diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,88722 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "eval_steps": 500,
+  "global_step": 14782,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.504504504504505e-08,
+      "loss": 0.6833,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.00900900900901e-08,
+      "loss": 1.457,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3513513513513515e-07,
+      "loss": 0.74,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.801801801801802e-07,
+      "loss": 1.4741,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2522522522522524e-07,
+      "loss": 0.69,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.702702702702703e-07,
+      "loss": 1.4181,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.153153153153153e-07,
+      "loss": 0.6507,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.603603603603604e-07,
+      "loss": 1.4396,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.0540540540540546e-07,
+      "loss": 1.417,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.504504504504505e-07,
+      "loss": 1.446,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.954954954954956e-07,
+      "loss": 1.4679,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.405405405405406e-07,
+      "loss": 1.4074,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.855855855855856e-07,
+      "loss": 1.4067,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.306306306306306e-07,
+      "loss": 0.6538,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.756756756756758e-07,
+      "loss": 1.4149,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.207207207207208e-07,
+      "loss": 1.3854,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.657657657657658e-07,
+      "loss": 1.3775,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.108108108108109e-07,
+      "loss": 1.3297,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.55855855855856e-07,
+      "loss": 0.6855,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.00900900900901e-07,
+      "loss": 1.4059,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.459459459459461e-07,
+      "loss": 0.6884,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.909909909909911e-07,
+      "loss": 1.3885,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0360360360360361e-06,
+      "loss": 1.3595,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0810810810810812e-06,
+      "loss": 1.3063,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1261261261261262e-06,
+      "loss": 0.7016,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1711711711711712e-06,
+      "loss": 1.3171,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2162162162162164e-06,
+      "loss": 1.3749,
+      "step": 27
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2612612612612613e-06,
+      "loss": 1.2981,
+      "step": 28
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3063063063063065e-06,
+      "loss": 1.2472,
+      "step": 29
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3513513513513515e-06,
+      "loss": 1.3135,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3963963963963963e-06,
+      "loss": 1.316,
+      "step": 31
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4414414414414416e-06,
+      "loss": 1.3363,
+      "step": 32
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4864864864864868e-06,
+      "loss": 1.3089,
+      "step": 33
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5315315315315316e-06,
+      "loss": 1.2976,
+      "step": 34
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5765765765765766e-06,
+      "loss": 1.2252,
+      "step": 35
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6216216216216219e-06,
+      "loss": 1.2427,
+      "step": 36
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6666666666666667e-06,
+      "loss": 1.2497,
+      "step": 37
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.711711711711712e-06,
+      "loss": 1.255,
+      "step": 38
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.756756756756757e-06,
+      "loss": 0.6539,
+      "step": 39
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.801801801801802e-06,
+      "loss": 1.252,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.846846846846847e-06,
+      "loss": 0.6266,
+      "step": 41
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8918918918918922e-06,
+      "loss": 1.2665,
+      "step": 42
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9369369369369372e-06,
+      "loss": 1.2481,
+      "step": 43
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.9819819819819822e-06,
+      "loss": 1.1746,
+      "step": 44
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0270270270270273e-06,
+      "loss": 1.2215,
+      "step": 45
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0720720720720723e-06,
+      "loss": 1.2374,
+      "step": 46
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1171171171171173e-06,
+      "loss": 1.2151,
+      "step": 47
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1621621621621623e-06,
+      "loss": 1.2059,
+      "step": 48
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2072072072072073e-06,
+      "loss": 1.189,
+      "step": 49
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2522522522522524e-06,
+      "loss": 1.1983,
+      "step": 50
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.297297297297298e-06,
+      "loss": 1.1874,
+      "step": 51
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3423423423423424e-06,
+      "loss": 1.1667,
+      "step": 52
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3873873873873874e-06,
+      "loss": 1.2376,
+      "step": 53
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.432432432432433e-06,
+      "loss": 0.594,
+      "step": 54
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.4774774774774775e-06,
+      "loss": 1.2235,
+      "step": 55
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5225225225225225e-06,
+      "loss": 1.1987,
+      "step": 56
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5675675675675675e-06,
+      "loss": 1.1628,
+      "step": 57
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.612612612612613e-06,
+      "loss": 1.2283,
+      "step": 58
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.657657657657658e-06,
+      "loss": 1.1663,
+      "step": 59
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.702702702702703e-06,
+      "loss": 1.1823,
+      "step": 60
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.747747747747748e-06,
+      "loss": 0.5836,
+      "step": 61
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7927927927927926e-06,
+      "loss": 0.5982,
+      "step": 62
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.837837837837838e-06,
+      "loss": 1.1183,
+      "step": 63
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.882882882882883e-06,
+      "loss": 1.1116,
+      "step": 64
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.927927927927928e-06,
+      "loss": 1.1305,
+      "step": 65
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9729729729729736e-06,
+      "loss": 1.1745,
+      "step": 66
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.0180180180180186e-06,
+      "loss": 1.1524,
+      "step": 67
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.063063063063063e-06,
+      "loss": 0.6141,
+      "step": 68
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.1081081081081082e-06,
+      "loss": 0.5807,
+      "step": 69
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.1531531531531532e-06,
+      "loss": 1.1482,
+      "step": 70
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.1981981981981987e-06,
+      "loss": 1.1478,
+      "step": 71
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.2432432432432437e-06,
+      "loss": 1.1448,
+      "step": 72
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.2882882882882887e-06,
+      "loss": 0.5883,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3333333333333333e-06,
+      "loss": 1.1232,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.3783783783783788e-06,
+      "loss": 1.1141,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.423423423423424e-06,
+      "loss": 1.0783,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.468468468468469e-06,
+      "loss": 1.1464,
+      "step": 77
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.513513513513514e-06,
+      "loss": 1.1324,
+      "step": 78
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.5585585585585584e-06,
+      "loss": 1.122,
+      "step": 79
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.603603603603604e-06,
+      "loss": 1.1008,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.648648648648649e-06,
+      "loss": 1.1206,
+      "step": 81
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.693693693693694e-06,
+      "loss": 1.199,
+      "step": 82
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7387387387387394e-06,
+      "loss": 0.6439,
+      "step": 83
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7837837837837844e-06,
+      "loss": 0.5962,
+      "step": 84
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.828828828828829e-06,
+      "loss": 1.1115,
+      "step": 85
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.8738738738738744e-06,
+      "loss": 1.0961,
+      "step": 86
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.918918918918919e-06,
+      "loss": 1.0818,
+      "step": 87
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.9639639639639645e-06,
+      "loss": 0.609,
+      "step": 88
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.009009009009009e-06,
+      "loss": 1.0876,
+      "step": 89
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.0540540540540545e-06,
+      "loss": 0.5566,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.099099099099099e-06,
+      "loss": 1.1085,
+      "step": 91
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.1441441441441446e-06,
+      "loss": 0.5679,
+      "step": 92
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.189189189189189e-06,
+      "loss": 1.0802,
+      "step": 93
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.234234234234235e-06,
+      "loss": 1.0762,
+      "step": 94
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.27927927927928e-06,
+      "loss": 1.1263,
+      "step": 95
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.324324324324325e-06,
+      "loss": 1.0997,
+      "step": 96
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.369369369369369e-06,
+      "loss": 1.0761,
+      "step": 97
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.414414414414415e-06,
+      "loss": 1.0615,
+      "step": 98
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.45945945945946e-06,
+      "loss": 1.1107,
+      "step": 99
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.504504504504505e-06,
+      "loss": 1.0899,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.54954954954955e-06,
+      "loss": 1.0932,
+      "step": 101
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.594594594594596e-06,
+      "loss": 1.0784,
+      "step": 102
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.63963963963964e-06,
+      "loss": 1.0862,
+      "step": 103
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.684684684684685e-06,
+      "loss": 1.0684,
+      "step": 104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.72972972972973e-06,
+      "loss": 1.06,
+      "step": 105
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.774774774774775e-06,
+      "loss": 1.1246,
+      "step": 106
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.81981981981982e-06,
+      "loss": 1.0681,
+      "step": 107
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.864864864864866e-06,
+      "loss": 0.5572,
+      "step": 108
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.90990990990991e-06,
+      "loss": 1.1135,
+      "step": 109
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.954954954954955e-06,
+      "loss": 1.0791,
+      "step": 110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5e-06,
+      "loss": 0.5739,
+      "step": 111
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.045045045045045e-06,
+      "loss": 1.0823,
+      "step": 112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.0900900900900905e-06,
+      "loss": 1.0908,
+      "step": 113
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.135135135135135e-06,
+      "loss": 1.0825,
+      "step": 114
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.180180180180181e-06,
+      "loss": 1.0757,
+      "step": 115
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.225225225225226e-06,
+      "loss": 1.0921,
+      "step": 116
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.2702702702702705e-06,
+      "loss": 1.0933,
+      "step": 117
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.315315315315316e-06,
+      "loss": 1.0765,
+      "step": 118
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.360360360360361e-06,
+      "loss": 1.068,
+      "step": 119
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.405405405405406e-06,
+      "loss": 1.0521,
+      "step": 120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.450450450450451e-06,
+      "loss": 1.0788,
+      "step": 121
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.495495495495496e-06,
+      "loss": 1.1092,
+      "step": 122
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.540540540540541e-06,
+      "loss": 1.0251,
+      "step": 123
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.585585585585585e-06,
+      "loss": 1.1167,
+      "step": 124
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.6306306306306316e-06,
+      "loss": 1.0492,
+      "step": 125
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.675675675675676e-06,
+      "loss": 1.0519,
+      "step": 126
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.720720720720722e-06,
+      "loss": 0.5496,
+      "step": 127
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.765765765765766e-06,
+      "loss": 1.0631,
+      "step": 128
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.810810810810811e-06,
+      "loss": 1.0674,
+      "step": 129
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.855855855855856e-06,
+      "loss": 1.0335,
+      "step": 130
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.900900900900901e-06,
+      "loss": 1.0548,
+      "step": 131
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.945945945945947e-06,
+      "loss": 0.5331,
+      "step": 132
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.990990990990992e-06,
+      "loss": 0.5431,
+      "step": 133
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.036036036036037e-06,
+      "loss": 1.0838,
+      "step": 134
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.081081081081082e-06,
+      "loss": 1.058,
+      "step": 135
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.126126126126126e-06,
+      "loss": 1.054,
+      "step": 136
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.171171171171172e-06,
+      "loss": 0.5661,
+      "step": 137
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.2162162162162164e-06,
+      "loss": 1.0904,
+      "step": 138
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.261261261261262e-06,
+      "loss": 1.0193,
+      "step": 139
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.3063063063063065e-06,
+      "loss": 1.0392,
+      "step": 140
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.351351351351351e-06,
+      "loss": 0.5626,
+      "step": 141
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.396396396396397e-06,
+      "loss": 1.0428,
+      "step": 142
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.441441441441442e-06,
+      "loss": 1.0705,
+      "step": 143
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.486486486486487e-06,
+      "loss": 1.0349,
+      "step": 144
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.531531531531532e-06,
+      "loss": 1.0368,
+      "step": 145
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.5765765765765775e-06,
+      "loss": 1.0527,
+      "step": 146
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.621621621621622e-06,
+      "loss": 0.5669,
+      "step": 147
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 1.0262,
+      "step": 148
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.711711711711713e-06,
+      "loss": 0.9785,
+      "step": 149
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.7567567567567575e-06,
+      "loss": 1.0605,
+      "step": 150
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.801801801801803e-06,
+      "loss": 1.073,
+      "step": 151
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.846846846846848e-06,
+      "loss": 1.0669,
+      "step": 152
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.891891891891892e-06,
+      "loss": 0.9806,
+      "step": 153
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.936936936936938e-06,
+      "loss": 0.5566,
+      "step": 154
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.981981981981982e-06,
+      "loss": 1.031,
+      "step": 155
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.027027027027028e-06,
+      "loss": 1.0296,
+      "step": 156
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.072072072072072e-06,
+      "loss": 0.5762,
+      "step": 157
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.117117117117117e-06,
+      "loss": 0.5567,
+      "step": 158
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.162162162162163e-06,
+      "loss": 1.066,
+      "step": 159
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.207207207207208e-06,
+      "loss": 1.0309,
+      "step": 160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.252252252252253e-06,
+      "loss": 1.0068,
+      "step": 161
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.297297297297298e-06,
+      "loss": 1.0778,
+      "step": 162
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.342342342342343e-06,
+      "loss": 0.5598,
+      "step": 163
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.387387387387388e-06,
+      "loss": 1.0166,
