372038e-06,
+      "loss": 17.5661,
+      "step": 16966
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.5887603427367285e-06,
+      "loss": 0.7733,
+      "step": 16967
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.5845770606956355e-06,
+      "loss": 18.9501,
+      "step": 16968
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.580394837353188e-06,
+      "loss": 0.7584,
+      "step": 16969
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.576213672813776e-06,
+      "loss": 0.7542,
+      "step": 16970
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.572033567181747e-06,
+      "loss": 17.2534,
+      "step": 16971
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.567854520561457e-06,
+      "loss": 0.7633,
+      "step": 16972
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.5636765330572065e-06,
+      "loss": 17.9857,
+      "step": 16973
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.559499604773279e-06,
+      "loss": 0.7363,
+      "step": 16974
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.555323735813935e-06,
+      "loss": 0.759,
+      "step": 16975
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.551148926283385e-06,
+      "loss": 17.8489,
+      "step": 16976
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.54697517628587e-06,
+      "loss": 0.7636,
+      "step": 16977
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.542802485925543e-06,
+      "loss": 0.7082,
+      "step": 16978
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.5386308553065445e-06,
+      "loss": 16.5729,
+      "step": 16979
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.534460284533046e-06,
+      "loss": 17.9058,
+      "step": 16980
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.530290773709087e-06,
+      "loss": 0.7325,
+      "step": 16981
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.526122322938778e-06,
+      "loss": 15.9049,
+      "step": 16982
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.521954932326153e-06,
+      "loss": 18.543,
+      "step": 16983
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.517788601975223e-06,
+      "loss": 17.9902,
+      "step": 16984
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.513623331990005e-06,
+      "loss": 0.7656,
+      "step": 16985
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.509459122474444e-06,
+      "loss": 17.5147,
+      "step": 16986
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.5052959735324894e-06,
+      "loss": 0.7285,
+      "step": 16987
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.501133885268047e-06,
+      "loss": 18.5567,
+      "step": 16988
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.496972857784995e-06,
+      "loss": 0.8209,
+      "step": 16989
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.492812891187212e-06,
+      "loss": 0.7894,
+      "step": 16990
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.4886539855785305e-06,
+      "loss": 16.6506,
+      "step": 16991
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.484496141062741e-06,
+      "loss": 16.1243,
+      "step": 16992
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.480339357743649e-06,
+      "loss": 17.0609,
+      "step": 16993
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.476183635724993e-06,
+      "loss": 0.77,
+      "step": 16994
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.472028975110501e-06,
+      "loss": 16.5245,
+      "step": 16995
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.467875376003863e-06,
+      "loss": 0.7305,
+      "step": 16996
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.46372283850878e-06,
+      "loss": 19.0928,
+      "step": 16997
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.45957136272889e-06,
+      "loss": 0.7338,
+      "step": 16998
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.4554209487678e-06,
+      "loss": 17.3415,
+      "step": 16999
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.451271596729126e-06,
+      "loss": 16.8221,
+      "step": 17000
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.4471233067164305e-06,
+      "loss": 0.7757,
+      "step": 17001
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.442976078833252e-06,
+      "loss": 0.7848,
+      "step": 17002
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.438829913183109e-06,
+      "loss": 16.0911,
+      "step": 17003
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.434684809869475e-06,
+      "loss": 18.0243,
+      "step": 17004
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.430540768995841e-06,
+      "loss": 17.7254,
+      "step": 17005
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.426397790665629e-06,
+      "loss": 18.0517,
+      "step": 17006
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.422255874982237e-06,
+      "loss": 16.8722,
+      "step": 17007
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.418115022049083e-06,
+      "loss": 0.6869,
+      "step": 17008
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.413975231969478e-06,
+      "loss": 18.6201,
+      "step": 17009
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.40983650484679e-06,
+      "loss": 0.7909,
+      "step": 17010
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.4056988407843035e-06,
+      "loss": 17.126,
+      "step": 17011
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.4015622398852845e-06,
+      "loss": 18.1909,
+      "step": 17012
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.397426702253018e-06,
+      "loss": 0.776,
+      "step": 17013
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.393292227990701e-06,
+      "loss": 17.591,
+      "step": 17014
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.389158817201542e-06,
+      "loss": 17.4735,
+      "step": 17015
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.385026469988709e-06,
+      "loss": 0.7597,
+      "step": 17016
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.380895186455334e-06,
+      "loss": 17.8901,
+      "step": 17017
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.376764966704558e-06,
+      "loss": 0.768,
+      "step": 17018
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.372635810839457e-06,
+      "loss": 0.7609,
+      "step": 17019
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.368507718963091e-06,
+      "loss": 16.2514,
+      "step": 17020
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.364380691178519e-06,
+      "loss": 19.6016,
+      "step": 17021
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.360254727588739e-06,
+      "loss": 17.8466,
+      "step": 17022
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.356129828296732e-06,
+      "loss": 17.7942,
+      "step": 17023
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.352005993405453e-06,
+      "loss": 0.8063,
+      "step": 17024
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.34788322301786e-06,
+      "loss": 18.07,
+      "step": 17025
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.34376151723683e-06,
+      "loss": 16.3749,
+      "step": 17026
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.339640876165249e-06,
+      "loss": 19.4582,
+      "step": 17027
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.335521299905979e-06,
+      "loss": 17.892,
+      "step": 17028
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.331402788561842e-06,
+      "loss": 0.7879,
+      "step": 17029
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.327285342235629e-06,
+      "loss": 0.7233,
+      "step": 17030
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.323168961030125e-06,
+      "loss": 16.7817,
+      "step": 17031
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.3190536450480516e-06,
+      "loss": 17.9211,
+      "step": 17032
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.314939394392156e-06,
+      "loss": 16.1579,
+      "step": 17033
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.310826209165117e-06,
+      "loss": 18.9456,
+      "step": 17034
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.306714089469607e-06,
+      "loss": 0.7585,
+      "step": 17035
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.302603035408262e-06,
+      "loss": 0.7748,
+      "step": 17036
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.298493047083682e-06,
+      "loss": 0.7658,
+      "step": 17037
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.294384124598475e-06,
+      "loss": 0.7653,
+      "step": 17038
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.290276268055196e-06,
+      "loss": 0.8223,
+      "step": 17039
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.286169477556354e-06,
+      "loss": 18.6379,
+      "step": 17040
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.2820637532044935e-06,
+      "loss": 0.7933,
+      "step": 17041
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.277959095102072e-06,
+      "loss": 18.0836,
+      "step": 17042
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.273855503351546e-06,
+      "loss": 17.5757,
+      "step": 17043
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.269752978055344e-06,
+      "loss": 18.3117,
+      "step": 17044
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.265651519315852e-06,
+      "loss": 16.8065,
+      "step": 17045
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.261551127235466e-06,
+      "loss": 17.5734,
+      "step": 17046
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.2574518019165264e-06,
+      "loss": 0.7393,
+      "step": 17047
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.253353543461333e-06,
+      "loss": 17.3936,
+      "step": 17048
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.249256351972217e-06,
+      "loss": 0.7977,
+      "step": 17049
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.245160227551406e-06,
+      "loss": 0.7524,
+      "step": 17050
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.24106517030117e-06,
+      "loss": 0.7506,
+      "step": 17051
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.236971180323693e-06,
+      "loss": 18.3444,
+      "step": 17052
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.232878257721193e-06,
+      "loss": 18.7389,
+      "step": 17053
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.228786402595822e-06,
+      "loss": 0.7524,
+      "step": 17054
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.224695615049692e-06,
+      "loss": 0.7442,
+      "step": 17055
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.220605895184945e-06,
+      "loss": 0.7898,
+      "step": 17056
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.216517243103626e-06,
+      "loss": 0.7785,
+      "step": 17057
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.212429658907816e-06,
+      "loss": 18.2741,
+      "step": 17058
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.208343142699531e-06,
+      "loss": 0.7087,
+      "step": 17059
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.204257694580763e-06,
+      "loss": 0.7331,
+      "step": 17060
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.2001733146535025e-06,
+      "loss": 16.5905,
+      "step": 17061
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.19609000301969e-06,
+      "loss": 0.7603,
+      "step": 17062
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.192007759781239e-06,
+      "loss": 17.6218,
+      "step": 17063
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.187926585040056e-06,
+      "loss": 0.7201,
+      "step": 17064
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.183846478897988e-06,
+      "loss": 18.4633,
+      "step": 17065
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.179767441456897e-06,
+      "loss": 0.705,
+      "step": 17066
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.175689472818587e-06,
+      "loss": 17.5281,
+      "step": 17067
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.171612573084835e-06,
+      "loss": 0.8153,
+      "step": 17068
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.167536742357428e-06,
+      "loss": 18.2758,
+      "step": 17069
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.163461980738079e-06,
+      "loss": 0.788,
+      "step": 17070
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.1593882883285e-06,
+      "loss": 0.7784,
+      "step": 17071
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.155315665230372e-06,
+      "loss": 0.7287,
+      "step": 17072
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.151244111545341e-06,
+      "loss": 18.3163,
+      "step": 17073
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.147173627375048e-06,
+      "loss": 17.4224,
+      "step": 17074
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.143104212821072e-06,
+      "loss": 17.6936,
+      "step": 17075
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.139035867985011e-06,
+      "loss": 0.7515,
+      "step": 17076
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.134968592968416e-06,
+      "loss": 0.7863,
+      "step": 17077
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.1309023878727735e-06,
+      "loss": 15.9149,
+      "step": 17078
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.126837252799601e-06,
+      "loss": 0.7728,
+      "step": 17079
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.122773187850351e-06,
+      "loss": 16.3781,
+      "step": 17080
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.1187101931264825e-06,
+      "loss": 16.6045,
+      "step": 17081
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.114648268729396e-06,
+      "loss": 17.2077,
+      "step": 17082
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.1105874147604726e-06,
+      "loss": 0.7281,
+      "step": 17083
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.106527631321103e-06,
+      "loss": 0.718,
+      "step": 17084
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.102468918512578e-06,
+      "loss": 0.7586,
+      "step": 17085
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.098411276436229e-06,
+      "loss": 0.7251,
+      "step": 17086
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.094354705193335e-06,
+      "loss": 18.123,
+      "step": 17087
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.0902992048851255e-06,
+      "loss": 17.6602,
+      "step": 17088
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.086244775612866e-06,
+      "loss": 0.7112,
+      "step": 17089
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.082191417477729e-06,
+      "loss": 0.7725,
+      "step": 17090
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.078139130580896e-06,
+      "loss": 0.7569,
+      "step": 17091
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.074087915023509e-06,
+      "loss": 16.3721,
+      "step": 17092
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.070037770906679e-06,
+      "loss": 0.7593,
+      "step": 17093
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.065988698331521e-06,
+      "loss": 0.738,
+      "step": 17094
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.061940697399089e-06,
+      "loss": 20.7881,
+      "step": 17095
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.057893768210405e-06,
+      "loss": 0.7404,
+      "step": 17096
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.053847910866513e-06,
+      "loss": 16.8955,
+      "step": 17097
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.049803125468385e-06,
+      "loss": 0.7381,
+      "step": 17098
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.04575941211697e-06,
+      "loss": 17.5646,
+      "step": 17099
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.0417167709132145e-06,
+      "loss": 0.7609,
+      "step": 17100
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.037675201958005e-06,
+      "loss": 0.726,
+      "step": 17101
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.033634705352238e-06,
+      "loss": 0.7261,
+      "step": 17102
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.0295952811967545e-06,
+      "loss": 19.3607,
+      "step": 17103
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.025556929592392e-06,
+      "loss": 17.0544,
+      "step": 17104
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.0215196506399515e-06,
+      "loss": 0.7173,
+      "step": 17105
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.017483444440176e-06,
+      "loss": 0.7601,
+      "step": 17106
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.013448311093834e-06,
+      "loss": 0.7574,
+      "step": 17107
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.0094142507016296e-06,
+      "loss": 0.7263,
+      "step": 17108
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.005381263364269e-06,
+      "loss": 16.2766,
+      "step": 17109
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.001349349182412e-06,
+      "loss": 0.7523,
+      "step": 17110
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.99731850825669e-06,
+      "loss": 0.795,
+      "step": 17111
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.99328874068772e-06,
+      "loss": 0.7284,
+      "step": 17112
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.989260046576074e-06,
+      "loss": 17.7075,
+      "step": 17113
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.985232426022325e-06,
+      "loss": 18.9786,
+      "step": 17114
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.981205879126995e-06,
+      "loss": 0.7833,
+      "step": 17115
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.9771804059905795e-06,
+      "loss": 0.7585,
+      "step": 17116
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.973156006713577e-06,
+      "loss": 0.7309,
+      "step": 17117
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.9691326813964235e-06,
+      "loss": 18.3242,
+      "step": 17118
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.965110430139543e-06,
+      "loss": 15.0703,
+      "step": 17119
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.961089253043329e-06,
+      "loss": 16.2782,
+      "step": 17120
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.957069150208151e-06,
+      "loss": 18.1948,
+      "step": 17121
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.953050121734362e-06,
+      "loss": 18.1894,
+      "step": 17122
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.9490321677222694e-06,
+      "loss": 18.1762,
+      "step": 17123
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.945015288272155e-06,
+      "loss": 15.3188,
+      "step": 17124
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.9409994834843115e-06,
+      "loss": 19.6334,
+      "step": 17125
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.936984753458925e-06,
+      "loss": 16.7789,
+      "step": 17126
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.9329710982962484e-06,
+      "loss": 17.2457,
+      "step": 17127
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.9289585180964454e-06,
+      "loss": 0.7371,
+      "step": 17128
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.924947012959665e-06,
+      "loss": 0.7373,
+      "step": 17129
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.920936582986048e-06,
+      "loss": 16.0513,
+      "step": 17130
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.916927228275683e-06,
+      "loss": 16.1304,
+      "step": 17131
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.912918948928671e-06,
+      "loss": 0.6511,
+      "step": 17132
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.908911745045027e-06,
+      "loss": 0.7826,
+      "step": 17133
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.904905616724777e-06,
+      "loss": 17.8613,
+      "step": 17134
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.900900564067936e-06,
+      "loss": 0.7956,
+      "step": 17135
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.896896587174445e-06,
+      "loss": 0.7125,
+      "step": 17136
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.892893686144264e-06,
+      "loss": 0.7382,
+      "step": 17137
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.888891861077301e-06,
+      "loss": 17.159,
+      "step": 17138
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.8848911120734436e-06,
+      "loss": 0.7581,
+      "step": 17139
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.880891439232545e-06,
+      "loss": 0.7359,
+      "step": 17140
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.876892842654431e-06,
+      "loss": 16.4617,
+      "step": 17141
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.872895322438932e-06,
+      "loss": 0.746,
+      "step": 17142
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.868898878685809e-06,
+      "loss": 16.787,
+      "step": 17143
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.864903511494814e-06,
+      "loss": 18.0666,
+      "step": 17144
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.8609092209656775e-06,
+      "loss": 0.7696,
+      "step": 17145
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.8569160071981054e-06,
+      "loss": 0.7143,
+      "step": 17146
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.852923870291755e-06,
+      "loss": 16.1693,
+      "step": 17147
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.8489328103462815e-06,
+      "loss": 17.328,
+      "step": 17148
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.844942827461281e-06,
+      "loss": 0.7414,
+      "step": 17149
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.840953921736376e-06,
+      "loss": 16.6929,
+      "step": 17150
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.836966093271113e-06,
+      "loss": 16.6716,
+      "step": 17151
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.832979342165024e-06,
+      "loss": 17.775,
+      "step": 17152
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.828993668517647e-06,
+      "loss": 17.4603,
+      "step": 17153
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.825009072428423e-06,
+      "loss": 0.7355,
+      "step": 17154
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.82102555399684e-06,
+      "loss": 0.6956,
+      "step": 17155
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.817043113322319e-06,
+      "loss": 17.656,
+      "step": 17156
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.813061750504252e-06,
+      "loss": 0.7303,
+      "step": 17157
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.809081465642036e-06,
+      "loss": 0.7573,
+      "step": 17158
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.805102258835e-06,
+      "loss": 19.8074,
+      "step": 17159
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.80112413018249e-06,
+      "loss": 0.7008,
+      "step": 17160
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.797147079783778e-06,
+      "loss": 16.8401,
+      "step": 17161
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.793171107738127e-06,
+      "loss": 0.7466,
+      "step": 17162
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.789196214144805e-06,
+      "loss": 16.9527,
+      "step": 17163
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.785222399102997e-06,
+      "loss": 16.8128,
+      "step": 17164
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.78124966271192e-06,
+      "loss": 16.091,
+      "step": 17165
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.777278005070714e-06,
+      "loss": 18.948,
+      "step": 17166
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.773307426278524e-06,
+      "loss": 0.7791,
+      "step": 17167
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.769337926434449e-06,
+      "loss": 0.7507,
+      "step": 17168
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.7653695056375535e-06,
+      "loss": 18.0949,
+      "step": 17169
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.761402163986924e-06,
+      "loss": 0.7284,
+      "step": 17170
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.757435901581566e-06,
+      "loss": 16.8163,
+      "step": 17171
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.753470718520472e-06,
+      "loss": 15.9899,
+      "step": 17172
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.749506614902634e-06,
+      "loss": 17.7434,
+      "step": 17173
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.745543590826986e-06,
+      "loss": 0.7677,
+      "step": 17174
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.741581646392448e-06,
+      "loss": 16.274,
+      "step": 17175
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.737620781697906e-06,
+      "loss": 16.8662,
+      "step": 17176
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.733660996842223e-06,
+      "loss": 0.7742,
+      "step": 17177
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.729702291924244e-06,
+      "loss": 18.0538,
+      "step": 17178
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.725744667042777e-06,
+      "loss": 0.7871,
+      "step": 17179
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.721788122296596e-06,
+      "loss": 17.8521,
+      "step": 17180
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.717832657784484e-06,
+      "loss": 0.7934,
+      "step": 17181
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.713878273605134e-06,
+      "loss": 0.7628,
+      "step": 17182
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.709924969857273e-06,
+      "loss": 0.7426,
+      "step": 17183
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.7059727466395716e-06,
+      "loss": 17.7017,
+      "step": 17184
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.702021604050662e-06,
+      "loss": 0.7953,
+      "step": 17185
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.698071542189188e-06,
+      "loss": 17.2666,
+      "step": 17186
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.694122561153742e-06,
+      "loss": 0.6978,
+      "step": 17187
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.69017466104288e-06,
+      "loss": 0.7593,
+      "step": 17188
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.686227841955151e-06,
+      "loss": 0.7631,
+      "step": 17189
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.682282103989052e-06,
+      "loss": 0.7212,
+      "step": 17190
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.6783374472430934e-06,
+      "loss": 15.6928,
+      "step": 17191
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.674393871815715e-06,
+      "loss": 0.7311,
+      "step": 17192
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.670451377805365e-06,
+      "loss": 15.3213,
+      "step": 17193
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.666509965310452e-06,
+      "loss": 0.7292,
+      "step": 17194
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.662569634429327e-06,
+      "loss": 0.756,
+      "step": 17195
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.658630385260367e-06,
+      "loss": 17.6082,
+      "step": 17196
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.654692217901881e-06,
+      "loss": 17.4078,
+      "step": 17197
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.650755132452185e-06,
+      "loss": 0.7572,
+      "step": 17198
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.646819129009535e-06,
+      "loss": 0.7644,
+      "step": 17199
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.6428842076721735e-06,
+      "loss": 0.7974,
+      "step": 17200
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.638950368538338e-06,
+      "loss": 0.7677,
+      "step": 17201
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.6350176117061845e-06,
+      "loss": 17.1639,
+      "step": 17202
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.631085937273901e-06,
+      "loss": 0.7704,
+      "step": 17203
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.6271553453396144e-06,
+      "loss": 0.7626,
+      "step": 17204
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.6232258360014235e-06,
+      "loss": 0.7274,
+      "step": 17205
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.619297409357428e-06,
+      "loss": 15.6669,
+      "step": 17206
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.615370065505677e-06,
+      "loss": 0.6989,
+      "step": 17207
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.611443804544198e-06,
+      "loss": 17.1309,
+      "step": 17208
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.607518626570985e-06,
+      "loss": 0.7247,
+      "step": 17209
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.603594531684004e-06,
+      "loss": 17.6528,
+      "step": 17210
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.599671519981226e-06,
+      "loss": 20.0391,
+      "step": 17211
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.595749591560557e-06,
+      "loss": 18.1485,
+      "step": 17212
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.591828746519873e-06,
+      "loss": 19.4496,
+      "step": 17213
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.58790898495707e-06,
+      "loss": 0.7486,
+      "step": 17214
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.583990306969967e-06,
+      "loss": 0.6686,
+      "step": 17215
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.580072712656382e-06,
+      "loss": 17.2222,
+      "step": 17216
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.576156202114098e-06,
+      "loss": 17.5607,
+      "step": 17217
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.5722407754408524e-06,
+      "loss": 0.7739,
+      "step": 17218
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.568326432734412e-06,
+      "loss": 16.6027,
+      "step": 17219
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.564413174092443e-06,
+      "loss": 0.7794,
+      "step": 17220
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.560500999612651e-06,
+      "loss": 16.252,
+      "step": 17221
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.556589909392685e-06,
+      "loss": 0.7022,
+      "step": 17222
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.552679903530135e-06,
+      "loss": 18.4143,
+      "step": 17223
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.548770982122621e-06,
+      "loss": 15.4046,
+      "step": 17224
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.544863145267699e-06,
+      "loss": 16.9024,
+      "step": 17225
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.5409563930629195e-06,
+      "loss": 17.8327,
+      "step": 17226
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.537050725605792e-06,
+      "loss": 16.471,
+      "step": 17227
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.533146142993796e-06,
+      "loss": 17.4032,
+      "step": 17228
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.529242645324418e-06,
+      "loss": 16.763,
+      "step": 17229
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.525340232695043e-06,
+      "loss": 17.3131,
+      "step": 17230
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.52143890520312e-06,
+      "loss": 18.3911,
+      "step": 17231
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.517538662946004e-06,
+      "loss": 0.6885,
+      "step": 17232
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.513639506021041e-06,
+      "loss": 16.6098,
+      "step": 17233
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.5097414345255796e-06,
+      "loss": 0.7651,
+      "step": 17234
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.505844448556897e-06,
+      "loss": 0.7928,
+      "step": 17235
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.501948548212272e-06,
+      "loss": 0.7533,
+      "step": 17236
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.498053733588943e-06,
+      "loss": 17.1014,
+      "step": 17237
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.494160004784111e-06,
+      "loss": 0.7944,
+      "step": 17238
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.490267361894997e-06,
+      "loss": 0.7782,
+      "step": 17239
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.486375805018735e-06,
+      "loss": 0.7505,
+      "step": 17240
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.482485334252464e-06,
+      "loss": 0.8238,
+      "step": 17241
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.4785959496933055e-06,
+      "loss": 0.7727,
+      "step": 17242
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.474707651438328e-06,
+      "loss": 17.4761,
+      "step": 17243
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.470820439584585e-06,
+      "loss": 19.9293,
+      "step": 17244
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.466934314229106e-06,
+      "loss": 21.3048,
+      "step": 17245
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.463049275468874e-06,
+      "loss": 16.6862,
+      "step": 17246
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.459165323400878e-06,
+      "loss": 0.7874,
+      "step": 17247
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.455282458122053e-06,
+      "loss": 16.904,
+      "step": 17248
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.45140067972933e-06,
+      "loss": 18.103,
+      "step": 17249
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.4475199883196005e-06,
+      "loss": 17.4219,
+      "step": 17250
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.443640383989691e-06,
+      "loss": 0.7271,
+      "step": 17251
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.43976186683648e-06,
+      "loss": 0.8176,
+      "step": 17252
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.435884436956746e-06,
+      "loss": 0.7362,
+      "step": 17253
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.43200809444729e-06,
+      "loss": 17.5775,
+      "step": 17254
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.428132839404865e-06,
+      "loss": 18.0601,
+      "step": 17255
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.424258671926181e-06,
+      "loss": 0.7596,
+      "step": 17256
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.420385592107969e-06,
+      "loss": 15.8888,
+      "step": 17257
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.416513600046864e-06,
+      "loss": 0.7698,
+      "step": 17258
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.412642695839538e-06,
+      "loss": 0.7764,
+      "step": 17259
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.408772879582609e-06,
+      "loss": 0.7264,
+      "step": 17260
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.4049041513726485e-06,
+      "loss": 17.317,
+      "step": 17261
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.401036511306241e-06,
+      "loss": 16.8294,
+      "step": 17262
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.397169959479921e-06,
+      "loss": 18.2526,
+      "step": 17263
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.3933044959901935e-06,
+      "loss": 17.5638,
+      "step": 17264
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.389440120933549e-06,
+      "loss": 0.745,
+      "step": 17265
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.385576834406415e-06,
+      "loss": 16.0323,
+      "step": 17266
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.38171463650526e-06,
+      "loss": 0.7059,
+      "step": 17267
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.377853527326466e-06,
+      "loss": 16.1912,
+      "step": 17268
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.373993506966397e-06,
+      "loss": 0.6833,
+      "step": 17269
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.37013457552143e-06,
+      "loss": 16.863,
+      "step": 17270
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.366276733087845e-06,
+      "loss": 18.1263,
+      "step": 17271
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.362419979761969e-06,
+      "loss": 0.7132,
+      "step": 17272
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.3585643156400535e-06,
+      "loss": 17.5123,
+      "step": 17273
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.354709740818327e-06,
+      "loss": 16.0648,
+      "step": 17274
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.350856255393017e-06,
+      "loss": 0.7047,
+      "step": 17275
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.347003859460293e-06,
+      "loss": 0.7286,
+      "step": 17276
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.343152553116344e-06,
+      "loss": 0.7619,
+      "step": 17277
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.339302336457264e-06,
+      "loss": 0.713,
+      "step": 17278
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.335453209579156e-06,
+      "loss": 17.9607,
+      "step": 17279
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.331605172578115e-06,
+      "loss": 17.8169,
+      "step": 17280
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.3277582255501676e-06,
+      "loss": 18.0698,
+      "step": 17281
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.323912368591362e-06,
+      "loss": 0.7972,
+      "step": 17282
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.320067601797674e-06,
+      "loss": 0.8023,
+      "step": 17283
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.316223925265069e-06,
+      "loss": 0.6733,
+      "step": 17284
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.312381339089496e-06,
+      "loss": 17.9873,
+      "step": 17285
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.308539843366839e-06,
+      "loss": 0.7593,
+      "step": 17286
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.304699438193023e-06,
+      "loss": 0.6862,
+      "step": 17287
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.300860123663882e-06,
+      "loss": 17.155,
+      "step": 17288
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.297021899875244e-06,
+      "loss": 0.8078,
+      "step": 17289
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.293184766922921e-06,
+      "loss": 0.7554,
+      "step": 17290
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.289348724902694e-06,
+      "loss": 0.7465,
+      "step": 17291
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.285513773910301e-06,
+      "loss": 0.8008,
+      "step": 17292
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.281679914041461e-06,
+      "loss": 16.9378,
+      "step": 17293
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.277847145391863e-06,
+      "loss": 15.6612,
+      "step": 17294
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.274015468057199e-06,
+      "loss": 18.1162,
+      "step": 17295
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.27018488213309e-06,
+      "loss": 0.7055,
+      "step": 17296
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.26635538771514e-06,
+      "loss": 0.7393,
+      "step": 17297
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.262526984898964e-06,
+      "loss": 19.2594,
+      "step": 17298
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.258699673780083e-06,
+      "loss": 17.8063,
+      "step": 17299
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.254873454454058e-06,
+      "loss": 0.7739,
+      "step": 17300
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.251048327016374e-06,
+      "loss": 0.7419,
+      "step": 17301
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.247224291562509e-06,
+      "loss": 0.7388,
+      "step": 17302
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.2434013481879215e-06,
+      "loss": 0.7302,
+      "step": 17303
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.239579496988019e-06,
+      "loss": 17.2104,
+      "step": 17304
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.235758738058223e-06,
+      "loss": 0.7696,
+      "step": 17305
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.231939071493864e-06,
+      "loss": 18.4154,
+      "step": 17306
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.228120497390305e-06,
+      "loss": 16.6799,
+      "step": 17307
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.224303015842858e-06,
+      "loss": 16.3927,
+      "step": 17308
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.220486626946792e-06,
+      "loss": 0.7463,
+      "step": 17309
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.21667133079738e-06,
+      "loss": 17.3332,
+      "step": 17310
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.212857127489857e-06,
+      "loss": 0.7675,
+      "step": 17311
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.209044017119414e-06,
+      "loss": 17.4752,
+      "step": 17312
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.20523199978123e-06,
+      "loss": 16.627,
+      "step": 17313
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.201421075570446e-06,
+      "loss": 16.4284,
+      "step": 17314
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.1976112445822025e-06,
+      "loss": 17.4096,
+      "step": 17315
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.193802506911583e-06,
+      "loss": 0.7481,
+      "step": 17316
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.189994862653647e-06,
+      "loss": 15.3711,
+      "step": 17317
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.186188311903451e-06,
+      "loss": 17.3238,
+      "step": 17318
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.182382854755997e-06,
+      "loss": 21.2141,
+      "step": 17319
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.178578491306275e-06,
+      "loss": 0.7067,
+      "step": 17320
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.174775221649243e-06,
+      "loss": 0.7195,
+      "step": 17321
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.1709730458798074e-06,
+      "loss": 20.1875,
+      "step": 17322
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.1671719640929106e-06,
+      "loss": 0.7709,
+      "step": 17323
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.163371976383409e-06,
+      "loss": 17.2052,
+      "step": 17324
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.1595730828461375e-06,
+      "loss": 16.8345,
+      "step": 17325
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.155775283575959e-06,
+      "loss": 17.9424,
+      "step": 17326
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.151978578667617e-06,
+      "loss": 0.7673,
+      "step": 17327
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.1481829682159094e-06,
+      "loss": 18.3519,
+      "step": 17328
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.144388452315569e-06,
+      "loss": 0.7574,
+      "step": 17329
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.140595031061297e-06,
+      "loss": 18.4674,
+      "step": 17330
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.136802704547795e-06,
+      "loss": 18.3701,
+      "step": 17331
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.133011472869704e-06,
+      "loss": 18.6043,
+      "step": 17332
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.129221336121682e-06,
+      "loss": 17.045,
+      "step": 17333
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.125432294398298e-06,
+      "loss": 0.7134,
+      "step": 17334
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.121644347794148e-06,
+      "loss": 0.7749,
+      "step": 17335
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.1178574964037724e-06,
+      "loss": 17.8427,
+      "step": 17336
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.114071740321681e-06,
+      "loss": 16.6678,
+      "step": 17337
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.110287079642396e-06,
+      "loss": 16.3895,
+      "step": 17338
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.10650351446036e-06,
+      "loss": 17.325,
+      "step": 17339
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.1027210448700245e-06,
+      "loss": 16.5828,
+      "step": 17340
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.098939670965792e-06,
+      "loss": 19.004,
+      "step": 17341
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.095159392842037e-06,
+      "loss": 16.0103,
+      "step": 17342
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.091380210593145e-06,
+      "loss": 16.0493,
+      "step": 17343
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.087602124313424e-06,
+      "loss": 0.7832,
+      "step": 17344
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.083825134097171e-06,
+      "loss": 17.005,
+      "step": 17345
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.0800492400386875e-06,
+      "loss": 17.2121,
+      "step": 17346
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.076274442232188e-06,
+      "loss": 0.7759,
+      "step": 17347
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.0725007407719134e-06,
+      "loss": 0.7919,
+      "step": 17348
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.06872813575205e-06,
+      "loss": 0.745,
+      "step": 17349
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.064956627266754e-06,
+      "loss": 0.779,
+      "step": 17350
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.061186215410186e-06,
+      "loss": 18.3201,
+      "step": 17351
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.057416900276442e-06,
+      "loss": 17.8646,
+      "step": 17352
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.053648681959601e-06,
+      "loss": 18.7497,
+      "step": 17353
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.049881560553727e-06,
+      "loss": 16.8503,
+      "step": 17354
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.04611553615283e-06,
+      "loss": 19.2862,
+      "step": 17355
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.042350608850938e-06,
+      "loss": 18.188,
+      "step": 17356
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.038586778742017e-06,
+      "loss": 16.4682,
+      "step": 17357
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.03482404591999e-06,
+      "loss": 18.0831,
+      "step": 17358
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.031062410478805e-06,
+      "loss": 0.7227,
+      "step": 17359
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.027301872512342e-06,
+      "loss": 18.5118,
+      "step": 17360
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.023542432114465e-06,
+      "loss": 0.7389,
+      "step": 17361
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.0197840893789995e-06,
+      "loss": 18.1382,
+      "step": 17362
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.016026844399775e-06,
+      "loss": 16.3464,
+      "step": 17363
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.012270697270567e-06,
+      "loss": 0.8052,
+      "step": 17364
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.008515648085111e-06,
+      "loss": 17.0561,
+      "step": 17365
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.004761696937161e-06,
+      "loss": 0.7595,
+      "step": 17366
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 6.001008843920408e-06,
+      "loss": 17.2883,
+      "step": 17367
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.997257089128516e-06,
+      "loss": 0.753,
+      "step": 17368
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.9935064326551325e-06,
+      "loss": 16.4915,
+      "step": 17369
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.989756874593872e-06,
+      "loss": 0.7307,
+      "step": 17370
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.9860084150383355e-06,
+      "loss": 0.7716,
+      "step": 17371
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.982261054082078e-06,
+      "loss": 0.7348,
+      "step": 17372
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.978514791818629e-06,
+      "loss": 0.7563,
+      "step": 17373
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.97476962834152e-06,
+      "loss": 17.3783,
+      "step": 17374
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.971025563744193e-06,
+      "loss": 19.0491,
+      "step": 17375
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.967282598120127e-06,
+      "loss": 0.7582,
+      "step": 17376
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.963540731562744e-06,
+      "loss": 16.4276,
+      "step": 17377
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.959799964165435e-06,
+      "loss": 15.551,
+      "step": 17378
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.956060296021576e-06,
+      "loss": 15.566,
+      "step": 17379
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.952321727224513e-06,
+      "loss": 17.8527,
+      "step": 17380
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.948584257867557e-06,
+      "loss": 0.7429,
+      "step": 17381
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.944847888043997e-06,
+      "loss": 17.7735,
+      "step": 17382
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.941112617847083e-06,
+      "loss": 0.747,
+      "step": 17383
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.937378447370067e-06,
+      "loss": 19.5157,
+      "step": 17384
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.933645376706143e-06,
+      "loss": 17.766,
+      "step": 17385
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.929913405948484e-06,
+      "loss": 17.2083,
+      "step": 17386
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.926182535190261e-06,
+      "loss": 0.7112,
+      "step": 17387
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.922452764524583e-06,
+      "loss": 0.8003,
+      "step": 17388
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.918724094044547e-06,
+      "loss": 0.77,
+      "step": 17389
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.914996523843214e-06,
+      "loss": 17.1241,
+      "step": 17390
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.911270054013645e-06,
+      "loss": 0.7402,
+      "step": 17391
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.907544684648841e-06,
+      "loss": 15.4297,
+      "step": 17392
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.903820415841782e-06,
+      "loss": 16.5597,
+      "step": 17393
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.900097247685438e-06,
+      "loss": 0.7756,
+      "step": 17394
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.89637518027274e-06,
+      "loss": 18.0067,
+      "step": 17395
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.892654213696592e-06,
+      "loss": 16.7571,
+      "step": 17396
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.888934348049857e-06,
+      "loss": 0.7965,
+      "step": 17397
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.885215583425385e-06,
+      "loss": 17.9833,
+      "step": 17398
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.881497919916018e-06,
+      "loss": 19.7726,
+      "step": 17399
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.877781357614537e-06,
+      "loss": 16.1319,
+      "step": 17400
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.874065896613695e-06,
+      "loss": 18.5123,
+      "step": 17401
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.870351537006263e-06,
+      "loss": 0.7054,
+      "step": 17402
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.8666382788849114e-06,
+      "loss": 0.8143,
+      "step": 17403
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.8629261223423605e-06,
+      "loss": 0.7895,
+      "step": 17404
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.859215067471246e-06,
+      "loss": 18.4314,
+      "step": 17405
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.855505114364196e-06,
+      "loss": 16.6947,
+      "step": 17406
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.851796263113829e-06,
+      "loss": 18.6553,
+      "step": 17407
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.848088513812705e-06,
+      "loss": 15.3697,
+      "step": 17408
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.844381866553372e-06,
+      "loss": 0.8007,
+      "step": 17409
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.840676321428351e-06,
+      "loss": 18.6003,
+      "step": 17410
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.836971878530123e-06,
+      "loss": 0.7811,
+      "step": 17411
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.833268537951164e-06,
+      "loss": 18.7452,
+      "step": 17412
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.829566299783917e-06,
+      "loss": 16.5726,
+      "step": 17413
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.825865164120764e-06,
+      "loss": 0.8095,
+      "step": 17414
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.822165131054125e-06,
+      "loss": 16.1845,
+      "step": 17415
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.818466200676309e-06,
+      "loss": 17.3254,
+      "step": 17416
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.814768373079676e-06,
+      "loss": 19.0639,
+      "step": 17417
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.811071648356503e-06,
+      "loss": 0.7627,
+      "step": 17418
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.8073760265990816e-06,
+      "loss": 15.2289,
+      "step": 17419
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.803681507899645e-06,
+      "loss": 17.1003,
+      "step": 17420
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.7999880923504005e-06,
+      "loss": 17.8892,
+      "step": 17421
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.796295780043565e-06,
+      "loss": 16.5738,
+      "step": 17422
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.792604571071258e-06,
+      "loss": 17.5627,
+      "step": 17423
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.788914465525647e-06,
+      "loss": 0.813,
+      "step": 17424
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.785225463498828e-06,
+      "loss": 0.7619,
+      "step": 17425
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.781537565082862e-06,
+      "loss": 15.2864,
+      "step": 17426
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.777850770369825e-06,
+      "loss": 0.7678,
+      "step": 17427
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.774165079451732e-06,
+      "loss": 17.77,
+      "step": 17428
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.7704804924205764e-06,
+      "loss": 16.9224,
+      "step": 17429
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.766797009368325e-06,
+      "loss": 0.7828,
+      "step": 17430
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.763114630386912e-06,
+      "loss": 0.7201,
+      "step": 17431
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.759433355568267e-06,
+      "loss": 0.7708,
+      "step": 17432
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.75575318500427e-06,
+      "loss": 0.7426,
+      "step": 17433
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.752074118786766e-06,
+      "loss": 0.7034,
+      "step": 17434
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.7483961570076015e-06,
+      "loss": 18.3391,
+      "step": 17435
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.744719299758572e-06,
+      "loss": 16.8785,
+      "step": 17436
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.741043547131458e-06,
+      "loss": 0.7645,
+      "step": 17437
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.737368899217999e-06,
+      "loss": 19.7048,
+      "step": 17438
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.733695356109908e-06,
+      "loss": 17.0133,
+      "step": 17439
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.730022917898903e-06,
+      "loss": 0.7777,
+      "step": 17440
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.7263515846766316e-06,
+      "loss": 0.7565,
+      "step": 17441
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.72268135653472e-06,
+      "loss": 0.7082,
+      "step": 17442
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.719012233564819e-06,
+      "loss": 17.4183,
+      "step": 17443
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.7153442158584596e-06,
+      "loss": 18.2744,
+      "step": 17444
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.711677303507229e-06,
+      "loss": 0.8526,
+      "step": 17445
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.708011496602633e-06,
+      "loss": 0.7955,
+      "step": 17446
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.704346795236199e-06,
+      "loss": 0.7418,
+      "step": 17447
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.700683199499379e-06,
+      "loss": 0.7812,
+      "step": 17448
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.697020709483608e-06,
+      "loss": 18.316,
+      "step": 17449
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.693359325280345e-06,
+      "loss": 17.2696,
+      "step": 17450
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.689699046980923e-06,
+      "loss": 0.7476,
+      "step": 17451
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.686039874676741e-06,
+      "loss": 0.771,
+      "step": 17452
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.682381808459125e-06,
+      "loss": 0.7449,
+      "step": 17453
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.678724848419364e-06,
+      "loss": 17.0705,
+      "step": 17454
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.675068994648764e-06,
+      "loss": 0.8139,
+      "step": 17455
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.67141424723856e-06,
+      "loss": 0.7404,
+      "step": 17456
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.6677606062799815e-06,
+      "loss": 0.8072,
+      "step": 17457
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.664108071864216e-06,
+      "loss": 0.7477,
+      "step": 17458
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.660456644082429e-06,
+      "loss": 17.9875,
+      "step": 17459
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.656806323025776e-06,
+      "loss": 0.7719,
+      "step": 17460
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.6531571087853675e-06,
+      "loss": 16.6768,
+      "step": 17461
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.649509001452269e-06,
+      "loss": 17.6672,
+      "step": 17462
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.645862001117563e-06,
+      "loss": 16.6041,
+      "step": 17463
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.642216107872267e-06,
+      "loss": 17.0317,
+      "step": 17464
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.63857132180739e-06,
+      "loss": 0.7609,
+      "step": 17465
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.634927643013899e-06,
+      "loss": 18.1201,
+      "step": 17466
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.631285071582731e-06,
+      "loss": 18.648,
+      "step": 17467
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.627643607604832e-06,
+      "loss": 17.055,
+      "step": 17468
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.6240032511710785e-06,
+      "loss": 19.5746,
+      "step": 17469
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.6203640023723244e-06,
+      "loss": 0.7511,
+      "step": 17470
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.6167258612994436e-06,
+      "loss": 0.8487,
+      "step": 17471
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.613088828043195e-06,
+      "loss": 18.3435,
+      "step": 17472
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.609452902694396e-06,
+      "loss": 0.7492,
+      "step": 17473
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.605818085343778e-06,
+      "loss": 0.7869,
+      "step": 17474
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.602184376082081e-06,
+      "loss": 19.1686,
+      "step": 17475
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.598551775000005e-06,
+      "loss": 19.7618,
+      "step": 17476
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.59492028218821e-06,
+      "loss": 16.742,
+      "step": 17477
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.591289897737356e-06,
+      "loss": 0.7528,
+      "step": 17478
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.5876606217380275e-06,
+      "loss": 15.5289,
+      "step": 17479
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.584032454280841e-06,
+      "loss": 0.7734,
+      "step": 17480
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.580405395456345e-06,
+      "loss": 0.744,
+      "step": 17481
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.576779445355068e-06,
+      "loss": 0.7281,
+      "step": 17482
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.573154604067521e-06,
+      "loss": 18.4565,
+      "step": 17483
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.569530871684186e-06,
+      "loss": 0.7876,
+      "step": 17484
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.565908248295504e-06,
+      "loss": 16.7144,
+      "step": 17485
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.562286733991895e-06,
+      "loss": 17.6166,
+      "step": 17486
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.558666328863743e-06,
+      "loss": 19.1759,
+      "step": 17487
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.5550470330014425e-06,
+      "loss": 0.7674,
+      "step": 17488
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.551428846495316e-06,
+      "loss": 18.4496,
+      "step": 17489
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.547811769435662e-06,
+      "loss": 17.6864,
+      "step": 17490
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.544195801912794e-06,
+      "loss": 17.7254,
+      "step": 17491
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.5405809440169325e-06,
+      "loss": 0.7423,
+      "step": 17492
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.536967195838333e-06,
+      "loss": 17.6602,
+      "step": 17493
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.53335455746718e-06,
+      "loss": 16.6376,
+      "step": 17494
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.529743028993645e-06,
+      "loss": 19.1638,
+      "step": 17495
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.526132610507883e-06,
+      "loss": 19.9895,
+      "step": 17496
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.5225233021000025e-06,
+      "loss": 16.9183,
+      "step": 17497
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.518915103860112e-06,
+      "loss": 17.8576,
+      "step": 17498
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.515308015878246e-06,
+      "loss": 0.7797,
+      "step": 17499
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.511702038244443e-06,
+      "loss": 0.773,
+      "step": 17500
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.50809717104872e-06,
+      "loss": 0.7628,
+      "step": 17501
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.50449341438104e-06,
+      "loss": 17.9607,
+      "step": 17502
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.500890768331379e-06,
+      "loss": 0.7169,
+      "step": 17503
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.497289232989644e-06,
+      "loss": 0.7305,
+      "step": 17504
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.4936888084457296e-06,
+      "loss": 16.1013,
+      "step": 17505
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.490089494789502e-06,
+      "loss": 0.7341,
+      "step": 17506
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.486491292110796e-06,
+      "loss": 18.3546,
+      "step": 17507
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.482894200499444e-06,
+      "loss": 15.5992,
+      "step": 17508
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.479298220045215e-06,
+      "loss": 17.9072,
+      "step": 17509
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.4757033508378575e-06,
+      "loss": 0.7349,
+      "step": 17510
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.472109592967117e-06,
+      "loss": 19.7428,
+      "step": 17511
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.468516946522695e-06,
+      "loss": 17.3121,
+      "step": 17512
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.464925411594252e-06,
+      "loss": 17.9543,
+      "step": 17513
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.4613349882714436e-06,
+      "loss": 18.9331,
+      "step": 17514
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.457745676643872e-06,
+      "loss": 18.0862,
+      "step": 17515
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.454157476801147e-06,
+      "loss": 16.4632,
+      "step": 17516
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.450570388832826e-06,
+      "loss": 17.9345,
+      "step": 17517
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.4469844128284265e-06,
+      "loss": 0.7602,
+      "step": 17518
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.443399548877492e-06,
+      "loss": 15.9987,
+      "step": 17519
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.439815797069453e-06,
+      "loss": 18.2175,
+      "step": 17520
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.436233157493803e-06,
+      "loss": 18.6764,
+      "step": 17521
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.432651630239943e-06,
+      "loss": 15.888,
+      "step": 17522
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.429071215397269e-06,
+      "loss": 18.364,
+      "step": 17523
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.4254919130551585e-06,
+      "loss": 0.7778,
+      "step": 17524
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.421913723302946e-06,
+      "loss": 0.7176,
+      "step": 17525
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.41833664622996e-06,
+      "loss": 0.7663,
+      "step": 17526
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.414760681925457e-06,
+      "loss": 17.3789,
+      "step": 17527
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.4111858304787045e-06,
+      "loss": 0.747,
+      "step": 17528
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.407612091978942e-06,
+      "loss": 0.7168,
+      "step": 17529
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.4040394665153536e-06,
+      "loss": 0.7886,
+      "step": 17530
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.40046795417713e-06,
+      "loss": 0.7512,
+      "step": 17531
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.396897555053415e-06,
+      "loss": 15.7042,
+      "step": 17532
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.3933282692333165e-06,
+      "loss": 18.9175,
+      "step": 17533
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.3897600968059295e-06,
+      "loss": 16.4269,
+      "step": 17534
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.38619303786031e-06,
+      "loss": 17.7501,
+      "step": 17535
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.3826270924855036e-06,
+      "loss": 16.4401,
+      "step": 17536
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.379062260770518e-06,
+      "loss": 18.1359,
+      "step": 17537
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.375498542804313e-06,
+      "loss": 0.7415,
+      "step": 17538
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.371935938675865e-06,
+      "loss": 0.7916,
+      "step": 17539
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.368374448474089e-06,
+      "loss": 0.8096,
+      "step": 17540
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.364814072287872e-06,
+      "loss": 0.738,
+      "step": 17541
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.36125481020609e-06,
+      "loss": 0.7701,
+      "step": 17542
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.357696662317568e-06,
+      "loss": 0.766,
+      "step": 17543
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.354139628711147e-06,
+      "loss": 0.7293,
+      "step": 17544
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.350583709475587e-06,
+      "loss": 0.6859,
+      "step": 17545
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.347028904699647e-06,
+      "loss": 19.6548,
+      "step": 17546
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.343475214472077e-06,
+      "loss": 17.5242,
+      "step": 17547
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.339922638881545e-06,
+      "loss": 16.7525,
+      "step": 17548
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.336371178016752e-06,
+      "loss": 17.7622,
+      "step": 17549
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.332820831966329e-06,
+      "loss": 15.6883,
+      "step": 17550
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.329271600818891e-06,
+      "loss": 0.7321,
+      "step": 17551
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.325723484663042e-06,
+      "loss": 19.3865,
+      "step": 17552
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.322176483587327e-06,
+      "loss": 0.7794,
+      "step": 17553
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.318630597680302e-06,
+      "loss": 0.7703,
+      "step": 17554
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.315085827030458e-06,
+      "loss": 18.3001,
+      "step": 17555
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.311542171726258e-06,
+      "loss": 18.6779,
+      "step": 17556
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.307999631856181e-06,
+      "loss": 18.0082,
+      "step": 17557
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.304458207508628e-06,
+      "loss": 16.1753,
+      "step": 17558
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.300917898772012e-06,
+      "loss": 0.7714,
+      "step": 17559
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.297378705734696e-06,
+      "loss": 0.713,
+      "step": 17560
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.293840628485008e-06,
+      "loss": 0.7738,
+      "step": 17561
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.290303667111269e-06,
+      "loss": 0.7317,
+      "step": 17562
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.286767821701749e-06,
+      "loss": 0.7535,
+      "step": 17563
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.2832330923447245e-06,
+      "loss": 19.5796,
+      "step": 17564
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.2796994791284135e-06,
+      "loss": 0.7277,
+      "step": 17565
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.276166982141001e-06,
+      "loss": 18.4953,
+      "step": 17566
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.272635601470694e-06,
+      "loss": 0.7657,
+      "step": 17567
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.269105337205593e-06,
+      "loss": 0.7733,
+      "step": 17568
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.265576189433846e-06,
+      "loss": 19.1873,
+      "step": 17569
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.262048158243537e-06,
+      "loss": 18.7952,
+      "step": 17570
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.258521243722703e-06,
+      "loss": 0.7126,
+      "step": 17571
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.25499544595941e-06,
+      "loss": 0.7273,
+      "step": 17572
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.25147076504165e-06,
+      "loss": 0.7284,
+      "step": 17573
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.2479472010573905e-06,
+      "loss": 18.2359,
+      "step": 17574
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.24442475409459e-06,
+      "loss": 16.4289,
+      "step": 17575
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.240903424241156e-06,
+      "loss": 19.3877,
+      "step": 17576
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.237383211585001e-06,
+      "loss": 0.6866,
+      "step": 17577
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.233864116213982e-06,
+      "loss": 0.776,
+      "step": 17578
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.23034613821593e-06,
+      "loss": 0.7499,
+      "step": 17579
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.226829277678669e-06,
+      "loss": 19.1011,
+      "step": 17580
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.223313534689972e-06,
+      "loss": 18.9884,
+      "step": 17581
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.219798909337592e-06,
+      "loss": 0.7436,
+      "step": 17582
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.216285401709253e-06,
+      "loss": 16.63,
+      "step": 17583
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.212773011892652e-06,
+      "loss": 18.1526,
+      "step": 17584
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.209261739975468e-06,
+      "loss": 0.7427,
+      "step": 17585
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.205751586045332e-06,
+      "loss": 18.0901,
+      "step": 17586
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.202242550189873e-06,
+      "loss": 15.086,
+      "step": 17587
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.1987346324966765e-06,
+      "loss": 18.017,
+      "step": 17588
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.195227833053274e-06,
+      "loss": 0.8239,
+      "step": 17589
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.191722151947226e-06,
+      "loss": 0.7212,
+      "step": 17590
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.188217589266015e-06,
+      "loss": 19.2304,
+      "step": 17591
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.184714145097131e-06,
+      "loss": 15.9308,
+      "step": 17592
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.181211819528014e-06,
+      "loss": 16.8866,
+      "step": 17593
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.177710612646081e-06,
+      "loss": 17.6703,
+      "step": 17594
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.174210524538742e-06,
+      "loss": 0.7318,
+      "step": 17595
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.170711555293317e-06,
+      "loss": 15.9634,
+      "step": 17596
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.167213704997187e-06,
+      "loss": 16.7593,
+      "step": 17597
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.163716973737631e-06,
+      "loss": 15.506,
+      "step": 17598
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.1602213616019355e-06,
+      "loss": 0.7706,
+      "step": 17599
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.1567268686773584e-06,
+      "loss": 16.9664,
+      "step": 17600
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.1532334950511186e-06,
+      "loss": 0.7162,
+      "step": 17601
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.149741240810407e-06,
+      "loss": 15.6146,
+      "step": 17602
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.146250106042399e-06,
+      "loss": 0.7379,
+      "step": 17603
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.142760090834215e-06,
+      "loss": 0.7294,
+      "step": 17604
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.139271195272999e-06,
+      "loss": 16.0739,
+      "step": 17605
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.135783419445811e-06,
+      "loss": 17.6248,
+      "step": 17606
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.1322967634397035e-06,
+      "loss": 19.6784,
+      "step": 17607
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.128811227341724e-06,
+      "loss": 17.9011,
+      "step": 17608
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.125326811238862e-06,
+      "loss": 0.7074,
+      "step": 17609
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.121843515218089e-06,
+      "loss": 0.7532,
+      "step": 17610
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.118361339366351e-06,
+      "loss": 17.1933,
+      "step": 17611
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.1148802837705455e-06,
+      "loss": 0.7684,
+      "step": 17612
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.111400348517592e-06,
+      "loss": 17.7939,
+      "step": 17613
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.107921533694326e-06,
+      "loss": 15.7601,
+      "step": 17614
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.104443839387596e-06,
+      "loss": 0.7096,
+      "step": 17615
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.100967265684209e-06,
+      "loss": 17.8604,
+      "step": 17616
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.097491812670912e-06,
+      "loss": 0.8118,
+      "step": 17617
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.094017480434482e-06,
+      "loss": 0.78,
+      "step": 17618
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.09054426906162e-06,
+      "loss": 0.7786,
+      "step": 17619
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.087072178639035e-06,
+      "loss": 18.4671,
+      "step": 17620
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.08360120925338e-06,
+      "loss": 19.3415,
+      "step": 17621
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.080131360991291e-06,
+      "loss": 0.7384,
+      "step": 17622
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.076662633939394e-06,
+      "loss": 17.5014,
+      "step": 17623
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.073195028184241e-06,
+      "loss": 0.7696,
+      "step": 17624
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.069728543812408e-06,
+      "loss": 0.8196,
+      "step": 17625
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.066263180910402e-06,
+      "loss": 16.5075,
+      "step": 17626
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.0627989395647225e-06,
+      "loss": 18.6799,
+      "step": 17627
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.059335819861855e-06,
+      "loss": 0.7523,
+      "step": 17628
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.055873821888218e-06,
+      "loss": 16.9234,
+      "step": 17629
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.05241294573024e-06,
+      "loss": 0.8071,
+      "step": 17630
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.048953191474292e-06,
+      "loss": 18.2857,
+      "step": 17631
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.045494559206732e-06,
+      "loss": 0.7381,
+      "step": 17632
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.042037049013898e-06,
+      "loss": 0.7627,
+      "step": 17633
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.038580660982084e-06,
+      "loss": 16.9397,
+      "step": 17634
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.035125395197554e-06,
+      "loss": 0.7616,
+      "step": 17635
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.031671251746573e-06,
+      "loss": 0.7348,
+      "step": 17636
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.028218230715343e-06,
+      "loss": 17.1749,
+      "step": 17637
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.024766332190056e-06,
+      "loss": 17.5736,
+      "step": 17638
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.021315556256873e-06,
+      "loss": 0.7682,
+      "step": 17639
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.017865903001906e-06,
+      "loss": 15.4816,
+      "step": 17640
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.014417372511293e-06,
+      "loss": 0.8182,
+      "step": 17641
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.010969964871082e-06,
+      "loss": 18.0727,
+      "step": 17642
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.007523680167353e-06,
+      "loss": 18.0022,
+      "step": 17643
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.004078518486094e-06,
+      "loss": 16.2136,
+      "step": 17644
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.000634479913302e-06,
+      "loss": 18.1205,
+      "step": 17645
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.9971915645349586e-06,
+      "loss": 0.7209,
+      "step": 17646
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.993749772436978e-06,
+      "loss": 0.7584,
+      "step": 17647
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.990309103705287e-06,
+      "loss": 17.2649,
+      "step": 17648
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.986869558425761e-06,
+      "loss": 0.7379,
+      "step": 17649
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.983431136684247e-06,
+      "loss": 0.7727,
+      "step": 17650
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.9799938385665715e-06,
+      "loss": 18.2502,
+      "step": 17651
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.976557664158515e-06,
+      "loss": 16.7184,
+      "step": 17652
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.973122613545878e-06,
+      "loss": 0.7595,
+      "step": 17653
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.969688686814372e-06,
+      "loss": 17.7615,
+      "step": 17654
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.966255884049714e-06,
+      "loss": 0.7758,
+      "step": 17655
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.962824205337602e-06,
+      "loss": 0.7245,
+      "step": 17656
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.959393650763677e-06,
+      "loss": 18.8346,
+      "step": 17657
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.955964220413572e-06,
+      "loss": 0.7318,
+      "step": 17658
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.952535914372891e-06,
+      "loss": 18.0684,
+      "step": 17659
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.949108732727181e-06,
+      "loss": 18.0851,
+      "step": 17660
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.945682675562019e-06,
+      "loss": 15.6737,
+      "step": 17661
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.942257742962902e-06,
+      "loss": 0.7386,
+      "step": 17662
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.938833935015308e-06,
+      "loss": 0.7868,
+      "step": 17663
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.9354112518047334e-06,
+      "loss": 17.2123,
+      "step": 17664
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.931989693416561e-06,
+      "loss": 16.2941,
+      "step": 17665
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.928569259936222e-06,
+      "loss": 18.5131,
+      "step": 17666
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.925149951449088e-06,
+      "loss": 16.8208,
+      "step": 17667
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.921731768040494e-06,
+      "loss": 0.7454,
+      "step": 17668
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.91831470979578e-06,
+      "loss": 0.7766,
+      "step": 17669
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.914898776800214e-06,
+      "loss": 18.7692,
+      "step": 17670
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.9114839691390854e-06,
+      "loss": 18.1286,
+      "step": 17671
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.908070286897604e-06,
+      "loss": 16.3537,
+      "step": 17672
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.904657730160972e-06,
+      "loss": 18.4194,
+      "step": 17673
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.9012462990143895e-06,
+      "loss": 0.7784,
+      "step": 17674
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.8978359935429875e-06,
+      "loss": 18.0046,
+      "step": 17675
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.894426813831909e-06,
+      "loss": 16.4896,
+      "step": 17676
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.891018759966237e-06,
+      "loss": 17.0478,
+      "step": 17677
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.887611832031036e-06,
+      "loss": 16.9749,
+      "step": 17678
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.884206030111343e-06,
+      "loss": 17.5023,
+      "step": 17679
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.880801354292158e-06,
+      "loss": 0.7704,
+      "step": 17680
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.8773978046584844e-06,
+      "loss": 17.0613,
+      "step": 17681
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.873995381295266e-06,
+      "loss": 0.6918,
+      "step": 17682
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.870594084287411e-06,
+      "loss": 0.7397,
+      "step": 17683
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.8671939137198476e-06,
+      "loss": 0.7387,
+      "step": 17684
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.863794869677424e-06,
+      "loss": 0.7407,
+      "step": 17685
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.860396952244989e-06,
+      "loss": 0.7627,
+      "step": 17686
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.857000161507353e-06,
+      "loss": 17.388,
+      "step": 17687
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.853604497549291e-06,
+      "loss": 16.8829,
+      "step": 17688
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.850209960455582e-06,
+      "loss": 0.7584,
+      "step": 17689
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.846816550310934e-06,
+      "loss": 0.7909,
+      "step": 17690
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.843424267200053e-06,
+      "loss": 16.5571,
+      "step": 17691
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.840033111207626e-06,
+      "loss": 18.1805,
+      "step": 17692
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.836643082418269e-06,
+      "loss": 16.1012,
+      "step": 17693
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.833254180916619e-06,
+      "loss": 19.4848,
+      "step": 17694
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.829866406787259e-06,
+      "loss": 16.6052,
+      "step": 17695
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.826479760114738e-06,
+      "loss": 0.7458,
+      "step": 17696
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.823094240983606e-06,
+      "loss": 0.7417,
+      "step": 17697
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.8197098494783486e-06,
+      "loss": 0.7115,
+      "step": 17698
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.816326585683467e-06,
+      "loss": 16.3195,
+      "step": 17699
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.812944449683382e-06,
+      "loss": 0.7618,
+      "step": 17700
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.809563441562509e-06,
+      "loss": 0.748,
+      "step": 17701
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.8061835614052585e-06,
+      "loss": 0.7749,
+      "step": 17702
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.8028048092959795e-06,
+      "loss": 16.5204,
+      "step": 17703
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.799427185319022e-06,
+      "loss": 18.3178,
+      "step": 17704
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.796050689558684e-06,
+      "loss": 17.4676,
+      "step": 17705
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.792675322099238e-06,
+      "loss": 17.6252,
+      "step": 17706
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.789301083024944e-06,
+      "loss": 18.0296,
+      "step": 17707
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.785927972420007e-06,
+      "loss": 18.4172,
+      "step": 17708
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.782555990368637e-06,
+      "loss": 17.0628,
+      "step": 17709
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.779185136954994e-06,
+      "loss": 19.3571,
+      "step": 17710
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.775815412263207e-06,
+      "loss": 18.4412,
+      "step": 17711
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.772446816377407e-06,
+      "loss": 0.7774,
+      "step": 17712
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.769079349381655e-06,
+      "loss": 0.7407,
+      "step": 17713
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7657130113600125e-06,
+      "loss": 0.7504,
+      "step": 17714
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.762347802396505e-06,
+      "loss": 17.1574,
+      "step": 17715
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.758983722575111e-06,
+      "loss": 19.3094,
+      "step": 17716
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.755620771979824e-06,
+      "loss": 18.0603,
+      "step": 17717
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.752258950694577e-06,
+      "loss": 0.7114,
+      "step": 17718
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.748898258803264e-06,
+      "loss": 0.7909,
+      "step": 17719
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.745538696389801e-06,
+      "loss": 0.7693,
+      "step": 17720
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.74218026353801e-06,
+      "loss": 19.3339,
+      "step": 17721
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.73882296033174e-06,
+      "loss": 17.9649,
+      "step": 17722
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.735466786854781e-06,
+      "loss": 0.726,
+      "step": 17723
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.732111743190903e-06,
+      "loss": 18.1573,
+      "step": 17724
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7287578294238576e-06,
+      "loss": 0.7637,
+      "step": 17725
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.725405045637354e-06,
+      "loss": 19.2341,
+      "step": 17726
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.722053391915077e-06,
+      "loss": 0.7849,
+      "step": 17727
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.71870286834068e-06,
+      "loss": 0.7528,
+      "step": 17728
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7153534749978025e-06,
+      "loss": 16.9601,
+      "step": 17729
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7120052119700445e-06,
+      "loss": 0.779,
+      "step": 17730
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.708658079340966e-06,
+      "loss": 16.7325,
+      "step": 17731
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.705312077194135e-06,
+      "loss": 18.5535,
+      "step": 17732
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.701967205613056e-06,
+      "loss": 0.79,
+      "step": 17733
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.6986234646812176e-06,
+      "loss": 0.793,
+      "step": 17734
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.695280854482087e-06,
+      "loss": 15.1392,
+      "step": 17735
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.691939375099075e-06,
+      "loss": 0.8017,
+      "step": 17736
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.688599026615609e-06,
+      "loss": 15.5688,
+      "step": 17737
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.685259809115061e-06,
+      "loss": 0.7558,
+      "step": 17738
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.681921722680765e-06,
+      "loss": 16.7895,
+      "step": 17739
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.678584767396072e-06,
+      "loss": 16.3309,
+      "step": 17740
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.675248943344229e-06,
+      "loss": 16.4664,
+      "step": 17741
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.671914250608528e-06,
+      "loss": 0.7462,
+      "step": 17742
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.6685806892722e-06,
+      "loss": 17.0006,
+      "step": 17743
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.665248259418442e-06,
+      "loss": 0.7577,
+      "step": 17744
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.661916961130447e-06,
+      "loss": 16.4019,
+      "step": 17745
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.658586794491354e-06,
+      "loss": 0.7782,
+      "step": 17746
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.655257759584286e-06,
+      "loss": 0.7436,
+      "step": 17747
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.651929856492343e-06,
+      "loss": 16.7963,
+      "step": 17748
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.648603085298575e-06,
+      "loss": 18.1794,
+      "step": 17749
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.645277446086033e-06,
+      "loss": 18.8572,
+      "step": 17750
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.641952938937727e-06,
+      "loss": 0.7606,
+      "step": 17751
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.638629563936619e-06,
+      "loss": 0.7536,
+      "step": 17752
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.635307321165688e-06,
+      "loss": 16.4688,
+      "step": 17753
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.631986210707845e-06,
+      "loss": 18.6276,
+      "step": 17754
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.628666232645978e-06,
+      "loss": 0.7374,
+      "step": 17755
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.625347387062961e-06,
+      "loss": 0.7881,
+      "step": 17756
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.622029674041639e-06,
+      "loss": 0.7448,
+      "step": 17757
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.618713093664817e-06,
+      "loss": 15.1917,
+      "step": 17758
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.615397646015268e-06,
+      "loss": 18.0882,
+      "step": 17759
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.612083331175765e-06,
+      "loss": 0.7673,
+      "step": 17760
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.608770149229025e-06,
+      "loss": 0.7453,
+      "step": 17761
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.605458100257748e-06,
+      "loss": 19.105,
+      "step": 17762
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.602147184344601e-06,
+      "loss": 16.578,
+      "step": 17763
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.59883740157222e-06,
+      "loss": 17.1401,
+      "step": 17764
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.595528752023231e-06,
+      "loss": 17.6693,
+      "step": 17765
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.592221235780214e-06,
+      "loss": 0.742,
+      "step": 17766
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.588914852925707e-06,
+      "loss": 16.1911,
+      "step": 17767
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.585609603542279e-06,
+      "loss": 0.708,
+      "step": 17768
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.582305487712385e-06,
+      "loss": 17.0097,
+      "step": 17769
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.57900250551852e-06,
+      "loss": 0.8013,
+      "step": 17770
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.57570065704313e-06,
+      "loss": 0.7426,
+      "step": 17771
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5723999423686106e-06,
+      "loss": 0.7489,
+      "step": 17772
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.569100361577372e-06,
+      "loss": 16.3197,
+      "step": 17773
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.565801914751761e-06,
+      "loss": 16.5886,
+      "step": 17774
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.56250460197411e-06,
+      "loss": 0.717,
+      "step": 17775
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.559208423326716e-06,
+      "loss": 0.7727,
+      "step": 17776
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.555913378891846e-06,
+      "loss": 16.3538,
+      "step": 17777
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.552619468751767e-06,
+      "loss": 16.4956,
+      "step": 17778
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.549326692988681e-06,
+      "loss": 0.7396,
+      "step": 17779
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.546035051684766e-06,
+      "loss": 18.89,
+      "step": 17780
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.542744544922206e-06,
+      "loss": 0.7341,
+      "step": 17781
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.539455172783125e-06,
+      "loss": 0.7534,
+      "step": 17782
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.536166935349617e-06,
+      "loss": 16.9265,
+      "step": 17783
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.532879832703757e-06,
+      "loss": 17.316,
+      "step": 17784
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.529593864927606e-06,
+      "loss": 16.035,
+      "step": 17785
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.526309032103176e-06,
+      "loss": 0.7628,
+      "step": 17786
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.523025334312447e-06,
+      "loss": 0.7937,
+      "step": 17787
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.519742771637398e-06,
+      "loss": 17.0519,
+      "step": 17788
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5164613441599506e-06,
+      "loss": 18.3733,
+      "step": 17789
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.513181051962018e-06,
+      "loss": 0.7422,
+      "step": 17790
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.509901895125468e-06,
+      "loss": 0.6852,
+      "step": 17791
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.506623873732146e-06,
+      "loss": 17.2346,
+      "step": 17792
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.503346987863888e-06,
+      "loss": 0.8,
+      "step": 17793
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.500071237602482e-06,
+      "loss": 14.8169,
+      "step": 17794
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.496796623029676e-06,
+      "loss": 0.7426,
+      "step": 17795
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.49352314422723e-06,
+      "loss": 0.7675,
+      "step": 17796
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.490250801276819e-06,
+      "loss": 0.7443,
+      "step": 17797
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.486979594260154e-06,
+      "loss": 0.7886,
+      "step": 17798
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.483709523258867e-06,
+      "loss": 18.6748,
+      "step": 17799
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.480440588354573e-06,
+      "loss": 0.7113,
+      "step": 17800
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.477172789628886e-06,
+      "loss": 18.8449,
+      "step": 17801
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4739061271633574e-06,
+      "loss": 16.3595,
+      "step": 17802
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.470640601039527e-06,
+      "loss": 17.6327,
+      "step": 17803
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.467376211338908e-06,
+      "loss": 0.833,
+      "step": 17804
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.464112958142963e-06,
+      "loss": 0.7558,
+      "step": 17805
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4608508415331655e-06,
+      "loss": 18.2296,
+      "step": 17806
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.457589861590927e-06,
+      "loss": 18.7115,
+      "step": 17807
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.454330018397635e-06,
+      "loss": 17.0155,
+      "step": 17808
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.451071312034688e-06,
+      "loss": 0.7481,
+      "step": 17809
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.447813742583379e-06,
+      "loss": 0.7817,
+      "step": 17810
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.444557310125047e-06,
+      "loss": 0.7147,
+      "step": 17811
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.441302014740962e-06,
+      "loss": 15.2699,
+      "step": 17812
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.438047856512389e-06,
+      "loss": 0.7587,
+      "step": 17813
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.434794835520539e-06,
+      "loss": 0.7359,
+      "step": 17814
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.431542951846612e-06,
+      "loss": 0.7484,
+      "step": 17815
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.428292205571794e-06,
+      "loss": 16.3932,
+      "step": 17816
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.42504259677719e-06,
+      "loss": 15.5833,
+      "step": 17817
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.421794125543938e-06,
+      "loss": 0.7241,
+      "step": 17818
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4185467919531096e-06,
+      "loss": 0.762,
+      "step": 17819
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.415300596085753e-06,
+      "loss": 0.8192,
+      "step": 17820
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.412055538022908e-06,
+      "loss": 0.7203,
+      "step": 17821
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.408811617845571e-06,
+      "loss": 0.7039,
+      "step": 17822
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.405568835634699e-06,
+      "loss": 0.7815,
+      "step": 17823
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.402327191471245e-06,
+      "loss": 16.8516,
+      "step": 17824
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.399086685436105e-06,
+      "loss": 16.5972,
+      "step": 17825
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3958473176101855e-06,
+      "loss": 0.781,
+      "step": 17826
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.392609088074323e-06,
+      "loss": 0.8363,
+      "step": 17827
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.389371996909347e-06,
+      "loss": 16.7392,
+      "step": 17828
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3861360441960705e-06,
+      "loss": 15.9866,
+      "step": 17829
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.382901230015251e-06,
+      "loss": 0.7139,
+      "step": 17830
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.37966755444763e-06,
+      "loss": 0.7266,
+      "step": 17831
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.376435017573926e-06,
+      "loss": 16.5514,
+      "step": 17832
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.373203619474814e-06,
+      "loss": 16.6248,
+      "step": 17833
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.369973360230961e-06,
+      "loss": 18.6765,
+      "step": 17834
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.366744239922998e-06,
+      "loss": 0.7113,
+      "step": 17835
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3635162586315055e-06,
+      "loss": 17.7765,
+      "step": 17836
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.360289416437085e-06,
+      "loss": 17.9691,
+      "step": 17837
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.357063713420245e-06,
+      "loss": 0.7456,
+      "step": 17838
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.353839149661526e-06,
+      "loss": 0.7839,
+      "step": 17839
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.350615725241397e-06,
+      "loss": 16.8351,
+      "step": 17840
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.347393440240327e-06,
+      "loss": 16.0641,
+      "step": 17841
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.344172294738741e-06,
+      "loss": 18.6287,
+      "step": 17842
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3409522888170354e-06,
+      "loss": 17.4585,
+      "step": 17843
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.337733422555601e-06,
+      "loss": 0.809,
+      "step": 17844
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.334515696034747e-06,
+      "loss": 0.7717,
+      "step": 17845
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.331299109334819e-06,
+      "loss": 0.7162,
+      "step": 17846
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.328083662536087e-06,
+      "loss": 17.2624,
+      "step": 17847
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.324869355718814e-06,
+      "loss": 17.0796,
+      "step": 17848
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.321656188963235e-06,
+      "loss": 0.7617,
+      "step": 17849
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.31844416234955e-06,