+      "step": 164
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.4324324324324324e-06,
+      "loss": 1.0483,
+      "step": 165
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.477477477477479e-06,
+      "loss": 1.0247,
+      "step": 166
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.5225225225225225e-06,
+      "loss": 1.0287,
+      "step": 167
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.567567567567569e-06,
+      "loss": 1.0508,
+      "step": 168
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.612612612612613e-06,
+      "loss": 1.02,
+      "step": 169
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.657657657657658e-06,
+      "loss": 1.0599,
+      "step": 170
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.702702702702704e-06,
+      "loss": 1.0189,
+      "step": 171
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.747747747747749e-06,
+      "loss": 1.0141,
+      "step": 172
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.792792792792793e-06,
+      "loss": 1.0132,
+      "step": 173
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.837837837837838e-06,
+      "loss": 1.0314,
+      "step": 174
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.882882882882884e-06,
+      "loss": 0.5657,
+      "step": 175
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.927927927927929e-06,
+      "loss": 0.9928,
+      "step": 176
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.972972972972974e-06,
+      "loss": 0.9691,
+      "step": 177
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.018018018018018e-06,
+      "loss": 0.5244,
+      "step": 178
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.063063063063063e-06,
+      "loss": 0.5397,
+      "step": 179
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.108108108108109e-06,
+      "loss": 1.0095,
+      "step": 180
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.153153153153154e-06,
+      "loss": 1.0439,
+      "step": 181
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.198198198198198e-06,
+      "loss": 1.0696,
+      "step": 182
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.243243243243245e-06,
+      "loss": 1.0477,
+      "step": 183
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.288288288288289e-06,
+      "loss": 0.5191,
+      "step": 184
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.333333333333334e-06,
+      "loss": 1.0074,
+      "step": 185
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.378378378378378e-06,
+      "loss": 1.0358,
+      "step": 186
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.423423423423423e-06,
+      "loss": 1.0051,
+      "step": 187
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.46846846846847e-06,
+      "loss": 1.0214,
+      "step": 188
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.513513513513514e-06,
+      "loss": 0.5544,
+      "step": 189
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.55855855855856e-06,
+      "loss": 0.5783,
+      "step": 190
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.603603603603605e-06,
+      "loss": 1.042,
+      "step": 191
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.64864864864865e-06,
+      "loss": 0.9997,
+      "step": 192
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.693693693693694e-06,
+      "loss": 0.5024,
+      "step": 193
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.738738738738739e-06,
+      "loss": 1.0108,
+      "step": 194
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.783783783783785e-06,
+      "loss": 1.0207,
+      "step": 195
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.82882882882883e-06,
+      "loss": 0.5708,
+      "step": 196
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.873873873873876e-06,
+      "loss": 1.0339,
+      "step": 197
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.91891891891892e-06,
+      "loss": 0.5255,
+      "step": 198
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.963963963963965e-06,
+      "loss": 0.9873,
+      "step": 199
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.00900900900901e-06,
+      "loss": 1.0251,
+      "step": 200
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.054054054054054e-06,
+      "loss": 1.0393,
+      "step": 201
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.0990990990991e-06,
+      "loss": 0.9822,
+      "step": 202
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.144144144144145e-06,
+      "loss": 0.9968,
+      "step": 203
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.189189189189191e-06,
+      "loss": 1.0814,
+      "step": 204
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.234234234234236e-06,
+      "loss": 0.9686,
+      "step": 205
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.27927927927928e-06,
+      "loss": 0.9666,
+      "step": 206
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.324324324324325e-06,
+      "loss": 0.5655,
+      "step": 207
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.36936936936937e-06,
+      "loss": 0.5485,
+      "step": 208
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.414414414414416e-06,
+      "loss": 0.9827,
+      "step": 209
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.45945945945946e-06,
+      "loss": 1.0411,
+      "step": 210
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.504504504504505e-06,
+      "loss": 1.0685,
+      "step": 211
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.54954954954955e-06,
+      "loss": 0.5652,
+      "step": 212
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.594594594594594e-06,
+      "loss": 1.0245,
+      "step": 213
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.63963963963964e-06,
+      "loss": 0.9867,
+      "step": 214
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.684684684684685e-06,
+      "loss": 0.9978,
+      "step": 215
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.729729729729732e-06,
+      "loss": 0.4874,
+      "step": 216
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.774774774774776e-06,
+      "loss": 1.0596,
+      "step": 217
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.81981981981982e-06,
+      "loss": 0.9982,
+      "step": 218
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.864864864864865e-06,
+      "loss": 0.9429,
+      "step": 219
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.90990990990991e-06,
+      "loss": 1.0113,
+      "step": 220
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.954954954954956e-06,
+      "loss": 0.9813,
+      "step": 221
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1e-05,
+      "loss": 0.9921,
+      "step": 222
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0045045045045045e-05,
+      "loss": 0.9792,
+      "step": 223
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.009009009009009e-05,
+      "loss": 0.5133,
+      "step": 224
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0135135135135136e-05,
+      "loss": 0.5442,
+      "step": 225
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0180180180180181e-05,
+      "loss": 0.5362,
+      "step": 226
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0225225225225226e-05,
+      "loss": 0.5429,
+      "step": 227
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.027027027027027e-05,
+      "loss": 1.0587,
+      "step": 228
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0315315315315315e-05,
+      "loss": 0.9965,
+      "step": 229
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0360360360360363e-05,
+      "loss": 0.9524,
+      "step": 230
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0405405405405407e-05,
+      "loss": 0.9671,
+      "step": 231
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0450450450450452e-05,
+      "loss": 1.0256,
+      "step": 232
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0495495495495496e-05,
+      "loss": 0.5439,
+      "step": 233
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0540540540540541e-05,
+      "loss": 0.9701,
+      "step": 234
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0585585585585587e-05,
+      "loss": 0.5219,
+      "step": 235
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0630630630630632e-05,
+      "loss": 0.9624,
+      "step": 236
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0675675675675677e-05,
+      "loss": 0.9527,
+      "step": 237
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0720720720720721e-05,
+      "loss": 1.0006,
+      "step": 238
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0765765765765767e-05,
+      "loss": 1.0309,
+      "step": 239
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0810810810810812e-05,
+      "loss": 0.9651,
+      "step": 240
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0855855855855857e-05,
+      "loss": 0.4814,
+      "step": 241
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0900900900900901e-05,
+      "loss": 0.9839,
+      "step": 242
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0945945945945946e-05,
+      "loss": 0.5158,
+      "step": 243
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0990990990990992e-05,
+      "loss": 1.0622,
+      "step": 244
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1036036036036037e-05,
+      "loss": 1.0262,
+      "step": 245
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1081081081081081e-05,
+      "loss": 1.0163,
+      "step": 246
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1126126126126126e-05,
+      "loss": 0.5529,
+      "step": 247
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.117117117117117e-05,
+      "loss": 1.0161,
+      "step": 248
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1216216216216219e-05,
+      "loss": 0.9878,
+      "step": 249
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1261261261261263e-05,
+      "loss": 0.5298,
+      "step": 250
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1306306306306308e-05,
+      "loss": 0.4961,
+      "step": 251
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1351351351351352e-05,
+      "loss": 1.0379,
+      "step": 252
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1396396396396397e-05,
+      "loss": 0.9799,
+      "step": 253
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1441441441441443e-05,
+      "loss": 0.9981,
+      "step": 254
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1486486486486488e-05,
+      "loss": 0.9953,
+      "step": 255
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1531531531531532e-05,
+      "loss": 1.0156,
+      "step": 256
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1576576576576577e-05,
+      "loss": 0.9432,
+      "step": 257
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1621621621621622e-05,
+      "loss": 1.0213,
+      "step": 258
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1666666666666668e-05,
+      "loss": 0.9845,
+      "step": 259
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1711711711711713e-05,
+      "loss": 0.9719,
+      "step": 260
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1756756756756757e-05,
+      "loss": 0.9786,
+      "step": 261
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1801801801801802e-05,
+      "loss": 0.5245,
+      "step": 262
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1846846846846846e-05,
+      "loss": 0.9962,
+      "step": 263
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1891891891891894e-05,
+      "loss": 0.5445,
+      "step": 264
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1936936936936939e-05,
+      "loss": 0.9917,
+      "step": 265
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1981981981981983e-05,
+      "loss": 0.555,
+      "step": 266
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2027027027027028e-05,
+      "loss": 0.9704,
+      "step": 267
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2072072072072074e-05,
+      "loss": 1.0191,
+      "step": 268
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2117117117117119e-05,
+      "loss": 0.9787,
+      "step": 269
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2162162162162164e-05,
+      "loss": 0.5186,
+      "step": 270
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2207207207207208e-05,
+      "loss": 0.9797,
+      "step": 271
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2252252252252253e-05,
+      "loss": 1.0196,
+      "step": 272
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2297297297297299e-05,
+      "loss": 0.9721,
+      "step": 273
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2342342342342344e-05,
+      "loss": 0.5432,
+      "step": 274
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2387387387387388e-05,
+      "loss": 0.9907,
+      "step": 275
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2432432432432433e-05,
+      "loss": 0.5386,
+      "step": 276
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2477477477477477e-05,
+      "loss": 0.9599,
+      "step": 277
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2522522522522524e-05,
+      "loss": 0.997,
+      "step": 278
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2567567567567568e-05,
+      "loss": 0.9596,
+      "step": 279
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2612612612612613e-05,
+      "loss": 0.9494,
+      "step": 280
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2657657657657658e-05,
+      "loss": 0.9429,
+      "step": 281
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2702702702702702e-05,
+      "loss": 0.9238,
+      "step": 282
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.274774774774775e-05,
+      "loss": 1.0195,
+      "step": 283
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2792792792792795e-05,
+      "loss": 1.0245,
+      "step": 284
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.283783783783784e-05,
+      "loss": 1.016,
+      "step": 285
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2882882882882884e-05,
+      "loss": 1.0026,
+      "step": 286
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2927927927927929e-05,
+      "loss": 0.5186,
+      "step": 287
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2972972972972975e-05,
+      "loss": 0.5351,
+      "step": 288
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.301801801801802e-05,
+      "loss": 0.9752,
+      "step": 289
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3063063063063064e-05,
+      "loss": 0.5208,
+      "step": 290
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3108108108108109e-05,
+      "loss": 0.5108,
+      "step": 291
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3153153153153155e-05,
+      "loss": 1.0019,
+      "step": 292
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.31981981981982e-05,
+      "loss": 0.9744,
+      "step": 293
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3243243243243244e-05,
+      "loss": 0.9777,
+      "step": 294
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3288288288288289e-05,
+      "loss": 0.971,
+      "step": 295
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 0.5387,
+      "step": 296
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3378378378378381e-05,
+      "loss": 1.0284,
+      "step": 297
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3423423423423426e-05,
+      "loss": 1.0385,
+      "step": 298
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.346846846846847e-05,
+      "loss": 0.9543,
+      "step": 299
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3513513513513515e-05,
+      "loss": 0.9705,
+      "step": 300
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.355855855855856e-05,
+      "loss": 0.9405,
+      "step": 301
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3603603603603606e-05,
+      "loss": 1.0085,
+      "step": 302
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.364864864864865e-05,
+      "loss": 0.9889,
+      "step": 303
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3693693693693695e-05,
+      "loss": 0.9474,
+      "step": 304
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.373873873873874e-05,
+      "loss": 0.9875,
+      "step": 305
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3783783783783784e-05,
+      "loss": 1.0188,
+      "step": 306
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.382882882882883e-05,
+      "loss": 0.9458,
+      "step": 307
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3873873873873875e-05,
+      "loss": 0.9286,
+      "step": 308
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.391891891891892e-05,
+      "loss": 0.974,
+      "step": 309
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3963963963963964e-05,
+      "loss": 1.0238,
+      "step": 310
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4009009009009009e-05,
+      "loss": 0.9955,
+      "step": 311
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4054054054054055e-05,
+      "loss": 0.5458,
+      "step": 312
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.40990990990991e-05,
+      "loss": 0.9723,
+      "step": 313
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4144144144144145e-05,
+      "loss": 0.978,
+      "step": 314
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4189189189189189e-05,
+      "loss": 0.9725,
+      "step": 315
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4234234234234234e-05,
+      "loss": 0.9428,
+      "step": 316
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4279279279279282e-05,
+      "loss": 0.916,
+      "step": 317
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4324324324324326e-05,
+      "loss": 1.032,
+      "step": 318
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4369369369369371e-05,
+      "loss": 0.5339,
+      "step": 319
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4414414414414416e-05,
+      "loss": 0.9083,
+      "step": 320
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4459459459459462e-05,
+      "loss": 0.9773,
+      "step": 321
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4504504504504506e-05,
+      "loss": 0.9482,
+      "step": 322
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4549549549549551e-05,