+      "loss": 18.5106,
+      "step": 17850
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.315233275957925e-06,
+      "loss": 14.4458,
+      "step": 17851
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.312023529868514e-06,
+      "loss": 0.8011,
+      "step": 17852
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.308814924161414e-06,
+      "loss": 16.3223,
+      "step": 17853
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.305607458916738e-06,
+      "loss": 19.0734,
+      "step": 17854
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3024011342145346e-06,
+      "loss": 0.7571,
+      "step": 17855
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.299195950134821e-06,
+      "loss": 0.769,
+      "step": 17856
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.2959919067576186e-06,
+      "loss": 17.8804,
+      "step": 17857
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.292789004162895e-06,
+      "loss": 0.7567,
+      "step": 17858
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.289587242430598e-06,
+      "loss": 0.7614,
+      "step": 17859
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.286386621640637e-06,
+      "loss": 0.7645,
+      "step": 17860
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.283187141872896e-06,
+      "loss": 18.2959,
+      "step": 17861
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.279988803207252e-06,
+      "loss": 18.4601,
+      "step": 17862
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.276791605723524e-06,
+      "loss": 0.7434,
+      "step": 17863
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.273595549501514e-06,
+      "loss": 19.1856,
+      "step": 17864
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.270400634621013e-06,
+      "loss": 0.742,
+      "step": 17865
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.267206861161732e-06,
+      "loss": 16.4665,
+      "step": 17866
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.264014229203428e-06,
+      "loss": 16.8482,
+      "step": 17867
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.260822738825754e-06,
+      "loss": 0.7234,
+      "step": 17868
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.257632390108401e-06,
+      "loss": 0.7338,
+      "step": 17869
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.25444318313099e-06,
+      "loss": 0.701,
+      "step": 17870
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.251255117973113e-06,
+      "loss": 17.5314,
+      "step": 17871
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.248068194714372e-06,
+      "loss": 16.7067,
+      "step": 17872
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.244882413434276e-06,
+      "loss": 0.7741,
+      "step": 17873
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.2416977742123666e-06,
+      "loss": 18.5758,
+      "step": 17874
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.238514277128136e-06,
+      "loss": 18.6184,
+      "step": 17875
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.235331922261026e-06,
+      "loss": 17.4746,
+      "step": 17876
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.232150709690491e-06,
+      "loss": 0.7606,
+      "step": 17877
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.228970639495921e-06,
+      "loss": 17.0376,
+      "step": 17878
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.225791711756699e-06,
+      "loss": 0.7263,
+      "step": 17879
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.22261392655216e-06,
+      "loss": 18.7322,
+      "step": 17880
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.21943728396163e-06,
+      "loss": 0.7559,
+      "step": 17881
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.216261784064402e-06,
+      "loss": 17.8245,
+      "step": 17882
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.2130874269397325e-06,
+      "loss": 16.0286,
+      "step": 17883
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.209914212666849e-06,
+      "loss": 16.1892,
+      "step": 17884
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.206742141324982e-06,
+      "loss": 16.4859,
+      "step": 17885
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.203571212993268e-06,
+      "loss": 18.5399,
+      "step": 17886
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.200401427750883e-06,
+      "loss": 0.836,
+      "step": 17887
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.19723278567693e-06,
+      "loss": 16.7362,
+      "step": 17888
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.194065286850501e-06,
+      "loss": 0.7207,
+      "step": 17889
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.190898931350668e-06,
+      "loss": 0.7324,
+      "step": 17890
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.187733719256448e-06,
+      "loss": 15.771,
+      "step": 17891
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.184569650646874e-06,
+      "loss": 0.7244,
+      "step": 17892
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1814067256008945e-06,
+      "loss": 17.6001,
+      "step": 17893
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1782449441974566e-06,
+      "loss": 16.6423,
+      "step": 17894
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.175084306515492e-06,
+      "loss": 0.745,
+      "step": 17895
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.171924812633882e-06,
+      "loss": 18.4547,
+      "step": 17896
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1687664626315015e-06,
+      "loss": 0.802,
+      "step": 17897
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.165609256587172e-06,
+      "loss": 17.296,
+      "step": 17898
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1624531945797015e-06,
+      "loss": 16.4512,
+      "step": 17899
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.159298276687862e-06,
+      "loss": 0.7597,
+      "step": 17900
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1561445029904e-06,
+      "loss": 15.9556,
+      "step": 17901
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.152991873566048e-06,
+      "loss": 0.7633,
+      "step": 17902
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1498403884934865e-06,
+      "loss": 0.7467,
+      "step": 17903
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.146690047851365e-06,
+      "loss": 0.7518,
+      "step": 17904
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.143540851718342e-06,
+      "loss": 15.8304,
+      "step": 17905
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.140392800173009e-06,
+      "loss": 0.747,
+      "step": 17906
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.137245893293945e-06,
+      "loss": 0.7702,
+      "step": 17907
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.134100131159691e-06,
+      "loss": 16.5811,
+      "step": 17908
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.130955513848767e-06,
+      "loss": 15.5725,
+      "step": 17909
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1278120414396716e-06,
+      "loss": 0.7443,
+      "step": 17910
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1246697140108595e-06,
+      "loss": 0.7653,
+      "step": 17911
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1215285316407615e-06,
+      "loss": 18.3185,
+      "step": 17912
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.118388494407804e-06,
+      "loss": 0.7375,
+      "step": 17913
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.115249602390326e-06,
+      "loss": 0.7734,
+      "step": 17914
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.112111855666706e-06,
+      "loss": 0.7631,
+      "step": 17915
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.108975254315251e-06,
+      "loss": 16.2838,
+      "step": 17916
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.10583979841424e-06,
+      "loss": 0.7553,
+      "step": 17917
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.102705488041958e-06,
+      "loss": 17.7549,
+      "step": 17918
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.099572323276618e-06,
+      "loss": 0.7607,
+      "step": 17919
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.096440304196447e-06,
+      "loss": 17.4465,
+      "step": 17920
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.093309430879605e-06,
+      "loss": 18.1232,
+      "step": 17921
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.090179703404229e-06,
+      "loss": 0.7465,
+      "step": 17922
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.087051121848456e-06,
+      "loss": 16.75,
+      "step": 17923
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.083923686290369e-06,
+      "loss": 19.4644,
+      "step": 17924
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.080797396808034e-06,
+      "loss": 15.7045,
+      "step": 17925
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.077672253479481e-06,
+      "loss": 16.3962,
+      "step": 17926
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.074548256382715e-06,
+      "loss": 0.7594,
+      "step": 17927
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.0714254055957134e-06,
+      "loss": 0.7723,
+      "step": 17928
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.068303701196408e-06,
+      "loss": 0.767,
+      "step": 17929
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.065183143262741e-06,
+      "loss": 0.7726,
+      "step": 17930
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.06206373187259e-06,
+      "loss": 17.739,
+      "step": 17931
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.058945467103808e-06,
+      "loss": 16.7731,
+      "step": 17932
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.055828349034246e-06,
+      "loss": 17.1373,
+      "step": 17933
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.0527123777416964e-06,
+      "loss": 0.7897,
+      "step": 17934
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.049597553303941e-06,
+      "loss": 20.3581,
+      "step": 17935
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.0464838757987235e-06,
+      "loss": 16.6242,
+      "step": 17936
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.043371345303748e-06,
+      "loss": 0.794,
+      "step": 17937
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.040259961896725e-06,
+      "loss": 0.7875,
+      "step": 17938
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.037149725655309e-06,
+      "loss": 0.7617,
+      "step": 17939
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.034040636657127e-06,
+      "loss": 0.7335,
+      "step": 17940
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.030932694979794e-06,
+      "loss": 0.8065,
+      "step": 17941
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.027825900700866e-06,
+      "loss": 0.7743,
+      "step": 17942
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.024720253897907e-06,
+      "loss": 0.7306,
+      "step": 17943
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.021615754648433e-06,
+      "loss": 0.77,
+      "step": 17944
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.018512403029911e-06,
+      "loss": 15.8807,
+      "step": 17945
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.015410199119835e-06,
+      "loss": 17.5116,
+      "step": 17946
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.012309142995608e-06,
+      "loss": 0.7882,
+      "step": 17947
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.009209234734657e-06,
+      "loss": 0.7484,
+      "step": 17948
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.006110474414343e-06,
+      "loss": 0.7321,
+      "step": 17949
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 4.003012862112005e-06,
+      "loss": 0.7622,
+      "step": 17950
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.999916397904974e-06,
+      "loss": 0.7298,
+      "step": 17951
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.996821081870522e-06,
+      "loss": 19.1945,
+      "step": 17952
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.993726914085933e-06,
+      "loss": 0.7811,
+      "step": 17953
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.990633894628431e-06,
+      "loss": 0.7992,
+      "step": 17954
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.987542023575197e-06,
+      "loss": 0.7533,
+      "step": 17955
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.984451301003422e-06,
+      "loss": 0.7937,
+      "step": 17956
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.981361726990252e-06,
+      "loss": 0.7966,
+      "step": 17957
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9782733016128005e-06,
+      "loss": 0.7854,
+      "step": 17958
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9751860249481574e-06,
+      "loss": 0.7284,
+      "step": 17959
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.972099897073372e-06,
+      "loss": 0.728,
+      "step": 17960
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9690149180655e-06,
+      "loss": 16.4355,
+      "step": 17961
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9659310880015125e-06,
+      "loss": 0.7611,
+      "step": 17962
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9628484069583975e-06,
+      "loss": 17.6595,
+      "step": 17963
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.959766875013099e-06,
+      "loss": 17.5376,
+      "step": 17964
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.956686492242529e-06,
+      "loss": 14.4455,
+      "step": 17965
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.953607258723579e-06,
+      "loss": 0.7502,
+      "step": 17966
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.950529174533113e-06,
+      "loss": 0.717,
+      "step": 17967
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.947452239747945e-06,
+      "loss": 19.2068,
+      "step": 17968
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.944376454444892e-06,
+      "loss": 0.7526,
+      "step": 17969
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.941301818700704e-06,
+      "loss": 17.7512,
+      "step": 17970
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.938228332592148e-06,
+      "loss": 16.5741,
+      "step": 17971
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.935155996195933e-06,
+      "loss": 0.7741,
+      "step": 17972
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.932084809588732e-06,
+      "loss": 0.7733,
+      "step": 17973
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9290147728472225e-06,
+      "loss": 17.5172,
+      "step": 17974
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9259458860480265e-06,
+      "loss": 16.0908,
+      "step": 17975
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.922878149267739e-06,
+      "loss": 0.7792,
+      "step": 17976
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.919811562582931e-06,
+      "loss": 17.4366,
+      "step": 17977
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.916746126070142e-06,
+      "loss": 0.7946,
+      "step": 17978
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9136818398058995e-06,
+      "loss": 17.5585,
+      "step": 17979
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.91061870386667e-06,
+      "loss": 17.7039,
+      "step": 17980
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.90755671832893e-06,
+      "loss": 0.7277,
+      "step": 17981
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.90449588326911e-06,
+      "loss": 16.2531,
+      "step": 17982
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.9014361987635806e-06,
+      "loss": 18.1634,
+      "step": 17983
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.898377664888742e-06,
+      "loss": 0.7523,
+      "step": 17984
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.89532028172091e-06,
+      "loss": 15.869,
+      "step": 17985
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.892264049336419e-06,
+      "loss": 0.7479,
+      "step": 17986
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.889208967811541e-06,
+      "loss": 19.0434,
+      "step": 17987
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.886155037222533e-06,
+      "loss": 0.7462,
+      "step": 17988
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.883102257645643e-06,
+      "loss": 17.3872,
+      "step": 17989
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.880050629157028e-06,
+      "loss": 18.0548,
+      "step": 17990
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.877000151832894e-06,
+      "loss": 0.7714,
+      "step": 17991
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.873950825749362e-06,
+      "loss": 17.0055,
+      "step": 17992
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.870902650982544e-06,
+      "loss": 17.2554,
+      "step": 17993
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8678556276085355e-06,
+      "loss": 16.4059,
+      "step": 17994
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.864809755703386e-06,
+      "loss": 17.4112,
+      "step": 17995
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.861765035343118e-06,
+      "loss": 0.7527,
+      "step": 17996
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.858721466603727e-06,
+      "loss": 16.6529,
+      "step": 17997
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.855679049561179e-06,
+      "loss": 0.6935,
+      "step": 17998
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.852637784291424e-06,
+      "loss": 17.6973,
+      "step": 17999
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.849597670870369e-06,
+      "loss": 0.7288,
+      "step": 18000
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.846558709373882e-06,
+      "loss": 16.3804,
+      "step": 18001
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.84352089987784e-06,
+      "loss": 17.3035,
+      "step": 18002
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8404842424580535e-06,
+      "loss": 0.716,
+      "step": 18003
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.837448737190319e-06,
+      "loss": 17.5058,
+      "step": 18004
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.834414384150403e-06,
+      "loss": 0.803,
+      "step": 18005
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.83138118341404e-06,
+      "loss": 0.7917,
+      "step": 18006
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.828349135056946e-06,
+      "loss": 0.7475,
+      "step": 18007
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8253182391548014e-06,
+      "loss": 0.7538,
+      "step": 18008
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.822288495783261e-06,
+      "loss": 0.693,
+      "step": 18009
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.819259905017947e-06,
+      "loss": 0.7125,
+      "step": 18010
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.816232466934438e-06,
+      "loss": 17.148,
+      "step": 18011
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8132061816083196e-06,
+      "loss": 16.891,
+      "step": 18012
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8101810491151136e-06,
+      "loss": 0.8212,
+      "step": 18013
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8071570695303427e-06,
+      "loss": 0.7849,
+      "step": 18014
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8041342429294802e-06,
+      "loss": 0.674,
+      "step": 18015
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8011125693879602e-06,
+      "loss": 0.79,
+      "step": 18016
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7980920489812445e-06,
+      "loss": 18.2765,
+      "step": 18017
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7950726817846784e-06,
+      "loss": 18.7016,
+      "step": 18018
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.792054467873657e-06,
+      "loss": 16.5198,
+      "step": 18019
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7890374073235034e-06,
+      "loss": 0.7616,
+      "step": 18020
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7860215002095245e-06,
+      "loss": 0.7668,
+      "step": 18021
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.78300674660701e-06,
+      "loss": 0.7792,
+      "step": 18022
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.779993146591193e-06,
+      "loss": 0.7308,
+      "step": 18023
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7769807002373093e-06,
+      "loss": 0.7259,
+      "step": 18024
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7739694076205366e-06,
+      "loss": 0.7596,
+      "step": 18025
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7709592688160324e-06,
+      "loss": 13.6955,
+      "step": 18026
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.767950283898952e-06,
+      "loss": 16.1753,
+      "step": 18027
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7649424529443867e-06,
+      "loss": 19.0128,
+      "step": 18028
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.761935776027409e-06,
+      "loss": 17.745,
+      "step": 18029
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7589302532230865e-06,
+      "loss": 17.1801,
+      "step": 18030
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.755925884606404e-06,
+      "loss": 15.5825,
+      "step": 18031
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.752922670252379e-06,
+      "loss": 18.4082,
+      "step": 18032
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7499206102359684e-06,
+      "loss": 15.7607,
+      "step": 18033
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7469197046320848e-06,
+      "loss": 0.764,
+      "step": 18034
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7439199535156623e-06,
+      "loss": 0.7372,
+      "step": 18035
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7409213569615463e-06,
+      "loss": 17.9375,
+      "step": 18036
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.737923915044611e-06,
+      "loss": 16.812,
+      "step": 18037
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.734927627839652e-06,
+      "loss": 0.7178,
+      "step": 18038
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.731932495421453e-06,
+      "loss": 18.1739,
+      "step": 18039
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.728938517864794e-06,
+      "loss": 0.7458,
+      "step": 18040
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7259456952443873e-06,
+      "loss": 17.2092,
+      "step": 18041
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.722954027634945e-06,
+      "loss": 0.7049,
+      "step": 18042
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.719963515111141e-06,
+      "loss": 19.5891,
+      "step": 18043
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7169741577476157e-06,
+      "loss": 17.7002,
+      "step": 18044
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7139859556189815e-06,
+      "loss": 0.7209,
+      "step": 18045
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7109989087998176e-06,
+      "loss": 15.8593,
+      "step": 18046
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7080130173646977e-06,
+      "loss": 16.1307,
+      "step": 18047
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.70502828138814e-06,
+      "loss": 0.7156,
+      "step": 18048
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7020447009446457e-06,
+      "loss": 16.1423,
+      "step": 18049
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.699062276108689e-06,
+      "loss": 0.7085,
+      "step": 18050
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6960810069547103e-06,
+      "loss": 16.8497,
+      "step": 18051
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6931008935571275e-06,
+      "loss": 0.7579,
+      "step": 18052
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.690121935990315e-06,
+      "loss": 14.9477,
+      "step": 18053
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.687144134328624e-06,
+      "loss": 16.2202,
+      "step": 18054
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6841674886464005e-06,
+      "loss": 0.7763,
+      "step": 18055
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.68119199901793e-06,
+      "loss": 16.8162,
+      "step": 18056
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.678217665517475e-06,
+      "loss": 17.0328,
+      "step": 18057
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6752444882193036e-06,
+      "loss": 16.5914,
+      "step": 18058
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6722724671975905e-06,
+      "loss": 18.0767,
+      "step": 18059
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.669301602526537e-06,
+      "loss": 0.7764,
+      "step": 18060
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6663318942803004e-06,
+      "loss": 0.7587,
+      "step": 18061
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6633633425329884e-06,
+      "loss": 16.5686,
+      "step": 18062
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6603959473587133e-06,
+      "loss": 0.8006,
+      "step": 18063
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6574297088315336e-06,
+      "loss": 17.3581,
+      "step": 18064
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6544646270255e-06,
+      "loss": 18.4441,
+      "step": 18065
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.651500702014604e-06,
+      "loss": 0.7659,
+      "step": 18066
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6485379338728308e-06,
+      "loss": 18.2338,
+      "step": 18067
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.645576322674138e-06,
+      "loss": 0.7241,
+      "step": 18068
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.642615868492433e-06,
+      "loss": 0.7636,
+      "step": 18069
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.639656571401634e-06,
+      "loss": 15.7774,
+      "step": 18070
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6366984314755935e-06,
+      "loss": 17.0808,
+      "step": 18071
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6337414487881418e-06,
+      "loss": 17.7264,
+      "step": 18072
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.630785623413091e-06,
+      "loss": 17.0157,
+      "step": 18073
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6278309554242106e-06,
+      "loss": 0.7438,
+      "step": 18074
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6248774448952695e-06,
+      "loss": 0.786,
+      "step": 18075
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6219250918999747e-06,
+      "loss": 16.732,
+      "step": 18076
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6189738965120066e-06,
+      "loss": 0.7558,
+      "step": 18077
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.616023858805051e-06,
+      "loss": 0.771,
+      "step": 18078
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6130749788527316e-06,
+      "loss": 16.8476,
+      "step": 18079
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6101272567286514e-06,
+      "loss": 16.4172,
+      "step": 18080
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.60718069250639e-06,
+      "loss": 0.8445,
+      "step": 18081
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6042352862594776e-06,
+      "loss": 18.4355,
+      "step": 18082
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6012910380614607e-06,
+      "loss": 16.9137,
+      "step": 18083
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5983479479858086e-06,
+      "loss": 0.734,
+      "step": 18084
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5954060161059788e-06,
+      "loss": 14.8765,
+      "step": 18085
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.59246524249543e-06,
+      "loss": 0.7513,
+      "step": 18086
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5895256272275245e-06,
+      "loss": 16.9232,
+      "step": 18087
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.586587170375666e-06,
+      "loss": 0.7462,
+      "step": 18088
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5836498720131895e-06,
+      "loss": 0.7188,
+      "step": 18089
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5807137322134033e-06,
+      "loss": 0.7669,
+      "step": 18090
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.57777875104961e-06,
+      "loss": 0.7964,
+      "step": 18091
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5748449285950504e-06,
+      "loss": 0.78,
+      "step": 18092
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.571912264922972e-06,
+      "loss": 0.7359,
+      "step": 18093
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5689807601065607e-06,
+      "loss": 0.8016,
+      "step": 18094
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5660504142189855e-06,
+      "loss": 15.9174,
+      "step": 18095
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5631212273333992e-06,
+      "loss": 17.5787,
+      "step": 18096
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.560193199522904e-06,
+      "loss": 16.8902,
+      "step": 18097
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.557266330860598e-06,
+      "loss": 0.7372,
+      "step": 18098
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5543406214195272e-06,
+      "loss": 0.7252,
+      "step": 18099
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5514160712727175e-06,
+      "loss": 0.698,
+      "step": 18100
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5484926804931707e-06,
+      "loss": 0.7395,
+      "step": 18101
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5455704491538456e-06,
+      "loss": 18.8508,
+      "step": 18102
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5426493773276946e-06,
+      "loss": 18.0468,
+      "step": 18103
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5397294650876266e-06,
+      "loss": 16.487,
+      "step": 18104
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5368107125065054e-06,
+      "loss": 0.7654,
+      "step": 18105
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5338931196572165e-06,
+      "loss": 0.7324,
+      "step": 18106
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5309766866125527e-06,
+      "loss": 17.158,
+      "step": 18107
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.528061413445327e-06,
+      "loss": 17.3045,
+      "step": 18108
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5251473002282987e-06,
+      "loss": 0.7341,
+      "step": 18109
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.522234347034198e-06,
+      "loss": 17.9352,
+      "step": 18110
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5193225539357443e-06,
+      "loss": 0.7948,
+      "step": 18111
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.516411921005619e-06,
+      "loss": 0.7022,
+      "step": 18112
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.513502448316458e-06,
+      "loss": 18.8312,
+      "step": 18113
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5105941359408977e-06,
+      "loss": 17.2947,
+      "step": 18114
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.507686983951508e-06,
+      "loss": 18.0672,
+      "step": 18115
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.504780992420881e-06,
+      "loss": 0.8228,
+      "step": 18116
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.501876161421536e-06,
+      "loss": 16.928,
+      "step": 18117
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.4989724910259647e-06,
+      "loss": 18.4054,
+      "step": 18118
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.496069981306671e-06,
+      "loss": 0.7263,
+      "step": 18119
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.4931686323360857e-06,
+      "loss": 16.3146,
+      "step": 18120
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.490268444186634e-06,
+      "loss": 17.1006,
+      "step": 18121
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.4873694169306913e-06,
+      "loss": 17.2683,
+      "step": 18122
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.4844715506406387e-06,
+      "loss": 20.4509,
+      "step": 18123
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.481574845388791e-06,
+      "loss": 15.7241,
+      "step": 18124
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.478679301247456e-06,
+      "loss": 0.7945,
+      "step": 18125
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.47578491828891e-06,
+      "loss": 0.7884,
+      "step": 18126
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.4728916965853952e-06,
+      "loss": 0.7556,
+      "step": 18127
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.4699996362091313e-06,
+      "loss": 16.1593,
+      "step": 18128
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.4671087372322997e-06,
+      "loss": 0.7383,
+      "step": 18129
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.464218999727048e-06,
+      "loss": 16.27,
+      "step": 18130
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.46133042376553e-06,
+      "loss": 14.8508,
+      "step": 18131
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.4584430094198262e-06,
+      "loss": 17.9711,
+      "step": 18132
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.4555567567620017e-06,
+      "loss": 18.9265,
+      "step": 18133
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.4526716658641265e-06,
+      "loss": 19.1218,
+      "step": 18134
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.449787736798177e-06,
+      "loss": 17.5397,
+      "step": 18135
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.4469049696361667e-06,
+      "loss": 0.8175,
+      "step": 18136
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.444023364450033e-06,
+      "loss": 17.1853,
+      "step": 18137
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.441142921311702e-06,
+      "loss": 18.3425,
+      "step": 18138
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.438263640293077e-06,
+      "loss": 0.751,
+      "step": 18139
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.435385521466028e-06,
+      "loss": 0.7547,
+      "step": 18140
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.432508564902387e-06,
+      "loss": 0.7634,
+      "step": 18141
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.4296327706739628e-06,
+      "loss": 16.5201,
+      "step": 18142
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.4267581388525373e-06,
+      "loss": 0.7499,
+      "step": 18143
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.4238846695098636e-06,
+      "loss": 17.7616,
+      "step": 18144
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.4210123627176627e-06,
+      "loss": 0.7657,
+      "step": 18145
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.418141218547627e-06,
+      "loss": 17.757,
+      "step": 18146
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.415271237071427e-06,
+      "loss": 0.7778,
+      "step": 18147
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.4124024183606885e-06,
+      "loss": 16.6357,
+      "step": 18148
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.4095347624870323e-06,
+      "loss": 18.6375,
+      "step": 18149
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.406668269522012e-06,
+      "loss": 19.6079,
+      "step": 18150
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.4038029395371986e-06,
+      "loss": 17.9198,
+      "step": 18151
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.4009387726041064e-06,
+      "loss": 18.2098,
+      "step": 18152
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3980757687942123e-06,
+      "loss": 0.8024,
+      "step": 18153
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3952139281789974e-06,
+      "loss": 18.7493,
+      "step": 18154
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3923532508298826e-06,
+      "loss": 0.7746,
+      "step": 18155
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.389493736818272e-06,
+      "loss": 17.2246,
+      "step": 18156
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.386635386215542e-06,
+      "loss": 19.135,
+      "step": 18157
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.383778199093024e-06,
+      "loss": 0.7079,
+      "step": 18158
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3809221755220554e-06,
+      "loss": 0.7778,
+      "step": 18159
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3780673155739186e-06,
+      "loss": 17.9935,
+      "step": 18160
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.375213619319856e-06,
+      "loss": 18.7594,
+      "step": 18161
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3723610868311216e-06,
+      "loss": 0.7304,
+      "step": 18162
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.369509718178887e-06,
+      "loss": 15.532,
+      "step": 18163
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.366659513434345e-06,
+      "loss": 17.3779,
+      "step": 18164
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3638104726686272e-06,
+      "loss": 0.7102,
+      "step": 18165
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.360962595952838e-06,
+      "loss": 15.7868,
+      "step": 18166
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.358115883358087e-06,
+      "loss": 15.2327,
+      "step": 18167
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.355270334955407e-06,
+      "loss": 17.7494,
+      "step": 18168
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3524259508158352e-06,
+      "loss": 0.7375,
+      "step": 18169
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.349582731010359e-06,
+      "loss": 0.7299,
+      "step": 18170
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3467406756099386e-06,
+      "loss": 17.6311,
+      "step": 18171
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3438997846855393e-06,
+      "loss": 0.7218,
+      "step": 18172
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3410600583080488e-06,
+      "loss": 16.934,
+      "step": 18173
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.338221496548344e-06,
+      "loss": 18.2113,
+      "step": 18174
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3353840994772958e-06,
+      "loss": 15.4963,
+      "step": 18175
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.332547867165714e-06,
+      "loss": 16.0708,
+      "step": 18176
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3297127996843925e-06,
+      "loss": 0.72,
+      "step": 18177
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.326878897104091e-06,
+      "loss": 16.6328,
+      "step": 18178
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.324046159495553e-06,
+      "loss": 0.7554,
+      "step": 18179
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.321214586929483e-06,
+      "loss": 0.747,
+      "step": 18180
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.318384179476547e-06,
+      "loss": 0.7681,
+      "step": 18181
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.315554937207416e-06,
+      "loss": 19.2273,
+      "step": 18182
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3127268601926776e-06,
+      "loss": 0.7295,
+      "step": 18183
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3098999485029424e-06,
+      "loss": 15.7356,
+      "step": 18184
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.307074202208765e-06,
+      "loss": 20.1784,
+      "step": 18185
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.3042496213806727e-06,
+      "loss": 16.4771,
+      "step": 18186
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.301426206089181e-06,
+      "loss": 17.999,
+      "step": 18187
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.29860395640475e-06,
+      "loss": 18.3696,
+      "step": 18188
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.295782872397829e-06,
+      "loss": 0.7565,
+      "step": 18189
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2929629541388284e-06,
+      "loss": 0.7392,
+      "step": 18190
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2901442016981366e-06,
+      "loss": 18.1415,
+      "step": 18191
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2873266151461135e-06,
+      "loss": 16.0612,
+      "step": 18192
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.284510194553081e-06,
+      "loss": 18.3427,
+      "step": 18193
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.281694939989338e-06,
+      "loss": 0.7675,
+      "step": 18194
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2788808515251623e-06,
+      "loss": 0.7467,
+      "step": 18195
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2760679292307863e-06,
+      "loss": 17.6914,
+      "step": 18196
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2732561731764253e-06,
+      "loss": 15.7105,
+      "step": 18197
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2704455834322633e-06,
+      "loss": 0.791,
+      "step": 18198
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.267636160068438e-06,
+      "loss": 0.7838,
+      "step": 18199
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2648279031550877e-06,
+      "loss": 16.5704,
+      "step": 18200
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.262020812762312e-06,
+      "loss": 17.289,
+      "step": 18201
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.259214888960155e-06,
+      "loss": 0.7017,
+      "step": 18202
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.256410131818682e-06,
+      "loss": 19.1651,
+      "step": 18203
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.253606541407872e-06,
+      "loss": 0.7556,
+      "step": 18204
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2508041177977232e-06,
+      "loss": 17.8886,
+      "step": 18205
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.248002861058175e-06,
+      "loss": 18.4422,
+      "step": 18206
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.245202771259154e-06,
+      "loss": 16.232,
+      "step": 18207
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2424038484705434e-06,
+      "loss": 17.2479,
+      "step": 18208
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2396060927622042e-06,
+      "loss": 17.1473,
+      "step": 18209
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2368095042039916e-06,
+      "loss": 18.7916,
+      "step": 18210
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.234014082865672e-06,
+      "loss": 17.9078,
+      "step": 18211
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2312198288170505e-06,
+      "loss": 0.7352,
+      "step": 18212
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.22842674212786e-06,
+      "loss": 0.8238,
+      "step": 18213
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.225634822867807e-06,
+      "loss": 18.5255,
+      "step": 18214
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2228440711065955e-06,
+      "loss": 0.7833,
+      "step": 18215
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2200544869138814e-06,
+      "loss": 0.8051,
+      "step": 18216
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2172660703592873e-06,
+      "loss": 18.6595,
+      "step": 18217
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.214478821512412e-06,
+      "loss": 0.7574,
+      "step": 18218
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2116927404428233e-06,
+      "loss": 16.703,
+      "step": 18219
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.20890782722007e-06,
+      "loss": 0.7311,
+      "step": 18220
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.206124081913664e-06,
+      "loss": 17.7463,
+      "step": 18221
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.203341504593077e-06,
+      "loss": 17.5391,
+      "step": 18222
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.200560095327776e-06,
+      "loss": 0.742,
+      "step": 18223
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1977798541871884e-06,
+      "loss": 16.6186,
+      "step": 18224
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1950007812406925e-06,
+      "loss": 17.1784,
+      "step": 18225
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1922228765576713e-06,
+      "loss": 0.7797,
+      "step": 18226
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1894461402074417e-06,
+      "loss": 16.2929,
+      "step": 18227
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1866705722593372e-06,
+      "loss": 0.8068,
+      "step": 18228
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1838961727826244e-06,
+      "loss": 0.7696,
+      "step": 18229
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1811229418465425e-06,
+      "loss": 17.6027,
+      "step": 18230
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1783508795203363e-06,
+      "loss": 14.9896,
+      "step": 18231
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.175579985873167e-06,
+      "loss": 0.7572,
+      "step": 18232
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1728102609742237e-06,
+      "loss": 18.3493,
+      "step": 18233
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.170041704892618e-06,
+      "loss": 0.7338,
+      "step": 18234
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1672743176974715e-06,
+      "loss": 0.7371,
+      "step": 18235
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1645080994578524e-06,
+      "loss": 17.3675,
+      "step": 18236
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1617430502427993e-06,
+      "loss": 17.1436,
+      "step": 18237
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1589791701213512e-06,
+      "loss": 0.7079,
+      "step": 18238
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.156216459162459e-06,
+      "loss": 18.3941,
+      "step": 18239
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.153454917435106e-06,
+      "loss": 0.7584,
+      "step": 18240
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1506945450082147e-06,
+      "loss": 18.9239,
+      "step": 18241
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1479353419506807e-06,
+      "loss": 16.6854,
+      "step": 18242
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1451773083313875e-06,
+      "loss": 17.9378,
+      "step": 18243
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.142420444219163e-06,
+      "loss": 16.5915,
+      "step": 18244
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1396647496828247e-06,
+      "loss": 16.2533,
+      "step": 18245
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.136910224791151e-06,
+      "loss": 0.7564,
+      "step": 18246
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.134156869612892e-06,
+      "loss": 0.7707,
+      "step": 18247
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1314046842167878e-06,
+      "loss": 0.7624,
+      "step": 18248
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1286536686715217e-06,
+      "loss": 0.7308,
+      "step": 18249
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.125903823045756e-06,
+      "loss": 17.324,
+      "step": 18250
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.123155147408141e-06,
+      "loss": 16.722,
+      "step": 18251
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1204076418272775e-06,
+      "loss": 0.7566,
+      "step": 18252
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.117661306371744e-06,
+      "loss": 16.4797,
+      "step": 18253
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.114916141110086e-06,
+      "loss": 16.0889,
+      "step": 18254
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.11217214611082e-06,
+      "loss": 0.7759,
+      "step": 18255
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.109429321442453e-06,
+      "loss": 17.9091,
+      "step": 18256
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1066876671734356e-06,
+      "loss": 18.3188,
+      "step": 18257
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.103947183372202e-06,
+      "loss": 17.0728,
+      "step": 18258
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1012078701071535e-06,
+      "loss": 0.7225,
+      "step": 18259
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0984697274466566e-06,
+      "loss": 0.8035,
+      "step": 18260
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0957327554590688e-06,
+      "loss": 0.8152,
+      "step": 18261
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0929969542127014e-06,
+      "loss": 0.7915,
+      "step": 18262
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0902623237758387e-06,
+      "loss": 18.2329,
+      "step": 18263
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.087528864216743e-06,
+      "loss": 0.7877,
+      "step": 18264
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0847965756036436e-06,
+      "loss": 0.732,
+      "step": 18265
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0820654580047294e-06,
+      "loss": 16.8545,
+      "step": 18266
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.079335511488163e-06,
+      "loss": 17.3162,
+      "step": 18267
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0766067361221063e-06,
+      "loss": 17.4181,
+      "step": 18268
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0738791319746606e-06,
+      "loss": 19.3556,
+      "step": 18269
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0711526991138996e-06,
+      "loss": 17.5739,
+      "step": 18270
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.06842743760789e-06,
+      "loss": 17.6478,
+      "step": 18271
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0657033475246455e-06,
+      "loss": 0.7495,
+      "step": 18272
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0629804289321605e-06,
+      "loss": 17.1455,
+      "step": 18273
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0602586818984035e-06,
+      "loss": 15.5231,
+      "step": 18274
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.057538106491303e-06,
+      "loss": 0.7686,
+      "step": 18275
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.054818702778772e-06,
+      "loss": 0.7837,
+      "step": 18276
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0521004708286894e-06,
+      "loss": 0.7791,
+      "step": 18277
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0493834107088837e-06,
+      "loss": 15.4632,
+      "step": 18278
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.046667522487212e-06,
+      "loss": 17.1803,
+      "step": 18279
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0439528062314205e-06,
+      "loss": 0.8198,
+      "step": 18280
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0412392620092933e-06,
+      "loss": 17.5421,
+      "step": 18281
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.03852688988856e-06,
+      "loss": 0.7552,
+      "step": 18282
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0358156899369106e-06,
+      "loss": 0.7218,
+      "step": 18283
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0331056622220355e-06,
+      "loss": 0.7556,
+      "step": 18284
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0303968068115527e-06,
+      "loss": 0.7299,
+      "step": 18285
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.027689123773109e-06,
+      "loss": 17.1094,
+      "step": 18286
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.02498261317426e-06,
+      "loss": 14.9297,
+      "step": 18287
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.022277275082569e-06,
+      "loss": 16.7936,
+      "step": 18288
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0195731095655656e-06,
+      "loss": 0.6943,
+      "step": 18289
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.01687011669074e-06,
+      "loss": 0.7646,
+      "step": 18290
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.014168296525566e-06,
+      "loss": 16.5585,
+      "step": 18291
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0114676491374838e-06,