+      "loss": 0.9815,
+      "step": 323
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4594594594594596e-05,
+      "loss": 0.9731,
+      "step": 324
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.463963963963964e-05,
+      "loss": 0.9511,
+      "step": 325
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4684684684684686e-05,
+      "loss": 0.9602,
+      "step": 326
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4729729729729731e-05,
+      "loss": 0.9627,
+      "step": 327
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4774774774774776e-05,
+      "loss": 1.001,
+      "step": 328
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.481981981981982e-05,
+      "loss": 0.9751,
+      "step": 329
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4864864864864865e-05,
+      "loss": 1.0053,
+      "step": 330
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4909909909909913e-05,
+      "loss": 0.9341,
+      "step": 331
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4954954954954957e-05,
+      "loss": 0.9369,
+      "step": 332
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 0.9211,
+      "step": 333
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5045045045045045e-05,
+      "loss": 0.9709,
+      "step": 334
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.509009009009009e-05,
+      "loss": 1.005,
+      "step": 335
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5135135135135138e-05,
+      "loss": 0.9676,
+      "step": 336
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5180180180180182e-05,
+      "loss": 0.549,
+      "step": 337
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5225225225225227e-05,
+      "loss": 1.0011,
+      "step": 338
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.527027027027027e-05,
+      "loss": 0.4826,
+      "step": 339
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5315315315315316e-05,
+      "loss": 0.9973,
+      "step": 340
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5360360360360362e-05,
+      "loss": 0.9326,
+      "step": 341
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.540540540540541e-05,
+      "loss": 1.0131,
+      "step": 342
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.545045045045045e-05,
+      "loss": 0.5205,
+      "step": 343
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5495495495495498e-05,
+      "loss": 0.969,
+      "step": 344
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.554054054054054e-05,
+      "loss": 0.9924,
+      "step": 345
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5585585585585587e-05,
+      "loss": 0.9744,
+      "step": 346
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5630630630630633e-05,
+      "loss": 0.4939,
+      "step": 347
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5675675675675676e-05,
+      "loss": 0.958,
+      "step": 348
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5720720720720722e-05,
+      "loss": 0.5502,
+      "step": 349
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.576576576576577e-05,
+      "loss": 0.9836,
+      "step": 350
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.581081081081081e-05,
+      "loss": 1.0205,
+      "step": 351
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5855855855855858e-05,
+      "loss": 0.9586,
+      "step": 352
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.59009009009009e-05,
+      "loss": 0.9439,
+      "step": 353
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5945945945945947e-05,
+      "loss": 0.8986,
+      "step": 354
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5990990990990993e-05,
+      "loss": 0.9653,
+      "step": 355
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6036036036036036e-05,
+      "loss": 0.9703,
+      "step": 356
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6081081081081083e-05,
+      "loss": 0.9216,
+      "step": 357
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6126126126126126e-05,
+      "loss": 1.0271,
+      "step": 358
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6171171171171172e-05,
+      "loss": 0.9849,
+      "step": 359
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6216216216216218e-05,
+      "loss": 0.9348,
+      "step": 360
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6261261261261264e-05,
+      "loss": 0.938,
+      "step": 361
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6306306306306307e-05,
+      "loss": 0.9518,
+      "step": 362
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6351351351351354e-05,
+      "loss": 0.9488,
+      "step": 363
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6396396396396396e-05,
+      "loss": 0.9628,
+      "step": 364
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6441441441441443e-05,
+      "loss": 0.97,
+      "step": 365
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.648648648648649e-05,
+      "loss": 0.9642,
+      "step": 366
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6531531531531532e-05,
+      "loss": 0.9597,
+      "step": 367
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6576576576576578e-05,
+      "loss": 0.965,
+      "step": 368
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.662162162162162e-05,
+      "loss": 1.0055,
+      "step": 369
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6666666666666667e-05,
+      "loss": 0.9602,
+      "step": 370
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6711711711711714e-05,
+      "loss": 0.9479,
+      "step": 371
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6756756756756757e-05,
+      "loss": 0.9239,
+      "step": 372
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6801801801801803e-05,
+      "loss": 0.4944,
+      "step": 373
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6846846846846846e-05,
+      "loss": 0.9758,
+      "step": 374
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6891891891891896e-05,
+      "loss": 0.9904,
+      "step": 375
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.693693693693694e-05,
+      "loss": 0.9975,
+      "step": 376
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6981981981981985e-05,
+      "loss": 0.9596,
+      "step": 377
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7027027027027028e-05,
+      "loss": 1.0113,
+      "step": 378
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7072072072072074e-05,
+      "loss": 0.9523,
+      "step": 379
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.711711711711712e-05,
+      "loss": 0.9943,
+      "step": 380
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7162162162162163e-05,
+      "loss": 0.973,
+      "step": 381
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.720720720720721e-05,
+      "loss": 0.968,
+      "step": 382
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7252252252252252e-05,
+      "loss": 0.9254,
+      "step": 383
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.72972972972973e-05,
+      "loss": 1.0031,
+      "step": 384
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7342342342342345e-05,
+      "loss": 0.5256,
+      "step": 385
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7387387387387388e-05,
+      "loss": 0.9766,
+      "step": 386
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7432432432432434e-05,
+      "loss": 0.5124,
+      "step": 387
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7477477477477477e-05,
+      "loss": 0.9722,
+      "step": 388
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7522522522522523e-05,
+      "loss": 0.9565,
+      "step": 389
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.756756756756757e-05,
+      "loss": 0.9574,
+      "step": 390
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7612612612612613e-05,
+      "loss": 0.9414,
+      "step": 391
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.765765765765766e-05,
+      "loss": 0.9512,
+      "step": 392
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7702702702702702e-05,
+      "loss": 0.5656,
+      "step": 393
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.774774774774775e-05,
+      "loss": 1.0525,
+      "step": 394
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7792792792792794e-05,
+      "loss": 0.986,
+      "step": 395
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.783783783783784e-05,
+      "loss": 0.9966,
+      "step": 396
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7882882882882883e-05,
+      "loss": 0.9692,
+      "step": 397
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.792792792792793e-05,
+      "loss": 0.9598,
+      "step": 398
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7972972972972976e-05,
+      "loss": 1.0017,
+      "step": 399
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.801801801801802e-05,
+      "loss": 0.9734,
+      "step": 400
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8063063063063065e-05,
+      "loss": 0.5185,
+      "step": 401
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8108108108108108e-05,
+      "loss": 0.9931,
+      "step": 402
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8153153153153154e-05,
+      "loss": 0.9842,
+      "step": 403
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.81981981981982e-05,
+      "loss": 0.9833,
+      "step": 404
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8243243243243244e-05,
+      "loss": 1.0213,
+      "step": 405
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.828828828828829e-05,
+      "loss": 0.9502,
+      "step": 406
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8333333333333333e-05,
+      "loss": 0.9731,
+      "step": 407
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8378378378378383e-05,
+      "loss": 0.9832,
+      "step": 408
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8423423423423425e-05,
+      "loss": 0.4806,
+      "step": 409
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8468468468468472e-05,
+      "loss": 0.5789,
+      "step": 410
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8513513513513515e-05,
+      "loss": 0.5237,
+      "step": 411
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.855855855855856e-05,
+      "loss": 1.0115,
+      "step": 412
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8603603603603607e-05,
+      "loss": 0.9448,
+      "step": 413
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.864864864864865e-05,
+      "loss": 1.0097,
+      "step": 414
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8693693693693696e-05,
+      "loss": 0.9798,
+      "step": 415
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.873873873873874e-05,
+      "loss": 1.0553,
+      "step": 416
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8783783783783786e-05,
+      "loss": 0.5367,
+      "step": 417
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8828828828828832e-05,
+      "loss": 0.5133,
+      "step": 418
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8873873873873875e-05,
+      "loss": 0.9635,
+      "step": 419
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.891891891891892e-05,
+      "loss": 0.8979,
+      "step": 420
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8963963963963964e-05,
+      "loss": 0.9414,
+      "step": 421
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.900900900900901e-05,
+      "loss": 0.5293,
+      "step": 422
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9054054054054057e-05,
+      "loss": 0.9555,
+      "step": 423
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.90990990990991e-05,
+      "loss": 0.9738,
+      "step": 424
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9144144144144146e-05,
+      "loss": 1.0299,
+      "step": 425
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.918918918918919e-05,
+      "loss": 0.925,
+      "step": 426
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9234234234234235e-05,
+      "loss": 0.5063,
+      "step": 427
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.927927927927928e-05,
+      "loss": 0.9597,
+      "step": 428
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9324324324324328e-05,
+      "loss": 0.4999,
+      "step": 429
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.936936936936937e-05,
+      "loss": 0.5005,
+      "step": 430
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9414414414414417e-05,
+      "loss": 0.9823,
+      "step": 431
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9459459459459463e-05,
+      "loss": 0.5175,
+      "step": 432
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9504504504504506e-05,
+      "loss": 0.9456,
+      "step": 433
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9549549549549552e-05,
+      "loss": 0.4561,
+      "step": 434
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9594594594594595e-05,
+      "loss": 1.0153,
+      "step": 435
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.963963963963964e-05,
+      "loss": 0.9525,
+      "step": 436
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9684684684684688e-05,
+      "loss": 0.9592,
+      "step": 437
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.972972972972973e-05,
+      "loss": 0.9004,
+      "step": 438
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9774774774774777e-05,
+      "loss": 0.9297,
+      "step": 439
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.981981981981982e-05,
+      "loss": 0.9858,
+      "step": 440
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9864864864864866e-05,
+      "loss": 0.9532,
+      "step": 441
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9909909909909912e-05,
+      "loss": 0.4939,
+      "step": 442
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.995495495495496e-05,
+      "loss": 0.9332,
+      "step": 443
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2e-05,
+      "loss": 0.9638,
+      "step": 444
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999999975995504e-05,
+      "loss": 0.9449,
+      "step": 445
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999999039820163e-05,
+      "loss": 0.5234,
+      "step": 446
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999997839595412e-05,
+      "loss": 0.9641,
+      "step": 447
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999996159280836e-05,
+      "loss": 0.5282,
+      "step": 448
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999993998876524e-05,
+      "loss": 0.977,
+      "step": 449
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999991358382574e-05,
+      "loss": 0.9743,
+      "step": 450
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999998823779911e-05,
+      "loss": 0.9595,
+      "step": 451
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999984637126295e-05,
+      "loss": 0.4972,
+      "step": 452
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999980556364288e-05,
+      "loss": 0.4821,
+      "step": 453
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999997599551329e-05,
+      "loss": 0.9511,
+      "step": 454
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999997095457352e-05,
+      "loss": 0.5096,
+      "step": 455
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999965433545224e-05,
+      "loss": 0.5546,
+      "step": 456
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999995943242866e-05,
+      "loss": 0.5288,
+      "step": 457
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999995295122412e-05,
+      "loss": 0.9594,
+      "step": 458
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999945989931913e-05,
+      "loss": 0.5127,
+      "step": 459
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999938548552372e-05,
+      "loss": 0.5197,
+      "step": 460
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999993062708586e-05,
+      "loss": 1.0886,
+      "step": 461
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999922225532756e-05,
+      "loss": 0.9455,
+      "step": 462
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999991334389346e-05,
+      "loss": 0.9871,
+      "step": 463
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.99999039821684e-05,
+      "loss": 0.9447,
+      "step": 464
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999894140358026e-05,
+      "loss": 0.9702,
+      "step": 465
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999883818462805e-05,
+      "loss": 0.8924,
+      "step": 466
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999873016483242e-05,
+      "loss": 0.9298,
+      "step": 467
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999986173441985e-05,
+      "loss": 0.8934,
+      "step": 468
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999849972273174e-05,
+      "loss": 0.5009,
+      "step": 469
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999837730043774e-05,
+      "loss": 0.9416,
+      "step": 470
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999825007732238e-05,
+      "loss": 0.9111,
+      "step": 471
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999981180533918e-05,
+      "loss": 0.9666,
+      "step": 472
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999979812286524e-05,
+      "loss": 0.5163,
+      "step": 473
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999783960311056e-05,
+      "loss": 0.9306,
+      "step": 474
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999769317677328e-05,
+      "loss": 0.9335,
+      "step": 475
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999754194964743e-05,
+      "loss": 0.993,
+      "step": 476
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999738592174037e-05,
+      "loss": 0.5029,
+      "step": 477
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999972250930596e-05,
+      "loss": 0.4747,
+      "step": 478
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999970594636128e-05,
+      "loss": 0.9305,
+      "step": 479
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999688903340792e-05,
+      "loss": 0.9721,
+      "step": 480
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999671380245313e-05,
+      "loss": 0.9741,
+      "step": 481
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999653377075686e-05,
+      "loss": 0.9002,
+      "step": 482
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999963489383278e-05,
+      "loss": 0.9443,
+      "step": 483
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999961593051748e-05,
+      "loss": 0.949,
+      "step": 484
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999596487130687e-05,
+      "loss": 0.9346,
+      "step": 485
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999576563673343e-05,
+      "loss": 0.5317,
+      "step": 486
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999556160146406e-05,
+      "loss": 0.9981,
+      "step": 487
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999535276550853e-05,
+      "loss": 0.9385,
+      "step": 488
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999513912887685e-05,
+      "loss": 0.9487,
+      "step": 489
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999949206915793e-05,
+      "loss": 0.5249,
+      "step": 490
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999469745362633e-05,
+      "loss": 0.9203,
+      "step": 491
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999446941502874e-05,
+      "loss": 0.9419,
+      "step": 492
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999942365757974e-05,
+      "loss": 0.9602,
+      "step": 493
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999399893594353e-05,
+      "loss": 0.9596,
+      "step": 494
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999375649547848e-05,
+      "loss": 0.925,
+      "step": 495
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999350925441395e-05,
+      "loss": 0.4882,
+      "step": 496
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999932572127618e-05,
+      "loss": 0.9277,
+      "step": 497
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999300037053412e-05,
+      "loss": 0.9381,
+      "step": 498
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999273872774322e-05,
+      "loss": 0.9207,
+      "step": 499
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999924722844017e-05,
+      "loss": 0.9599,
+      "step": 500
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999220104052232e-05,
+      "loss": 0.9651,
+      "step": 501
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999192499611814e-05,
+      "loss": 0.9791,
+      "step": 502
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999164415120237e-05,
+      "loss": 0.9844,
+      "step": 503
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999135850578853e-05,
+      "loss": 0.9586,
+      "step": 504
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999910680598903e-05,
+      "loss": 0.9399,
+      "step": 505
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999077281352165e-05,
+      "loss": 0.94,
+      "step": 506
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999047276669673e-05,
+      "loss": 0.9452,
+      "step": 507
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999016791942995e-05,
+      "loss": 0.949,
+      "step": 508
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998985827173598e-05,
+      "loss": 0.4914,
+      "step": 509
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998954382362965e-05,
+      "loss": 0.963,
+      "step": 510
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998922457512608e-05,
+      "loss": 0.4768,
+      "step": 511
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998890052624055e-05,
+      "loss": 0.5369,
+      "step": 512
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998857167698868e-05,
+      "loss": 0.9822,
+      "step": 513
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999882380273862e-05,
+      "loss": 0.8974,
+      "step": 514
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998789957744923e-05,
+      "loss": 0.9358,
+      "step": 515
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998755632719387e-05,
+      "loss": 0.5439,
+      "step": 516
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999872082766367e-05,
+      "loss": 0.4822,
+      "step": 517
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999868554257944e-05,
+      "loss": 0.4837,
+      "step": 518
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999864977746839e-05,
+      "loss": 1.0631,
+      "step": 519
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999861353233224e-05,
+      "loss": 0.5266,
+      "step": 520
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999857680717273e-05,
+      "loss": 0.9892,
+      "step": 521
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999853960199162e-05,
+      "loss": 0.9411,
+      "step": 522
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99985019167907e-05,
+      "loss": 0.993,
+      "step": 523
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998463751571772e-05,
+      "loss": 0.9335,
+      "step": 524
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998425106336675e-05,
+      "loss": 0.9564,
+      "step": 525
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999838598108727e-05,
+      "loss": 0.9236,
+      "step": 526
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999834637582542e-05,
+      "loss": 0.95,
+      "step": 527
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998306290553037e-05,
+      "loss": 0.9398,
+      "step": 528
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999826572527204e-05,
+      "loss": 0.532,
+      "step": 529
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998224679984382e-05,
+      "loss": 0.9775,
+      "step": 530
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998183154692033e-05,
+      "loss": 0.5565,
+      "step": 531
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999814114939698e-05,
+      "loss": 0.958,
+      "step": 532
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998098664101246e-05,
+      "loss": 1.0,
+      "step": 533
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998055698806872e-05,
+      "loss": 0.9525,
+      "step": 534
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998012253515914e-05,
+      "loss": 0.9239,
+      "step": 535
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999796832823046e-05,
+      "loss": 0.9277,
+      "step": 536
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997923922952626e-05,
+      "loss": 0.9257,
+      "step": 537
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997879037684533e-05,
+      "loss": 0.9429,
+      "step": 538
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997833672428344e-05,
+      "loss": 0.9119,
+      "step": 539
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999778782718623e-05,
+      "loss": 0.9245,
+      "step": 540
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99977415019604e-05,
+      "loss": 0.9337,
+      "step": 541
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999769469675307e-05,
+      "loss": 0.9721,
+      "step": 542
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997647411566494e-05,
+      "loss": 0.9217,
+      "step": 543
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997599646402935e-05,
+      "loss": 0.9678,
+      "step": 544
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997551401264697e-05,
+      "loss": 0.9275,
+      "step": 545
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997502676154084e-05,
+      "loss": 0.8626,
+      "step": 546
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999745347107344e-05,
+      "loss": 0.9264,
+      "step": 547
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997403786025132e-05,
+      "loss": 0.9135,
+      "step": 548
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999735362101154e-05,
+      "loss": 0.9621,
+      "step": 549
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999730297603507e-05,
+      "loss": 0.8949,
+      "step": 550
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999725185109816e-05,
+      "loss": 0.9305,
+      "step": 551
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997200246203264e-05,
+      "loss": 0.4809,
+      "step": 552
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997148161352854e-05,
+      "loss": 0.8981,
+      "step": 553
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997095596549433e-05,
+      "loss": 0.9433,
+      "step": 554
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997042551795524e-05,
+      "loss": 0.9735,
+      "step": 555
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999698902709368e-05,
+      "loss": 0.9225,
+      "step": 556
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999693502244646e-05,
+      "loss": 0.9448,
+      "step": 557
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996880537856462e-05,
+      "loss": 0.9291,
+      "step": 558
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996825573326305e-05,
+      "loss": 0.5178,
+      "step": 559
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996770128858624e-05,
+      "loss": 0.4955,
+      "step": 560
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999671420445608e-05,
+      "loss": 0.8996,
+      "step": 561
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996657800121363e-05,
+      "loss": 0.911,
+      "step": 562
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996600915857174e-05,
+      "loss": 0.4725,
+      "step": 563
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996543551666248e-05,
+      "loss": 0.9213,
+      "step": 564
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996485707551337e-05,
+      "loss": 0.8861,
+      "step": 565
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999642738351522e-05,
+      "loss": 0.9871,
+      "step": 566
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996368579560698e-05,
+      "loss": 0.9192,
+      "step": 567
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999630929569059e-05,
+      "loss": 0.9215,
+      "step": 568
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996249531907747e-05,
+      "loss": 1.0187,
+      "step": 569
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996189288215032e-05,
+      "loss": 0.9227,
+      "step": 570
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996128564615344e-05,
+      "loss": 0.5143,
+      "step": 571
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996067361111595e-05,
+      "loss": 0.5502,
+      "step": 572
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996005677706724e-05,
+      "loss": 0.9614,
+      "step": 573
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995943514403692e-05,
+      "loss": 0.5188,
+      "step": 574
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995880871205483e-05,
+      "loss": 0.988,
+      "step": 575
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995817748115104e-05,
+      "loss": 0.9139,
+      "step": 576
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999575414513559e-05,
+      "loss": 0.9323,
+      "step": 577
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995690062269985e-05,
+      "loss": 0.9476,
+      "step": 578
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995625499521374e-05,
+      "loss": 0.9231,
+      "step": 579
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995560456892854e-05,
+      "loss": 0.9232,
+      "step": 580
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999549493438755e-05,
+      "loss": 0.4997,
+      "step": 581
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995428932008602e-05,
+      "loss": 0.941,
+      "step": 582
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995362449759185e-05,
+      "loss": 0.9021,
+      "step": 583
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995295487642486e-05,
+      "loss": 0.5729,
+      "step": 584
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995228045661718e-05,
+      "loss": 0.9605,
+      "step": 585
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999516012382013e-05,
+      "loss": 0.9113,
+      "step": 586
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999509172212097e-05,
+      "loss": 0.9776,
+      "step": 587
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995022840567525e-05,
+      "loss": 0.9335,
+      "step": 588
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994953479163112e-05,
+      "loss": 0.9565,
+      "step": 589
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994883637911044e-05,
+      "loss": 0.9573,
+      "step": 590
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999481331681469e-05,
+      "loss": 0.9356,
+      "step": 591
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994742515877416e-05,
+      "loss": 0.9068,
+      "step": 592
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999467123510262e-05,
+      "loss": 0.9764,
+      "step": 593
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994599474493736e-05,
+      "loss": 0.9706,
+      "step": 594
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99945272340542e-05,
+      "loss": 0.9272,
+      "step": 595
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999445451378748e-05,
+      "loss": 0.9264,
+      "step": 596
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994381313697067e-05,
+      "loss": 0.5292,
+      "step": 597
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999430763378648e-05,
+      "loss": 0.4894,
+      "step": 598
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999423347405925e-05,
+      "loss": 0.9366,
+      "step": 599
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994158834518944e-05,
+      "loss": 0.9017,
+      "step": 600
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999408371516914e-05,
+      "loss": 0.9142,
+      "step": 601
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994008116013445e-05,
+      "loss": 0.9664,
+      "step": 602
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993932037055495e-05,
+      "loss": 0.9814,
+      "step": 603
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993855478298935e-05,
+      "loss": 0.8709,
+      "step": 604
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999377843974744e-05,
+      "loss": 0.9388,
+      "step": 605
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993700921404715e-05,
+      "loss": 0.9517,
+      "step": 606
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999362292327448e-05,
+      "loss": 1.0001,
+      "step": 607
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993544445360472e-05,
+      "loss": 0.9702,
+      "step": 608
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993465487666465e-05,
+      "loss": 0.8972,
+      "step": 609
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993386050196252e-05,
+      "loss": 0.9681,
+      "step": 610
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993306132953643e-05,
+      "loss": 0.9351,
+      "step": 611
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993225735942478e-05,
+      "loss": 0.9351,
+      "step": 612
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999314485916661e-05,
+      "loss": 0.4898,
+      "step": 613
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993063502629926e-05,
+      "loss": 0.4652,
+      "step": 614
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999298166633633e-05,
+      "loss": 0.9134,
+      "step": 615
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992899350289757e-05,
+      "loss": 0.5112,
+      "step": 616
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992816554494157e-05,
+      "loss": 0.9439,
+      "step": 617
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99927332789535e-05,
+      "loss": 0.523,
+      "step": 618
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992649523671784e-05,
+      "loss": 0.9542,
+      "step": 619
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992565288653033e-05,
+      "loss": 0.8893,
+      "step": 620
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992480573901292e-05,
+      "loss": 0.9056,
+      "step": 621
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992395379420628e-05,
+      "loss": 0.4781,
+      "step": 622
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999230970521513e-05,
+      "loss": 0.9315,
+      "step": 623
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992223551288904e-05,
+      "loss": 0.9381,
+      "step": 624
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992136917646103e-05,
+      "loss": 0.4644,
+      "step": 625
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999204980429087e-05,
+      "loss": 0.499,
+      "step": 626
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991962211227395e-05,
+      "loss": 0.9646,
+      "step": 627
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999187413845988e-05,
+      "loss": 0.9368,
+      "step": 628
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999178558599256e-05,
+      "loss": 0.98,
+      "step": 629
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991696553829676e-05,
+      "loss": 0.5558,
+      "step": 630
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999160704197551e-05,
+      "loss": 0.9858,
+      "step": 631
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991517050434357e-05,
+      "loss": 0.4891,
+      "step": 632
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999142657921054e-05,
+      "loss": 0.9633,
+      "step": 633
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99913356283084e-05,
+      "loss": 0.4943,
+      "step": 634
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99912441977323e-05,
+      "loss": 0.9342,
+      "step": 635
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999115228748663e-05,
+      "loss": 0.9734,
+      "step": 636
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999105989757581e-05,
+      "loss": 0.9046,
+      "step": 637
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999096702800427e-05,
+      "loss": 0.9612,
+      "step": 638
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999087367877647e-05,
+      "loss": 0.4806,
+      "step": 639
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990779849896892e-05,
+      "loss": 0.4869,
+      "step": 640
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999068554137004e-05,
+      "loss": 0.8919,
+      "step": 641