+      "loss": 17.3519,
+      "step": 18292
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0087681745939012e-06,
+      "loss": 17.4561,
+      "step": 18293
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0060698729621916e-06,
+      "loss": 0.748,
+      "step": 18294
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.003372744309696e-06,
+      "loss": 0.7329,
+      "step": 18295
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0006767887037603e-06,
+      "loss": 0.7532,
+      "step": 18296
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.997982006211664e-06,
+      "loss": 0.7027,
+      "step": 18297
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9952883969006584e-06,
+      "loss": 0.7403,
+      "step": 18298
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.992595960837996e-06,
+      "loss": 16.9859,
+      "step": 18299
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9899046980908674e-06,
+      "loss": 15.3367,
+      "step": 18300
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9872146087264574e-06,
+      "loss": 16.7028,
+      "step": 18301
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9845256928119014e-06,
+      "loss": 17.291,
+      "step": 18302
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9818379504143124e-06,
+      "loss": 18.8599,
+      "step": 18303
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.979151381600792e-06,
+      "loss": 0.7693,
+      "step": 18304
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.976465986438387e-06,
+      "loss": 0.7458,
+      "step": 18305
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9737817649941267e-06,
+      "loss": 0.7144,
+      "step": 18306
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.971098717335019e-06,
+      "loss": 16.6298,
+      "step": 18307
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.968416843528016e-06,
+      "loss": 0.7282,
+      "step": 18308
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9657361436400687e-06,
+      "loss": 16.7824,
+      "step": 18309
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9630566177380915e-06,
+      "loss": 17.1111,
+      "step": 18310
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.960378265888952e-06,
+      "loss": 15.2092,
+      "step": 18311
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9577010881595147e-06,
+      "loss": 20.3128,
+      "step": 18312
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9550250846165972e-06,
+      "loss": 0.7497,
+      "step": 18313
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.952350255327013e-06,
+      "loss": 17.2827,
+      "step": 18314
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9496766003574925e-06,
+      "loss": 0.7739,
+      "step": 18315
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.947004119774788e-06,
+      "loss": 0.7417,
+      "step": 18316
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9443328136456062e-06,
+      "loss": 15.415,
+      "step": 18317
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9416626820366165e-06,
+      "loss": 0.7678,
+      "step": 18318
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9389937250144772e-06,
+      "loss": 17.8563,
+      "step": 18319
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9363259426458012e-06,
+      "loss": 16.0356,
+      "step": 18320
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9336593349971742e-06,
+      "loss": 0.7288,
+      "step": 18321
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9309939021351594e-06,
+      "loss": 0.7703,
+      "step": 18322
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9283296441262765e-06,
+      "loss": 16.4939,
+      "step": 18323
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.925666561037038e-06,
+      "loss": 0.7965,
+      "step": 18324
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9230046529339137e-06,
+      "loss": 19.4933,
+      "step": 18325
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.920343919883334e-06,
+      "loss": 16.4975,
+      "step": 18326
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.917684361951728e-06,
+      "loss": 0.7299,
+      "step": 18327
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9150259792054656e-06,
+      "loss": 0.6973,
+      "step": 18328
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9123687717109107e-06,
+      "loss": 15.7035,
+      "step": 18329
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9097127395343817e-06,
+      "loss": 0.7438,
+      "step": 18330
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9070578827421702e-06,
+      "loss": 0.7856,
+      "step": 18331
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9044042014005513e-06,
+      "loss": 17.2026,
+      "step": 18332
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.901751695575755e-06,
+      "loss": 0.7831,
+      "step": 18333
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.89910036533399e-06,
+      "loss": 16.2366,
+      "step": 18334
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.896450210741436e-06,
+      "loss": 0.7961,
+      "step": 18335
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.893801231864235e-06,
+      "loss": 0.8173,
+      "step": 18336
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8911534287685172e-06,
+      "loss": 17.5276,
+      "step": 18337
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8885068015203632e-06,
+      "loss": 0.7856,
+      "step": 18338
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.885861350185831e-06,
+      "loss": 17.9908,
+      "step": 18339
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8832170748309627e-06,
+      "loss": 0.7854,
+      "step": 18340
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.880573975521761e-06,
+      "loss": 0.773,
+      "step": 18341
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.877932052324184e-06,
+      "loss": 18.2642,
+      "step": 18342
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8752913053041843e-06,
+      "loss": 0.7427,
+      "step": 18343
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8726517345276648e-06,
+      "loss": 17.2219,
+      "step": 18344
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8700133400605287e-06,
+      "loss": 0.7788,
+      "step": 18345
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.867376121968612e-06,
+      "loss": 0.7386,
+      "step": 18346
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8647400803177627e-06,
+      "loss": 17.8948,
+      "step": 18347
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8621052151737604e-06,
+      "loss": 15.8922,
+      "step": 18348
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8594715266023698e-06,
+      "loss": 18.3859,
+      "step": 18349
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8568390146693324e-06,
+      "loss": 16.6803,
+      "step": 18350
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8542076794403573e-06,
+      "loss": 0.7411,
+      "step": 18351
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8515775209811247e-06,
+      "loss": 0.7882,
+      "step": 18352
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8489485393572878e-06,
+      "loss": 0.7205,
+      "step": 18353
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.84632073463445e-06,
+      "loss": 19.2173,
+      "step": 18354
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8436941068782253e-06,
+      "loss": 16.6891,
+      "step": 18355
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8410686561541553e-06,
+      "loss": 0.723,
+      "step": 18356
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.838444382527783e-06,
+      "loss": 0.7657,
+      "step": 18357
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.835821286064605e-06,
+      "loss": 19.0225,
+      "step": 18358
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8331993668300972e-06,
+      "loss": 0.7512,
+      "step": 18359
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8305786248897014e-06,
+      "loss": 16.4748,
+      "step": 18360
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.827959060308838e-06,
+      "loss": 0.7211,
+      "step": 18361
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8253406731528877e-06,
+      "loss": 17.1653,
+      "step": 18362
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8227234634872034e-06,
+      "loss": 0.7406,
+      "step": 18363
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8201074313771057e-06,
+      "loss": 16.5142,
+      "step": 18364
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.817492576887909e-06,
+      "loss": 17.592,
+      "step": 18365
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8148789000848664e-06,
+      "loss": 16.4641,
+      "step": 18366
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8122664010332144e-06,
+      "loss": 0.7688,
+      "step": 18367
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8096550797981792e-06,
+      "loss": 17.6507,
+      "step": 18368
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8070449364449247e-06,
+      "loss": 0.7958,
+      "step": 18369
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8044359710386047e-06,
+      "loss": 17.1804,
+      "step": 18370
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.8018281836443395e-06,
+      "loss": 0.7239,
+      "step": 18371
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.79922157432721e-06,
+      "loss": 18.8184,
+      "step": 18372
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7966161431522976e-06,
+      "loss": 0.7299,
+      "step": 18373
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7940118901846112e-06,
+      "loss": 0.7282,
+      "step": 18374
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7914088154891827e-06,
+      "loss": 15.4861,
+      "step": 18375
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.788806919130971e-06,
+      "loss": 0.8238,
+      "step": 18376
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7862062011749013e-06,
+      "loss": 0.7264,
+      "step": 18377
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.783606661685917e-06,
+      "loss": 17.6529,
+      "step": 18378
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.781008300728882e-06,
+      "loss": 0.7693,
+      "step": 18379
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.778411118368668e-06,
+      "loss": 17.4648,
+      "step": 18380
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7758151146700994e-06,
+      "loss": 17.236,
+      "step": 18381
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7732202896979587e-06,
+      "loss": 0.7521,
+      "step": 18382
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7706266435170327e-06,
+      "loss": 16.2354,
+      "step": 18383
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7680341761920413e-06,
+      "loss": 17.1624,
+      "step": 18384
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.765442887787706e-06,
+      "loss": 16.1688,
+      "step": 18385
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.762852778368702e-06,
+      "loss": 16.3119,
+      "step": 18386
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.760263847999667e-06,
+      "loss": 18.7899,
+      "step": 18387
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7576760967452433e-06,
+      "loss": 0.7264,
+      "step": 18388
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7550895246700126e-06,
+      "loss": 16.989,
+      "step": 18389
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7525041318385293e-06,
+      "loss": 16.4169,
+      "step": 18390
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7499199183153357e-06,
+      "loss": 15.746,
+      "step": 18391
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7473368841649193e-06,
+      "loss": 18.1782,
+      "step": 18392
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7447550294517733e-06,
+      "loss": 16.9509,
+      "step": 18393
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7421743542403343e-06,
+      "loss": 0.7647,
+      "step": 18394
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7395948585950014e-06,
+      "loss": 0.7773,
+      "step": 18395
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7370165425801785e-06,
+      "loss": 0.7585,
+      "step": 18396
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7344394062602196e-06,
+      "loss": 0.7799,
+      "step": 18397
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.73186344969944e-06,
+      "loss": 0.7386,
+      "step": 18398
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7292886729621437e-06,
+      "loss": 0.7806,
+      "step": 18399
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.726715076112585e-06,
+      "loss": 0.7574,
+      "step": 18400
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7241426592150233e-06,
+      "loss": 0.7747,
+      "step": 18401
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7215714223336465e-06,
+      "loss": 0.7594,
+      "step": 18402
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7190013655326528e-06,
+      "loss": 0.7915,
+      "step": 18403
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7164324888761807e-06,
+      "loss": 0.8168,
+      "step": 18404
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7138647924283388e-06,
+      "loss": 0.76,
+      "step": 18405
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.711298276253238e-06,
+      "loss": 18.4907,
+      "step": 18406
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.708732940414921e-06,
+      "loss": 0.7467,
+      "step": 18407
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7061687849774365e-06,
+      "loss": 15.7647,
+      "step": 18408
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.703605810004772e-06,
+      "loss": 17.8551,
+      "step": 18409
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.701044015560911e-06,
+      "loss": 18.6053,
+      "step": 18410
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6984834017097894e-06,
+      "loss": 0.7497,
+      "step": 18411
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6959239685153183e-06,
+      "loss": 15.7658,
+      "step": 18412
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6933657160413907e-06,
+      "loss": 17.1044,
+      "step": 18413
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6908086443518555e-06,
+      "loss": 0.778,
+      "step": 18414
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6882527535105286e-06,
+      "loss": 19.8888,
+      "step": 18415
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6856980435812253e-06,
+      "loss": 0.7612,
+      "step": 18416
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6831445146277057e-06,
+      "loss": 17.3998,
+      "step": 18417
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.680592166713697e-06,
+      "loss": 17.0429,
+      "step": 18418
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.67804099990292e-06,
+      "loss": 18.3124,
+      "step": 18419
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6754910142590293e-06,
+      "loss": 16.0624,
+      "step": 18420
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6729422098457022e-06,
+      "loss": 0.7894,
+      "step": 18421
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6703945867265378e-06,
+      "loss": 0.7722,
+      "step": 18422
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6678481449651295e-06,
+      "loss": 18.1028,
+      "step": 18423
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6653028846250485e-06,
+      "loss": 0.7775,
+      "step": 18424
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.662758805769805e-06,
+      "loss": 0.7515,
+      "step": 18425
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6602159084629154e-06,
+      "loss": 0.7508,
+      "step": 18426
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.657674192767845e-06,
+      "loss": 0.7878,
+      "step": 18427
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6551336587480323e-06,
+      "loss": 0.7358,
+      "step": 18428
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.652594306466899e-06,
+      "loss": 16.7986,
+      "step": 18429
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6500561359878107e-06,
+      "loss": 17.1423,
+      "step": 18430
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.647519147374156e-06,
+      "loss": 15.5643,
+      "step": 18431
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.644983340689222e-06,
+      "loss": 0.7595,
+      "step": 18432
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6424487159963095e-06,
+      "loss": 0.7631,
+      "step": 18433
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.639915273358701e-06,
+      "loss": 17.3926,
+      "step": 18434
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6373830128396114e-06,
+      "loss": 0.7455,
+      "step": 18435
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6348519345022637e-06,
+      "loss": 19.688,
+      "step": 18436
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6323220384098236e-06,
+      "loss": 15.8851,
+      "step": 18437
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.629793324625446e-06,
+      "loss": 0.7499,
+      "step": 18438
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.627265793212241e-06,
+      "loss": 0.7641,
+      "step": 18439
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6247394442332873e-06,
+      "loss": 19.1082,
+      "step": 18440
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.622214277751667e-06,
+      "loss": 17.0368,
+      "step": 18441
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.619690293830396e-06,
+      "loss": 0.7715,
+      "step": 18442
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.617167492532463e-06,
+      "loss": 0.7046,
+      "step": 18443
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6146458739208567e-06,
+      "loss": 0.7878,
+      "step": 18444
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.61212543805851e-06,
+      "loss": 16.7081,
+      "step": 18445
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6096061850083287e-06,
+      "loss": 17.0883,
+      "step": 18446
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6070881148332003e-06,
+      "loss": 17.4995,
+      "step": 18447
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6045712275959698e-06,
+      "loss": 17.3726,
+      "step": 18448
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6020555233594644e-06,
+      "loss": 0.7604,
+      "step": 18449
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5995410021864787e-06,
+      "loss": 0.7215,
+      "step": 18450
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5970276641397675e-06,
+      "loss": 17.3234,
+      "step": 18451
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5945155092820807e-06,
+      "loss": 0.7709,
+      "step": 18452
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.592004537676096e-06,
+      "loss": 0.6957,
+      "step": 18453
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5894947493845134e-06,
+      "loss": 16.5527,
+      "step": 18454
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5869861444699607e-06,
+      "loss": 0.7145,
+      "step": 18455
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5844787229950597e-06,
+      "loss": 18.9529,
+      "step": 18456
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.581972485022399e-06,
+      "loss": 16.2783,
+      "step": 18457
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.579467430614524e-06,
+      "loss": 16.2991,
+      "step": 18458
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.576963559833989e-06,
+      "loss": 0.8074,
+      "step": 18459
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5744608727432617e-06,
+      "loss": 15.4079,
+      "step": 18460
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5719593694048085e-06,
+      "loss": 0.7281,
+      "step": 18461
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.56945904988109e-06,
+      "loss": 0.762,
+      "step": 18462
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5669599142344958e-06,
+      "loss": 0.7714,
+      "step": 18463
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5644619625274204e-06,
+      "loss": 0.7202,
+      "step": 18464
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.561965194822202e-06,
+      "loss": 18.0158,
+      "step": 18465
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.55946961118117e-06,
+      "loss": 17.6495,
+      "step": 18466
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.556975211666601e-06,
+      "loss": 16.8241,
+      "step": 18467
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.554481996340763e-06,
+      "loss": 16.5756,
+      "step": 18468
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5519899652658885e-06,
+      "loss": 17.6944,
+      "step": 18469
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.549499118504184e-06,
+      "loss": 0.7673,
+      "step": 18470
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.547009456117805e-06,
+      "loss": 15.6292,
+      "step": 18471
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.544520978168913e-06,
+      "loss": 17.3629,
+      "step": 18472
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5420336847196135e-06,
+      "loss": 16.6521,
+      "step": 18473
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.539547575831991e-06,
+      "loss": 16.5049,
+      "step": 18474
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5370626515680952e-06,
+      "loss": 0.7652,
+      "step": 18475
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.534578911989943e-06,
+      "loss": 0.7363,
+      "step": 18476
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5320963571595523e-06,
+      "loss": 16.6254,
+      "step": 18477
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5296149871388674e-06,
+      "loss": 19.4979,
+      "step": 18478
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5271348019898277e-06,
+      "loss": 18.1733,
+      "step": 18479
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5246558017743616e-06,
+      "loss": 16.7129,
+      "step": 18480
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5221779865543083e-06,
+      "loss": 0.7539,
+      "step": 18481
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5197013563915406e-06,
+      "loss": 16.9759,
+      "step": 18482
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.517225911347865e-06,
+      "loss": 15.6565,
+      "step": 18483
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.51475165148507e-06,
+      "loss": 18.3969,
+      "step": 18484
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.512278576864924e-06,
+      "loss": 0.7447,
+      "step": 18485
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.509806687549143e-06,
+      "loss": 18.1147,
+      "step": 18486
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.507335983599435e-06,
+      "loss": 0.7219,
+      "step": 18487
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.50486646507746e-06,
+      "loss": 17.9185,
+      "step": 18488
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.502398132044859e-06,
+      "loss": 16.4138,
+      "step": 18489
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.499930984563248e-06,
+      "loss": 16.0554,
+      "step": 18490
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4974650226942064e-06,
+      "loss": 20.4651,
+      "step": 18491
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.495000246499285e-06,
+      "loss": 0.7344,
+      "step": 18492
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4925366560400065e-06,
+      "loss": 0.7503,
+      "step": 18493
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.490074251377861e-06,
+      "loss": 17.2512,
+      "step": 18494
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.487613032574304e-06,
+      "loss": 18.4233,
+      "step": 18495
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4851529996907763e-06,
+      "loss": 0.7957,
+      "step": 18496
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4826941527886782e-06,
+      "loss": 0.767,
+      "step": 18497
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4802364919293886e-06,
+      "loss": 0.7575,
+      "step": 18498
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.477780017174236e-06,
+      "loss": 0.7894,
+      "step": 18499
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4753247285845605e-06,
+      "loss": 17.5124,
+      "step": 18500
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.472870626221618e-06,
+      "loss": 16.8171,
+      "step": 18501
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.470417710146683e-06,
+      "loss": 0.7647,
+      "step": 18502
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4679659804209777e-06,
+      "loss": 0.7951,
+      "step": 18503
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.465515437105681e-06,
+      "loss": 15.6947,
+      "step": 18504
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.463066080261983e-06,
+      "loss": 17.3998,
+      "step": 18505
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4606179099510074e-06,
+      "loss": 0.7933,
+      "step": 18506
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.458170926233866e-06,
+      "loss": 18.9425,
+      "step": 18507
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4557251291716322e-06,
+      "loss": 15.3659,
+      "step": 18508
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4532805188253516e-06,
+      "loss": 15.8237,
+      "step": 18509
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4508370952560478e-06,
+      "loss": 18.581,
+      "step": 18510
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4483948585247106e-06,
+      "loss": 0.687,
+      "step": 18511
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.445953808692286e-06,
+      "loss": 0.7941,
+      "step": 18512
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.44351394581972e-06,
+      "loss": 0.7582,
+      "step": 18513
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4410752699679018e-06,
+      "loss": 18.8168,
+      "step": 18514
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4386377811977055e-06,
+      "loss": 16.6913,
+      "step": 18515
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.436201479569966e-06,
+      "loss": 18.3377,
+      "step": 18516
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.433766365145501e-06,
+      "loss": 0.7314,
+      "step": 18517
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4313324379850897e-06,
+      "loss": 18.3414,
+      "step": 18518
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4288996981494726e-06,
+      "loss": 16.1535,
+      "step": 18519
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4264681456993955e-06,
+      "loss": 15.9027,
+      "step": 18520
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.424037780695526e-06,
+      "loss": 17.8054,
+      "step": 18521
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.421608603198544e-06,
+      "loss": 0.6987,
+      "step": 18522
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.419180613269073e-06,
+      "loss": 0.8021,
+      "step": 18523
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.416753810967709e-06,
+      "loss": 0.7404,
+      "step": 18524
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4143281963550425e-06,
+      "loss": 18.7743,
+      "step": 18525
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.411903769491608e-06,
+      "loss": 16.8307,
+      "step": 18526
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.409480530437919e-06,
+      "loss": 0.7875,
+      "step": 18527
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4070584792544703e-06,
+      "loss": 17.2439,
+      "step": 18528
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.404637616001698e-06,
+      "loss": 0.781,
+      "step": 18529
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.402217940740048e-06,
+      "loss": 16.8302,
+      "step": 18530
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3997994535298995e-06,
+      "loss": 0.7694,
+      "step": 18531
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.397382154431621e-06,
+      "loss": 16.9423,
+      "step": 18532
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3949660435055644e-06,
+      "loss": 0.7066,
+      "step": 18533
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.39255112081202e-06,
+      "loss": 0.7955,
+      "step": 18534
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.390137386411273e-06,
+      "loss": 0.8185,
+      "step": 18535
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3877248403635643e-06,
+      "loss": 17.3208,
+      "step": 18536
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3853134827291067e-06,
+      "loss": 0.7402,
+      "step": 18537
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3829033135681077e-06,
+      "loss": 0.7518,
+      "step": 18538
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3804943329407136e-06,
+      "loss": 0.6774,
+      "step": 18539
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.378086540907043e-06,
+      "loss": 0.7452,
+      "step": 18540
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.37567993752722e-06,
+      "loss": 0.76,
+      "step": 18541
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3732745228612905e-06,
+      "loss": 0.7869,
+      "step": 18542
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.370870296969313e-06,
+      "loss": 16.6685,
+      "step": 18543
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3684672599112776e-06,
+      "loss": 17.3661,
+      "step": 18544
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.366065411747176e-06,
+      "loss": 20.2471,
+      "step": 18545
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3636647525369646e-06,
+      "loss": 17.9149,
+      "step": 18546
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3612652823405524e-06,
+      "loss": 18.9746,
+      "step": 18547
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3588670012178404e-06,
+      "loss": 17.2098,
+      "step": 18548
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3564699092286813e-06,
+      "loss": 19.6392,
+      "step": 18549
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.354074006432916e-06,
+      "loss": 16.6315,
+      "step": 18550
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3516792928903465e-06,
+      "loss": 16.9876,
+      "step": 18551
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3492857686607252e-06,
+      "loss": 17.041,
+      "step": 18552
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3468934338038264e-06,
+      "loss": 19.6852,
+      "step": 18553
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3445022883793467e-06,
+      "loss": 16.9665,
+      "step": 18554
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.342112332446961e-06,
+      "loss": 17.4603,
+      "step": 18555
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3397235660663485e-06,
+      "loss": 15.1891,
+      "step": 18556
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3373359892971015e-06,
+      "loss": 17.1517,
+      "step": 18557
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3349496021988435e-06,
+      "loss": 0.7889,
+      "step": 18558
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3325644048311222e-06,
+      "loss": 0.7372,
+      "step": 18559
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.330180397253473e-06,
+      "loss": 0.7871,
+      "step": 18560
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3277975795254092e-06,
+      "loss": 15.7664,
+      "step": 18561
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3254159517064057e-06,
+      "loss": 17.225,
+      "step": 18562
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3230355138559035e-06,
+      "loss": 0.7588,
+      "step": 18563
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.320656266033322e-06,
+      "loss": 0.697,
+      "step": 18564
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3182782082980413e-06,
+      "loss": 19.5607,
+      "step": 18565
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.315901340709431e-06,
+      "loss": 17.6761,
+      "step": 18566
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.31352566332681e-06,
+      "loss": 16.6701,
+      "step": 18567
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.311151176209464e-06,
+      "loss": 0.7752,
+      "step": 18568
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.308777879416696e-06,
+      "loss": 0.7151,
+      "step": 18569
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.3064057730077026e-06,
+      "loss": 0.7286,
+      "step": 18570
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.30403485704172e-06,
+      "loss": 15.8172,
+      "step": 18571
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.301665131577907e-06,
+      "loss": 0.7573,
+      "step": 18572
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2992965966754377e-06,
+      "loss": 16.7305,
+      "step": 18573
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2969292523934093e-06,
+      "loss": 15.9547,
+      "step": 18574
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.294563098790914e-06,
+      "loss": 17.3309,
+      "step": 18575
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2921981359270318e-06,
+      "loss": 18.3334,
+      "step": 18576
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2898343638607654e-06,
+      "loss": 0.8103,
+      "step": 18577
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2874717826511295e-06,
+      "loss": 17.3457,
+      "step": 18578
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2851103923570925e-06,
+      "loss": 0.7949,
+      "step": 18579
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2827501930375918e-06,
+      "loss": 0.7633,
+      "step": 18580
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2803911847515457e-06,
+      "loss": 17.3989,
+      "step": 18581
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.27803336755783e-06,
+      "loss": 0.7683,
+      "step": 18582
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.275676741515298e-06,
+      "loss": 0.7634,
+      "step": 18583
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.273321306682774e-06,
+      "loss": 17.9776,
+      "step": 18584
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2709670631190395e-06,
+      "loss": 0.6932,
+      "step": 18585
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.268614010882869e-06,
+      "loss": 0.7245,
+      "step": 18586
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.266262150032994e-06,
+      "loss": 0.8153,
+      "step": 18587
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2639114806281115e-06,
+      "loss": 16.4293,
+      "step": 18588
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2615620027269026e-06,
+      "loss": 16.5453,
+      "step": 18589
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2592137163880035e-06,
+      "loss": 16.5135,
+      "step": 18590
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2568666216700284e-06,
+      "loss": 0.7405,
+      "step": 18591
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.25452071863157e-06,
+      "loss": 0.7783,
+      "step": 18592
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2521760073311637e-06,
+      "loss": 17.5678,
+      "step": 18593
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.249832487827358e-06,
+      "loss": 18.7585,
+      "step": 18594
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2474901601786337e-06,
+      "loss": 0.8092,
+      "step": 18595
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2451490244434546e-06,
+      "loss": 0.7708,
+      "step": 18596
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2428090806802694e-06,
+      "loss": 16.8584,
+      "step": 18597
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.240470328947464e-06,
+      "loss": 18.0138,
+      "step": 18598
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.238132769303425e-06,
+      "loss": 17.5056,
+      "step": 18599
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2357964018064946e-06,
+      "loss": 0.7829,
+      "step": 18600
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.233461226514999e-06,
+      "loss": 16.9284,
+      "step": 18601
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.231127243487219e-06,
+      "loss": 0.7255,
+      "step": 18602
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.228794452781402e-06,
+      "loss": 16.0429,
+      "step": 18603
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2264628544557964e-06,
+      "loss": 17.7734,
+      "step": 18604
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.224132448568572e-06,
+      "loss": 17.0394,
+      "step": 18605
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.221803235177916e-06,
+      "loss": 14.2959,
+      "step": 18606
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.219475214341965e-06,
+      "loss": 16.4949,
+      "step": 18607
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2171483861188115e-06,
+      "loss": 0.774,
+      "step": 18608
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.214822750566553e-06,
+      "loss": 0.75,
+      "step": 18609
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2124983077432273e-06,
+      "loss": 16.954,
+      "step": 18610
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2101750577068592e-06,
+      "loss": 17.9875,
+      "step": 18611
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.207853000515431e-06,
+      "loss": 0.7582,
+      "step": 18612
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2055321362268956e-06,
+      "loss": 17.2814,
+      "step": 18613
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2032124648992015e-06,
+      "loss": 17.2309,
+      "step": 18614
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2008939865902354e-06,
+      "loss": 0.772,
+      "step": 18615
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.198576701357863e-06,
+      "loss": 16.8392,
+      "step": 18616
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.196260609259937e-06,
+      "loss": 0.7848,
+      "step": 18617
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1939457103542617e-06,
+      "loss": 0.7961,
+      "step": 18618
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.191632004698613e-06,
+      "loss": 0.7645,
+      "step": 18619
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.18931949235075e-06,
+      "loss": 17.9817,
+      "step": 18620
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.187008173368382e-06,
+      "loss": 0.7916,
+      "step": 18621
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.184698047809214e-06,
+      "loss": 18.3873,
+      "step": 18622
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1823891157308984e-06,
+      "loss": 0.742,
+      "step": 18623
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.180081377191068e-06,
+      "loss": 0.7981,
+      "step": 18624
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1777748322473314e-06,
+      "loss": 0.7018,
+      "step": 18625
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.175469480957243e-06,
+      "loss": 0.7851,
+      "step": 18626
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1731653233783623e-06,
+      "loss": 15.3698,
+      "step": 18627
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1708623595681933e-06,
+      "loss": 0.7719,
+      "step": 18628
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.168560589584223e-06,
+      "loss": 0.6997,
+      "step": 18629
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1662600134839006e-06,
+      "loss": 17.9402,
+      "step": 18630
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1639606313246465e-06,
+      "loss": 0.7698,
+      "step": 18631
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1616624431638698e-06,
+      "loss": 0.7156,
+      "step": 18632
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.159365449058909e-06,
+      "loss": 0.7388,
+      "step": 18633
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1570696490671117e-06,
+      "loss": 17.3663,
+      "step": 18634
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1547750432457826e-06,
+      "loss": 0.7332,
+      "step": 18635
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1524816316521867e-06,
+      "loss": 0.7761,
+      "step": 18636
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.150189414343584e-06,
+      "loss": 0.7688,
+      "step": 18637
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1478983913771787e-06,
+      "loss": 18.6279,
+      "step": 18638
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.145608562810153e-06,
+      "loss": 16.0391,
+      "step": 18639
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.143319928699661e-06,
+      "loss": 16.8059,
+      "step": 18640
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.141032489102829e-06,
+      "loss": 16.3583,
+      "step": 18641
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1387462440767615e-06,
+      "loss": 0.7067,
+      "step": 18642
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1364611936785127e-06,
+      "loss": 16.1605,
+      "step": 18643
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1341773379651154e-06,
+      "loss": 18.0608,
+      "step": 18644
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.13189467699359e-06,
+      "loss": 0.774,
+      "step": 18645
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1296132108208967e-06,
+      "loss": 17.8047,
+      "step": 18646
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.12733293950399e-06,
+      "loss": 19.6372,
+      "step": 18647
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.125053863099785e-06,
+      "loss": 16.8712,
+      "step": 18648
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.122775981665165e-06,
+      "loss": 0.7554,
+      "step": 18649
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.120499295256989e-06,
+      "loss": 17.51,
+      "step": 18650
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1182238039320843e-06,
+      "loss": 0.7536,
+      "step": 18651
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1159495077472445e-06,
+      "loss": 15.9932,
+      "step": 18652
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.113676406759241e-06,
+      "loss": 0.7431,
+      "step": 18653
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1114045010248053e-06,
+      "loss": 16.9106,
+      "step": 18654
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1091337906006482e-06,
+      "loss": 17.1594,
+      "step": 18655
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.106864275543441e-06,
+      "loss": 0.7473,
+      "step": 18656
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.104595955909844e-06,
+      "loss": 16.1745,
+      "step": 18657
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.102328831756467e-06,
+      "loss": 16.6595,
+      "step": 18658
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1000629031399042e-06,
+      "loss": 0.7862,
+      "step": 18659
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0977981701166995e-06,
+      "loss": 17.3285,
+      "step": 18660
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.09553463274339e-06,
+      "loss": 17.0825,
+      "step": 18661
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0932722910764815e-06,
+      "loss": 0.7114,
+      "step": 18662
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.091011145172428e-06,
+      "loss": 18.1211,
+      "step": 18663
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0887511950876737e-06,
+      "loss": 0.7293,
+      "step": 18664
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0864924408786344e-06,
+      "loss": 16.9038,
+      "step": 18665
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0842348826016877e-06,
+      "loss": 0.7624,
+      "step": 18666
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.081978520313177e-06,
+      "loss": 0.6995,
+      "step": 18667
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.079723354069418e-06,
+      "loss": 19.381,
+      "step": 18668
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0774693839267057e-06,
+      "loss": 15.3811,
+      "step": 18669
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0752166099413052e-06,
+      "loss": 0.7811,
+      "step": 18670
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0729650321694385e-06,
+      "loss": 16.6402,
+      "step": 18671
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0707146506673048e-06,
+      "loss": 0.7202,
+      "step": 18672
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0684654654910874e-06,
+      "loss": 0.7423,
+      "step": 18673
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0662174766969024e-06,
+      "loss": 16.8821,
+      "step": 18674
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0639706843408825e-06,
+      "loss": 17.5823,
+      "step": 18675
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.061725088479094e-06,
+      "loss": 18.0887,
+      "step": 18676
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0594806891675923e-06,
+      "loss": 19.2365,
+      "step": 18677
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0572374864624044e-06,
+      "loss": 18.1732,
+      "step": 18678
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0549954804195082e-06,
+      "loss": 19.3016,
+      "step": 18679
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0527546710948864e-06,
+      "loss": 0.7656,
+      "step": 18680
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.05051505854445e-06,
+      "loss": 0.7773,
+      "step": 18681
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0482766428240984e-06,
+      "loss": 19.818,
+      "step": 18682
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.046039423989715e-06,
+      "loss": 17.0073,
+      "step": 18683
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.043803402097133e-06,
+      "loss": 18.9202,
+      "step": 18684
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0415685772021744e-06,
+      "loss": 17.2115,
+      "step": 18685
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0393349493606105e-06,
+      "loss": 16.3681,
+      "step": 18686
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0371025186282032e-06,
+      "loss": 17.3815,
+      "step": 18687
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0348712850606633e-06,
+      "loss": 0.7683,
+      "step": 18688
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.032641248713685e-06,
+      "loss": 0.7717,
+      "step": 18689
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.03041240964294e-06,
+      "loss": 15.5334,
+      "step": 18690
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.028184767904051e-06,
+      "loss": 0.7624,
+      "step": 18691
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.025958323552618e-06,
+      "loss": 0.7891,
+      "step": 18692
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0237330766442244e-06,
+      "loss": 0.7702,
+      "step": 18693
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.021509027234408e-06,
+      "loss": 18.6749,
+      "step": 18694
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0192861753786875e-06,
+      "loss": 0.6908,
+      "step": 18695
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.017064521132533e-06,
+      "loss": 17.0423,
+      "step": 18696
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0148440645514023e-06,
+      "loss": 19.6618,
+      "step": 18697
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0126248056907216e-06,
+      "loss": 0.7778,
+      "step": 18698
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0104067446058862e-06,
+      "loss": 16.2881,
+      "step": 18699
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.008189881352246e-06,
+      "loss": 0.7037,
+      "step": 18700
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.005974215985157e-06,
+      "loss": 0.7665,
+      "step": 18701
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0037597485599026e-06,
+      "loss": 18.5588,
+      "step": 18702
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.001546479131766e-06,
+      "loss": 0.7719,
+      "step": 18703
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.999334407755987e-06,
+      "loss": 0.7231,
+      "step": 18704
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.997123534487777e-06,
+      "loss": 17.7911,
+      "step": 18705
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9949138593823356e-06,
+      "loss": 0.6933,
+      "step": 18706
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9927053824947916e-06,
+      "loss": 0.7425,
+      "step": 18707
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9904981038803004e-06,
+      "loss": 0.8083,
+      "step": 18708
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9882920235939294e-06,
+      "loss": 0.721,
+      "step": 18709
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.986087141690746e-06,
+      "loss": 0.8142,
+      "step": 18710
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9838834582257937e-06,
+      "loss": 0.7748,
+      "step": 18711
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9816809732540742e-06,
+      "loss": 19.2654,
+      "step": 18712
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9794796868305654e-06,
+      "loss": 17.5222,
+      "step": 18713
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.977279599010212e-06,
+      "loss": 17.2783,
+      "step": 18714
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9750807098479153e-06,
+      "loss": 0.7483,
+      "step": 18715
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9728830193985694e-06,
+      "loss": 15.9601,
+      "step": 18716
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9706865277170307e-06,
+      "loss": 16.7005,
+      "step": 18717
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9684912348581274e-06,
+      "loss": 18.7041,
+      "step": 18718
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9662971408766494e-06,
+      "loss": 0.8005,
+      "step": 18719
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.964104245827353e-06,
+      "loss": 0.7609,
+      "step": 18720
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9619125497649994e-06,
+      "loss": 18.7481,
+      "step": 18721
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9597220527442616e-06,
+      "loss": 17.2196,
+      "step": 18722
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.957532754819835e-06,
+      "loss": 17.1448,
+      "step": 18723
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9553446560463652e-06,
+      "loss": 0.7099,
+      "step": 18724
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.953157756478452e-06,
+      "loss": 0.7452,
+      "step": 18725
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9509720561707024e-06,
+      "loss": 0.7417,
+      "step": 18726
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9487875551776615e-06,
+      "loss": 16.6803,
+      "step": 18727
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.946604253553852e-06,
+      "loss": 16.7579,
+      "step": 18728
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9444221513537696e-06,
+      "loss": 15.9325,
+      "step": 18729
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.942241248631882e-06,
+      "loss": 19.0379,
+      "step": 18730
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9400615454426285e-06,
+      "loss": 0.8181,
+      "step": 18731
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.937883041840416e-06,
+      "loss": 0.8205,
+      "step": 18732
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9357057378796063e-06,
+      "loss": 0.7162,
+      "step": 18733
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.933529633614567e-06,
+      "loss": 0.7292,
+      "step": 18734
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.931354729099599e-06,
+      "loss": 0.7826,
+      "step": 18735
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.929181024388993e-06,
+      "loss": 18.3772,
+      "step": 18736
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9270085195370045e-06,
+      "loss": 17.3258,
+      "step": 18737
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9248372145978577e-06,
+      "loss": 16.7853,
+      "step": 18738
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.922667109625753e-06,
+      "loss": 17.6957,
+      "step": 18739
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9204982046748476e-06,
+      "loss": 0.7637,
+      "step": 18740
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.918330499799292e-06,
+      "loss": 0.76,
+      "step": 18741
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.916163995053194e-06,
+      "loss": 0.7506,
+      "step": 18742
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9139986904906094e-06,
+      "loss": 0.7633,
+      "step": 18743
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.911834586165601e-06,
+      "loss": 17.6206,
+      "step": 18744
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.909671682132175e-06,
+      "loss": 19.6997,
+      "step": 18745
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.907509978444333e-06,
+      "loss": 17.7663,
+      "step": 18746
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9053494751560207e-06,
+      "loss": 0.7761,
+      "step": 18747
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.903190172321162e-06,
+      "loss": 16.3564,
+      "step": 18748
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.901032069993669e-06,
+      "loss": 17.209,
+      "step": 18749
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8988751682273875e-06,
+      "loss": 0.7678,
+      "step": 18750
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.896719467076169e-06,
+      "loss": 18.4549,
+      "step": 18751
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8945649665938203e-06,
+      "loss": 17.4442,
+      "step": 18752
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.892411666834104e-06,
+      "loss": 17.487,
+      "step": 18753
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8902595678507829e-06,
+      "loss": 16.6009,
+      "step": 18754
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.888108669697569e-06,
+      "loss": 0.7562,
+      "step": 18755
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.885958972428148e-06,
+      "loss": 17.7074,
+      "step": 18756
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.883810476096176e-06,
+      "loss": 17.616,
+      "step": 18757
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8816631807552777e-06,
+      "loss": 16.7385,
+      "step": 18758
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8795170864590538e-06,
+      "loss": 0.8079,
+      "step": 18759
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8773721932610732e-06,
+      "loss": 19.5431,
+      "step": 18760
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8752285012148652e-06,
+      "loss": 17.0169,
+      "step": 18761
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8730860103739534e-06,
+      "loss": 15.8613,
+      "step": 18762
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8709447207917951e-06,
+      "loss": 17.4119,
+      "step": 18763
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8688046325218478e-06,
+      "loss": 20.1723,
+      "step": 18764
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8666657456175296e-06,
+      "loss": 0.8203,
+      "step": 18765
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8645280601322202e-06,
+      "loss": 18.2488,
+      "step": 18766
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8623915761192822e-06,
+      "loss": 16.3454,
+      "step": 18767
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8602562936320346e-06,
+      "loss": 18.2914,
+      "step": 18768
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.858122212723795e-06,
+      "loss": 16.7822,
+      "step": 18769
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8559893334478162e-06,
+      "loss": 16.6285,
+      "step": 18770
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8538576558573273e-06,
+      "loss": 18.0293,
+      "step": 18771
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8517271800055524e-06,
+      "loss": 0.7457,
+      "step": 18772
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8495979059456491e-06,
+      "loss": 16.778,
+      "step": 18773
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8474698337307862e-06,
+      "loss": 0.671,
+      "step": 18774
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8453429634140706e-06,
+      "loss": 18.6198,
+      "step": 18775
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8432172950485826e-06,
+      "loss": 0.7053,
+      "step": 18776
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8410928286874018e-06,
+      "loss": 0.766,
+      "step": 18777
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8389695643835247e-06,
+      "loss": 16.7838,
+      "step": 18778
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.83684750218997e-06,
+      "loss": 17.0124,
+      "step": 18779
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.834726642159701e-06,
+      "loss": 0.762,
+      "step": 18780
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8326069843456417e-06,
+      "loss": 19.7221,
+      "step": 18781
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8304885288007223e-06,
+      "loss": 0.6863,
+      "step": 18782
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8283712755778059e-06,
+      "loss": 18.7572,
+      "step": 18783
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8262552247297394e-06,
+      "loss": 18.3515,
+      "step": 18784
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8241403763093413e-06,
+      "loss": 16.7515,
+      "step": 18785
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8220267303693973e-06,
+      "loss": 16.4919,
+      "step": 18786
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8199142869626706e-06,
+      "loss": 15.1862,
+      "step": 18787
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.817803046141886e-06,
+      "loss": 0.7511,
+      "step": 18788
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.815693007959729e-06,
+      "loss": 16.7116,
+      "step": 18789
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8135841724688906e-06,
+      "loss": 18.4927,
+      "step": 18790
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.811476539721979e-06,
+      "loss": 16.5751,
+      "step": 18791
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8093701097716242e-06,
+      "loss": 17.478,
+      "step": 18792
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8072648826703954e-06,
+      "loss": 0.7188,
+      "step": 18793
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8051608584708334e-06,
+      "loss": 16.3271,
+      "step": 18794
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.803058037225469e-06,
+      "loss": 18.8604,
+      "step": 18795
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8009564189867712e-06,
+      "loss": 17.2013,
+      "step": 18796
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7988560038072144e-06,
+      "loss": 16.62,
+      "step": 18797
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.796756791739218e-06,
+      "loss": 18.6742,
+      "step": 18798
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.794658782835168e-06,
+      "loss": 0.7551,
+      "step": 18799
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.79256197714745e-06,
+      "loss": 17.0816,
+      "step": 18800
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7904663747283889e-06,
+      "loss": 0.7374,
+      "step": 18801
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7883719756302986e-06,
+      "loss": 0.7516,
+      "step": 18802
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.786278779905448e-06,
+      "loss": 17.1699,
+      "step": 18803
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7841867876060902e-06,
+      "loss": 0.7929,
+      "step": 18804
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7820959987844444e-06,
+      "loss": 16.4885,
+      "step": 18805
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7800064134926797e-06,
+      "loss": 17.0166,
+      "step": 18806
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7779180317829712e-06,
+      "loss": 16.6072,
+      "step": 18807
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7758308537074443e-06,
+      "loss": 15.5842,
+      "step": 18808
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7737448793181787e-06,
+      "loss": 16.7941,
+      "step": 18809
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7716601086672613e-06,
+      "loss": 0.8166,
+      "step": 18810
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7695765418067168e-06,
+      "loss": 14.2132,
+      "step": 18811
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7674941787885534e-06,
+      "loss": 18.9456,
+      "step": 18812
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7654130196647523e-06,
+      "loss": 18.083,
+      "step": 18813
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7633330644872491e-06,
+      "loss": 0.7063,
+      "step": 18814
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7612543133079695e-06,
+      "loss": 17.0627,
+      "step": 18815
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7591767661787994e-06,
+      "loss": 0.758,
+      "step": 18816
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7571004231515863e-06,
+      "loss": 0.7426,
+      "step": 18817
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7550252842781723e-06,
+      "loss": 16.6008,
+      "step": 18818
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7529513496103323e-06,
+      "loss": 0.7196,
+      "step": 18819
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7508786191998471e-06,
+      "loss": 0.7367,
+      "step": 18820
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7488070930984535e-06,
+      "loss": 0.7964,
+      "step": 18821
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7467367713578376e-06,
+      "loss": 17.4121,
+      "step": 18822
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7446676540297025e-06,
+      "loss": 19.4887,
+      "step": 18823
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7425997411656736e-06,
+      "loss": 16.2448,
+      "step": 18824
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7405330328173818e-06,
+      "loss": 15.5745,
+      "step": 18825
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7384675290364027e-06,
+      "loss": 16.8259,
+      "step": 18826
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7364032298742839e-06,
+      "loss": 0.8244,
+      "step": 18827
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7343401353825672e-06,
+      "loss": 0.7175,
+      "step": 18828
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7322782456127395e-06,
+      "loss": 0.7595,
+      "step": 18829
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7302175606162708e-06,
+      "loss": 19.8609,
+      "step": 18830
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7281580804445917e-06,
+      "loss": 0.7612,
+      "step": 18831
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7260998051491118e-06,
+      "loss": 0.7841,
+      "step": 18832
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7240427347812004e-06,
+      "loss": 16.0958,
+      "step": 18833
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7219868693922003e-06,
+      "loss": 0.7299,
+      "step": 18834
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.719932209033437e-06,
+      "loss": 0.768,
+      "step": 18835
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7178787537561914e-06,
+      "loss": 17.4522,
+      "step": 18836
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.715826503611706e-06,
+      "loss": 0.7237,
+      "step": 18837
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.713775458651229e-06,
+      "loss": 17.8119,
+      "step": 18838
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7117256189259356e-06,
+      "loss": 0.7429,
+      "step": 18839
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7096769844869964e-06,
+      "loss": 17.9298,
+      "step": 18840
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7076295553855481e-06,
+      "loss": 0.766,
+      "step": 18841
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7055833316726888e-06,
+      "loss": 0.7483,
+      "step": 18842
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7035383133994997e-06,
+      "loss": 0.7369,
+      "step": 18843
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.701494500617018e-06,
+      "loss": 0.7445,
+      "step": 18844
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6994518933762638e-06,
+      "loss": 15.2842,
+      "step": 18845
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6974104917282242e-06,
+      "loss": 17.1634,
+      "step": 18846
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6953702957238416e-06,
+      "loss": 17.5977,
+      "step": 18847
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6933313054140476e-06,
+      "loss": 16.8558,
+      "step": 18848
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.691293520849735e-06,
+      "loss": 0.7762,
+      "step": 18849
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6892569420817573e-06,
+      "loss": 15.5101,
+      "step": 18850
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6872215691609684e-06,
+      "loss": 0.7537,
+      "step": 18851
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6851874021381498e-06,
+      "loss": 0.766,
+      "step": 18852
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6831544410641e-06,
+      "loss": 0.7116,
+      "step": 18853
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.681122685989539e-06,
+      "loss": 0.7744,
+      "step": 18854
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.679092136965188e-06,
+      "loss": 0.7155,
+      "step": 18855
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6770627940417284e-06,
+      "loss": 0.7842,
+      "step": 18856
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.675034657269814e-06,
+      "loss": 17.8757,
+      "step": 18857
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6730077267000765e-06,
+      "loss": 18.4512,
+      "step": 18858
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.670982002383098e-06,
+      "loss": 0.7685,
+      "step": 18859
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6689574843694433e-06,
+      "loss": 18.4945,
+      "step": 18860
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6669341727096444e-06,
+      "loss": 17.4866,
+      "step": 18861
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6649120674541996e-06,
+      "loss": 0.7688,
+      "step": 18862
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6628911686535909e-06,
+      "loss": 0.753,
+      "step": 18863
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6608714763582555e-06,
+      "loss": 0.7691,
+      "step": 18864
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6588529906186035e-06,
+      "loss": 17.5448,
+      "step": 18865
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6568357114850274e-06,
+      "loss": 17.6597,
+      "step": 18866
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6548196390078541e-06,
+      "loss": 18.2795,
+      "step": 18867
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.652804773237432e-06,
+      "loss": 0.8185,
+      "step": 18868
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6507911142240429e-06,
+      "loss": 16.8098,
+      "step": 18869
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6487786620179413e-06,
+      "loss": 16.6102,
+      "step": 18870
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6467674166693648e-06,
+      "loss": 16.3831,
+      "step": 18871
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6447573782285175e-06,
+      "loss": 16.8919,
+      "step": 18872
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6427485467455706e-06,
+      "loss": 0.7771,
+      "step": 18873
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.640740922270656e-06,
+      "loss": 17.3501,
+      "step": 18874
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6387345048538838e-06,
+      "loss": 16.6124,
+      "step": 18875
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6367292945453472e-06,
+      "loss": 0.7771,
+      "step": 18876
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.634725291395095e-06,
+      "loss": 18.0964,
+      "step": 18877
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6327224954531317e-06,
+      "loss": 0.7519,
+      "step": 18878
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6307209067694673e-06,
+      "loss": 0.7514,
+      "step": 18879
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6287205253940507e-06,
+      "loss": 0.7434,
+      "step": 18880
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6267213513768143e-06,
+      "loss": 17.7355,
+      "step": 18881
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6247233847676624e-06,
+      "loss": 0.7357,
+      "step": 18882
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6227266256164497e-06,
+      "loss": 18.3691,
+      "step": 18883
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6207310739730365e-06,
+      "loss": 16.8641,
+      "step": 18884
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6187367298872103e-06,
+      "loss": 0.7365,
+      "step": 18885
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.616743593408776e-06,
+      "loss": 17.6313,
+      "step": 18886
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6147516645874605e-06,
+      "loss": 0.7978,
+      "step": 18887
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6127609434729963e-06,
+      "loss": 0.6985,
+      "step": 18888
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6107714301150656e-06,
+      "loss": 0.7604,
+      "step": 18889
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6087831245633234e-06,
+      "loss": 17.1326,
+      "step": 18890
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6067960268674077e-06,
+      "loss": 0.8104,
+      "step": 18891
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6048101370769175e-06,
+      "loss": 0.7692,
+      "step": 18892
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6028254552414078e-06,
+      "loss": 15.7049,
+      "step": 18893
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6008419814104335e-06,
+      "loss": 0.7367,
+      "step": 18894
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5988597156334882e-06,
+      "loss": 0.7682,
+      "step": 18895
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5968786579600603e-06,
+      "loss": 0.736,
+      "step": 18896
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5948988084395877e-06,
+      "loss": 0.8137,
+      "step": 18897
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.592920167121492e-06,
+      "loss": 0.7979,
+      "step": 18898
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5909427340551676e-06,
+      "loss": 17.5086,
+      "step": 18899
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5889665092899685e-06,
+      "loss": 0.7232,
+      "step": 18900
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5869914928752117e-06,
+      "loss": 16.6309,
+      "step": 18901
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5850176848602071e-06,
+      "loss": 0.7528,
+      "step": 18902
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5830450852942102e-06,
+      "loss": 18.1797,
+      "step": 18903
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5810736942264647e-06,
+      "loss": 15.3068,
+      "step": 18904
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5791035117061758e-06,
+      "loss": 16.9914,
+      "step": 18905
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5771345377825152e-06,
+      "loss": 18.8517,
+      "step": 18906
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5751667725046382e-06,
+      "loss": 0.7616,
+      "step": 18907
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5732002159216551e-06,
+      "loss": 0.7862,
+      "step": 18908
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.571234868082655e-06,
+      "loss": 0.7736,
+      "step": 18909
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5692707290366816e-06,
+      "loss": 15.4395,
+      "step": 18910
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5673077988327733e-06,
+      "loss": 0.7095,
+      "step": 18911
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5653460775199247e-06,
+      "loss": 18.3715,
+      "step": 18912
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5633855651470851e-06,
+      "loss": 17.299,
+      "step": 18913
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.561426261763216e-06,
+      "loss": 18.2406,
+      "step": 18914
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5594681674171996e-06,
+      "loss": 0.7932,
+      "step": 18915
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5575112821579197e-06,
+      "loss": 16.2678,
+      "step": 18916
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5555556060342202e-06,
+      "loss": 15.9697,
+      "step": 18917
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5536011390949124e-06,
+      "loss": 18.4193,
+      "step": 18918
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5516478813887848e-06,
+      "loss": 0.8122,
+      "step": 18919
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5496958329645872e-06,
+      "loss": 19.4893,
+      "step": 18920
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5477449938710365e-06,
+      "loss": 0.788,
+      "step": 18921
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.545795364156849e-06,
+      "loss": 0.7032,
+      "step": 18922
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5438469438706583e-06,
+      "loss": 0.7122,
+      "step": 18923
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5418997330611196e-06,
+      "loss": 0.7593,
+      "step": 18924
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5399537317768275e-06,
+      "loss": 16.9946,
+      "step": 18925
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.538008940066349e-06,
+      "loss": 0.7376,
+      "step": 18926
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5360653579782392e-06,
+      "loss": 19.4701,
+      "step": 18927
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.534122985561004e-06,
+      "loss": 15.7714,
+      "step": 18928
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5321818228631213e-06,
+      "loss": 0.6992,
+      "step": 18929
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5302418699330522e-06,
+      "loss": 0.7588,
+      "step": 18930
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5283031268192028e-06,
+      "loss": 0.7937,
+      "step": 18931
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5263655935699839e-06,
+      "loss": 0.7549,
+      "step": 18932
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.524429270233746e-06,
+      "loss": 14.5162,
+      "step": 18933
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5224941568588169e-06,
+      "loss": 18.7366,
+      "step": 18934
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5205602534935138e-06,
+      "loss": 16.8841,
+      "step": 18935
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5186275601860866e-06,
+      "loss": 0.7549,
+      "step": 18936
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5166960769847916e-06,
+      "loss": 19.1477,
+      "step": 18937
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5147658039378233e-06,
+      "loss": 17.0435,
+      "step": 18938
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.512836741093382e-06,
+      "loss": 0.702,
+      "step": 18939
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5109088884996015e-06,
+      "loss": 0.7128,
+      "step": 18940
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5089822462046045e-06,
+      "loss": 17.7401,
+      "step": 18941
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.507056814256491e-06,
+      "loss": 15.2031,
+      "step": 18942
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.505132592703301e-06,
+      "loss": 0.8288,
+      "step": 18943
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5032095815930791e-06,
+      "loss": 16.989,
+      "step": 18944
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.50128778097382e-06,
+      "loss": 17.2387,
+      "step": 18945
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.499367190893486e-06,
+      "loss": 0.7371,
+      "step": 18946
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4974478114000267e-06,
+      "loss": 16.5142,
+      "step": 18947
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.495529642541349e-06,
+      "loss": 16.4377,
+      "step": 18948
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4936126843653198e-06,
+      "loss": 18.6202,
+      "step": 18949
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4916969369197954e-06,
+      "loss": 0.7097,
+      "step": 18950
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4897824002525874e-06,
+      "loss": 16.4659,
+      "step": 18951
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.487869074411491e-06,
+      "loss": 16.9613,
+      "step": 18952
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4859569594442623e-06,
+      "loss": 16.1719,
+      "step": 18953
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4840460553986135e-06,
+      "loss": 17.2065,
+      "step": 18954
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4821363623222617e-06,
+      "loss": 0.7511,
+      "step": 18955
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4802278802628688e-06,
+      "loss": 16.0052,
+      "step": 18956
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4783206092680635e-06,
+      "loss": 0.7332,
+      "step": 18957
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.476414549385452e-06,
+      "loss": 18.5253,
+      "step": 18958
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4745097006626075e-06,
+      "loss": 17.3797,
+      "step": 18959
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4726060631470862e-06,
+      "loss": 16.5467,
+      "step": 18960
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4707036368864003e-06,
+      "loss": 0.7564,
+      "step": 18961
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4688024219280284e-06,
+      "loss": 0.7555,
+      "step": 18962
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.466902418319438e-06,
+      "loss": 0.7746,
+      "step": 18963
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4650036261080301e-06,
+      "loss": 18.2838,
+      "step": 18964
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4631060453412226e-06,
+      "loss": 17.2645,
+      "step": 18965
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4612096760663663e-06,
+      "loss": 17.0184,
+      "step": 18966
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.459314518330801e-06,
+      "loss": 19.0439,
+      "step": 18967
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4574205721818335e-06,
+      "loss": 18.0859,
+      "step": 18968
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4555278376667203e-06,
+      "loss": 17.2039,
+      "step": 18969
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4536363148327348e-06,
+      "loss": 17.7824,
+      "step": 18970
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4517460037270557e-06,
+      "loss": 0.784,
+      "step": 18971
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4498569043968902e-06,
+      "loss": 16.9896,
+      "step": 18972
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4479690168893778e-06,
+      "loss": 0.7183,
+      "step": 18973
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4460823412516423e-06,
+      "loss": 18.2284,
+      "step": 18974
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.444196877530779e-06,
+      "loss": 16.646,
+      "step": 18975
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4423126257738507e-06,
+      "loss": 18.9037,
+      "step": 18976
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4404295860278915e-06,
+      "loss": 16.2018,
+      "step": 18977
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4385477583398922e-06,
+      "loss": 17.5994,
+      "step": 18978
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4366671427568256e-06,
+      "loss": 0.7104,
+      "step": 18979
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4347877393256381e-06,
+      "loss": 0.7277,
+      "step": 18980
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4329095480932365e-06,
+      "loss": 0.696,
+      "step": 18981
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4310325691064996e-06,
+      "loss": 17.5011,
+      "step": 18982
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4291568024122848e-06,
+      "loss": 18.1238,
+      "step": 18983
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4272822480574043e-06,
+      "loss": 16.6273,
+      "step": 18984
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4254089060886543e-06,
+      "loss": 0.7568,
+      "step": 18985
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4235367765527862e-06,
+      "loss": 0.7749,
+      "step": 18986
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4216658594965237e-06,
+      "loss": 0.7301,
+      "step": 18987
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4197961549665795e-06,
+      "loss": 0.7674,
+      "step": 18988
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.417927663009616e-06,
+      "loss": 16.1464,
+      "step": 18989
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4160603836722685e-06,
+      "loss": 0.7654,
+      "step": 18990
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4141943170011551e-06,
+      "loss": 16.887,
+      "step": 18991
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.412329463042833e-06,
+      "loss": 16.7352,
+      "step": 18992
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4104658218438704e-06,
+      "loss": 16.1242,
+      "step": 18993
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4086033934507692e-06,
+      "loss": 17.3588,
+      "step": 18994
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4067421779100253e-06,
+      "loss": 0.7538,
+      "step": 18995
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4048821752680962e-06,
+      "loss": 17.9794,
+      "step": 18996
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4030233855713948e-06,
+      "loss": 15.9648,
+      "step": 18997
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4011658088663337e-06,
+      "loss": 17.2542,
+      "step": 18998
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.399309445199265e-06,
+      "loss": 17.2641,
+      "step": 18999
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3974542946165347e-06,
+      "loss": 16.7016,
+      "step": 19000
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3956003571644393e-06,
+      "loss": 17.8655,
+      "step": 19001
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.393747632889253e-06,
+      "loss": 0.765,
+      "step": 19002
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3918961218372272e-06,
+      "loss": 0.739,
+      "step": 19003
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3900458240545754e-06,
+      "loss": 16.4467,
+      "step": 19004
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3881967395874828e-06,
+      "loss": 16.8084,
+      "step": 19005
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3863488684820903e-06,
+      "loss": 0.7917,
+      "step": 19006
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.384502210784533e-06,
+      "loss": 0.728,
+      "step": 19007
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3826567665409019e-06,
+      "loss": 0.798,
+      "step": 19008
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.38081253579726e-06,
+      "loss": 0.7811,
+      "step": 19009
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.378969518599632e-06,
+      "loss": 0.7754,
+      "step": 19010
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.377127714994031e-06,
+      "loss": 16.4089,
+      "step": 19011
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3752871250264255e-06,
+      "loss": 15.5047,
+      "step": 19012
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3734477487427567e-06,
+      "loss": 0.7117,
+      "step": 19013
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3716095861889321e-06,
+      "loss": 0.7303,
+      "step": 19014
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.369772637410832e-06,
+      "loss": 0.7724,
+      "step": 19015
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3679369024543198e-06,
+      "loss": 18.0783,
+      "step": 19016
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3661023813651974e-06,
+      "loss": 0.7216,
+      "step": 19017
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3642690741892727e-06,
+      "loss": 0.7787,
+      "step": 19018
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3624369809722926e-06,
+      "loss": 0.6942,
+      "step": 19019
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.360606101759987e-06,
+      "loss": 0.7496,
+      "step": 19020
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3587764365980638e-06,
+      "loss": 18.3423,
+      "step": 19021
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.356947985532181e-06,
+      "loss": 0.7692,
+      "step": 19022
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.355120748607991e-06,
+      "loss": 17.5436,
+      "step": 19023
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3532947258710904e-06,
+      "loss": 0.7712,
+      "step": 19024
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3514699173670653e-06,
+      "loss": 0.7949,
+      "step": 19025
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3496463231414568e-06,
+      "loss": 15.6873,
+      "step": 19026
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3478239432397787e-06,
+      "loss": 0.7321,
+      "step": 19027
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.346002777707528e-06,
+      "loss": 0.7415,
+      "step": 19028
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3441828265901568e-06,
+      "loss": 17.1046,
+      "step": 19029
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3423640899330904e-06,
+      "loss": 18.6744,
+      "step": 19030
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.340546567781731e-06,
+      "loss": 0.7433,
+      "step": 19031
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3387302601814366e-06,
+      "loss": 0.7432,
+      "step": 19032
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3369151671775493e-06,
+      "loss": 16.5837,
+      "step": 19033
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3351012888153657e-06,
+      "loss": 0.7602,
+      "step": 19034
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.333288625140161e-06,
+      "loss": 0.8632,
+      "step": 19035
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3314771761971934e-06,
+      "loss": 17.0245,
+      "step": 19036
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3296669420316654e-06,
+      "loss": 0.7105,
+      "step": 19037
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3278579226887577e-06,
+      "loss": 0.7476,
+      "step": 19038
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3260501182136454e-06,
+      "loss": 18.3131,
+      "step": 19039
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3242435286514199e-06,
+      "loss": 16.985,
+      "step": 19040
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3224381540471954e-06,
+      "loss": 18.8632,
+      "step": 19041
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.32063399444603e-06,
+      "loss": 14.9868,
+      "step": 19042
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.318831049892949e-06,
+      "loss": 0.817,
+      "step": 19043
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3170293204329665e-06,
+      "loss": 20.9438,
+      "step": 19044
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3152288061110518e-06,
+      "loss": 17.0095,
+      "step": 19045
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3134295069721358e-06,
+      "loss": 0.7562,
+      "step": 19046
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3116314230611382e-06,
+      "loss": 0.7571,
+      "step": 19047
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.309834554422934e-06,
+      "loss": 0.7611,
+      "step": 19048
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.308038901102382e-06,
+      "loss": 0.774,
+      "step": 19049
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3062444631442905e-06,
+      "loss": 16.6297,
+      "step": 19050
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3044512405934572e-06,
+      "loss": 16.8538,
+      "step": 19051
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3026592334946408e-06,
+      "loss": 0.7472,
+      "step": 19052
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.300868441892572e-06,
+      "loss": 16.2271,
+      "step": 19053
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2990788658319486e-06,
+      "loss": 16.1667,
+      "step": 19054
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2972905053574236e-06,
+      "loss": 0.7301,
+      "step": 19055
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2955033605136614e-06,
+      "loss": 16.3964,
+      "step": 19056
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2937174313452483e-06,
+      "loss": 0.747,
+      "step": 19057
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.291932717896771e-06,
+      "loss": 0.7077,
+      "step": 19058
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2901492202127774e-06,
+      "loss": 0.7724,
+      "step": 19059
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2883669383377761e-06,
+      "loss": 0.7194,
+      "step": 19060
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2865858723162648e-06,
+      "loss": 0.7413,
+      "step": 19061
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2848060221926917e-06,
+      "loss": 18.121,
+      "step": 19062
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2830273880114763e-06,
+      "loss": 17.1607,
+      "step": 19063
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2812499698170277e-06,
+      "loss": 16.9009,
+      "step": 19064
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2794737676536994e-06,
+      "loss": 17.0665,
+      "step": 19065
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2776987815658282e-06,
+      "loss": 0.7496,
+      "step": 19066
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.275925011597734e-06,
+      "loss": 0.7794,
+      "step": 19067
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2741524577936593e-06,
+      "loss": 16.1264,
+      "step": 19068
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.272381120197874e-06,
+      "loss": 0.7491,
+      "step": 19069
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2706109988545822e-06,
+      "loss": 0.7413,
+      "step": 19070
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2688420938079593e-06,
+      "loss": 0.7852,
+      "step": 19071
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2670744051021699e-06,
+      "loss": 18.7041,
+      "step": 19072
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2653079327813288e-06,
+      "loss": 0.7706,
+      "step": 19073
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2635426768895342e-06,
+      "loss": 17.5126,
+      "step": 19074
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2617786374708451e-06,
+      "loss": 0.7635,
+      "step": 19075
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2600158145692764e-06,
+      "loss": 17.0498,
+      "step": 19076
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2582542082288484e-06,
+      "loss": 18.7411,
+      "step": 19077
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2564938184935203e-06,
+      "loss": 18.7555,
+      "step": 19078
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2547346454072463e-06,
+      "loss": 0.733,
+      "step": 19079
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2529766890139184e-06,
+      "loss": 16.2188,
+      "step": 19080
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.251219949357424e-06,
+      "loss": 0.798,
+      "step": 19081
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2494644264816113e-06,
+      "loss": 17.124,
+      "step": 19082
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2477101204302955e-06,
+      "loss": 0.8046,
+      "step": 19083
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2459570312472691e-06,
+      "loss": 0.7648,
+      "step": 19084
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2442051589762859e-06,
+      "loss": 17.0562,
+      "step": 19085
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2424545036610724e-06,
+      "loss": 15.906,
+      "step": 19086
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2407050653453323e-06,
+      "loss": 16.6533,
+      "step": 19087
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.238956844072725e-06,
+      "loss": 0.7776,
+      "step": 19088
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2372098398868882e-06,
+      "loss": 0.7834,
+      "step": 19089
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.235464052831431e-06,
+      "loss": 13.9702,
+      "step": 19090
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2337194829499187e-06,
+      "loss": 17.2805,
+      "step": 19091
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2319761302859056e-06,
+      "loss": 17.2064,
+      "step": 19092
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2302339948829011e-06,
+      "loss": 0.7701,
+      "step": 19093
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2284930767843983e-06,
+      "loss": 15.6873,
+      "step": 19094
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2267533760338401e-06,
+      "loss": 0.7626,
+      "step": 19095
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.225014892674653e-06,
+      "loss": 16.1856,
+      "step": 19096
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.22327762675023e-06,
+      "loss": 0.7318,
+      "step": 19097
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2215415783039418e-06,
+      "loss": 17.6846,
+      "step": 19098
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2198067473791042e-06,
+      "loss": 18.8874,
+      "step": 19099
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2180731340190322e-06,
+      "loss": 18.5203,
+      "step": 19100
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2163407382669967e-06,
+      "loss": 15.9007,
+      "step": 19101
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.21460956016623e-06,
+      "loss": 0.7568,
+      "step": 19102
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.212879599759953e-06,
+      "loss": 0.7445,
+      "step": 19103
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.211150857091331e-06,
+      "loss": 17.2725,
+      "step": 19104
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2094233322035297e-06,
+      "loss": 17.0795,
+      "step": 19105
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.207697025139659e-06,
+      "loss": 16.8136,
+      "step": 19106
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2059719359428179e-06,
+      "loss": 16.9634,
+      "step": 19107
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2042480646560606e-06,
+      "loss": 0.8004,
+      "step": 19108
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2025254113224084e-06,
+      "loss": 0.7465,
+      "step": 19109
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.20080397598486e-06,
+      "loss": 0.727,
+      "step": 19110
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1990837586863868e-06,
+      "loss": 17.3882,
+      "step": 19111
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1973647594699321e-06,
+      "loss": 0.7859,
+      "step": 19112
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1956469783783952e-06,
+      "loss": 0.6916,
+      "step": 19113
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.193930415454647e-06,
+      "loss": 0.7968,
+      "step": 19114
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.192215070741548e-06,
+      "loss": 0.7373,
+      "step": 19115
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1905009442818971e-06,
+      "loss": 18.676,
+      "step": 19116
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1887880361184878e-06,
+      "loss": 0.7749,
+      "step": 19117
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1870763462940803e-06,
+      "loss": 14.7632,
+      "step": 19118
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1853658748513852e-06,
+      "loss": 16.3844,
+      "step": 19119
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.183656621833107e-06,
+      "loss": 0.7288,
+      "step": 19120
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1819485872819114e-06,
+      "loss": 0.7859,
+      "step": 19121
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.18024177124042e-06,
+      "loss": 16.435,
+      "step": 19122
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1785361737512435e-06,
+      "loss": 17.539,
+      "step": 19123
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1768317948569473e-06,
+      "loss": 0.7486,
+      "step": 19124
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1751286346000812e-06,
+      "loss": 18.2774,
+      "step": 19125
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1734266930231496e-06,
+      "loss": 0.7219,
+      "step": 19126
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1717259701686357e-06,
+      "loss": 17.554,
+      "step": 19127
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.170026466078994e-06,
+      "loss": 17.7519,
+      "step": 19128
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1683281807966407e-06,
+      "loss": 17.5341,
+      "step": 19129
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.166631114363964e-06,
+      "loss": 16.8713,
+      "step": 19130
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1649352668233305e-06,
+      "loss": 0.7964,
+      "step": 19131
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1632406382170503e-06,
+      "loss": 18.6406,
+      "step": 19132
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1615472285874451e-06,
+      "loss": 0.7684,
+      "step": 19133
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1598550379767647e-06,
+      "loss": 17.9983,
+      "step": 19134
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1581640664272642e-06,
+      "loss": 0.7516,
+      "step": 19135
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1564743139811373e-06,
+      "loss": 0.7502,
+      "step": 19136
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1547857806805618e-06,
+      "loss": 19.7107,
+      "step": 19137
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.153098466567687e-06,
+      "loss": 0.8008,
+      "step": 19138
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1514123716846237e-06,
+      "loss": 0.7389,
+      "step": 19139
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1497274960734661e-06,
+      "loss": 0.7548,
+      "step": 19140
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.148043839776264e-06,
+      "loss": 0.7331,
+      "step": 19141
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1463614028350389e-06,
+      "loss": 0.7916,
+      "step": 19142
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.144680185291791e-06,
+      "loss": 19.698,
+      "step": 19143
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.143000187188481e-06,
+      "loss": 18.3948,
+      "step": 19144
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1413214085670421e-06,
+      "loss": 0.7585,
+      "step": 19145
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.139643849469374e-06,
+      "loss": 0.7061,
+      "step": 19146
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1379675099373488e-06,
+      "loss": 18.3923,
+      "step": 19147
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1362923900128164e-06,
+      "loss": 15.3694,
+      "step": 19148
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1346184897375823e-06,
+      "loss": 0.7324,
+      "step": 19149
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1329458091534295e-06,
+      "loss": 0.77,
+      "step": 19150
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1312743483021083e-06,
+      "loss": 16.677,
+      "step": 19151
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1296041072253293e-06,
+      "loss": 0.7858,
+      "step": 19152
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1279350859647931e-06,
+      "loss": 0.733,
+      "step": 19153
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1262672845621603e-06,
+      "loss": 16.7223,
+      "step": 19154
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1246007030590533e-06,
+      "loss": 15.6894,
+      "step": 19155
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1229353414970722e-06,
+      "loss": 16.3752,
+      "step": 19156
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1212711999177894e-06,
+      "loss": 0.7847,
+      "step": 19157
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1196082783627325e-06,
+      "loss": 0.6935,
+      "step": 19158
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1179465768734188e-06,
+      "loss": 0.7318,
+      "step": 19159
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.116286095491309e-06,
+      "loss": 0.7966,
+      "step": 19160
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1146268342578703e-06,
+      "loss": 0.7132,
+      "step": 19161
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1129687932145028e-06,
+      "loss": 17.9364,
+      "step": 19162
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1113119724025955e-06,
+      "loss": 0.7118,
+      "step": 19163
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1096563718635155e-06,
+      "loss": 0.7666,
+      "step": 19164
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.108001991638563e-06,
+      "loss": 19.0402,
+      "step": 19165
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1063488317690495e-06,
+      "loss": 0.7729,
+      "step": 19166
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1046968922962309e-06,
+      "loss": 17.4068,
+      "step": 19167
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1030461732613405e-06,
+      "loss": 17.5046,
+      "step": 19168
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.1013966747055848e-06,
+      "loss": 0.8224,
+      "step": 19169
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.099748396670136e-06,
+      "loss": 17.6005,
+      "step": 19170
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0981013391961337e-06,
+      "loss": 0.7228,
+      "step": 19171
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.096455502324678e-06,
+      "loss": 0.7622,
+      "step": 19172
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0948108860968641e-06,
+      "loss": 17.4756,
+      "step": 19173
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0931674905537425e-06,
+      "loss": 0.7682,
+      "step": 19174
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0915253157363193e-06,
+      "loss": 0.7643,
+      "step": 19175
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0898843616855947e-06,
+      "loss": 0.7473,
+      "step": 19176