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990590753200437e-05,
+      "loss": 0.516,
+      "step": 642
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999049548539264e-05,
+      "loss": 0.9188,
+      "step": 643
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990399737951223e-05,
+      "loss": 0.8728,
+      "step": 644
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999030351088078e-05,
+      "loss": 0.9488,
+      "step": 645
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999020680418593e-05,
+      "loss": 0.8985,
+      "step": 646
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990109617871316e-05,
+      "loss": 0.4797,
+      "step": 647
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990011951941607e-05,
+      "loss": 0.9927,
+      "step": 648
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998991380640149e-05,
+      "loss": 0.9882,
+      "step": 649
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989815181255676e-05,
+      "loss": 0.9233,
+      "step": 650
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99897160765089e-05,
+      "loss": 0.9383,
+      "step": 651
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989616492165922e-05,
+      "loss": 0.943,
+      "step": 652
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989516428231523e-05,
+      "loss": 0.9348,
+      "step": 653
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99894158847105e-05,
+      "loss": 0.9524,
+      "step": 654
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989314861607687e-05,
+      "loss": 0.9298,
+      "step": 655
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989213358927935e-05,
+      "loss": 0.8946,
+      "step": 656
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989111376676112e-05,
+      "loss": 0.9491,
+      "step": 657
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989008914857115e-05,
+      "loss": 0.949,
+      "step": 658
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998890597347587e-05,
+      "loss": 0.8938,
+      "step": 659
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988802552537308e-05,
+      "loss": 0.9454,
+      "step": 660
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99886986520464e-05,
+      "loss": 0.9298,
+      "step": 661
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988594272008138e-05,
+      "loss": 0.947,
+      "step": 662
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988489412427528e-05,
+      "loss": 0.9339,
+      "step": 663
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988384073309603e-05,
+      "loss": 0.909,
+      "step": 664
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988278254659422e-05,
+      "loss": 0.9663,
+      "step": 665
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998817195648207e-05,
+      "loss": 0.9584,
+      "step": 666
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9988065178782645e-05,
+      "loss": 0.9082,
+      "step": 667
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998795792156627e-05,
+      "loss": 0.5089,
+      "step": 668
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987850184838105e-05,
+      "loss": 0.953,
+      "step": 669
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987741968603314e-05,
+      "loss": 0.9137,
+      "step": 670
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987633272867092e-05,
+      "loss": 0.9071,
+      "step": 671
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998752409763466e-05,
+      "loss": 0.9295,
+      "step": 672
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998741444291126e-05,
+      "loss": 0.9445,
+      "step": 673
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998730430870216e-05,
+      "loss": 0.9584,
+      "step": 674
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987193695012636e-05,
+      "loss": 0.9492,
+      "step": 675
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987082601848007e-05,
+      "loss": 0.4639,
+      "step": 676
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986971029213608e-05,
+      "loss": 0.9531,
+      "step": 677
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986858977114787e-05,
+      "loss": 0.9293,
+      "step": 678
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986746445556934e-05,
+      "loss": 0.9013,
+      "step": 679
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986633434545445e-05,
+      "loss": 0.9731,
+      "step": 680
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986519944085746e-05,
+      "loss": 0.9428,
+      "step": 681
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986405974183288e-05,
+      "loss": 0.5262,
+      "step": 682
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998629152484354e-05,
+      "loss": 0.4851,
+      "step": 683
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986176596072e-05,
+      "loss": 0.9275,
+      "step": 684
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998606118787418e-05,
+      "loss": 0.4969,
+      "step": 685
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985945300255623e-05,
+      "loss": 0.9656,
+      "step": 686
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985828933221895e-05,
+      "loss": 0.9427,
+      "step": 687
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985712086778583e-05,
+      "loss": 0.8569,
+      "step": 688
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985594760931293e-05,
+      "loss": 0.951,
+      "step": 689
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985476955685663e-05,
+      "loss": 0.9118,
+      "step": 690
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985358671047343e-05,
+      "loss": 0.9845,
+      "step": 691
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985239907022012e-05,
+      "loss": 0.5225,
+      "step": 692
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985120663615377e-05,
+      "loss": 0.9273,
+      "step": 693
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985000940833158e-05,
+      "loss": 0.5078,
+      "step": 694
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984880738681104e-05,
+      "loss": 0.9587,
+      "step": 695
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984760057164986e-05,
+      "loss": 0.4895,
+      "step": 696
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99846388962906e-05,
+      "loss": 0.9453,
+      "step": 697
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998451725606376e-05,
+      "loss": 0.9327,
+      "step": 698
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984395136490307e-05,
+      "loss": 0.8928,
+      "step": 699
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984272537576102e-05,
+      "loss": 0.9081,
+      "step": 700
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984149459327032e-05,
+      "loss": 0.9397,
+      "step": 701
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984025901749008e-05,
+      "loss": 0.9178,
+      "step": 702
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983901864847957e-05,
+      "loss": 0.5059,
+      "step": 703
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998377734862984e-05,
+      "loss": 0.9444,
+      "step": 704
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983652353100628e-05,
+      "loss": 0.8988,
+      "step": 705
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983526878266332e-05,
+      "loss": 0.9387,
+      "step": 706
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983400924132965e-05,
+      "loss": 0.9322,
+      "step": 707
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998327449070658e-05,
+      "loss": 0.9419,
+      "step": 708
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983147577993244e-05,
+      "loss": 0.4944,
+      "step": 709
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998302018599905e-05,
+      "loss": 0.4597,
+      "step": 710
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982892314730117e-05,
+      "loss": 0.9284,
+      "step": 711
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982763964192586e-05,
+      "loss": 0.933,
+      "step": 712
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982635134392612e-05,
+      "loss": 0.9253,
+      "step": 713
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982505825336382e-05,
+      "loss": 0.919,
+      "step": 714
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998237603703011e-05,
+      "loss": 0.9373,
+      "step": 715
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982245769480016e-05,
+      "loss": 0.9947,
+      "step": 716
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982115022692366e-05,
+      "loss": 0.9453,
+      "step": 717
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981983796673426e-05,
+      "loss": 0.9504,
+      "step": 718
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981852091429507e-05,
+      "loss": 0.9498,
+      "step": 719
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981719906966922e-05,
+      "loss": 0.9364,
+      "step": 720
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981587243292023e-05,
+      "loss": 0.9248,
+      "step": 721
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981454100411178e-05,
+      "loss": 0.9674,
+      "step": 722
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981320478330777e-05,
+      "loss": 0.8906,
+      "step": 723
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981186377057238e-05,
+      "loss": 0.923,
+      "step": 724
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981051796597e-05,
+      "loss": 0.9201,
+      "step": 725
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998091673695652e-05,
+      "loss": 0.9128,
+      "step": 726
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980781198142282e-05,
+      "loss": 0.9514,
+      "step": 727
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980645180160794e-05,
+      "loss": 0.9283,
+      "step": 728
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998050868301859e-05,
+      "loss": 0.9347,
+      "step": 729
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998037170672222e-05,
+      "loss": 0.4908,
+      "step": 730
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998023425127826e-05,
+      "loss": 0.5105,
+      "step": 731
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980096316693308e-05,
+      "loss": 0.9333,
+      "step": 732
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997995790297399e-05,
+      "loss": 0.9593,
+      "step": 733
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979819010126946e-05,
+      "loss": 0.971,
+      "step": 734
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979679638158848e-05,
+      "loss": 0.9251,
+      "step": 735
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997953978707638e-05,
+      "loss": 0.9179,
+      "step": 736
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997939945688627e-05,
+      "loss": 0.9004,
+      "step": 737
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979258647595244e-05,
+      "loss": 0.9671,
+      "step": 738
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979117359210064e-05,
+      "loss": 0.9444,
+      "step": 739
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978975591737516e-05,
+      "loss": 0.473,
+      "step": 740
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978833345184404e-05,
+      "loss": 0.9468,
+      "step": 741
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978690619557555e-05,
+      "loss": 0.5156,
+      "step": 742
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978547414863826e-05,
+      "loss": 0.9653,
+      "step": 743
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978403731110088e-05,
+      "loss": 0.9573,
+      "step": 744
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978259568303243e-05,
+      "loss": 0.9244,
+      "step": 745
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978114926450207e-05,
+      "loss": 0.875,
+      "step": 746
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977969805557928e-05,
+      "loss": 0.8998,
+      "step": 747
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977824205633375e-05,
+      "loss": 0.9818,
+      "step": 748
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997767812668353e-05,
+      "loss": 0.9302,
+      "step": 749
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977531568715415e-05,
+      "loss": 0.9588,
+      "step": 750
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997738453173606e-05,
+      "loss": 0.4948,
+      "step": 751
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997723701575253e-05,
+      "loss": 0.4805,
+      "step": 752
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99770890207719e-05,
+      "loss": 0.9256,
+      "step": 753
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997694054680128e-05,
+      "loss": 0.5025,
+      "step": 754
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976791593847796e-05,
+      "loss": 0.4671,
+      "step": 755
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99766421619186e-05,
+      "loss": 0.91,
+      "step": 756
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976492251020866e-05,
+      "loss": 0.5,
+      "step": 757
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997634186116179e-05,
+      "loss": 0.9172,
+      "step": 758
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976190992348598e-05,
+      "loss": 0.9049,
+      "step": 759
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976039644588522e-05,
+      "loss": 0.981,
+      "step": 760
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975887817888838e-05,
+      "loss": 0.9273,
+      "step": 761
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975735512256827e-05,
+      "loss": 0.8943,
+      "step": 762
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997558272769981e-05,
+      "loss": 0.9382,
+      "step": 763
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975429464225115e-05,
+      "loss": 0.4883,
+      "step": 764
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975275721840105e-05,
+      "loss": 0.5412,
+      "step": 765
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975121500552154e-05,
+      "loss": 0.9297,
+      "step": 766
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997496680036867e-05,
+      "loss": 0.9151,
+      "step": 767
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9974811621297085e-05,
+      "loss": 0.8896,
+      "step": 768
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9974655963344842e-05,
+      "loss": 0.9212,
+      "step": 769
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9974499826519412e-05,
+      "loss": 0.8922,
+      "step": 770
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9974343210828304e-05,
+      "loss": 0.9462,
+      "step": 771
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997418611627902e-05,
+      "loss": 0.934,
+      "step": 772
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9974028542879115e-05,
+      "loss": 0.4886,
+      "step": 773
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997387049063615e-05,
+      "loss": 0.4907,
+      "step": 774
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9973711959557708e-05,
+      "loss": 0.4928,
+      "step": 775
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9973552949651406e-05,
+      "loss": 0.9425,
+      "step": 776
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9973393460924876e-05,
+      "loss": 0.9394,
+      "step": 777
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9973233493385773e-05,
+      "loss": 0.9146,
+      "step": 778
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997307304704178e-05,
+      "loss": 0.953,
+      "step": 779
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972912121900597e-05,
+      "loss": 0.9671,
+      "step": 780
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972750717969954e-05,
+      "loss": 0.916,
+      "step": 781
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972588835257597e-05,
+      "loss": 0.9332,
+      "step": 782
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972426473771297e-05,
+      "loss": 0.4954,
+      "step": 783
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972263633518852e-05,
+      "loss": 0.9148,
+      "step": 784
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972100314508075e-05,
+      "loss": 0.9645,
+      "step": 785
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9971936516746808e-05,
+      "loss": 0.5286,
+      "step": 786
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997177224024292e-05,
+      "loss": 0.9439,
+      "step": 787
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997160748500429e-05,
+      "loss": 0.9245,
+      "step": 788
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9971442251038832e-05,
+      "loss": 0.8938,
+      "step": 789
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9971276538354484e-05,
+      "loss": 0.9377,
+      "step": 790
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997111034695919e-05,
+      "loss": 0.8531,
+      "step": 791
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9970943676860938e-05,
+      "loss": 0.9058,
+      "step": 792
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9970776528067722e-05,
+      "loss": 0.906,
+      "step": 793
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9970608900587578e-05,
+      "loss": 0.9617,
+      "step": 794
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997044079442854e-05,
+      "loss": 0.9461,
+      "step": 795
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997027220959869e-05,
+      "loss": 0.9603,
+      "step": 796
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9970103146106114e-05,
+      "loss": 0.928,
+      "step": 797
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9969933603958934e-05,
+      "loss": 0.913,
+      "step": 798
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9969763583165286e-05,
+      "loss": 0.8798,
+      "step": 799
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996959308373333e-05,
+      "loss": 0.9069,
+      "step": 800
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9969422105671262e-05,
+      "loss": 0.907,