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.088244628442525e-06,
+      "loss": 0.739,
+      "step": 19177
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.086606116048039e-06,
+      "loss": 0.7202,
+      "step": 19178
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.084968824543031e-06,
+      "loss": 16.3176,
+      "step": 19179
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0833327539683636e-06,
+      "loss": 17.287,
+      "step": 19180
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0816979043648867e-06,
+      "loss": 19.3835,
+      "step": 19181
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0800642757734014e-06,
+      "loss": 0.7544,
+      "step": 19182
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.078431868234675e-06,
+      "loss": 18.0222,
+      "step": 19183
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0768006817894693e-06,
+      "loss": 0.7648,
+      "step": 19184
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.075170716478474e-06,
+      "loss": 0.7345,
+      "step": 19185
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0735419723424012e-06,
+      "loss": 0.7533,
+      "step": 19186
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0719144494218902e-06,
+      "loss": 17.1637,
+      "step": 19187
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0702881477575588e-06,
+      "loss": 0.7214,
+      "step": 19188
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0686630673900134e-06,
+      "loss": 0.7665,
+      "step": 19189
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0670392083598101e-06,
+      "loss": 18.7909,
+      "step": 19190
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0654165707074836e-06,
+      "loss": 0.7807,
+      "step": 19191
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0637951544735292e-06,
+      "loss": 17.5665,
+      "step": 19192
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0621749596984198e-06,
+      "loss": 17.5491,
+      "step": 19193
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0605559864226011e-06,
+      "loss": 18.2809,
+      "step": 19194
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.058938234686474e-06,
+      "loss": 0.7898,
+      "step": 19195
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0573217045304285e-06,
+      "loss": 18.2916,
+      "step": 19196
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0557063959948098e-06,
+      "loss": 16.8229,
+      "step": 19197
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0540923091199307e-06,
+      "loss": 0.7042,
+      "step": 19198
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0524794439460862e-06,
+      "loss": 0.7443,
+      "step": 19199
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0508678005135275e-06,
+      "loss": 17.6455,
+      "step": 19200
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0492573788624837e-06,
+      "loss": 18.5151,
+      "step": 19201
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0476481790331617e-06,
+      "loss": 17.4361,
+      "step": 19202
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0460402010657067e-06,
+      "loss": 17.6712,
+      "step": 19203
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0444334450002702e-06,
+      "loss": 0.735,
+      "step": 19204
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.042827910876959e-06,
+      "loss": 14.9635,
+      "step": 19205
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0412235987358354e-06,
+      "loss": 0.7562,
+      "step": 19206
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0396205086169564e-06,
+      "loss": 0.8,
+      "step": 19207
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0380186405603177e-06,
+      "loss": 0.7538,
+      "step": 19208
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0364179946059206e-06,
+      "loss": 0.7934,
+      "step": 19209
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0348185707937108e-06,
+      "loss": 17.7637,
+      "step": 19210
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.033220369163601e-06,
+      "loss": 0.7429,
+      "step": 19211
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0316233897555039e-06,
+      "loss": 15.906,
+      "step": 19212
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0300276326092595e-06,
+      "loss": 16.0739,
+      "step": 19213
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0284330977647084e-06,
+      "loss": 14.4675,
+      "step": 19214
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0268397852616517e-06,
+      "loss": 18.0736,
+      "step": 19215
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.025247695139847e-06,
+      "loss": 0.748,
+      "step": 19216
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0236568274390513e-06,
+      "loss": 16.9111,
+      "step": 19217
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0220671821989603e-06,
+      "loss": 16.8952,
+      "step": 19218
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.020478759459259e-06,
+      "loss": 17.991,
+      "step": 19219
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0188915592595882e-06,
+      "loss": 17.6276,
+      "step": 19220
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.017305581639566e-06,
+      "loss": 16.952,
+      "step": 19221
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0157208266387884e-06,
+      "loss": 16.6141,
+      "step": 19222
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0141372942967907e-06,
+      "loss": 17.3783,
+      "step": 19223
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0125549846531247e-06,
+      "loss": 0.7579,
+      "step": 19224
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.010973897747264e-06,
+      "loss": 0.6911,
+      "step": 19225
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0093940336186825e-06,
+      "loss": 17.9356,
+      "step": 19226
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0078153923068157e-06,
+      "loss": 19.2332,
+      "step": 19227
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.006237973851054e-06,
+      "loss": 16.5205,
+      "step": 19228
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.004661778290783e-06,
+      "loss": 0.7781,
+      "step": 19229
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.003086805665343e-06,
+      "loss": 18.5879,
+      "step": 19230
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0015130560140417e-06,
+      "loss": 16.6144,
+      "step": 19231
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.999405293761643e-07,
+      "loss": 15.8912,
+      "step": 19232
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.983692257909627e-07,
+      "loss": 0.8093,
+      "step": 19233
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.9679914529765e-07,
+      "loss": 18.2013,
+      "step": 19234
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.95230287935417e-07,
+      "loss": 17.181,
+      "step": 19235
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.93662653743427e-07,
+      "loss": 0.8165,
+      "step": 19236
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.920962427608094e-07,
+      "loss": 16.4465,
+      "step": 19237
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.905310550266555e-07,
+      "loss": 0.7433,
+      "step": 19238
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.889670905800397e-07,
+      "loss": 18.3411,
+      "step": 19239
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.874043494599972e-07,
+      "loss": 17.8041,
+      "step": 19240
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.858428317055302e-07,
+      "loss": 16.3442,
+      "step": 19241
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.842825373556186e-07,
+      "loss": 17.5472,
+      "step": 19242
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.82723466449209e-07,
+      "loss": 0.7255,
+      "step": 19243
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.811656190252038e-07,
+      "loss": 15.6487,
+      "step": 19244
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.796089951225051e-07,
+      "loss": 16.0723,
+      "step": 19245
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.780535947799596e-07,
+      "loss": 15.8034,
+      "step": 19246
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.764994180363918e-07,
+      "loss": 0.7649,
+      "step": 19247
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.749464649305872e-07,
+      "loss": 0.7198,
+      "step": 19248
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.733947355013096e-07,
+      "loss": 0.732,
+      "step": 19249
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.718442297872997e-07,
+      "loss": 17.7313,
+      "step": 19250
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.702949478272495e-07,
+      "loss": 0.8011,
+      "step": 19251
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.687468896598328e-07,
+      "loss": 18.2121,
+      "step": 19252
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.672000553236914e-07,
+      "loss": 17.1255,
+      "step": 19253
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.656544448574333e-07,
+      "loss": 17.6902,
+      "step": 19254
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.641100582996331e-07,
+      "loss": 0.7321,
+      "step": 19255
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.625668956888379e-07,
+      "loss": 18.9563,
+      "step": 19256
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.610249570635777e-07,
+      "loss": 17.8912,
+      "step": 19257
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.594842424623274e-07,
+      "loss": 17.8215,
+      "step": 19258
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.579447519235452e-07,
+      "loss": 17.6931,
+      "step": 19259
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.564064854856669e-07,
+      "loss": 19.0038,
+      "step": 19260
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.548694431870786e-07,
+      "loss": 20.3391,
+      "step": 19261
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.53333625066144e-07,
+      "loss": 0.6898,
+      "step": 19262
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.517990311612046e-07,
+      "loss": 17.6452,
+      "step": 19263
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.502656615105577e-07,
+      "loss": 0.7414,
+      "step": 19264
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.487335161524779e-07,
+      "loss": 17.0973,
+      "step": 19265
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.472025951252129e-07,
+      "loss": 15.5536,
+      "step": 19266
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.456728984669761e-07,
+      "loss": 0.7301,
+      "step": 19267
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.441444262159372e-07,
+      "loss": 17.9509,
+      "step": 19268
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.426171784102544e-07,
+      "loss": 0.7642,
+      "step": 19269
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.410911550880475e-07,
+      "loss": 14.9218,
+      "step": 19270
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.395663562874135e-07,
+      "loss": 0.761,
+      "step": 19271
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.380427820463944e-07,
+      "loss": 16.5427,
+      "step": 19272
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.365204324030374e-07,
+      "loss": 0.7565,
+      "step": 19273
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.349993073953345e-07,
+      "loss": 16.9322,
+      "step": 19274
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.334794070612496e-07,
+      "loss": 18.7744,
+      "step": 19275
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.319607314387246e-07,
+      "loss": 17.4919,
+      "step": 19276
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.304432805656571e-07,
+      "loss": 0.7672,
+      "step": 19277
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.289270544799333e-07,
+      "loss": 16.6221,
+      "step": 19278
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.274120532193897e-07,
+      "loss": 19.3391,
+      "step": 19279
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.258982768218516e-07,
+      "loss": 0.7813,
+      "step": 19280
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.243857253250943e-07,
+      "loss": 17.2754,
+      "step": 19281
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.228743987668764e-07,
+      "loss": 0.7114,
+      "step": 19282
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.213642971849234e-07,
+      "loss": 0.7652,
+      "step": 19283
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.198554206169164e-07,
+      "loss": 0.7416,
+      "step": 19284
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.183477691005304e-07,
+      "loss": 17.8233,
+      "step": 19285
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.168413426733857e-07,
+      "loss": 0.7649,
+      "step": 19286
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.15336141373091e-07,
+      "loss": 0.8161,
+      "step": 19287
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.138321652372162e-07,
+      "loss": 0.7399,
+      "step": 19288
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.123294143032923e-07,
+      "loss": 17.0816,
+      "step": 19289
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.108278886088395e-07,
+      "loss": 0.7748,
+      "step": 19290
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.093275881913332e-07,
+      "loss": 17.5547,
+      "step": 19291
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.078285130882102e-07,
+      "loss": 0.8284,
+      "step": 19292
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.063306633369017e-07,
+      "loss": 19.187,
+      "step": 19293
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.048340389747889e-07,
+      "loss": 0.7847,
+      "step": 19294
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.033386400392307e-07,
+      "loss": 19.1877,
+      "step": 19295
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.01844466567553e-07,
+      "loss": 0.829,
+      "step": 19296
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.003515185970368e-07,
+      "loss": 18.0954,
+      "step": 19297
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.988597961649691e-07,
+      "loss": 0.8093,
+      "step": 19298
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.973692993085648e-07,
+      "loss": 0.7164,
+      "step": 19299
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.958800280650381e-07,
+      "loss": 17.1689,
+      "step": 19300
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.943919824715541e-07,
+      "loss": 0.7348,
+      "step": 19301
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.929051625652662e-07,
+      "loss": 17.9362,
+      "step": 19302
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.914195683832727e-07,
+      "loss": 15.4142,
+      "step": 19303
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.899351999626549e-07,
+      "loss": 0.738,
+      "step": 19304
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.88452057340472e-07,
+      "loss": 17.2327,
+      "step": 19305
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.869701405537445e-07,
+      "loss": 17.3877,
+      "step": 19306
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.854894496394483e-07,
+      "loss": 0.7502,
+      "step": 19307
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.84009984634554e-07,
+      "loss": 0.7916,
+      "step": 19308
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.825317455759874e-07,
+      "loss": 0.7459,
+      "step": 19309
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.810547325006413e-07,
+      "loss": 17.5117,
+      "step": 19310
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.795789454453862e-07,
+      "loss": 15.5854,
+      "step": 19311
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.781043844470482e-07,
+      "loss": 0.7954,
+      "step": 19312
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.766310495424479e-07,
+      "loss": 15.3554,
+      "step": 19313
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.751589407683503e-07,
+      "loss": 19.4541,
+      "step": 19314
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.736880581615036e-07,
+      "loss": 19.6883,
+      "step": 19315
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.722184017586233e-07,
+      "loss": 0.7599,
+      "step": 19316
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.707499715963796e-07,
+      "loss": 17.304,
+      "step": 19317
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.692827677114435e-07,
+      "loss": 17.3504,
+      "step": 19318
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.678167901404244e-07,
+      "loss": 0.7529,
+      "step": 19319
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.663520389199098e-07,
+      "loss": 17.5049,
+      "step": 19320
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.648885140864759e-07,
+      "loss": 16.7279,
+      "step": 19321
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.634262156766437e-07,
+      "loss": 16.5774,
+      "step": 19322
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.619651437269117e-07,
+      "loss": 16.2081,
+      "step": 19323
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.60505298273745e-07,
+      "loss": 0.756,
+      "step": 19324
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.590466793535868e-07,
+      "loss": 0.7195,
+      "step": 19325
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.575892870028468e-07,
+      "loss": 15.7427,
+      "step": 19326
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.561331212579015e-07,
+      "loss": 18.3135,
+      "step": 19327
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.546781821550832e-07,
+      "loss": 16.3621,
+      "step": 19328
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.532244697307346e-07,
+      "loss": 15.2002,
+      "step": 19329
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.517719840211102e-07,
+      "loss": 0.7604,
+      "step": 19330
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.503207250624867e-07,
+      "loss": 19.5223,
+      "step": 19331
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.488706928910794e-07,
+      "loss": 17.9451,
+      "step": 19332
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.474218875430818e-07,
+      "loss": 20.0514,
+      "step": 19333
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.45974309054659e-07,
+      "loss": 15.8522,
+      "step": 19334
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.445279574619325e-07,
+      "loss": 0.713,
+      "step": 19335
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.430828328010232e-07,
+      "loss": 0.7757,
+      "step": 19336
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.416389351079856e-07,
+      "loss": 0.7576,
+      "step": 19337
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.401962644188632e-07,
+      "loss": 18.4926,
+      "step": 19338
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.38754820769666e-07,
+      "loss": 16.2576,
+      "step": 19339
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.373146041963764e-07,
+      "loss": 18.5036,
+      "step": 19340
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.358756147349378e-07,
+      "loss": 0.7975,
+      "step": 19341
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.344378524212715e-07,
+      "loss": 15.7101,
+      "step": 19342
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.330013172912598e-07,
+      "loss": 18.612,
+      "step": 19343
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.315660093807631e-07,
+      "loss": 17.3787,
+      "step": 19344
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.301319287256026e-07,
+      "loss": 18.1621,
+      "step": 19345
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.286990753615775e-07,
+      "loss": 0.7637,
+      "step": 19346
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.272674493244536e-07,
+      "loss": 18.0145,
+      "step": 19347
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.25837050649958e-07,
+      "loss": 0.8142,
+      "step": 19348
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.244078793738008e-07,
+      "loss": 0.8086,
+      "step": 19349
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.229799355316481e-07,
+      "loss": 17.4023,
+      "step": 19350
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.21553219159149e-07,
+      "loss": 0.7353,
+      "step": 19351
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.201277302919086e-07,
+      "loss": 0.7616,
+      "step": 19352
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.187034689655037e-07,
+      "loss": 16.7024,
+      "step": 19353
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.17280435215495e-07,
+      "loss": 0.8152,
+      "step": 19354
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.158586290773984e-07,
+      "loss": 16.8638,
+      "step": 19355
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.144380505866967e-07,
+      "loss": 0.7128,
+      "step": 19356
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.130186997788559e-07,
+      "loss": 16.8806,
+      "step": 19357
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.116005766892976e-07,
+      "loss": 18.8894,
+      "step": 19358
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.101836813534158e-07,
+      "loss": 15.3426,
+      "step": 19359
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.087680138065823e-07,
+      "loss": 15.2256,
+      "step": 19360
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.073535740841354e-07,
+      "loss": 19.0056,
+      "step": 19361
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.059403622213746e-07,
+      "loss": 0.7162,
+      "step": 19362
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.045283782535662e-07,
+      "loss": 17.8665,
+      "step": 19363
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.031176222159764e-07,
+      "loss": 0.7965,
+      "step": 19364
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.017080941437938e-07,
+      "loss": 15.7755,
+      "step": 19365
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.002997940722123e-07,
+      "loss": 16.7765,
+      "step": 19366
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 7.988927220363818e-07,
+      "loss": 17.285,
+      "step": 19367
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 7.97486878071424e-07,
+      "loss": 17.1294,
+      "step": 19368
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 7.960822622124275e-07,
+      "loss": 0.8026,
+      "step": 19369
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 7.946788744944534e-07,
+      "loss": 16.6614,
+      "step": 19370
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 7.932767149525289e-07,
+      "loss": 17.0392,
+      "step": 19371
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 7.918757836216595e-07,
+      "loss": 17.505,
+      "step": 19372
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.90476080536795e-07,
+      "loss": 0.8269,
+      "step": 19373
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.890776057328908e-07,
+      "loss": 0.7308,
+      "step": 19374
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.876803592448412e-07,
+      "loss": 0.7329,
+      "step": 19375
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.862843411075294e-07,
+      "loss": 0.766,
+      "step": 19376
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.848895513557941e-07,
+      "loss": 16.7894,
+      "step": 19377
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.834959900244576e-07,
+      "loss": 17.4595,
+      "step": 19378
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.821036571482976e-07,
+      "loss": 15.5043,
+      "step": 19379
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.807125527620695e-07,
+      "loss": 0.809,
+      "step": 19380
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.793226769004902e-07,
+      "loss": 0.7514,
+      "step": 19381
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.779340295982596e-07,
+      "loss": 18.5765,
+      "step": 19382
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.765466108900332e-07,
+      "loss": 0.7352,
+      "step": 19383
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.751604208104335e-07,
+      "loss": 15.9382,
+      "step": 19384
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.737754593940827e-07,
+      "loss": 17.8517,
+      "step": 19385
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.72391726675531e-07,
+      "loss": 17.2082,
+      "step": 19386
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.710092226893228e-07,
+      "loss": 16.7897,
+      "step": 19387
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.696279474699586e-07,
+      "loss": 18.099,
+      "step": 19388
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.682479010519273e-07,
+      "loss": 0.7393,
+      "step": 19389
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.668690834696734e-07,
+      "loss": 17.9936,
+      "step": 19390
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.654914947575975e-07,
+      "loss": 17.6979,
+      "step": 19391
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.641151349501108e-07,
+      "loss": 15.8879,
+      "step": 19392
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.627400040815413e-07,
+      "loss": 0.7563,
+      "step": 19393
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.613661021862228e-07,
+      "loss": 17.4608,
+      "step": 19394
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.599934292984557e-07,
+      "loss": 17.1557,
+      "step": 19395
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.5862198545249e-07,
+      "loss": 15.6418,
+      "step": 19396
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.572517706825655e-07,
+      "loss": 0.7722,
+      "step": 19397
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.558827850228822e-07,
+      "loss": 0.745,
+      "step": 19398
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.545150285076075e-07,
+      "loss": 18.4364,
+      "step": 19399
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.531485011708805e-07,
+      "loss": 15.7912,
+      "step": 19400
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.517832030468075e-07,
+      "loss": 0.7777,
+      "step": 19401
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.504191341694777e-07,
+      "loss": 0.7627,
+      "step": 19402
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.490562945729252e-07,
+      "loss": 17.3762,
+      "step": 19403
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.476946842911725e-07,
+      "loss": 0.763,
+      "step": 19404
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.463343033582149e-07,
+      "loss": 0.8113,
+      "step": 19405
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.449751518079917e-07,
+      "loss": 20.5127,
+      "step": 19406
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.436172296744426e-07,
+      "loss": 16.2421,
+      "step": 19407
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.422605369914459e-07,
+      "loss": 16.7712,
+      "step": 19408
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.409050737928747e-07,
+      "loss": 0.7347,
+      "step": 19409
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.395508401125628e-07,
+      "loss": 17.0244,
+      "step": 19410
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.381978359843111e-07,
+      "loss": 17.202,
+      "step": 19411
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.36846061441887e-07,
+      "loss": 0.7384,
+      "step": 19412
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.354955165190358e-07,
+      "loss": 17.3345,
+      "step": 19413
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.341462012494582e-07,
+      "loss": 17.7141,
+      "step": 19414
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.32798115666844e-07,
+      "loss": 17.2991,
+      "step": 19415
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.314512598048384e-07,
+      "loss": 0.7563,
+      "step": 19416
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.301056336970591e-07,
+      "loss": 0.7333,
+      "step": 19417
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.287612373770958e-07,
+      "loss": 19.6864,
+      "step": 19418
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.274180708784994e-07,
+      "loss": 18.4211,
+      "step": 19419
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.260761342347989e-07,
+      "loss": 0.7197,
+      "step": 19420
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.247354274794838e-07,
+      "loss": 0.757,
+      "step": 19421
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.233959506460275e-07,
+      "loss": 18.1084,
+      "step": 19422
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.220577037678589e-07,
+      "loss": 16.3701,
+      "step": 19423
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.207206868783845e-07,
+      "loss": 16.798,
+      "step": 19424
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.193849000109721e-07,
+      "loss": 18.9109,
+      "step": 19425
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.180503431989616e-07,
+      "loss": 0.7426,
+      "step": 19426
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.167170164756709e-07,
+      "loss": 0.7854,
+      "step": 19427
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.15384919874379e-07,
+      "loss": 0.7224,
+      "step": 19428
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.140540534283258e-07,
+      "loss": 16.9436,
+      "step": 19429
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.127244171707404e-07,
+      "loss": 15.3603,
+      "step": 19430
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.113960111348072e-07,
+      "loss": 0.7527,
+      "step": 19431
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.100688353536778e-07,
+      "loss": 0.7437,
+      "step": 19432
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.087428898604975e-07,
+      "loss": 17.0225,
+      "step": 19433
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.074181746883401e-07,
+      "loss": 19.4615,
+      "step": 19434
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.060946898702791e-07,
+      "loss": 0.7348,
+      "step": 19435
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.047724354393548e-07,
+      "loss": 18.7906,
+      "step": 19436
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.034514114285628e-07,
+      "loss": 0.8158,
+      "step": 19437
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.021316178708826e-07,
+      "loss": 16.838,
+      "step": 19438
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.008130547992486e-07,
+      "loss": 0.7344,
+      "step": 19439
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.994957222465848e-07,
+      "loss": 0.7618,
+      "step": 19440
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.981796202457591e-07,
+      "loss": 16.5063,
+      "step": 19441
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.968647488296231e-07,
+      "loss": 16.6989,
+      "step": 19442
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.955511080310062e-07,
+      "loss": 15.2217,
+      "step": 19443
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.942386978826876e-07,
+      "loss": 0.7914,
+      "step": 19444
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.929275184174356e-07,
+      "loss": 0.7468,
+      "step": 19445
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.916175696679627e-07,
+      "loss": 0.7945,
+      "step": 19446
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.90308851666982e-07,
+      "loss": 18.2223,
+      "step": 19447
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.89001364447145e-07,
+      "loss": 15.613,
+      "step": 19448
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.876951080410977e-07,
+      "loss": 18.9238,
+      "step": 19449
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.863900824814362e-07,
+      "loss": 0.7681,
+      "step": 19450
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.850862878007402e-07,
+      "loss": 16.0804,
+      "step": 19451
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.837837240315447e-07,
+      "loss": 18.9206,
+      "step": 19452
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.824823912063738e-07,
+      "loss": 15.7705,
+      "step": 19453
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.811822893577069e-07,
+      "loss": 15.8046,
+      "step": 19454
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.798834185179847e-07,
+      "loss": 15.699,
+      "step": 19455
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.78585778719637e-07,
+      "loss": 0.7679,
+      "step": 19456
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.772893699950433e-07,
+      "loss": 0.7392,
+      "step": 19457
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.759941923765778e-07,
+      "loss": 16.9112,
+      "step": 19458
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.747002458965535e-07,
+      "loss": 0.7423,
+      "step": 19459
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.734075305872723e-07,
+      "loss": 0.7885,
+      "step": 19460
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.721160464810084e-07,
+      "loss": 0.7419,
+      "step": 19461
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.70825793609986e-07,
+      "loss": 18.6698,
+      "step": 19462
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.695367720064183e-07,
+      "loss": 16.1417,
+      "step": 19463
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.68248981702474e-07,
+      "loss": 0.8065,
+      "step": 19464
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.66962422730294e-07,
+      "loss": 17.2241,
+      "step": 19465
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.656770951220026e-07,
+      "loss": 17.9749,
+      "step": 19466
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.643929989096798e-07,
+      "loss": 0.7343,
+      "step": 19467
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.631101341253665e-07,
+      "loss": 0.7675,
+      "step": 19468
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.618285008010871e-07,
+      "loss": 0.8259,
+      "step": 19469
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.605480989688384e-07,
+      "loss": 16.7316,
+      "step": 19470
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.592689286605724e-07,
+      "loss": 0.739,
+      "step": 19471
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.579909899082192e-07,
+      "loss": 16.7498,
+      "step": 19472
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.567142827436756e-07,
+      "loss": 0.7965,
+      "step": 19473
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.55438807198816e-07,
+      "loss": 0.7354,
+      "step": 19474
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.54164563305465e-07,
+      "loss": 17.3483,
+      "step": 19475
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.528915510954359e-07,
+      "loss": 17.0171,
+      "step": 19476
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.516197706005033e-07,
+      "loss": 16.1508,
+      "step": 19477
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.503492218524032e-07,
+      "loss": 0.7846,
+      "step": 19478
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.4907990488286e-07,
+      "loss": 0.7532,
+      "step": 19479
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.478118197235483e-07,
+      "loss": 0.7267,
+      "step": 19480
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.465449664061263e-07,
+      "loss": 0.804,
+      "step": 19481
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.452793449622019e-07,
+      "loss": 0.7999,
+      "step": 19482
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.440149554233832e-07,
+      "loss": 0.74,
+      "step": 19483
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.427517978212171e-07,
+      "loss": 16.5449,
+      "step": 19484
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.414898721872342e-07,
+      "loss": 17.5336,
+      "step": 19485
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.402291785529369e-07,
+      "loss": 0.7468,
+      "step": 19486
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.389697169497888e-07,
+      "loss": 16.9569,
+      "step": 19487
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.377114874092261e-07,
+      "loss": 0.7554,
+      "step": 19488
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.364544899626568e-07,
+      "loss": 0.7875,
+      "step": 19489
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.351987246414503e-07,
+      "loss": 19.0897,
+      "step": 19490
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.339441914769595e-07,
+      "loss": 16.8538,
+      "step": 19491
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.326908905004925e-07,
+      "loss": 0.764,
+      "step": 19492
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.314388217433298e-07,
+      "loss": 0.7096,
+      "step": 19493
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.301879852367299e-07,
+      "loss": 0.7867,
+      "step": 19494
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.289383810119121e-07,
+      "loss": 16.7045,
+      "step": 19495
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.276900091000626e-07,
+      "loss": 0.7713,
+      "step": 19496
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.264428695323454e-07,
+      "loss": 17.0765,
+      "step": 19497
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.251969623398857e-07,
+      "loss": 17.8519,
+      "step": 19498
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.239522875537862e-07,
+      "loss": 17.6335,
+      "step": 19499
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.227088452051055e-07,
+      "loss": 0.8254,
+      "step": 19500
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.21466635324891e-07,
+      "loss": 16.089,
+      "step": 19501
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.202256579441457e-07,
+      "loss": 18.1391,
+      "step": 19502
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.189859130938391e-07,
+      "loss": 0.7903,
+      "step": 19503
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.177474008049189e-07,
+      "loss": 0.7892,
+      "step": 19504
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.165101211082991e-07,
+      "loss": 17.3934,
+      "step": 19505
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.152740740348661e-07,
+      "loss": 17.5064,
+      "step": 19506
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.140392596154676e-07,
+      "loss": 18.1587,
+      "step": 19507
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.128056778809232e-07,
+      "loss": 0.709,
+      "step": 19508
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.115733288620306e-07,
+      "loss": 0.745,
+      "step": 19509
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.103422125895375e-07,
+      "loss": 0.7432,
+      "step": 19510
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.091123290941803e-07,
+      "loss": 17.693,
+      "step": 19511
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.078836784066621e-07,
+      "loss": 16.79,
+      "step": 19512
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.066562605576365e-07,
+      "loss": 15.8163,
+      "step": 19513
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.054300755777564e-07,
+      "loss": 0.7657,
+      "step": 19514
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.042051234976198e-07,
+      "loss": 0.7352,
+      "step": 19515
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.029814043478022e-07,
+      "loss": 0.7961,
+      "step": 19516
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.017589181588456e-07,
+      "loss": 0.7262,
+      "step": 19517
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.005376649612593e-07,
+      "loss": 0.7622,
+      "step": 19518
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.993176447855409e-07,
+      "loss": 17.3982,
+      "step": 19519
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.980988576621327e-07,
+      "loss": 19.0454,
+      "step": 19520
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.968813036214493e-07,
+      "loss": 0.8081,
+      "step": 19521
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.956649826938943e-07,
+      "loss": 15.7265,
+      "step": 19522
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.944498949098265e-07,
+      "loss": 0.7777,
+      "step": 19523
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.932360402995663e-07,
+      "loss": 0.7554,
+      "step": 19524
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.920234188934115e-07,
+      "loss": 16.1943,
+      "step": 19525
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.90812030721638e-07,
+      "loss": 0.7175,
+      "step": 19526
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.896018758144772e-07,
+      "loss": 0.6876,
+      "step": 19527
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.883929542021327e-07,
+      "loss": 18.972,
+      "step": 19528
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.871852659147858e-07,
+      "loss": 17.2786,
+      "step": 19529
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.859788109825793e-07,
+      "loss": 16.3177,
+      "step": 19530
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.847735894356166e-07,
+      "loss": 17.9293,
+      "step": 19531
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.835696013039904e-07,
+      "loss": 19.1587,
+      "step": 19532
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.82366846617749e-07,
+      "loss": 0.7555,
+      "step": 19533
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.811653254069183e-07,
+      "loss": 0.7724,
+      "step": 19534
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.799650377014853e-07,
+      "loss": 0.761,
+      "step": 19535
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.787659835314041e-07,
+      "loss": 17.0923,
+      "step": 19536
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.775681629266172e-07,
+      "loss": 17.8582,
+      "step": 19537
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.763715759170064e-07,
+      "loss": 16.9577,
+      "step": 19538
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.751762225324475e-07,
+      "loss": 19.9036,
+      "step": 19539
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.739821028027781e-07,
+      "loss": 17.4926,
+      "step": 19540
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.727892167577964e-07,
+      "loss": 0.7695,
+      "step": 19541
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.715975644272842e-07,
+      "loss": 17.1947,
+      "step": 19542
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.704071458409843e-07,
+      "loss": 0.7177,
+      "step": 19543
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.692179610286063e-07,
+      "loss": 17.6055,
+      "step": 19544
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.680300100198321e-07,
+      "loss": 0.7584,
+      "step": 19545
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.668432928443157e-07,
+      "loss": 16.1878,
+      "step": 19546
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.656578095316833e-07,
+      "loss": 18.9955,
+      "step": 19547
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.644735601115169e-07,
+      "loss": 18.5416,
+      "step": 19548
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.632905446133707e-07,
+      "loss": 16.1165,
+      "step": 19549
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.621087630667932e-07,
+      "loss": 17.6863,
+      "step": 19550
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.609282155012607e-07,
+      "loss": 18.11,
+      "step": 19551
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.597489019462499e-07,
+      "loss": 16.6497,
+      "step": 19552
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.585708224311981e-07,
+      "loss": 17.961,
+      "step": 19553
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.573939769854986e-07,
+      "loss": 0.7908,
+      "step": 19554
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.562183656385445e-07,
+      "loss": 0.71,
+      "step": 19555
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.550439884196624e-07,
+      "loss": 16.942,
+      "step": 19556
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.538708453581787e-07,
+      "loss": 17.3294,
+      "step": 19557
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.526989364833701e-07,
+      "loss": 0.7719,
+      "step": 19558
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.515282618244744e-07,
+      "loss": 0.7628,
+      "step": 19559
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.503588214107347e-07,
+      "loss": 0.7323,
+      "step": 19560
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.491906152713222e-07,
+      "loss": 15.9163,
+      "step": 19561
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.480236434354081e-07,
+      "loss": 18.5522,
+      "step": 19562
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.468579059321133e-07,
+      "loss": 16.7772,
+      "step": 19563
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.456934027905425e-07,
+      "loss": 16.5337,
+      "step": 19564
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.445301340397502e-07,
+      "loss": 0.7209,
+      "step": 19565
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.433680997087798e-07,
+      "loss": 0.7619,
+      "step": 19566
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.422072998266303e-07,
+      "loss": 16.3474,
+      "step": 19567
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.410477344222842e-07,
+      "loss": 18.316,
+      "step": 19568
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.398894035246738e-07,
+      "loss": 16.322,
+      "step": 19569
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.387323071627259e-07,
+      "loss": 18.4646,
+      "step": 19570
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.375764453653065e-07,
+      "loss": 15.6861,
+      "step": 19571
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.364218181612757e-07,
+      "loss": 16.8446,
+      "step": 19572
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.352684255794493e-07,
+      "loss": 18.2213,
+      "step": 19573
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.341162676486155e-07,
+      "loss": 18.475,
+      "step": 19574
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.329653443975347e-07,
+      "loss": 16.9348,
+      "step": 19575
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.318156558549336e-07,
+      "loss": 16.1204,
+      "step": 19576
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.306672020495118e-07,
+      "loss": 18.1432,
+      "step": 19577
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.295199830099351e-07,
+      "loss": 19.511,
+      "step": 19578
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.28373998764825e-07,
+      "loss": 0.7441,
+      "step": 19579
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.272292493428033e-07,
+      "loss": 18.4181,
+      "step": 19580
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.260857347724302e-07,
+      "loss": 0.7479,
+      "step": 19581
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.249434550822552e-07,
+      "loss": 0.7149,
+      "step": 19582
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.238024103007888e-07,
+      "loss": 17.958,
+      "step": 19583
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.226626004565139e-07,
+      "loss": 17.0408,
+      "step": 19584
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.215240255778797e-07,
+      "loss": 16.8624,
+      "step": 19585
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.203866856932971e-07,
+      "loss": 17.4033,
+      "step": 19586
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.192505808311599e-07,
+      "loss": 0.7375,
+      "step": 19587
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.181157110198287e-07,
+      "loss": 15.8218,
+      "step": 19588
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.169820762876254e-07,
+      "loss": 0.8141,
+      "step": 19589
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.158496766628495e-07,
+      "loss": 0.7819,
+      "step": 19590
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.147185121737675e-07,
+      "loss": 17.5366,
+      "step": 19591
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.13588582848612e-07,
+      "loss": 17.12,
+      "step": 19592
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.124598887155774e-07,
+      "loss": 19.0444,
+      "step": 19593
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.113324298028466e-07,
+      "loss": 0.7463,
+      "step": 19594
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.102062061385581e-07,
+      "loss": 0.7375,
+      "step": 19595
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.090812177508231e-07,
+      "loss": 0.7469,
+      "step": 19596
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.079574646677187e-07,
+      "loss": 17.4984,
+      "step": 19597
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.068349469173006e-07,
+      "loss": 17.6542,
+      "step": 19598
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.057136645275796e-07,
+      "loss": 0.7374,
+      "step": 19599
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.045936175265498e-07,
+      "loss": 0.7297,
+      "step": 19600
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.034748059421612e-07,
+      "loss": 18.8581,
+      "step": 19601
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.023572298023416e-07,
+      "loss": 0.7349,
+      "step": 19602
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.012408891349906e-07,
+      "loss": 0.7498,
+      "step": 19603
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.001257839679696e-07,
+      "loss": 17.4395,
+      "step": 19604
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.990119143291116e-07,
+      "loss": 16.5251,
+      "step": 19605
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.978992802462168e-07,
+      "loss": 17.744,
+      "step": 19606
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.967878817470573e-07,
+      "loss": 16.3046,
+      "step": 19607
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.956777188593775e-07,
+      "loss": 17.7891,
+      "step": 19608
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.945687916108887e-07,
+      "loss": 0.7853,
+      "step": 19609
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.934611000292577e-07,
+      "loss": 0.7384,
+      "step": 19610
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.923546441421511e-07,
+      "loss": 0.7889,
+      "step": 19611
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.912494239771693e-07,
+      "loss": 0.7556,
+      "step": 19612
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.901454395619121e-07,
+      "loss": 16.164,
+      "step": 19613
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.890426909239243e-07,
+      "loss": 18.8473,
+      "step": 19614
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.879411780907395e-07,
+      "loss": 18.1548,
+      "step": 19615
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.868409010898467e-07,
+      "loss": 0.7358,
+      "step": 19616
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.857418599487074e-07,
+      "loss": 16.7579,
+      "step": 19617
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.846440546947551e-07,
+      "loss": 16.6954,
+      "step": 19618