+      "step": 801
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9969250648987278e-05,
+      "loss": 0.9071,
+      "step": 802
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9969078713689617e-05,
+      "loss": 0.8826,
+      "step": 803
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9968906299786533e-05,
+      "loss": 0.5103,
+      "step": 804
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9968733407286297e-05,
+      "loss": 0.8979,
+      "step": 805
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996856003619722e-05,
+      "loss": 0.4753,
+      "step": 806
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9968386186527614e-05,
+      "loss": 0.5002,
+      "step": 807
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9968211858285834e-05,
+      "loss": 0.8911,
+      "step": 808
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996803705148025e-05,
+      "loss": 0.9312,
+      "step": 809
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9967861766119246e-05,
+      "loss": 0.9023,
+      "step": 810
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996768600221124e-05,
+      "loss": 0.9227,
+      "step": 811
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9967509759764677e-05,
+      "loss": 0.9486,
+      "step": 812
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9967333038788012e-05,
+      "loss": 0.5129,
+      "step": 813
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996715583928973e-05,
+      "loss": 0.9035,
+      "step": 814
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966978161278336e-05,
+      "loss": 0.9202,
+      "step": 815
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966800004762365e-05,
+      "loss": 0.9381,
+      "step": 816
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966621369750366e-05,
+      "loss": 0.8726,
+      "step": 817
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996644225625092e-05,
+      "loss": 0.916,
+      "step": 818
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996626266427262e-05,
+      "loss": 0.8913,
+      "step": 819
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966082593824095e-05,
+      "loss": 0.8977,
+      "step": 820
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9965902044913987e-05,
+      "loss": 0.9348,
+      "step": 821
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9965721017550958e-05,
+      "loss": 0.9063,
+      "step": 822
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9965539511743707e-05,
+      "loss": 0.9158,
+      "step": 823
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9965357527500945e-05,
+      "loss": 0.4536,
+      "step": 824
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996517506483141e-05,
+      "loss": 0.9169,
+      "step": 825
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964992123743857e-05,
+      "loss": 0.8943,
+      "step": 826
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964808704247073e-05,
+      "loss": 0.85,
+      "step": 827
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964624806349864e-05,
+      "loss": 0.8841,
+      "step": 828
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964440430061058e-05,
+      "loss": 0.4891,
+      "step": 829
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996425557538951e-05,
+      "loss": 0.9277,
+      "step": 830
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964070242344087e-05,
+      "loss": 0.4965,
+      "step": 831
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9963884430933694e-05,
+      "loss": 0.903,
+      "step": 832
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9963698141167247e-05,
+      "loss": 0.8918,
+      "step": 833
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9963511373053693e-05,
+      "loss": 0.9507,
+      "step": 834
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9963324126601994e-05,
+      "loss": 0.9619,
+      "step": 835
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9963136401821145e-05,
+      "loss": 0.4968,
+      "step": 836
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9962948198720156e-05,
+      "loss": 0.5151,
+      "step": 837
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996275951730806e-05,
+      "loss": 0.9387,
+      "step": 838
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9962570357593918e-05,
+      "loss": 0.9352,
+      "step": 839
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996238071958681e-05,
+      "loss": 0.905,
+      "step": 840
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9962190603295845e-05,
+      "loss": 0.8639,
+      "step": 841
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9962000008730145e-05,
+      "loss": 0.9061,
+      "step": 842
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996180893589886e-05,
+      "loss": 0.9327,
+      "step": 843
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9961617384811166e-05,
+      "loss": 0.9247,
+      "step": 844
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9961425355476262e-05,
+      "loss": 0.9502,
+      "step": 845
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996123284790336e-05,
+      "loss": 0.8979,
+      "step": 846
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9961039862101708e-05,
+      "loss": 0.8919,
+      "step": 847
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9960846398080565e-05,
+      "loss": 0.917,
+      "step": 848
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9960652455849224e-05,
+      "loss": 0.9116,
+      "step": 849
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9960458035416997e-05,
+      "loss": 0.8877,
+      "step": 850
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9960263136793214e-05,
+      "loss": 0.9118,
+      "step": 851
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9960067759987232e-05,
+      "loss": 0.9393,
+      "step": 852
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995987190500843e-05,
+      "loss": 0.4683,
+      "step": 853
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9959675571866217e-05,
+      "loss": 0.4931,
+      "step": 854
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9959478760570013e-05,
+      "loss": 0.9186,
+      "step": 855
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995928147112927e-05,
+      "loss": 0.9127,
+      "step": 856
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995908370355346e-05,
+      "loss": 0.9485,
+      "step": 857
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9958885457852074e-05,
+      "loss": 0.5339,
+      "step": 858
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995868673403463e-05,
+      "loss": 0.8626,
+      "step": 859
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995848753211067e-05,
+      "loss": 0.9388,
+      "step": 860
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9958287852089757e-05,
+      "loss": 1.0069,
+      "step": 861
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995808769398148e-05,
+      "loss": 0.5031,
+      "step": 862
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9957887057795445e-05,
+      "loss": 0.9475,
+      "step": 863
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9957685943541282e-05,
+      "loss": 0.963,
+      "step": 864
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9957484351228655e-05,
+      "loss": 0.4398,
+      "step": 865
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9957282280867235e-05,
+      "loss": 0.9276,
+      "step": 866
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9957079732466725e-05,
+      "loss": 0.9143,
+      "step": 867
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9956876706036852e-05,
+      "loss": 0.8859,
+      "step": 868
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9956673201587358e-05,
+      "loss": 0.8889,
+      "step": 869
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9956469219128016e-05,
+      "loss": 0.9062,
+      "step": 870
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995626475866862e-05,
+      "loss": 0.8998,
+      "step": 871
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9956059820218982e-05,
+      "loss": 0.8996,
+      "step": 872
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9955854403788945e-05,
+      "loss": 0.8553,
+      "step": 873
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995564850938837e-05,
+      "loss": 0.9328,
+      "step": 874
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995544213702714e-05,
+      "loss": 0.5073,
+      "step": 875
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995523528671516e-05,
+      "loss": 0.9318,
+      "step": 876
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9955027958462373e-05,
+      "loss": 0.9704,
+      "step": 877
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9954820152278715e-05,
+      "loss": 0.8966,
+      "step": 878
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995461186817418e-05,
+      "loss": 0.9514,
+      "step": 879
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9954403106158753e-05,
+      "loss": 0.8971,
+      "step": 880
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9954193866242462e-05,
+      "loss": 0.9398,
+      "step": 881
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9953984148435353e-05,
+      "loss": 0.9231,
+      "step": 882
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99537739527475e-05,
+      "loss": 0.9312,
+      "step": 883
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9953563279188985e-05,
+      "loss": 0.9365,
+      "step": 884
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9953352127769926e-05,
+      "loss": 0.9593,
+      "step": 885
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9953140498500458e-05,
+      "loss": 0.9125,
+      "step": 886
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9952928391390743e-05,
+      "loss": 0.8518,
+      "step": 887
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995271580645097e-05,
+      "loss": 0.9157,
+      "step": 888
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9952502743691334e-05,
+      "loss": 0.9008,
+      "step": 889
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995228920312207e-05,
+      "loss": 0.8919,
+      "step": 890
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995207518475343e-05,
+      "loss": 0.5214,
+      "step": 891
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9951860688595687e-05,
+      "loss": 0.4949,
+      "step": 892
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9951645714659137e-05,
+      "loss": 0.8912,
+      "step": 893
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9951430262954106e-05,
+      "loss": 0.9314,
+      "step": 894
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9951214333490933e-05,
+      "loss": 0.8874,
+      "step": 895
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995099792627999e-05,
+      "loss": 0.4931,
+      "step": 896
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995078104133166e-05,
+      "loss": 0.5593,
+      "step": 897
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9950563678656357e-05,
+      "loss": 0.9182,
+      "step": 898
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995034583826452e-05,
+      "loss": 0.9035,
+      "step": 899
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9950127520166603e-05,
+      "loss": 0.4609,
+      "step": 900
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9949908724373092e-05,
+      "loss": 0.8857,
+      "step": 901
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9949689450894486e-05,
+      "loss": 0.4611,
+      "step": 902
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9949469699741315e-05,
+      "loss": 0.9597,
+      "step": 903
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9949249470924128e-05,
+      "loss": 0.9164,
+      "step": 904
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9949028764453498e-05,
+      "loss": 0.9278,
+      "step": 905
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9948807580340023e-05,
+      "loss": 0.915,
+      "step": 906
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9948585918594315e-05,
+      "loss": 0.889,
+      "step": 907
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9948363779227027e-05,
+      "loss": 0.8842,
+      "step": 908
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9948141162248814e-05,
+      "loss": 0.5019,
+      "step": 909
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9947918067670372e-05,
+      "loss": 0.8885,
+      "step": 910
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99476944955024e-05,
+      "loss": 0.8533,
+      "step": 911
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9947470445755642e-05,
+      "loss": 0.9139,
+      "step": 912
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994724591844085e-05,
+      "loss": 0.8843,
+      "step": 913
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9947020913568807e-05,
+      "loss": 0.9241,
+      "step": 914
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9946795431150307e-05,
+      "loss": 0.4919,
+      "step": 915
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9946569471196183e-05,
+      "loss": 0.5164,
+      "step": 916
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9946343033717278e-05,
+      "loss": 0.921,
+      "step": 917
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994611611872447e-05,
+      "loss": 0.9036,
+      "step": 918
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9945888726228646e-05,
+      "loss": 0.5026,
+      "step": 919
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9945660856240726e-05,
+      "loss": 0.4733,
+      "step": 920
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9945432508771646e-05,
+      "loss": 0.878,
+      "step": 921
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9945203683832373e-05,
+      "loss": 0.8909,
+      "step": 922
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9944974381433897e-05,
+      "loss": 0.9313,
+      "step": 923
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9944744601587217e-05,
+      "loss": 0.9237,
+      "step": 924
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994451434430337e-05,
+      "loss": 0.4737,
+      "step": 925
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9944283609593405e-05,
+      "loss": 0.9134,
+      "step": 926
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9944052397468408e-05,
+      "loss": 0.909,
+      "step": 927
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9943820707939473e-05,
+      "loss": 0.9191,
+      "step": 928
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994358854101772e-05,
+      "loss": 0.9056,
+      "step": 929
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9943355896714306e-05,
+      "loss": 0.4938,
+      "step": 930
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9943122775040394e-05,
+      "loss": 0.5055,
+      "step": 931
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9942889176007173e-05,
+      "loss": 0.9719,
+      "step": 932
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9942655099625862e-05,
+      "loss": 0.9021,
+      "step": 933
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99424205459077e-05,
+      "loss": 0.8533,
+      "step": 934
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994218551486394e-05,
+      "loss": 0.4961,
+      "step": 935
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9941950006505873e-05,
+      "loss": 0.4941,
+      "step": 936
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9941714020844807e-05,
+      "loss": 0.5082,
+      "step": 937
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9941477557892063e-05,
+      "loss": 0.9179,
+      "step": 938
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9941240617658998e-05,
+      "loss": 0.9184,
+      "step": 939
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9941003200156988e-05,
+      "loss": 0.9205,
+      "step": 940
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9940765305397436e-05,
+      "loss": 0.8982,
+      "step": 941
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9940526933391752e-05,
+      "loss": 0.9391,
+      "step": 942
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994028808415139e-05,
+      "loss": 0.951,
+      "step": 943
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9940048757687807e-05,
+      "loss": 0.8842,
+      "step": 944
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99398089540125e-05,
+      "loss": 0.8781,
+      "step": 945
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9939568673136982e-05,
+      "loss": 0.9086,
+      "step": 946
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9939327915072784e-05,
+      "loss": 0.9032,
+      "step": 947
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993908667983147e-05,
+      "loss": 0.8766,
+      "step": 948
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9938844967424617e-05,
+      "loss": 0.8982,
+      "step": 949
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993860277786383e-05,
+      "loss": 0.9712,
+      "step": 950
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993836011116074e-05,
+      "loss": 0.9111,
+      "step": 951
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9938116967326996e-05,
+      "loss": 0.4735,
+      "step": 952
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9937873346374266e-05,
+      "loss": 0.9518,
+      "step": 953
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993762924831425e-05,
+      "loss": 0.9179,
+      "step": 954
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9937384673158672e-05,
+      "loss": 0.4214,
+      "step": 955
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993713962091926e-05,
+      "loss": 0.5368,
+      "step": 956
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993689409160779e-05,
+      "loss": 0.4878,
+      "step": 957
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993664808523605e-05,
+      "loss": 0.8428,
+      "step": 958
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9936401601815846e-05,
+      "loss": 0.4538,
+      "step": 959
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993615464135901e-05,