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.835474853554011e-07,
+      "loss": 0.7394,
+      "step": 19619
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.824521519580017e-07,
+      "loss": 16.7707,
+      "step": 19620
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.813580545299068e-07,
+      "loss": 18.6004,
+      "step": 19621
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.80265193098417e-07,
+      "loss": 0.7078,
+      "step": 19622
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.791735676908216e-07,
+      "loss": 17.3338,
+      "step": 19623
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.780831783343543e-07,
+      "loss": 15.3452,
+      "step": 19624
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.769940250562377e-07,
+      "loss": 18.7307,
+      "step": 19625
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.7590610788366685e-07,
+      "loss": 18.1305,
+      "step": 19626
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.7481942684378113e-07,
+      "loss": 0.7502,
+      "step": 19627
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.7373398196370875e-07,
+      "loss": 16.6092,
+      "step": 19628
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.726497732705504e-07,
+      "loss": 0.753,
+      "step": 19629
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.71566800791351e-07,
+      "loss": 16.5328,
+      "step": 19630
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.7048506455316663e-07,
+      "loss": 0.8,
+      "step": 19631
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.6940456458297587e-07,
+      "loss": 16.1774,
+      "step": 19632
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.6832530090775707e-07,
+      "loss": 0.7568,
+      "step": 19633
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.672472735544553e-07,
+      "loss": 19.4923,
+      "step": 19634
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.6617048254996023e-07,
+      "loss": 17.5196,
+      "step": 19635
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.6509492792116693e-07,
+      "loss": 16.9889,
+      "step": 19636
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.6402060969491513e-07,
+      "loss": 0.7792,
+      "step": 19637
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.6294752789801663e-07,
+      "loss": 16.6437,
+      "step": 19638
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.6187568255726124e-07,
+      "loss": 0.7432,
+      "step": 19639
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.608050736993941e-07,
+      "loss": 0.7906,
+      "step": 19640
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.597357013511494e-07,
+      "loss": 0.735,
+      "step": 19641
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.586675655392114e-07,
+      "loss": 17.0122,
+      "step": 19642
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.5760066629024214e-07,
+      "loss": 17.2144,
+      "step": 19643
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.5653500363087577e-07,
+      "loss": 20.8521,
+      "step": 19644
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.554705775877022e-07,
+      "loss": 0.7464,
+      "step": 19645
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.544073881873001e-07,
+      "loss": 18.3901,
+      "step": 19646
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.533454354561983e-07,
+      "loss": 16.8285,
+      "step": 19647
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.52284719420909e-07,
+      "loss": 18.8355,
+      "step": 19648
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.512252401079109e-07,
+      "loss": 0.7259,
+      "step": 19649
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.501669975436384e-07,
+      "loss": 0.726,
+      "step": 19650
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.491099917545094e-07,
+      "loss": 17.9172,
+      "step": 19651
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.480542227669138e-07,
+      "loss": 17.9738,
+      "step": 19652
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.4699969060719716e-07,
+      "loss": 17.2188,
+      "step": 19653
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.4594639530168846e-07,
+      "loss": 18.7554,
+      "step": 19654
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.4489433687666673e-07,
+      "loss": 0.6942,
+      "step": 19655
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.4384351535839417e-07,
+      "loss": 16.8204,
+      "step": 19656
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.4279393077311103e-07,
+      "loss": 0.7292,
+      "step": 19657
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.4174558314699633e-07,
+      "loss": 16.669,
+      "step": 19658
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.406984725062346e-07,
+      "loss": 0.7561,
+      "step": 19659
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.3965259887695507e-07,
+      "loss": 16.8342,
+      "step": 19660
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.3860796228526455e-07,
+      "loss": 17.5944,
+      "step": 19661
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.3756456275723114e-07,
+      "loss": 18.1101,
+      "step": 19662
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.3652240031890056e-07,
+      "loss": 0.7652,
+      "step": 19663
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.35481474996291e-07,
+      "loss": 16.8243,
+      "step": 19664
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.344417868153816e-07,
+      "loss": 17.0368,
+      "step": 19665
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.334033358021183e-07,
+      "loss": 0.8012,
+      "step": 19666
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.3236612198243043e-07,
+      "loss": 17.3364,
+      "step": 19667
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.3133014538220274e-07,
+      "loss": 0.7624,
+      "step": 19668
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.3029540602729235e-07,
+      "loss": 16.2007,
+      "step": 19669
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.2926190394352306e-07,
+      "loss": 17.1158,
+      "step": 19670
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.2822963915669647e-07,
+      "loss": 17.6064,
+      "step": 19671
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.271986116925752e-07,
+      "loss": 0.7047,
+      "step": 19672
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.261688215768944e-07,
+      "loss": 0.7054,
+      "step": 19673
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.2514026883536674e-07,
+      "loss": 0.7342,
+      "step": 19674
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.2411295349364945e-07,
+      "loss": 16.5594,
+      "step": 19675
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.230868755773998e-07,
+      "loss": 0.781,
+      "step": 19676
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.22062035112214e-07,
+      "loss": 16.0178,
+      "step": 19677
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.2103843212368263e-07,
+      "loss": 0.7607,
+      "step": 19678
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.2001606663735205e-07,
+      "loss": 0.7423,
+      "step": 19679
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.189949386787462e-07,
+      "loss": 16.502,
+      "step": 19680
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.179750482733391e-07,
+      "loss": 0.7024,
+      "step": 19681
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.1695639544660495e-07,
+      "loss": 0.7635,
+      "step": 19682
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.159389802239511e-07,
+      "loss": 17.1546,
+      "step": 19683
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.1492280263079053e-07,
+      "loss": 18.441,
+      "step": 19684
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.139078626924753e-07,
+      "loss": 0.7698,
+      "step": 19685
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.1289416043433503e-07,
+      "loss": 0.7764,
+      "step": 19686
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.1188169588168847e-07,
+      "loss": 0.7766,
+      "step": 19687
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.108704690597931e-07,
+      "loss": 15.2719,
+      "step": 19688
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.098604799938954e-07,
+      "loss": 0.7617,
+      "step": 19689
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.0885172870919753e-07,
+      "loss": 18.6802,
+      "step": 19690
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.078442152308848e-07,
+      "loss": 16.093,
+      "step": 19691
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.068379395841093e-07,
+      "loss": 17.3797,
+      "step": 19692
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.058329017939788e-07,
+      "loss": 16.6315,
+      "step": 19693
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.048291018855843e-07,
+      "loss": 0.752,
+      "step": 19694
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.03826539883978e-07,
+      "loss": 0.7714,
+      "step": 19695
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.028252158141843e-07,
+      "loss": 0.7459,
+      "step": 19696
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.0182512970120543e-07,
+      "loss": 15.7039,
+      "step": 19697
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.0082628156998814e-07,
+      "loss": 16.8987,
+      "step": 19698
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.998286714454735e-07,
+      "loss": 0.768,
+      "step": 19699
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.988322993525584e-07,
+      "loss": 0.7151,
+      "step": 19700
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.9783716531611726e-07,
+      "loss": 0.758,
+      "step": 19701
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.9684326936099136e-07,
+      "loss": 0.7588,
+      "step": 19702
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.95850611511972e-07,
+      "loss": 16.7969,
+      "step": 19703
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.94859191793856e-07,
+      "loss": 0.6848,
+      "step": 19704
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.938690102313791e-07,
+      "loss": 19.9607,
+      "step": 19705
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.928800668492549e-07,
+      "loss": 20.1323,
+      "step": 19706
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.918923616721748e-07,
+      "loss": 14.7989,
+      "step": 19707
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.909058947247857e-07,
+      "loss": 0.7915,
+      "step": 19708
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.8992066603171786e-07,
+      "loss": 17.3501,
+      "step": 19709
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.8893667561755165e-07,
+      "loss": 17.4359,
+      "step": 19710
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.8795392350685633e-07,
+      "loss": 0.8213,
+      "step": 19711
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.869724097241567e-07,
+      "loss": 0.7609,
+      "step": 19712
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.859921342939554e-07,
+      "loss": 0.7296,
+      "step": 19713
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.850130972407162e-07,
+      "loss": 16.4086,
+      "step": 19714
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.840352985888862e-07,
+      "loss": 15.4799,
+      "step": 19715
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.8305873836285697e-07,
+      "loss": 16.3356,
+      "step": 19716
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.8208341658701463e-07,
+      "loss": 18.9567,
+      "step": 19717
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.8110933328570074e-07,
+      "loss": 0.7838,
+      "step": 19718
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.8013648848321814e-07,
+      "loss": 0.7305,
+      "step": 19719
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.791648822038696e-07,
+      "loss": 0.6682,
+      "step": 19720
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.7819451447189126e-07,
+      "loss": 16.1246,
+      "step": 19721
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.7722538531150265e-07,
+      "loss": 0.8092,
+      "step": 19722
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.7625749474690664e-07,
+      "loss": 17.2219,
+      "step": 19723
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.752908428022506e-07,
+      "loss": 0.7857,
+      "step": 19724
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.7432542950166514e-07,
+      "loss": 16.7549,
+      "step": 19725
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.7336125486924776e-07,
+      "loss": 19.5493,
+      "step": 19726
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.7239831892906805e-07,
+      "loss": 0.8015,
+      "step": 19727
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.714366217051568e-07,
+      "loss": 16.0299,
+      "step": 19728
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.704761632215115e-07,
+      "loss": 16.9767,
+      "step": 19729
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.6951694350212395e-07,
+      "loss": 18.4543,
+      "step": 19730
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.685589625709196e-07,
+      "loss": 0.719,
+      "step": 19731
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.676022204518181e-07,
+      "loss": 0.7807,
+      "step": 19732
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.6664671716870045e-07,
+      "loss": 17.4396,
+      "step": 19733
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.656924527454086e-07,
+      "loss": 0.7167,
+      "step": 19734
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.6473942720576803e-07,
+      "loss": 17.8961,
+      "step": 19735
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.637876405735652e-07,
+      "loss": 0.7511,
+      "step": 19736
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.628370928725533e-07,
+      "loss": 18.648,
+      "step": 19737
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.618877841264634e-07,
+      "loss": 15.942,
+      "step": 19738
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.609397143589877e-07,
+      "loss": 0.7819,
+      "step": 19739
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.599928835937905e-07,
+      "loss": 18.238,
+      "step": 19740
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.590472918545085e-07,
+      "loss": 0.7328,
+      "step": 19741
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.581029391647339e-07,
+      "loss": 15.0611,
+      "step": 19742
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.5715982554804795e-07,
+      "loss": 0.7434,
+      "step": 19743
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.562179510279928e-07,
+      "loss": 18.1682,
+      "step": 19744
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.5527731562806645e-07,
+      "loss": 18.055,
+      "step": 19745
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.543379193717555e-07,
+      "loss": 0.7631,
+      "step": 19746
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.5339976228250805e-07,
+      "loss": 0.7339,
+      "step": 19747
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.52462844383733e-07,
+      "loss": 0.7302,
+      "step": 19748
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.515271656988284e-07,
+      "loss": 0.811,
+      "step": 19749
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.505927262511366e-07,
+      "loss": 0.732,
+      "step": 19750
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.49659526063989e-07,
+      "loss": 0.7282,
+      "step": 19751
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.487275651606725e-07,
+      "loss": 15.4319,
+      "step": 19752
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.477968435644574e-07,
+      "loss": 0.7763,
+      "step": 19753
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.46867361298564e-07,
+      "loss": 15.6811,
+      "step": 19754
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4593911838620707e-07,
+      "loss": 19.536,
+      "step": 19755
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4501211485054584e-07,
+      "loss": 0.7514,
+      "step": 19756
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4408635071471184e-07,
+      "loss": 0.7384,
+      "step": 19757
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.43161826001831e-07,
+      "loss": 0.7799,
+      "step": 19758
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.422385407349626e-07,
+      "loss": 0.7487,
+      "step": 19759
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4131649493716056e-07,
+      "loss": 16.8533,
+      "step": 19760
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4039568863143414e-07,
+      "loss": 17.7024,
+      "step": 19761
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.394761218407705e-07,
+      "loss": 0.7673,
+      "step": 19762
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.385577945881291e-07,
+      "loss": 19.3569,
+      "step": 19763
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.376407068964194e-07,
+      "loss": 0.713,
+      "step": 19764
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.367248587885341e-07,
+      "loss": 0.7901,
+      "step": 19765
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.358102502873439e-07,
+      "loss": 16.3698,
+      "step": 19766
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.348968814156583e-07,
+      "loss": 17.7151,
+      "step": 19767
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.3398475219629243e-07,
+      "loss": 0.726,
+      "step": 19768
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.330738626520113e-07,
+      "loss": 0.7294,
+      "step": 19769
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.321642128055413e-07,
+      "loss": 16.1739,
+      "step": 19770
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.3125580267959754e-07,
+      "loss": 17.6549,
+      "step": 19771
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.303486322968452e-07,
+      "loss": 0.7383,
+      "step": 19772
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.294427016799384e-07,
+      "loss": 0.729,
+      "step": 19773
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.285380108514757e-07,
+      "loss": 15.239,
+      "step": 19774
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.276345598340502e-07,
+      "loss": 18.227,
+      "step": 19775
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.2673234865020495e-07,
+      "loss": 17.3168,
+      "step": 19776
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.2583137732246636e-07,
+      "loss": 0.7548,
+      "step": 19777
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.2493164587331646e-07,
+      "loss": 0.714,
+      "step": 19778
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.240331543252151e-07,
+      "loss": 0.7584,
+      "step": 19779
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.231359027005887e-07,
+      "loss": 15.6756,
+      "step": 19780
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.222398910218305e-07,
+      "loss": 0.7942,
+      "step": 19781
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.21345119311306e-07,
+      "loss": 0.788,
+      "step": 19782
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.204515875913583e-07,
+      "loss": 0.739,
+      "step": 19783
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.1955929588427526e-07,
+      "loss": 0.7865,
+      "step": 19784
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.18668244212339e-07,
+      "loss": 15.6771,
+      "step": 19785
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.1777843259778726e-07,
+      "loss": 15.2593,
+      "step": 19786
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.1688986106283015e-07,
+      "loss": 18.8283,
+      "step": 19787
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.1600252962964426e-07,
+      "loss": 17.8512,
+      "step": 19788
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.1511643832037865e-07,
+      "loss": 0.7642,
+      "step": 19789
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.142315871571544e-07,
+      "loss": 16.9063,
+      "step": 19790
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.1334797616204836e-07,
+      "loss": 0.7877,
+      "step": 19791
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.124656053571318e-07,
+      "loss": 15.0951,
+      "step": 19792
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.1158447476441475e-07,
+      "loss": 0.7019,
+      "step": 19793
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.107045844058909e-07,
+      "loss": 0.7068,
+      "step": 19794
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.0982593430352593e-07,
+      "loss": 14.8606,
+      "step": 19795
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.089485244792578e-07,
+      "loss": 18.3107,
+      "step": 19796
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.0807235495497464e-07,
+      "loss": 18.0368,
+      "step": 19797
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.071974257525534e-07,
+      "loss": 19.0422,
+      "step": 19798
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.063237368938321e-07,
+      "loss": 18.2334,
+      "step": 19799
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.0545128840061555e-07,
+      "loss": 19.6611,
+      "step": 19800
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.0458008029468077e-07,
+      "loss": 0.8119,
+      "step": 19801
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.0371011259777703e-07,
+      "loss": 16.261,
+      "step": 19802
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.0284138533160924e-07,
+      "loss": 0.7098,
+      "step": 19803
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.0197389851787663e-07,
+      "loss": 16.6608,
+      "step": 19804
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.0110765217821746e-07,
+      "loss": 18.1963,
+      "step": 19805
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.0024264633426446e-07,
+      "loss": 0.7372,
+      "step": 19806
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.993788810075948e-07,
+      "loss": 0.8047,
+      "step": 19807
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.9851635621978013e-07,
+      "loss": 0.7224,
+      "step": 19808
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.9765507199234764e-07,
+      "loss": 16.4599,
+      "step": 19809
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.967950283467913e-07,
+      "loss": 0.7592,
+      "step": 19810
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.959362253045772e-07,
+      "loss": 16.6568,
+      "step": 19811
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.9507866288714936e-07,
+      "loss": 18.5547,
+      "step": 19812
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.942223411159073e-07,
+      "loss": 0.7347,
+      "step": 19813
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.933672600122228e-07,
+      "loss": 16.5384,
+      "step": 19814
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.925134195974344e-07,
+      "loss": 0.7755,
+      "step": 19815
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.916608198928694e-07,
+      "loss": 17.4462,
+      "step": 19816
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.908094609197998e-07,
+      "loss": 17.4085,
+      "step": 19817
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8995934269946957e-07,
+      "loss": 0.787,
+      "step": 19818
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.891104652531118e-07,
+      "loss": 0.7467,
+      "step": 19819
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.88262828601904e-07,
+      "loss": 0.7404,
+      "step": 19820
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.87416432767007e-07,
+      "loss": 0.7602,
+      "step": 19821
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.865712777695484e-07,
+      "loss": 0.7612,
+      "step": 19822
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.857273636306168e-07,
+      "loss": 0.7463,
+      "step": 19823
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.848846903712843e-07,
+      "loss": 17.0577,
+      "step": 19824
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.84043258012584e-07,
+      "loss": 0.7553,
+      "step": 19825
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8320306657551034e-07,
+      "loss": 17.5139,
+      "step": 19826
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.823641160810464e-07,
+      "loss": 0.7583,
+      "step": 19827
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8152640655012e-07,
+      "loss": 15.8538,
+      "step": 19828
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8068993800364764e-07,
+      "loss": 16.6216,
+      "step": 19829
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.798547104625071e-07,
+      "loss": 0.6727,
+      "step": 19830
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7902072394754844e-07,
+      "loss": 0.8,
+      "step": 19831
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7818797847958265e-07,
+      "loss": 17.0625,
+      "step": 19832
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.773564740793988e-07,
+      "loss": 16.3249,
+      "step": 19833
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7652621076775243e-07,
+      "loss": 16.9962,
+      "step": 19834
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.756971885653659e-07,
+      "loss": 0.8088,
+      "step": 19835
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7486940749292256e-07,
+      "loss": 15.0461,
+      "step": 19836
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7404286757110043e-07,
+      "loss": 0.8106,
+      "step": 19837
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7321756882052185e-07,
+      "loss": 16.7732,
+      "step": 19838
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7239351126178146e-07,
+      "loss": 0.7256,
+      "step": 19839
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7157069491546285e-07,
+      "loss": 17.0194,
+      "step": 19840
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7074911980208837e-07,
+      "loss": 17.5872,
+      "step": 19841
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6992878594216954e-07,
+      "loss": 0.7936,
+      "step": 19842
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.691096933561843e-07,
+      "loss": 0.7479,
+      "step": 19843
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.682918420645719e-07,
+      "loss": 0.7669,
+      "step": 19844
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.674752320877549e-07,
+      "loss": 18.8123,
+      "step": 19845
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.666598634461115e-07,
+      "loss": 0.7809,
+      "step": 19846
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.65845736159992e-07,
+      "loss": 16.6447,
+      "step": 19847
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.650328502497246e-07,
+      "loss": 0.7347,
+      "step": 19848
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.642212057355875e-07,
+      "loss": 0.7466,
+      "step": 19849
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.634108026378479e-07,
+      "loss": 16.2037,
+      "step": 19850
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6260164097672846e-07,
+      "loss": 17.1555,
+      "step": 19851
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.617937207724297e-07,
+      "loss": 0.7271,
+      "step": 19852
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6098704204511327e-07,
+      "loss": 0.7677,
+      "step": 19853
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6018160481492414e-07,
+      "loss": 0.779,
+      "step": 19854
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5937740910195186e-07,
+      "loss": 16.8632,
+      "step": 19855
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.585744549262803e-07,
+      "loss": 16.4973,
+      "step": 19856
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.57772742307949e-07,
+      "loss": 17.4418,
+      "step": 19857
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5697227126696423e-07,
+      "loss": 0.8008,
+      "step": 19858
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.561730418233099e-07,
+      "loss": 0.7948,
+      "step": 19859
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.553750539969368e-07,
+      "loss": 15.6347,
+      "step": 19860
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.545783078077568e-07,
+      "loss": 15.4474,
+      "step": 19861
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5378280327566506e-07,
+      "loss": 0.7499,
+      "step": 19862
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5298854042051235e-07,
+      "loss": 18.6376,
+      "step": 19863
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.521955192621217e-07,
+      "loss": 18.6781,
+      "step": 19864
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5140373982029397e-07,
+      "loss": 0.7092,
+      "step": 19865
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5061320211478e-07,
+      "loss": 0.7278,
+      "step": 19866
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.498239061653307e-07,
+      "loss": 0.785,
+      "step": 19867
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.4903585199163026e-07,
+      "loss": 0.7393,
+      "step": 19868
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.482490396133519e-07,
+      "loss": 0.7888,
+      "step": 19869
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.4746346905014097e-07,
+      "loss": 16.4937,
+      "step": 19870
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.46679140321604e-07,
+      "loss": 18.5063,
+      "step": 19871
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.458960534473087e-07,
+      "loss": 17.4848,
+      "step": 19872
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.451142084468172e-07,
+      "loss": 0.754,
+      "step": 19873
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.443336053396306e-07,
+      "loss": 17.991,
+      "step": 19874
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.4355424414523875e-07,
+      "loss": 0.7332,
+      "step": 19875
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.4277612488309286e-07,
+      "loss": 18.2807,
+      "step": 19876
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.419992475726163e-07,
+      "loss": 19.0696,
+      "step": 19877
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.41223612233199e-07,
+      "loss": 17.6312,
+      "step": 19878
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.4044921888420333e-07,
+      "loss": 0.7422,
+      "step": 19879
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3967606754495274e-07,
+      "loss": 16.6548,
+      "step": 19880
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3890415823475396e-07,
+      "loss": 0.7565,
+      "step": 19881
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3813349097286942e-07,
+      "loss": 0.7327,
+      "step": 19882
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3736406577853366e-07,
+      "loss": 18.4587,
+      "step": 19883
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3659588267095357e-07,
+      "loss": 0.7492,
+      "step": 19884
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3582894166930268e-07,
+      "loss": 0.7513,
+      "step": 19885
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.350632427927213e-07,
+      "loss": 0.7448,
+      "step": 19886
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.342987860603274e-07,
+      "loss": 0.7618,
+      "step": 19887
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3353557149119466e-07,
+      "loss": 16.1513,
+      "step": 19888
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3277359910438555e-07,
+      "loss": 17.9262,
+      "step": 19889
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3201286891890163e-07,
+      "loss": 16.6974,
+      "step": 19890
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3125338095374426e-07,
+      "loss": 17.6441,
+      "step": 19891
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3049513522786503e-07,
+      "loss": 0.7855,
+      "step": 19892
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.297381317601932e-07,
+      "loss": 0.7388,
+      "step": 19893
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2898237056961925e-07,
+      "loss": 17.0364,
+      "step": 19894
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2822785167501137e-07,
+      "loss": 0.7906,
+      "step": 19895
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2747457509520453e-07,
+      "loss": 17.9685,
+      "step": 19896
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2672254084898926e-07,
+      "loss": 0.7314,
+      "step": 19897
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2597174895514494e-07,
+      "loss": 16.8643,
+      "step": 19898
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2522219943241218e-07,
+      "loss": 18.4656,
+      "step": 19899
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2447389229949823e-07,
+      "loss": 17.2291,
+      "step": 19900
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2372682757508257e-07,
+      "loss": 16.7148,
+      "step": 19901
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.229810052778114e-07,
+      "loss": 0.746,
+      "step": 19902
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2223642542629764e-07,
+      "loss": 0.7502,
+      "step": 19903
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2149308803913193e-07,
+      "loss": 16.7627,
+      "step": 19904
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2075099313486058e-07,
+      "loss": 0.7417,
+      "step": 19905
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.200101407320132e-07,
+      "loss": 0.7688,
+      "step": 19906
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.19270530849075e-07,
+      "loss": 18.0952,
+      "step": 19907
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1853216350451455e-07,
+      "loss": 16.205,
+      "step": 19908
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.17795038716756e-07,
+      "loss": 0.7412,
+      "step": 19909
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1705915650420128e-07,
+      "loss": 18.4717,
+      "step": 19910
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1632451688521904e-07,
+      "loss": 17.9904,
+      "step": 19911
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1559111987813906e-07,
+      "loss": 16.0576,
+      "step": 19912
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1485896550127448e-07,
+      "loss": 17.5377,
+      "step": 19913
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1412805377289957e-07,
+      "loss": 16.7116,
+      "step": 19914
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1339838471124973e-07,
+      "loss": 17.6593,
+      "step": 19915
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1266995833454929e-07,
+      "loss": 0.7304,
+      "step": 19916
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1194277466097256e-07,
+      "loss": 0.7598,
+      "step": 19917
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.112168337086773e-07,
+      "loss": 0.7414,
+      "step": 19918
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1049213549577118e-07,
+      "loss": 18.2515,
+      "step": 19919
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0976868004035088e-07,
+      "loss": 16.4122,
+      "step": 19920
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0904646736047417e-07,
+      "loss": 0.7813,
+      "step": 19921
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0832549747416553e-07,
+      "loss": 0.79,
+      "step": 19922
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.076057703994272e-07,
+      "loss": 0.7643,
+      "step": 19923
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0688728615421704e-07,
+      "loss": 18.768,
+      "step": 19924
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0617004475645963e-07,
+      "loss": 0.7488,
+      "step": 19925
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0545404622407393e-07,
+      "loss": 17.5016,
+      "step": 19926
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0473929057492347e-07,
+      "loss": 17.8126,
+      "step": 19927
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0402577782685505e-07,
+      "loss": 18.0928,
+      "step": 19928
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0331350799766558e-07,
+      "loss": 0.7613,
+      "step": 19929
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0260248110514636e-07,
+      "loss": 0.7754,
+      "step": 19930
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0189269716703874e-07,
+      "loss": 0.714,
+      "step": 19931
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0118415620105634e-07,
+      "loss": 16.114,
+      "step": 19932
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.004768582248906e-07,
+      "loss": 0.7676,
+      "step": 19933
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.99770803256194e-07,
+      "loss": 16.1918,
+      "step": 19934
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9906599131258585e-07,
+      "loss": 16.687,
+      "step": 19935
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9836242241166313e-07,
+      "loss": 15.0813,
+      "step": 19936
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.976600965709896e-07,
+      "loss": 0.7667,
+      "step": 19937
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9695901380808458e-07,
+      "loss": 0.7643,
+      "step": 19938
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9625917414045625e-07,
+      "loss": 17.0503,
+      "step": 19939
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.95560577585574e-07,
+      "loss": 17.4626,
+      "step": 19940
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9486322416086832e-07,
+      "loss": 0.7583,
+      "step": 19941
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9416711388374752e-07,
+      "loss": 16.9557,
+      "step": 19942
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9347224677159214e-07,
+      "loss": 0.8254,
+      "step": 19943
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9277862284173832e-07,
+      "loss": 0.7512,
+      "step": 19944
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.920862421115055e-07,
+      "loss": 0.7622,
+      "step": 19945
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9139510459816877e-07,
+      "loss": 20.0567,
+      "step": 19946
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9070521031898658e-07,
+      "loss": 0.725,
+      "step": 19947
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.900165592911729e-07,
+      "loss": 16.6542,
+      "step": 19948
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8932915153191954e-07,
+      "loss": 0.7682,
+      "step": 19949
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8864298705837945e-07,
+      "loss": 18.3992,
+      "step": 19950
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.879580658876945e-07,
+      "loss": 18.3425,
+      "step": 19951
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8727438803693987e-07,
+      "loss": 0.6911,
+      "step": 19952
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8659195352319637e-07,
+      "loss": 17.9996,
+      "step": 19953
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8591076236348925e-07,
+      "loss": 17.1686,
+      "step": 19954
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.852308145748216e-07,
+      "loss": 18.7734,
+      "step": 19955
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8455211017416874e-07,
+      "loss": 0.7572,
+      "step": 19956
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8387464917846708e-07,
+      "loss": 15.7161,
+      "step": 19957
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8319843160463647e-07,
+      "loss": 0.7166,
+      "step": 19958
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8252345746954114e-07,
+      "loss": 17.1961,
+      "step": 19959
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8184972679003432e-07,
+      "loss": 16.6823,
+      "step": 19960
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8117723958293586e-07,
+      "loss": 15.7967,
+      "step": 19961
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.805059958650268e-07,
+      "loss": 17.5071,
+      "step": 19962
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7983599565306598e-07,
+      "loss": 15.1921,
+      "step": 19963
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.791672389637733e-07,
+      "loss": 0.7895,
+      "step": 19964
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.78499725813841e-07,
+      "loss": 16.5288,
+      "step": 19965
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7783345621993352e-07,
+      "loss": 17.774,
+      "step": 19966
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7716843019867646e-07,
+      "loss": 18.2006,
+      "step": 19967
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7650464776667318e-07,
+      "loss": 0.7665,
+      "step": 19968
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7584210894048825e-07,
+      "loss": 17.3547,
+      "step": 19969
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.751808137366584e-07,
+      "loss": 20.0808,
+      "step": 19970
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7452076217169267e-07,
+      "loss": 0.7869,
+      "step": 19971
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.738619542620723e-07,
+      "loss": 0.7547,
+      "step": 19972
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7320439002422862e-07,
+      "loss": 0.741,
+      "step": 19973
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.725480694745818e-07,
+      "loss": 19.0289,
+      "step": 19974
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.718929926295132e-07,
+      "loss": 0.8224,
+      "step": 19975
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7123915950536528e-07,
+      "loss": 17.7431,
+      "step": 19976
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7058657011847502e-07,
+      "loss": 0.7726,
+      "step": 19977
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6993522448511823e-07,
+      "loss": 18.4209,
+      "step": 19978
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6928512262155417e-07,
+      "loss": 0.7988,
+      "step": 19979
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6863626454401428e-07,
+      "loss": 15.7683,
+      "step": 19980
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.679886502686856e-07,
+      "loss": 17.9081,
+      "step": 19981
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.673422798117441e-07,
+      "loss": 16.4248,
+      "step": 19982
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6669715318931578e-07,
+      "loss": 17.8662,
+      "step": 19983
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.660532704175044e-07,
+      "loss": 0.7643,
+      "step": 19984
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6541063151238046e-07,
+      "loss": 0.7745,
+      "step": 19985
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.647692364899922e-07,
+      "loss": 0.7268,
+      "step": 19986
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6412908536633798e-07,
+      "loss": 17.6195,
+      "step": 19987
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6349017815739942e-07,
+      "loss": 17.2068,
+      "step": 19988
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.628525148791249e-07,
+      "loss": 20.1395,
+      "step": 19989
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.62216095547435e-07,
+      "loss": 17.2434,
+      "step": 19990
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.615809201782059e-07,
+      "loss": 0.7381,
+      "step": 19991
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6094698878730275e-07,
+      "loss": 18.1676,
+      "step": 19992
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6031430139053504e-07,
+      "loss": 16.8888,
+      "step": 19993
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5968285800370686e-07,
+      "loss": 17.7856,
+      "step": 19994
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.590526586425778e-07,
+      "loss": 0.7754,
+      "step": 19995
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.584237033228686e-07,
+      "loss": 0.7285,
+      "step": 19996
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.577959920602834e-07,
+      "loss": 0.761,
+      "step": 19997
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5716952487049853e-07,
+      "loss": 17.9992,
+      "step": 19998
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.565443017691348e-07,
+      "loss": 0.7624,
+      "step": 19999
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5592032277181313e-07,
+      "loss": 16.9759,
+      "step": 20000
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.552975878940932e-07,
+      "loss": 0.704,
+      "step": 20001
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.546760971515293e-07,
+      "loss": 17.5156,
+      "step": 20002
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5405585055963678e-07,
+      "loss": 0.7305,
+      "step": 20003
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5343684813388105e-07,
+      "loss": 19.1434,
+      "step": 20004
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5281908988973303e-07,
+      "loss": 0.7258,
+      "step": 20005
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.522025758425971e-07,
+      "loss": 18.9239,
+      "step": 20006
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5158730600786652e-07,
+      "loss": 0.7415,
+      "step": 20007
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.509732804009012e-07,
+      "loss": 16.9168,
+      "step": 20008
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.503604990370222e-07,
+      "loss": 19.8659,
+      "step": 20009
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4974896193152842e-07,
+      "loss": 0.7955,
+      "step": 20010
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4913866909967988e-07,
+      "loss": 19.4019,
+      "step": 20011
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4852962055671994e-07,
+      "loss": 18.4637,
+      "step": 20012
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4792181631783642e-07,
+      "loss": 0.7527,
+      "step": 20013
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4731525639821165e-07,
+      "loss": 17.6254,
+      "step": 20014
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4670994081297795e-07,
+      "loss": 16.5679,
+      "step": 20015
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.461058695772455e-07,
+      "loss": 0.7565,
+      "step": 20016
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4550304270609661e-07,
+      "loss": 0.7802,
+      "step": 20017
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4490146021457484e-07,
+      "loss": 17.3419,
+      "step": 20018
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4430112211769598e-07,
+      "loss": 18.5785,
+      "step": 20019
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.43702028430448e-07,
+      "loss": 0.7754,
+      "step": 20020
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.431041791677745e-07,
+      "loss": 0.7802,
+      "step": 20021
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4250757434460803e-07,
+      "loss": 0.7579,
+      "step": 20022
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4191221397583665e-07,
+      "loss": 18.2025,
+      "step": 20023
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4131809807631513e-07,
+      "loss": 17.3586,
+      "step": 20024
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4072522666088717e-07,
+      "loss": 0.7302,
+      "step": 20025
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4013359974433537e-07,
+      "loss": 0.7784,
+      "step": 20026
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3954321734143682e-07,
+      "loss": 15.6778,
+      "step": 20027
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3895407946692417e-07,
+      "loss": 17.3019,
+      "step": 20028
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3836618613549679e-07,
+      "loss": 16.2287,
+      "step": 20029
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3777953736184291e-07,
+      "loss": 0.7243,
+      "step": 20030
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.371941331605897e-07,
+      "loss": 17.7366,
+      "step": 20031
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3660997354635884e-07,
+      "loss": 15.9025,
+      "step": 20032
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3602705853372756e-07,
+      "loss": 0.7863,
+      "step": 20033
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.354453881372453e-07,
+      "loss": 17.3809,
+      "step": 20034
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3486496237143375e-07,
+      "loss": 14.5153,
+      "step": 20035
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.342857812507814e-07,
+      "loss": 0.7899,
+      "step": 20036
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3370784478973774e-07,
+      "loss": 0.7598,
+      "step": 20037
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3313115300273015e-07,
+      "loss": 18.3681,
+      "step": 20038
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3255570590415823e-07,
+      "loss": 17.3498,
+      "step": 20039
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3198150350838268e-07,
+      "loss": 0.8309,
+      "step": 20040
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.314085458297365e-07,
+      "loss": 0.7386,
+      "step": 20041
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.308368328825138e-07,
+      "loss": 0.7639,
+      "step": 20042
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.302663646809976e-07,
+      "loss": 0.7139,
+      "step": 20043
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2969714123941546e-07,
+      "loss": 19.4696,
+      "step": 20044
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2912916257197816e-07,
+      "loss": 0.7751,
+      "step": 20045
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2856242869286882e-07,
+      "loss": 17.0885,
+      "step": 20046
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2799693961622062e-07,
+      "loss": 0.7394,
+      "step": 20047
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2743269535616664e-07,
+      "loss": 0.7436,
+      "step": 20048
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.268696959267679e-07,
+      "loss": 0.736,
+      "step": 20049
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2630794134209646e-07,
+      "loss": 20.6369,
+      "step": 20050
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2574743161616887e-07,
+      "loss": 16.27,
+      "step": 20051
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.251881667629684e-07,
+      "loss": 18.8015,
+      "step": 20052
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.246301467964617e-07,
+      "loss": 18.0011,
+      "step": 20053
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2407337173057086e-07,
+      "loss": 0.7287,
+      "step": 20054
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2351784157920154e-07,
+      "loss": 18.9443,
+      "step": 20055
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2296355635620927e-07,
+      "loss": 18.4776,
+      "step": 20056
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2241051607543852e-07,
+      "loss": 0.7481,
+      "step": 20057
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2185872075068938e-07,
+      "loss": 0.735,
+      "step": 20058
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2130817039573972e-07,
+      "loss": 18.4213,
+      "step": 20059
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2075886502431743e-07,
+      "loss": 16.1337,
+      "step": 20060