+      "loss": 0.9134,
+      "step": 960
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9935907203877404e-05,
+      "loss": 0.8984,
+      "step": 961
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99356592893829e-05,
+      "loss": 0.5191,
+      "step": 962
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993541089788741e-05,
+      "loss": 0.9754,
+      "step": 963
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993516202940285e-05,
+      "loss": 0.4797,
+      "step": 964
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9934912683941172e-05,
+      "loss": 0.9298,
+      "step": 965
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9934662861514344e-05,
+      "loss": 0.9426,
+      "step": 966
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9934412562134367e-05,
+      "loss": 0.9234,
+      "step": 967
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9934161785813248e-05,
+      "loss": 0.517,
+      "step": 968
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993391053256303e-05,
+      "loss": 0.9828,
+      "step": 969
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9933658802395782e-05,
+      "loss": 0.9573,
+      "step": 970
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9933406595323576e-05,
+      "loss": 0.8529,
+      "step": 971
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9933153911358532e-05,
+      "loss": 0.9065,
+      "step": 972
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9932900750512777e-05,
+      "loss": 0.931,
+      "step": 973
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9932647112798468e-05,
+      "loss": 0.9498,
+      "step": 974
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9932392998227773e-05,
+      "loss": 0.9028,
+      "step": 975
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99321384068129e-05,
+      "loss": 0.9087,
+      "step": 976
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993188333856607e-05,
+      "loss": 0.8718,
+      "step": 977
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993162779349953e-05,
+      "loss": 0.4912,
+      "step": 978
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9931371771625545e-05,
+      "loss": 0.9003,
+      "step": 979
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9931115272956405e-05,
+      "loss": 0.9526,
+      "step": 980
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993085829750443e-05,
+      "loss": 0.8551,
+      "step": 981
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9930600845281953e-05,
+      "loss": 0.9047,
+      "step": 982
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9930342916301333e-05,
+      "loss": 0.9153,
+      "step": 983
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9930084510574957e-05,
+      "loss": 0.888,
+      "step": 984
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992982562811523e-05,
+      "loss": 0.4946,
+      "step": 985
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9929566268934577e-05,
+      "loss": 0.9624,
+      "step": 986
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9929306433045454e-05,
+      "loss": 0.8802,
+      "step": 987
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9929046120460334e-05,
+      "loss": 0.4757,
+      "step": 988
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9928785331191712e-05,
+      "loss": 0.8687,
+      "step": 989
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992852406525211e-05,
+      "loss": 0.9614,
+      "step": 990
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9928262322654074e-05,
+      "loss": 0.4937,
+      "step": 991
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9928000103410166e-05,
+      "loss": 0.8891,
+      "step": 992
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9927737407532972e-05,
+      "loss": 0.9112,
+      "step": 993
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9927474235035113e-05,
+      "loss": 0.4496,
+      "step": 994
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992721058592922e-05,
+      "loss": 0.8941,
+      "step": 995
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9926946460227942e-05,
+      "loss": 0.9436,
+      "step": 996
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992668185794397e-05,
+      "loss": 0.5096,
+      "step": 997
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9926416779090005e-05,
+      "loss": 0.4613,
+      "step": 998
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992615122367877e-05,
+      "loss": 0.5115,
+      "step": 999
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9925885191723018e-05,
+      "loss": 0.9063,
+      "step": 1000
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9925618683235515e-05,
+      "loss": 0.4459,
+      "step": 1001
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9925351698229065e-05,
+      "loss": 0.951,
+      "step": 1002
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992508423671648e-05,
+      "loss": 0.5253,
+      "step": 1003
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9924816298710597e-05,
+      "loss": 0.9021,
+      "step": 1004
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9924547884224284e-05,
+      "loss": 0.5181,
+      "step": 1005
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992427899327043e-05,
+      "loss": 0.8945,
+      "step": 1006
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992400962586194e-05,
+      "loss": 0.8983,
+      "step": 1007
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9923739782011744e-05,
+      "loss": 0.9331,
+      "step": 1008
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9923469461732803e-05,
+      "loss": 0.9149,
+      "step": 1009
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992319866503809e-05,
+      "loss": 0.5132,
+      "step": 1010
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9922927391940606e-05,
+      "loss": 0.4728,
+      "step": 1011
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9922655642453377e-05,
+      "loss": 0.9282,
+      "step": 1012
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992238341658945e-05,
+      "loss": 0.8959,
+      "step": 1013
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992211071436189e-05,
+      "loss": 0.9295,
+      "step": 1014
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9921837535783793e-05,
+      "loss": 0.5043,
+      "step": 1015
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9921563880868273e-05,
+      "loss": 0.8912,
+      "step": 1016
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9921289749628466e-05,
+      "loss": 0.5045,
+      "step": 1017
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9921015142077538e-05,
+      "loss": 0.5013,
+      "step": 1018
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9920740058228667e-05,
+      "loss": 0.4723,
+      "step": 1019
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9920464498095063e-05,
+      "loss": 0.9077,
+      "step": 1020
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992018846168995e-05,
+      "loss": 0.8736,
+      "step": 1021
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9919911949026586e-05,
+      "loss": 0.9007,
+      "step": 1022
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9919634960118247e-05,
+      "loss": 0.9077,
+      "step": 1023
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9919357494978228e-05,
+      "loss": 0.9316,
+      "step": 1024
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9919079553619846e-05,
+      "loss": 0.9124,
+      "step": 1025
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9918801136056452e-05,
+      "loss": 0.9165,
+      "step": 1026
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9918522242301407e-05,
+      "loss": 0.9639,
+      "step": 1027
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9918242872368103e-05,
+      "loss": 0.9445,
+      "step": 1028
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9917963026269955e-05,
+      "loss": 0.519,
+      "step": 1029
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991768270402039e-05,
+      "loss": 0.8787,
+      "step": 1030
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9917401905632875e-05,
+      "loss": 0.9437,
+      "step": 1031
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991712063112089e-05,
+      "loss": 0.4735,
+      "step": 1032
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9916838880497928e-05,
+      "loss": 0.8533,
+      "step": 1033
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9916556653777527e-05,
+      "loss": 0.876,
+      "step": 1034
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991627395097323e-05,
+      "loss": 0.9128,
+      "step": 1035
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9915990772098615e-05,
+      "loss": 0.9063,
+      "step": 1036
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9915707117167275e-05,
+      "loss": 0.9168,
+      "step": 1037
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9915422986192823e-05,
+      "loss": 0.496,
+      "step": 1038
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9915138379188908e-05,
+      "loss": 0.9184,
+      "step": 1039
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9914853296169186e-05,
+      "loss": 0.5343,
+      "step": 1040
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991456773714735e-05,
+      "loss": 0.9135,
+      "step": 1041
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9914281702137104e-05,
+      "loss": 0.9067,
+      "step": 1042
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9913995191152184e-05,
+      "loss": 0.8967,
+      "step": 1043
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9913708204206342e-05,
+      "loss": 0.8792,
+      "step": 1044
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9913420741313362e-05,
+      "loss": 0.9218,
+      "step": 1045
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9913132802487036e-05,
+      "loss": 0.9232,
+      "step": 1046
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9912844387741194e-05,
+      "loss": 0.8743,
+      "step": 1047
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991255549708968e-05,
+      "loss": 0.9806,
+      "step": 1048
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9912266130546367e-05,
+      "loss": 0.5264,
+      "step": 1049
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9911976288125145e-05,
+      "loss": 0.5368,
+      "step": 1050
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991168596983992e-05,
+      "loss": 0.9002,
+      "step": 1051
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9911395175704645e-05,
+      "loss": 0.9328,
+      "step": 1052
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9911103905733277e-05,
+      "loss": 0.898,
+      "step": 1053
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991081215993979e-05,
+      "loss": 0.5028,
+      "step": 1054
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99105199383382e-05,
+      "loss": 0.9206,
+      "step": 1055
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9910227240942532e-05,
+      "loss": 0.9452,
+      "step": 1056
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990993406776684e-05,
+      "loss": 0.8867,
+      "step": 1057
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9909640418825198e-05,
+      "loss": 0.9041,
+      "step": 1058
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9909346294131705e-05,
+      "loss": 0.8816,
+      "step": 1059
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990905169370048e-05,
+      "loss": 0.8934,
+      "step": 1060
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9908756617545665e-05,
+      "loss": 0.9007,
+      "step": 1061
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9908461065681433e-05,
+      "loss": 0.8592,
+      "step": 1062
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9908165038121966e-05,
+      "loss": 0.4614,
+      "step": 1063
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9907868534881478e-05,
+      "loss": 0.8976,
+      "step": 1064
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9907571555974205e-05,
+      "loss": 0.8756,
+      "step": 1065
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9907274101414405e-05,
+      "loss": 0.8865,
+      "step": 1066
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9906976171216355e-05,
+      "loss": 0.88,
+      "step": 1067
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9906677765394364e-05,
+      "loss": 0.5588,
+      "step": 1068
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9906378883962753e-05,
+      "loss": 0.9024,
+      "step": 1069
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9906079526935874e-05,
+      "loss": 0.5088,
+      "step": 1070
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99057796943281e-05,
+      "loss": 0.8489,
+      "step": 1071
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990547938615382e-05,
+      "loss": 0.9221,
+      "step": 1072
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9905178602427457e-05,
+      "loss": 0.856,
+      "step": 1073
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9904877343163452e-05,
+      "loss": 0.9412,
+      "step": 1074
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990457560837626e-05,
+      "loss": 0.4993,
+      "step": 1075
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9904273398080374e-05,
+      "loss": 0.9118,
+      "step": 1076
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9903970712290303e-05,
+      "loss": 0.9216,
+      "step": 1077
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9903667551020577e-05,
+      "loss": 0.9016,
+      "step": 1078
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990336391428575e-05,
+      "loss": 0.8747,
+      "step": 1079
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9903059802100398e-05,
+      "loss": 0.5115,
+      "step": 1080
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9902755214479126e-05,
+      "loss": 0.9138,
+      "step": 1081
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9902450151436553e-05,
+      "loss": 0.9045,
+      "step": 1082
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9902144612987323e-05,
+      "loss": 0.8941,
+      "step": 1083
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990183859914611e-05,
+      "loss": 0.8849,
+      "step": 1084
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99015321099276e-05,
+      "loss": 0.8485,
+      "step": 1085
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990122514534651e-05,
+      "loss": 0.4747,
+      "step": 1086
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990091770541758e-05,
+      "loss": 0.8871,
+      "step": 1087
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9900609790155564e-05,
+      "loss": 0.4851,
+      "step": 1088
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990030139957525e-05,
+      "loss": 0.4907,
+      "step": 1089
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9899992533691436e-05,
+      "loss": 0.8904,
+      "step": 1090
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989968319251896e-05,
+      "loss": 0.4614,
+      "step": 1091
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9899373376072665e-05,
+      "loss": 0.9046,
+      "step": 1092
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9899063084367435e-05,
+      "loss": 0.8966,
+      "step": 1093
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9898752317418152e-05,
+      "loss": 0.4744,
+      "step": 1094
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989844107523975e-05,
+      "loss": 0.9431,
+      "step": 1095
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9898129357847162e-05,
+      "loss": 0.8722,
+      "step": 1096
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9897817165255357e-05,
+      "loss": 0.4752,
+      "step": 1097
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9897504497479324e-05,
+      "loss": 0.8308,
+      "step": 1098
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9897191354534068e-05,
+      "loss": 0.9071,
+      "step": 1099
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9896877736434632e-05,
+      "loss": 0.9308,
+      "step": 1100
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989656364319607e-05,
+      "loss": 0.9207,
+      "step": 1101
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9896249074833452e-05,
+      "loss": 0.8723,
+      "step": 1102
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989593403136189e-05,
+      "loss": 0.466,
+      "step": 1103
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9895618512796507e-05,
+      "loss": 0.4797,
+      "step": 1104
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9895302519152448e-05,
+      "loss": 0.4486,
+      "step": 1105
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9894986050444885e-05,
+      "loss": 0.8857,
+      "step": 1106
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989466910668901e-05,
+      "loss": 0.8439,
+      "step": 1107
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9894351687900047e-05,
+      "loss": 0.8802,
+      "step": 1108
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9894033794093222e-05,
+      "loss": 0.9097,
+      "step": 1109
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9893715425283804e-05,
+      "loss": 0.9409,
+      "step": 1110
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9893396581487082e-05,
+      "loss": 0.92,
+      "step": 1111
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9893077262718356e-05,
+      "loss": 0.4901,
+      "step": 1112
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9892757468992957e-05,
+      "loss": 0.4769,
+      "step": 1113
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989243720032624e-05,
+      "loss": 0.8767,
+      "step": 1114
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989211645673358e-05,
+      "loss": 0.9136,
+      "step": 1115
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9891795238230376e-05,
+      "loss": 0.912,
+      "step": 1116
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989147354483205e-05,
+      "loss": 0.9314,
+      "step": 1117