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2021080465015046e-07,
+      "loss": 16.0388,
+      "step": 20061
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1966398928691115e-07,
+      "loss": 15.952,
+      "step": 20062
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1911841894824416e-07,
+      "loss": 17.7152,
+      "step": 20063
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1857409364777195e-07,
+      "loss": 18.029,
+      "step": 20064
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1803101339907808e-07,
+      "loss": 0.7835,
+      "step": 20065
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1748917821572391e-07,
+      "loss": 17.0681,
+      "step": 20066
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1694858811122645e-07,
+      "loss": 17.4673,
+      "step": 20067
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1640924309908041e-07,
+      "loss": 0.7616,
+      "step": 20068
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1587114319275282e-07,
+      "loss": 18.7474,
+      "step": 20069
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1533428840567184e-07,
+      "loss": 16.4312,
+      "step": 20070
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.147986787512323e-07,
+      "loss": 16.395,
+      "step": 20071
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1426431424281237e-07,
+      "loss": 16.8456,
+      "step": 20072
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1373119489374584e-07,
+      "loss": 0.8161,
+      "step": 20073
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1319932071733875e-07,
+      "loss": 0.8211,
+      "step": 20074
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1266869172686378e-07,
+      "loss": 0.764,
+      "step": 20075
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1213930793557149e-07,
+      "loss": 16.6219,
+      "step": 20076
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1161116935667349e-07,
+      "loss": 0.7869,
+      "step": 20077
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.110842760033537e-07,
+      "loss": 17.5614,
+      "step": 20078
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1055862788875714e-07,
+      "loss": 18.2997,
+      "step": 20079
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1003422502600669e-07,
+      "loss": 0.7133,
+      "step": 20080
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0951106742819739e-07,
+      "loss": 15.9072,
+      "step": 20081
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0898915510838548e-07,
+      "loss": 16.5403,
+      "step": 20082
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0846848807959387e-07,
+      "loss": 0.7024,
+      "step": 20083
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0794906635481772e-07,
+      "loss": 0.7537,
+      "step": 20084
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0743088994702999e-07,
+      "loss": 17.0153,
+      "step": 20085
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0691395886915367e-07,
+      "loss": 0.7658,
+      "step": 20086
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0639827313410066e-07,
+      "loss": 0.7855,
+      "step": 20087
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0588383275473846e-07,
+      "loss": 16.3267,
+      "step": 20088
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0537063774390677e-07,
+      "loss": 16.2712,
+      "step": 20089
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0485868811441757e-07,
+      "loss": 18.4352,
+      "step": 20090
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0434798387904954e-07,
+      "loss": 17.9537,
+      "step": 20091
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0383852505054247e-07,
+      "loss": 0.8022,
+      "step": 20092
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.033303116416251e-07,
+      "loss": 16.7746,
+      "step": 20093
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.028233436649706e-07,
+      "loss": 0.752,
+      "step": 20094
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0231762113324105e-07,
+      "loss": 18.7902,
+      "step": 20095
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0181314405905972e-07,
+      "loss": 18.2275,
+      "step": 20096
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0130991245501098e-07,
+      "loss": 0.7452,
+      "step": 20097
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0080792633366254e-07,
+      "loss": 0.7492,
+      "step": 20098
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0030718570754327e-07,
+      "loss": 16.0989,
+      "step": 20099
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.980769058914874e-08,
+      "loss": 0.7457,
+      "step": 20100
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.930944099094674e-08,
+      "loss": 15.7337,
+      "step": 20101
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.881243692537733e-08,
+      "loss": 0.7369,
+      "step": 20102
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.831667840484172e-08,
+      "loss": 0.8137,
+      "step": 20103
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.78221654417133e-08,
+      "loss": 18.3741,
+      "step": 20104
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.732889804834333e-08,
+      "loss": 17.7765,
+      "step": 20105
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.683687623703309e-08,
+      "loss": 0.7225,
+      "step": 20106
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.634610002007272e-08,
+      "loss": 15.8709,
+      "step": 20107
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.585656940970245e-08,
+      "loss": 18.4223,
+      "step": 20108
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.536828441815138e-08,
+      "loss": 17.6929,
+      "step": 20109
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.488124505760421e-08,
+      "loss": 0.7244,
+      "step": 20110
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.439545134021233e-08,
+      "loss": 0.8337,
+      "step": 20111
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.391090327811048e-08,
+      "loss": 15.5444,
+      "step": 20112
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.342760088337787e-08,
+      "loss": 17.3562,
+      "step": 20113
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.294554416809376e-08,
+      "loss": 0.7852,
+      "step": 20114
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.246473314428739e-08,
+      "loss": 0.7379,
+      "step": 20115
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.198516782394917e-08,
+      "loss": 15.8563,
+      "step": 20116
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.150684821906397e-08,
+      "loss": 15.8399,
+      "step": 20117
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.102977434155557e-08,
+      "loss": 17.7806,
+      "step": 20118
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.055394620334779e-08,
+      "loss": 0.7914,
+      "step": 20119
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.007936381630334e-08,
+      "loss": 15.951,
+      "step": 20120
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.960602719227384e-08,
+      "loss": 15.7852,
+      "step": 20121
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.913393634307765e-08,
+      "loss": 15.9054,
+      "step": 20122
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.866309128049421e-08,
+      "loss": 0.7548,
+      "step": 20123
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.819349201627525e-08,
+      "loss": 16.1424,
+      "step": 20124
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.772513856213916e-08,
+      "loss": 0.7888,
+      "step": 20125
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.725803092978768e-08,
+      "loss": 18.3151,
+      "step": 20126
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.679216913086707e-08,
+      "loss": 0.7751,
+      "step": 20127
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.632755317701247e-08,
+      "loss": 16.2938,
+      "step": 20128
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.586418307981459e-08,
+      "loss": 17.6257,
+      "step": 20129
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.54020588508475e-08,
+      "loss": 17.643,
+      "step": 20130
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.494118050164646e-08,
+      "loss": 17.814,
+      "step": 20131
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.448154804370223e-08,
+      "loss": 16.8564,
+      "step": 20132
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.402316148850009e-08,
+      "loss": 16.6163,
+      "step": 20133
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.356602084747534e-08,
+      "loss": 20.0291,
+      "step": 20134
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.311012613204106e-08,
+      "loss": 0.7567,
+      "step": 20135
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.265547735357703e-08,
+      "loss": 0.7475,
+      "step": 20136
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.220207452342977e-08,
+      "loss": 0.7879,
+      "step": 20137
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.17499176529124e-08,
+      "loss": 16.2608,
+      "step": 20138
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.129900675332147e-08,
+      "loss": 16.1255,
+      "step": 20139
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.084934183589799e-08,
+      "loss": 0.7509,
+      "step": 20140
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.04009229118774e-08,
+      "loss": 0.7439,
+      "step": 20141
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.99537499924452e-08,
+      "loss": 19.9732,
+      "step": 20142
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.950782308877025e-08,
+      "loss": 0.7341,
+      "step": 20143
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.906314221197697e-08,
+      "loss": 0.7311,
+      "step": 20144
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.861970737316204e-08,
+      "loss": 16.9172,
+      "step": 20145
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.817751858339995e-08,
+      "loss": 15.5019,
+      "step": 20146
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.773657585372629e-08,
+      "loss": 16.3681,
+      "step": 20147
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.729687919514894e-08,
+      "loss": 0.7298,
+      "step": 20148
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.68584286186369e-08,
+      "loss": 0.7978,
+      "step": 20149
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.642122413513697e-08,
+      "loss": 0.7959,
+      "step": 20150
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.598526575556264e-08,
+      "loss": 18.0573,
+      "step": 20151
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.555055349079964e-08,
+      "loss": 16.9633,
+      "step": 20152
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.511708735168932e-08,
+      "loss": 0.7789,
+      "step": 20153
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.468486734905633e-08,
+      "loss": 0.7635,
+      "step": 20154
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.42538934936865e-08,
+      "loss": 0.7781,
+      "step": 20155
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.382416579634344e-08,
+      "loss": 17.8237,
+      "step": 20156
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.339568426774634e-08,
+      "loss": 0.7376,
+      "step": 20157
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.296844891859223e-08,
+      "loss": 18.3791,
+      "step": 20158
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.254245975955033e-08,
+      "loss": 0.7997,
+      "step": 20159
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.211771680123991e-08,
+      "loss": 18.1853,
+      "step": 20160
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.169422005428028e-08,
+      "loss": 0.7242,
+      "step": 20161
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.12719695292241e-08,
+      "loss": 0.7461,
+      "step": 20162
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.08509652366296e-08,
+      "loss": 0.7659,
+      "step": 20163
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.043120718698837e-08,
+      "loss": 0.7542,
+      "step": 20164
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.001269539078647e-08,
+      "loss": 19.9528,
+      "step": 20165
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.959542985846557e-08,
+      "loss": 15.9064,
+      "step": 20166
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.917941060045064e-08,
+      "loss": 16.9669,
+      "step": 20167
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.876463762711117e-08,
+      "loss": 16.7018,
+      "step": 20168
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.835111094881108e-08,
+      "loss": 0.759,
+      "step": 20169
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.793883057586436e-08,
+      "loss": 0.746,
+      "step": 20170
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.752779651856834e-08,
+      "loss": 0.7928,
+      "step": 20171
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.711800878718144e-08,
+      "loss": 0.7716,
+      "step": 20172
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.670946739192885e-08,
+      "loss": 0.7416,
+      "step": 20173
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.63021723430024e-08,
+      "loss": 17.5842,
+      "step": 20174
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.589612365058284e-08,
+      "loss": 0.7882,
+      "step": 20175
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.54913213247954e-08,
+      "loss": 0.7608,
+      "step": 20176
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.508776537574867e-08,
+      "loss": 0.7501,
+      "step": 20177
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.468545581350682e-08,
+      "loss": 16.2981,
+      "step": 20178
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.428439264812292e-08,
+      "loss": 0.7384,
+      "step": 20179
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.388457588960006e-08,
+      "loss": 0.7525,
+      "step": 20180
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.348600554791917e-08,
+      "loss": 0.7532,
+      "step": 20181
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.308868163303339e-08,
+      "loss": 0.7626,
+      "step": 20182
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.269260415485145e-08,
+      "loss": 0.7518,
+      "step": 20183
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.229777312326545e-08,
+      "loss": 17.0726,
+      "step": 20184
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.190418854813417e-08,
+      "loss": 17.1419,
+      "step": 20185
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.151185043927199e-08,
+      "loss": 18.151,
+      "step": 20186
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.11207588064766e-08,
+      "loss": 17.0466,
+      "step": 20187
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.073091365951244e-08,
+      "loss": 17.97,
+      "step": 20188
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.034231500810505e-08,
+      "loss": 0.7447,
+      "step": 20189
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.995496286195779e-08,
+      "loss": 15.4713,
+      "step": 20190
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.956885723073513e-08,
+      "loss": 0.739,
+      "step": 20191
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.9183998124079377e-08,
+      "loss": 0.7159,
+      "step": 20192
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.880038555159395e-08,
+      "loss": 16.7072,
+      "step": 20193
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.841801952284898e-08,
+      "loss": 16.6725,
+      "step": 20194
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.8036900047397926e-08,
+      "loss": 0.8158,
+      "step": 20195
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.765702713474985e-08,
+      "loss": 17.2792,
+      "step": 20196
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.727840079438051e-08,
+      "loss": 0.7591,
+      "step": 20197
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.690102103574902e-08,
+      "loss": 0.7758,
+      "step": 20198
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.652488786827004e-08,
+      "loss": 15.4085,
+      "step": 20199
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.6150001301336074e-08,
+      "loss": 0.7423,
+      "step": 20200
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.5776361344300757e-08,
+      "loss": 16.2294,
+      "step": 20201
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.540396800648995e-08,
+      "loss": 0.759,
+      "step": 20202
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.503282129720178e-08,
+      "loss": 19.142,
+      "step": 20203
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.46629212256955e-08,
+      "loss": 19.2532,
+      "step": 20204
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.429426780121372e-08,
+      "loss": 0.7397,
+      "step": 20205
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.392686103294353e-08,
+      "loss": 17.0548,
+      "step": 20206
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.3560700930066485e-08,
+      "loss": 19.5424,
+      "step": 20207
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.319578750171972e-08,
+      "loss": 0.7135,
+      "step": 20208
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.2832120757007054e-08,
+      "loss": 17.7488,
+      "step": 20209
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.2469700705010115e-08,
+      "loss": 18.095,
+      "step": 20210
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.2108527354771675e-08,
+      "loss": 15.6753,
+      "step": 20211
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.174860071531229e-08,
+      "loss": 17.2599,
+      "step": 20212
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.1389920795613667e-08,
+      "loss": 17.7414,
+      "step": 20213
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.1032487604624204e-08,
+      "loss": 17.6674,
+      "step": 20214
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.067630115127009e-08,
+      "loss": 16.2587,
+      "step": 20215
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.032136144443866e-08,
+      "loss": 17.1561,
+      "step": 20216
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.9967668492995054e-08,
+      "loss": 0.7982,
+      "step": 20217
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.961522230575999e-08,
+      "loss": 18.7684,
+      "step": 20218
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.926402289153753e-08,
+      "loss": 18.3586,
+      "step": 20219
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.891407025908734e-08,
+      "loss": 0.7434,
+      "step": 20220
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.8565364417146874e-08,
+      "loss": 0.7802,
+      "step": 20221
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.821790537442028e-08,
+      "loss": 0.7735,
+      "step": 20222
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.7871693139583954e-08,
+      "loss": 0.6948,
+      "step": 20223
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.752672772126987e-08,
+      "loss": 0.7605,
+      "step": 20224
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.718300912809892e-08,
+      "loss": 16.3818,
+      "step": 20225
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.6840537368647576e-08,
+      "loss": 17.8964,
+      "step": 20226
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.6499312451459e-08,
+      "loss": 16.8384,
+      "step": 20227
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.6159334385054154e-08,
+      "loss": 0.7824,
+      "step": 20228
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.582060317791515e-08,
+      "loss": 17.6375,
+      "step": 20229
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.548311883850187e-08,
+      "loss": 17.1992,
+      "step": 20230
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.514688137523537e-08,
+      "loss": 16.703,
+      "step": 20231
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.4811890796508937e-08,
+      "loss": 17.5178,
+      "step": 20232
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.4478147110688095e-08,
+      "loss": 0.7623,
+      "step": 20233
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.4145650326093965e-08,
+      "loss": 0.8017,
+      "step": 20234
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.3814400451036574e-08,
+      "loss": 16.8724,
+      "step": 20235
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.3484397493770424e-08,
+      "loss": 17.1455,
+      "step": 20236
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.315564146254447e-08,
+      "loss": 0.7009,
+      "step": 20237
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.282813236556327e-08,
+      "loss": 0.7633,
+      "step": 20238
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.25018702109925e-08,
+      "loss": 16.765,
+      "step": 20239
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.217685500698676e-08,
+      "loss": 0.7255,
+      "step": 20240
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.1853086761650675e-08,
+      "loss": 0.7681,
+      "step": 20241
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.1530565483072214e-08,
+      "loss": 17.1772,
+      "step": 20242
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.120929117929495e-08,
+      "loss": 0.7112,
+      "step": 20243
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.0889263858340246e-08,
+      "loss": 15.81,
+      "step": 20244
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.0570483528196144e-08,
+      "loss": 0.754,
+      "step": 20245
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.0252950196822955e-08,
+      "loss": 0.7558,
+      "step": 20246
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.9936663872142125e-08,
+      "loss": 0.7596,
+      "step": 20247
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.9621624562047324e-08,
+      "loss": 16.8765,
+      "step": 20248
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.9307832274404486e-08,
+      "loss": 17.2158,
+      "step": 20249
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.899528701704624e-08,
+      "loss": 0.7665,
+      "step": 20250
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.868398879777746e-08,
+      "loss": 0.7313,
+      "step": 20251
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.837393762435859e-08,
+      "loss": 19.0919,
+      "step": 20252
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.8065133504533444e-08,
+      "loss": 18.8895,
+      "step": 20253
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.775757644601807e-08,
+      "loss": 0.7264,
+      "step": 20254
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.745126645647301e-08,
+      "loss": 16.5157,
+      "step": 20255
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.71462035435588e-08,
+      "loss": 0.6884,
+      "step": 20256
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.684238771488047e-08,
+      "loss": 0.8153,
+      "step": 20257
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.653981897802083e-08,
+      "loss": 17.7971,
+      "step": 20258
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.6238497340540524e-08,
+      "loss": 0.7164,
+      "step": 20259
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.593842280995019e-08,
+      "loss": 17.6913,
+      "step": 20260
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.5639595393749393e-08,
+      "loss": 16.7295,
+      "step": 20261
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.5342015099393276e-08,
+      "loss": 0.8064,
+      "step": 20262
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.504568193430369e-08,
+      "loss": 0.7183,
+      "step": 20263
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.475059590589136e-08,
+      "loss": 0.7926,
+      "step": 20264
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.445675702150597e-08,
+      "loss": 17.4324,
+      "step": 20265
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.416416528848609e-08,
+      "loss": 17.0245,
+      "step": 20266
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.387282071414255e-08,
+      "loss": 17.8108,
+      "step": 20267
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.358272330574175e-08,
+      "loss": 18.8004,
+      "step": 20268
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.329387307052234e-08,
+      "loss": 16.0321,
+      "step": 20269
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.300627001570078e-08,
+      "loss": 16.9503,
+      "step": 20270
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.2719914148449105e-08,
+      "loss": 19.0958,
+      "step": 20271
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.2434805475917154e-08,
+      "loss": 19.304,
+      "step": 20272
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.215094400522145e-08,
+      "loss": 17.8899,
+      "step": 20273
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.1868329743445225e-08,
+      "loss": 0.7084,
+      "step": 20274
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.1586962697649493e-08,
+      "loss": 16.8829,
+      "step": 20275
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.130684287485086e-08,
+      "loss": 0.7665,
+      "step": 20276
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.1027970282043736e-08,
+      "loss": 18.5116,
+      "step": 20277
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.0750344926183674e-08,
+      "loss": 17.1299,
+      "step": 20278
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.0473966814209555e-08,
+      "loss": 0.7595,
+      "step": 20279
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.0198835953010316e-08,
+      "loss": 19.5994,
+      "step": 20280
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.992495234945825e-08,
+      "loss": 0.7446,
+      "step": 20281
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.965231601039231e-08,
+      "loss": 0.7419,
+      "step": 20282
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.938092694261263e-08,
+      "loss": 18.6926,
+      "step": 20283
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.911078515289156e-08,
+      "loss": 0.762,
+      "step": 20284
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8841890647979264e-08,
+      "loss": 17.0894,
+      "step": 20285
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8574243434581483e-08,
+      "loss": 15.5212,
+      "step": 20286
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.830784351938176e-08,
+      "loss": 0.7367,
+      "step": 20287
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8042690909024783e-08,
+      "loss": 16.7963,
+      "step": 20288
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.777878561013858e-08,
+      "loss": 18.3151,
+      "step": 20289
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.7516127629301226e-08,
+      "loss": 0.7438,
+      "step": 20290
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.7254716973068585e-08,
+      "loss": 0.7472,
+      "step": 20291
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.699455364797432e-08,
+      "loss": 15.7426,
+      "step": 20292
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.6735637660502133e-08,
+      "loss": 0.7905,
+      "step": 20293
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.6477969017119076e-08,
+      "loss": 15.8561,
+      "step": 20294
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.6221547724253337e-08,
+      "loss": 0.7289,
+      "step": 20295
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5966373788310905e-08,
+      "loss": 18.0043,
+      "step": 20296
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.571244721566446e-08,
+      "loss": 0.7819,
+      "step": 20297
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.545976801263672e-08,
+      "loss": 16.3679,
+      "step": 20298
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5208336185550406e-08,
+      "loss": 17.315,
+      "step": 20299
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.495815174067273e-08,
+      "loss": 0.7698,
+      "step": 20300
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.4709214684248695e-08,
+      "loss": 18.836,
+      "step": 20301
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.4461525022489994e-08,
+      "loss": 17.854,
+      "step": 20302
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.421508276159168e-08,
+      "loss": 17.5221,
+      "step": 20303
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.396988790768773e-08,
+      "loss": 17.4435,
+      "step": 20304
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.3725940466912123e-08,
+      "loss": 16.9089,
+      "step": 20305
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.3483240445343334e-08,
+      "loss": 0.6968,
+      "step": 20306
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.3241787849048736e-08,
+      "loss": 0.7747,
+      "step": 20307
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.300158268404573e-08,
+      "loss": 16.7809,
+      "step": 20308
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.276262495634063e-08,
+      "loss": 16.7173,
+      "step": 20309
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.2524914671884223e-08,
+      "loss": 0.7563,
+      "step": 20310
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.228845183662176e-08,
+      "loss": 0.7501,
+      "step": 20311
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.2053236456448524e-08,
+      "loss": 0.7781,
+      "step": 20312
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.181926853724314e-08,
+      "loss": 17.5417,
+      "step": 20313
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1586548084834292e-08,
+      "loss": 17.1095,
+      "step": 20314
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1355075105039536e-08,
+      "loss": 16.7748,
+      "step": 20315
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1124849603632035e-08,
+      "loss": 17.4463,
+      "step": 20316
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0895871586362746e-08,
+      "loss": 0.7603,
+      "step": 20317
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0668141058943768e-08,
+      "loss": 18.0539,
+      "step": 20318
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.044165802705944e-08,
+      "loss": 0.7722,
+      "step": 20319
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0216422496360797e-08,
+      "loss": 0.749,
+      "step": 20320
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.999243447247112e-08,
+      "loss": 0.7644,
+      "step": 20321
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9769693960985936e-08,
+      "loss": 16.5624,
+      "step": 20322
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9548200967461906e-08,
+      "loss": 15.2054,
+      "step": 20323
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.932795549742239e-08,
+      "loss": 0.7519,
+      "step": 20324
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9108957556374096e-08,
+      "loss": 17.1798,
+      "step": 20325
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8891207149773772e-08,
+      "loss": 0.7759,
+      "step": 20326
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8674704283067057e-08,
+      "loss": 18.0709,
+      "step": 20327
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.845944896164409e-08,
+      "loss": 16.2494,
+      "step": 20328
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8245441190889447e-08,
+      "loss": 0.8391,
+      "step": 20329
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.80326809761433e-08,
+      "loss": 15.2361,
+      "step": 20330
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7821168322712524e-08,
+      "loss": 0.7821,
+      "step": 20331
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.761090323587622e-08,
+      "loss": 0.8007,
+      "step": 20332
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7401885720885746e-08,
+      "loss": 16.7217,
+      "step": 20333
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7194115782953603e-08,
+      "loss": 16.592,
+      "step": 20334
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6987593427270076e-08,
+      "loss": 0.746,
+      "step": 20335
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.678231865899216e-08,
+      "loss": 0.7329,
+      "step": 20336
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6578291483237972e-08,
+      "loss": 0.7985,
+      "step": 20337
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6375511905103445e-08,
+      "loss": 16.4429,
+      "step": 20338
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6173979929645645e-08,
+      "loss": 0.7538,
+      "step": 20339
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5973695561899426e-08,
+      "loss": 0.7022,
+      "step": 20340
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.57746588068608e-08,
+      "loss": 17.5244,
+      "step": 20341
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.557686966950356e-08,
+      "loss": 16.5886,
+      "step": 20342
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5380328154757096e-08,
+      "loss": 0.7627,
+      "step": 20343
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.518503426753415e-08,
+      "loss": 17.827,
+      "step": 20344
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4990988012703044e-08,
+      "loss": 17.5402,
+      "step": 20345
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4798189395115458e-08,
+      "loss": 0.7606,
+      "step": 20346
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4606638419573104e-08,
+      "loss": 0.8094,
+      "step": 20347
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4416335090861045e-08,
+      "loss": 0.744,
+      "step": 20348
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4227279413731031e-08,
+      "loss": 0.7848,
+      "step": 20349
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4039471392901515e-08,
+      "loss": 0.7483,
+      "step": 20350
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3852911033063188e-08,
+      "loss": 19.2952,
+      "step": 20351
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3667598338867882e-08,
+      "loss": 0.7243,
+      "step": 20352
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.348353331494523e-08,
+      "loss": 19.2164,
+      "step": 20353
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.330071596588045e-08,
+      "loss": 0.7342,
+      "step": 20354
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3119146296247664e-08,
+      "loss": 0.7866,
+      "step": 20355
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.293882431057658e-08,
+      "loss": 0.732,
+      "step": 20356
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2759750013358052e-08,
+      "loss": 0.7751,
+      "step": 20357
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.258192340907738e-08,
+      "loss": 17.8384,
+      "step": 20358
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2405344502164351e-08,
+      "loss": 0.7936,
+      "step": 20359
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2230013297026554e-08,
+      "loss": 18.1219,
+      "step": 20360
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2055929798038268e-08,
+      "loss": 16.8726,
+      "step": 20361
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1883094009551565e-08,
+      "loss": 18.4392,
+      "step": 20362
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1711505935874112e-08,
+      "loss": 0.8043,
+      "step": 20363
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1541165581291369e-08,
+      "loss": 0.7947,
+      "step": 20364
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.137207295005549e-08,
+      "loss": 0.7783,
+      "step": 20365
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1204228046390875e-08,
+      "loss": 0.7449,
+      "step": 20366
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1037630874477512e-08,
+      "loss": 17.9873,
+      "step": 20367
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0872281438478738e-08,
+      "loss": 15.1926,
+      "step": 20368
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0708179742524583e-08,
+      "loss": 0.7346,
+      "step": 20369
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0545325790706218e-08,
+      "loss": 15.7424,
+      "step": 20370
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0383719587092611e-08,
+      "loss": 18.4133,
+      "step": 20371
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0223361135713871e-08,
+      "loss": 0.6986,
+      "step": 20372
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0064250440572354e-08,
+      "loss": 0.727,
+      "step": 20373
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.906387505648207e-09,
+      "loss": 0.7008,
+      "step": 20374
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.74977233486607e-09,
+      "loss": 17.8899,
+      "step": 20375
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.59440493215058e-09,
+      "loss": 18.02,
+      "step": 20376
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.440285301370865e-09,
+      "loss": 0.7673,
+      "step": 20377
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.28741344637385e-09,
+      "loss": 17.6058,
+      "step": 20378
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.135789370978698e-09,
+      "loss": 0.787,
+      "step": 20379
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.985413078971272e-09,
+      "loss": 0.7546,
+      "step": 20380
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.836284574098575e-09,
+      "loss": 0.7636,
+      "step": 20381
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.688403860090955e-09,
+      "loss": 0.7582,
+      "step": 20382
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.541770940628802e-09,
+      "loss": 17.4834,
+      "step": 20383
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.396385819381403e-09,
+      "loss": 15.5738,
+      "step": 20384
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.252248499979186e-09,
+      "loss": 16.4851,
+      "step": 20385
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.109358986013726e-09,
+      "loss": 0.7416,
+      "step": 20386
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.967717281048837e-09,
+      "loss": 17.2514,
+      "step": 20387
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.82732338862613e-09,
+      "loss": 0.7381,
+      "step": 20388
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.688177312248357e-09,
+      "loss": 0.7675,
+      "step": 20389
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.55027905539052e-09,
+      "loss": 17.5992,
+      "step": 20390
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.413628621488755e-09,
+      "loss": 0.7482,
+      "step": 20391
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.278226013962552e-09,
+      "loss": 16.826,
+      "step": 20392
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.144071236181438e-09,
+      "loss": 0.8099,
+      "step": 20393
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.011164291503835e-09,
+      "loss": 16.1097,
+      "step": 20394
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.879505183238211e-09,
+      "loss": 0.7351,
+      "step": 20395
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.7490939146819256e-09,
+      "loss": 0.7927,
+      "step": 20396
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.619930489076831e-09,
+      "loss": 15.2357,
+      "step": 20397
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.492014909653677e-09,
+      "loss": 0.7833,
+      "step": 20398
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.365347179604353e-09,
+      "loss": 15.6162,
+      "step": 20399
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.239927302092996e-09,
+      "loss": 0.7556,
+      "step": 20400
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.115755280250435e-09,
+      "loss": 17.078,
+      "step": 20401
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.99283111716864e-09,
+      "loss": 0.784,
+      "step": 20402
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.8711548159229305e-09,
+      "loss": 18.8016,
+      "step": 20403
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.750726379549764e-09,
+      "loss": 16.9695,
+      "step": 20404
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.631545811052296e-09,
+      "loss": 0.7978,
+      "step": 20405
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.513613113405924e-09,
+      "loss": 16.9293,
+      "step": 20406
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.39692828955829e-09,
+      "loss": 0.7622,
+      "step": 20407
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.281491342418176e-09,
+      "loss": 0.7327,
+      "step": 20408
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.167302274866615e-09,
+      "loss": 0.7079,
+      "step": 20409
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.054361089751325e-09,
+      "loss": 0.7846,
+      "step": 20410
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.942667789897826e-09,
+      "loss": 0.7708,
+      "step": 20411
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.832222378087226e-09,
+      "loss": 16.9661,
+      "step": 20412
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.72302485708398e-09,
+      "loss": 0.8131,
+      "step": 20413
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.615075229602583e-09,
+      "loss": 16.0785,
+      "step": 20414
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.508373498351981e-09,
+      "loss": 20.8175,
+      "step": 20415
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.402919665985605e-09,
+      "loss": 0.7347,
+      "step": 20416
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.298713735134685e-09,
+      "loss": 0.741,
+      "step": 20417
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.1957557084082445e-09,
+      "loss": 17.4726,
+      "step": 20418
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.094045588365347e-09,
+      "loss": 0.7193,
+      "step": 20419
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.993583377559507e-09,
+      "loss": 0.7778,
+      "step": 20420
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.894369078483173e-09,
+      "loss": 0.696,
+      "step": 20421
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.796402693617695e-09,
+      "loss": 0.718,
+      "step": 20422
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.6996842254111153e-09,
+      "loss": 0.7293,
+      "step": 20423
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.6042136762781676e-09,
+      "loss": 0.7311,
+      "step": 20424
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.5099910486002807e-09,
+      "loss": 0.7456,
+      "step": 20425
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.4170163447255766e-09,
+      "loss": 0.8066,
+      "step": 20426
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.325289566979972e-09,
+      "loss": 0.7645,
+      "step": 20427
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.234810717644976e-09,
+      "loss": 0.7615,
+      "step": 20428
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.145579798990994e-09,
+      "loss": 13.9696,
+      "step": 20429
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.0575968132384724e-09,
+      "loss": 18.2898,
+      "step": 20430
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9708617625856526e-09,
+      "loss": 16.7503,
+      "step": 20431
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.8853746491919187e-09,
+      "loss": 0.7394,
+      "step": 20432
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.80113547519445e-09,
+      "loss": 0.7438,
+      "step": 20433
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7181442427026693e-09,
+      "loss": 17.7432,
+      "step": 20434
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.6364009537760415e-09,
+      "loss": 0.7237,
+      "step": 20435
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.5559056104629277e-09,
+      "loss": 0.788,
+      "step": 20436
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.4766582147728314e-09,
+      "loss": 17.3408,
+      "step": 20437
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.3986587686763983e-09,
+      "loss": 16.063,
+      "step": 20438
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.3219072741276217e-09,
+      "loss": 17.9007,
+      "step": 20439
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.2464037330416352e-09,
+      "loss": 0.7578,
+      "step": 20440
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.172148147300268e-09,
+      "loss": 0.7786,
+      "step": 20441
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0991405187575918e-09,
+      "loss": 17.0835,
+      "step": 20442
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0273808492343727e-09,
+      "loss": 0.8331,
+      "step": 20443
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9568691405236206e-09,
+      "loss": 19.6103,
+      "step": 20444
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.88760539439059e-09,
+      "loss": 0.7285,
+      "step": 20445
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8195896125561273e-09,
+      "loss": 15.9563,
+      "step": 20446
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7528217967188732e-09,
+      "loss": 18.1767,
+      "step": 20447
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.6873019485497133e-09,
+      "loss": 0.7251,
+      "step": 20448
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.6230300696806754e-09,
+      "loss": 0.7219,
+      "step": 20449
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5600061617160322e-09,
+      "loss": 17.236,
+      "step": 20450
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4982302262323e-09,
+      "loss": 16.4231,
+      "step": 20451
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4377022647671379e-09,
+      "loss": 15.9952,
+      "step": 20452
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.378422278830449e-09,
+      "loss": 0.7699,
+      "step": 20453
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3203902699099324e-09,
+      "loss": 18.3157,
+      "step": 20454
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2636062394433268e-09,
+      "loss": 19.0213,
+      "step": 20455
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2080701888572688e-09,
+      "loss": 0.7755,
+      "step": 20456
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.153782119533986e-09,
+      "loss": 17.7574,
+      "step": 20457
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1007420328279506e-09,
+      "loss": 0.7494,
+      "step": 20458
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0489499300603278e-09,
+      "loss": 18.2688,
+      "step": 20459
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.984058125300787e-10,
+      "loss": 0.6918,
+      "step": 20460
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.491096814973066e-10,
+      "loss": 0.6963,
+      "step": 20461
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.010615381888076e-10,
+      "loss": 16.1385,
+      "step": 20462
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.542613838036229e-10,
+      "loss": 0.7639,
+      "step": 20463
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.087092195185886e-10,
+      "loss": 18.1357,
+      "step": 20464
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.644050464605812e-10,
+      "loss": 0.7501,
+      "step": 20465
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.213488657398237e-10,
+      "loss": 17.0901,
+      "step": 20466
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.795406784332326e-10,
+      "loss": 0.7335,
+      "step": 20467
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.389804855788662e-10,
+      "loss": 17.942,
+      "step": 20468
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.996682881870275e-10,
+      "loss": 0.7481,
+      "step": 20469
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.616040872513661e-10,
+      "loss": 0.8269,
+      "step": 20470
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.247878837100206e-10,
+      "loss": 0.7997,
+      "step": 20471
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.892196784900271e-10,
+      "loss": 0.6876,
+      "step": 20472
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.54899472474013e-10,
+      "loss": 0.781,
+      "step": 20473
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.218272665168499e-10,
+      "loss": 0.7284,
+      "step": 20474
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.9000306145120513e-10,
+      "loss": 17.8693,
+      "step": 20475
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.5942685807088814e-10,
+      "loss": 0.7676,
+      "step": 20476
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.3009865713085064e-10,
+      "loss": 15.7483,
+      "step": 20477
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.0201845936383976e-10,
+      "loss": 16.4467,
+      "step": 20478
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.751862654803983e-10,
+      "loss": 0.7551,
+      "step": 20479
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.4960207614666e-10,
+      "loss": 16.0833,
+      "step": 20480
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.25265891995452e-10,
+      "loss": 18.1489,
+      "step": 20481
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.02177713637397e-10,
+      "loss": 16.901,
+      "step": 20482
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8033754164981098e-10,
+      "loss": 0.7622,
+      "step": 20483
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.597453765822543e-10,
+      "loss": 14.8928,
+      "step": 20484
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4040121893987846e-10,
+      "loss": 0.7827,
+      "step": 20485
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2230506920563046e-10,
+      "loss": 17.7174,
+      "step": 20486
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0545692784025285e-10,
+      "loss": 17.8713,
+      "step": 20487
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.985679526007929e-11,
+      "loss": 0.7595,
+      "step": 20488
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.550467185368781e-11,
+      "loss": 0.7624,
+      "step": 20489
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.24005579763498e-11,
+      "loss": 0.7572,
+      "step": 20490
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.054445395558105e-11,
+      "loss": 0.7799,
+      "step": 20491
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.993636009669288e-11,
+      "loss": 18.7057,
+      "step": 20492
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.057627664948548e-11,
+      "loss": 0.7628,
+      "step": 20493
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.2464203863759026e-11,
+      "loss": 17.2996,
+      "step": 20494
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5600141928251434e-11,
+      "loss": 0.7332,
+      "step": 20495
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.984091020598386e-12,
+      "loss": 0.7183,
+      "step": 20496
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.6160512795777605e-12,
+      "loss": 0.7276,
+      "step": 20497
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.4960228162118624e-12,
+      "loss": 19.1638,
+      "step": 20498
+    },
+    {
+      "epoch": 1.0,
+      "step": 20498,
+      "total_flos": 7.41694861365513e+20,
+      "train_loss": 140.81203460384256,
+      "train_runtime": 155343.9584,
+      "train_samples_per_second": 25.337,
+      "train_steps_per_second": 0.132
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 20500,
+  "num_train_epochs": 1,
+  "save_steps": 1000,
+  "total_flos": 7.41694861365513e+20,
+  "trial_name": null,
+  "trial_params": null
+}