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9891151376554046e-05,
+      "loss": 0.9354,
+      "step": 1118
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989082873341183e-05,
+      "loss": 0.9186,
+      "step": 1119
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9890505615420887e-05,
+      "loss": 0.4886,
+      "step": 1120
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9890182022596744e-05,
+      "loss": 0.8991,
+      "step": 1121
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988985795495492e-05,
+      "loss": 0.892,
+      "step": 1122
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988953341251098e-05,
+      "loss": 0.8734,
+      "step": 1123
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988920839528051e-05,
+      "loss": 0.9346,
+      "step": 1124
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9888882903279106e-05,
+      "loss": 0.8961,
+      "step": 1125
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98885569365224e-05,
+      "loss": 0.8872,
+      "step": 1126
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9888230495026037e-05,
+      "loss": 0.9408,
+      "step": 1127
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9887903578805688e-05,
+      "loss": 0.9557,
+      "step": 1128
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9887576187877054e-05,
+      "loss": 0.9115,
+      "step": 1129
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988724832225585e-05,
+      "loss": 0.8707,
+      "step": 1130
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9886919981957814e-05,
+      "loss": 0.8263,
+      "step": 1131
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9886591166998714e-05,
+      "loss": 0.8988,
+      "step": 1132
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988626187739433e-05,
+      "loss": 0.5045,
+      "step": 1133
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9885932113160478e-05,
+      "loss": 0.8942,
+      "step": 1134
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9885601874312983e-05,
+      "loss": 0.839,
+      "step": 1135
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9885271160867706e-05,
+      "loss": 0.8409,
+      "step": 1136
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9884939972840515e-05,
+      "loss": 0.8857,
+      "step": 1137
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988460831024732e-05,
+      "loss": 0.9143,
+      "step": 1138
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9884276173104037e-05,
+      "loss": 0.4629,
+      "step": 1139
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9883943561426616e-05,
+      "loss": 0.8788,
+      "step": 1140
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9883610475231025e-05,
+      "loss": 0.9381,
+      "step": 1141
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988327691453325e-05,
+      "loss": 0.8768,
+      "step": 1142
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9882942879349308e-05,
+      "loss": 0.8502,
+      "step": 1143
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9882608369695235e-05,
+      "loss": 0.8872,
+      "step": 1144
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9882273385587094e-05,
+      "loss": 0.5087,
+      "step": 1145
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9881937927040967e-05,
+      "loss": 0.4954,
+      "step": 1146
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9881601994072953e-05,
+      "loss": 0.9241,
+      "step": 1147
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9881265586699187e-05,
+      "loss": 0.4911,
+      "step": 1148
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9880928704935813e-05,
+      "loss": 0.9208,
+      "step": 1149
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988059134879901e-05,
+      "loss": 0.9429,
+      "step": 1150
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9880253518304968e-05,
+      "loss": 0.9039,
+      "step": 1151
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987991521346991e-05,
+      "loss": 0.5209,
+      "step": 1152
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9879576434310082e-05,
+      "loss": 0.8863,
+      "step": 1153
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987923718084174e-05,
+      "loss": 0.8791,
+      "step": 1154
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9878897453081175e-05,
+      "loss": 0.8742,
+      "step": 1155
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9878557251044698e-05,
+      "loss": 0.8978,
+      "step": 1156
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987821657474864e-05,
+      "loss": 0.4569,
+      "step": 1157
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987787542420936e-05,
+      "loss": 0.9149,
+      "step": 1158
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987753379944323e-05,
+      "loss": 0.9463,
+      "step": 1159
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9877191700466654e-05,
+      "loss": 0.4729,
+      "step": 1160
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9876849127296055e-05,
+      "loss": 0.4873,
+      "step": 1161
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9876506079947884e-05,
+      "loss": 0.9129,
+      "step": 1162
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987616255843861e-05,
+      "loss": 0.8866,
+      "step": 1163
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987581856278472e-05,
+      "loss": 0.8831,
+      "step": 1164
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9875474093002725e-05,
+      "loss": 0.882,
+      "step": 1165
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9875129149109174e-05,
+      "loss": 0.4961,
+      "step": 1166
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9874783731120627e-05,
+      "loss": 0.8623,
+      "step": 1167
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9874437839053658e-05,
+      "loss": 0.8864,
+      "step": 1168
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9874091472924877e-05,
+      "loss": 0.8545,
+      "step": 1169
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9873744632750914e-05,
+      "loss": 0.9013,
+      "step": 1170
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987339731854842e-05,
+      "loss": 0.9353,
+      "step": 1171
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987304953033407e-05,
+      "loss": 0.4703,
+      "step": 1172
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9872701268124556e-05,
+      "loss": 0.4735,
+      "step": 1173
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9872352531936605e-05,
+      "loss": 0.8938,
+      "step": 1174
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9872003321786953e-05,
+      "loss": 0.9233,
+      "step": 1175
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9871653637692372e-05,
+      "loss": 0.462,
+      "step": 1176
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9871303479669644e-05,
+      "loss": 0.8826,
+      "step": 1177
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987095284773558e-05,
+      "loss": 0.9075,
+      "step": 1178
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9870601741907016e-05,
+      "loss": 0.9562,
+      "step": 1179
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987025016220081e-05,
+      "loss": 0.4709,
+      "step": 1180
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9869898108633834e-05,
+      "loss": 0.8888,
+      "step": 1181
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9869545581223e-05,
+      "loss": 0.5308,
+      "step": 1182
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9869192579985225e-05,
+      "loss": 0.864,
+      "step": 1183
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9868839104937457e-05,
+      "loss": 0.941,
+      "step": 1184
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986848515609667e-05,
+      "loss": 0.5449,
+      "step": 1185
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986813073347985e-05,
+      "loss": 0.9501,
+      "step": 1186
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9867775837104018e-05,
+      "loss": 0.8815,
+      "step": 1187
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986742046698621e-05,
+      "loss": 0.5096,
+      "step": 1188
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9867064623143487e-05,
+      "loss": 0.4748,
+      "step": 1189
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9866708305592936e-05,
+      "loss": 0.4507,
+      "step": 1190
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9866351514351656e-05,
+      "loss": 0.9027,
+      "step": 1191
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9865994249436783e-05,
+      "loss": 0.8593,
+      "step": 1192
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9865636510865466e-05,
+      "loss": 0.4974,
+      "step": 1193
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9865278298654884e-05,
+      "loss": 0.9119,
+      "step": 1194
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9864919612822224e-05,
+      "loss": 0.8508,
+      "step": 1195
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986456045338472e-05,
+      "loss": 0.8233,
+      "step": 1196
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9864200820359604e-05,
+      "loss": 0.5448,
+      "step": 1197
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9863840713764145e-05,
+      "loss": 0.8799,
+      "step": 1198
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986348013361563e-05,
+      "loss": 0.9685,
+      "step": 1199
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9863119079931377e-05,
+      "loss": 0.8863,
+      "step": 1200
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9862757552728712e-05,
+      "loss": 0.8509,
+      "step": 1201
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9862395552024997e-05,
+      "loss": 0.5074,
+      "step": 1202
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9862033077837602e-05,
+      "loss": 0.4882,
+      "step": 1203
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986167013018394e-05,
+      "loss": 0.8807,
+      "step": 1204
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986130670908143e-05,
+      "loss": 0.9451,
+      "step": 1205
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9860942814547522e-05,
+      "loss": 0.4626,
+      "step": 1206
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9860578446599684e-05,
+      "loss": 0.4956,
+      "step": 1207
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986021360525541e-05,
+      "loss": 0.8736,
+      "step": 1208
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9859848290532217e-05,
+      "loss": 0.8791,
+      "step": 1209
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985948250244764e-05,
+      "loss": 0.4933,
+      "step": 1210
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9859116241019246e-05,
+      "loss": 0.4973,
+      "step": 1211
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985874950626461e-05,
+      "loss": 0.8935,
+      "step": 1212
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9858382298201348e-05,
+      "loss": 0.8532,
+      "step": 1213
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9858014616847082e-05,
+      "loss": 0.8903,
+      "step": 1214
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9857646462219466e-05,
+      "loss": 0.4847,
+      "step": 1215
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9857277834336175e-05,
+      "loss": 0.5232,
+      "step": 1216
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985690873321491e-05,
+      "loss": 0.8844,
+      "step": 1217
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9856539158873388e-05,
+      "loss": 0.8903,
+      "step": 1218
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985616911132935e-05,
+      "loss": 0.8432,
+      "step": 1219
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9855798590600565e-05,
+      "loss": 0.9175,
+      "step": 1220
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9855427596704818e-05,
+      "loss": 0.9525,
+      "step": 1221
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9855056129659926e-05,
+      "loss": 0.9533,
+      "step": 1222
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9854684189483714e-05,
+      "loss": 0.9035,
+      "step": 1223
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9854311776194048e-05,
+      "loss": 0.9071,
+      "step": 1224
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98539388898088e-05,
+      "loss": 0.5176,
+      "step": 1225
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985356553034587e-05,
+      "loss": 0.9403,
+      "step": 1226
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9853191697823195e-05,
+      "loss": 0.8988,
+      "step": 1227
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985281739225871e-05,
+      "loss": 0.9005,
+      "step": 1228
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985244261367039e-05,
+      "loss": 0.8927,
+      "step": 1229
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985206736207623e-05,
+      "loss": 0.8811,
+      "step": 1230
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9851691637494235e-05,
+      "loss": 0.8961,
+      "step": 1231
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9851315439942457e-05,
+      "loss": 0.9009,
+      "step": 1232
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985093876943895e-05,
+      "loss": 0.9357,
+      "step": 1233
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9850561626001796e-05,
+      "loss": 0.9487,
+      "step": 1234
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9850184009649106e-05,
+      "loss": 0.5115,
+      "step": 1235
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9849805920399002e-05,
+      "loss": 0.9388,
+      "step": 1236
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9849427358269645e-05,
+      "loss": 0.8424,
+      "step": 1237
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98490483232792e-05,
+      "loss": 0.9353,
+      "step": 1238
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9848668815445873e-05,
+      "loss": 0.952,
+      "step": 1239
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9848288834787878e-05,
+      "loss": 0.8888,
+      "step": 1240
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9847908381323457e-05,
+      "loss": 0.817,
+      "step": 1241
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9847527455070877e-05,
+      "loss": 0.4842,
+      "step": 1242
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984714605604843e-05,
+      "loss": 0.8934,
+      "step": 1243
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9846764184274416e-05,
+      "loss": 0.8754,
+      "step": 1244
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9846381839767178e-05,
+      "loss": 0.451,
+      "step": 1245
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984599902254507e-05,
+      "loss": 0.9495,
+      "step": 1246
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9845615732626467e-05,
+      "loss": 0.946,
+      "step": 1247
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9845231970029774e-05,
+      "loss": 0.9865,
+      "step": 1248
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9844847734773413e-05,
+      "loss": 0.9082,
+      "step": 1249
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9844463026875832e-05,
+      "loss": 0.8597,
+      "step": 1250
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98440778463555e-05,
+      "loss": 0.5087,
+      "step": 1251
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984369219323091e-05,
+      "loss": 0.4826,
+      "step": 1252
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9843306067520574e-05,
+      "loss": 0.9182,
+      "step": 1253
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9842919469243034e-05,
+      "loss": 0.5173,
+      "step": 1254
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9842532398416844e-05,
+      "loss": 0.8919,
+      "step": 1255
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9842144855060595e-05,
+      "loss": 0.4708,
+      "step": 1256
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9841756839192885e-05,
+      "loss": 0.896,
+      "step": 1257
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9841368350832345e-05,
+      "loss": 0.442,
+      "step": 1258
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.984097938999763e-05,
+      "loss": 0.8879,
+      "step": 1259
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9840589956707404e-05,
+      "loss": 0.9024,
+      "step": 1260
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9840200050980372e-05,
+      "loss": 0.8748,
+      "step": 1261
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9839809672835252e-05,
+      "loss": 0.9042,
+      "step": 1262
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9839418822290784e-05,
+      "loss": 0.9447,
+      "step": 1263
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983902749936573e-05,
+      "loss": 0.9104,
+      "step": 1264
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983863570407888e-05,
+      "loss": 0.4959,
+      "step": 1265
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9838243436449044e-05,
+      "loss": 0.4759,
+      "step": 1266
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9837850696495052e-05,
+      "loss": 0.8838,
+      "step": 1267
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983745748423576e-05,
+      "loss": 0.4968,
+      "step": 1268
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9837063799690045e-05,
+      "loss": 0.4515,
+      "step": 1269
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983666964287681e-05,
+      "loss": 0.8902,
+      "step": 1270
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9836275013814975e-05,
+      "loss": 0.8821,
+      "step": 1271
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983587991252349e-05,
+      "loss": 0.9088,
+      "step": 1272
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983548433902132e-05,
+      "loss": 0.8683,
+      "step": 1273
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9835088293327455e-05,
+      "loss": 0.9272,
+      "step": 1274
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9834691775460913e-05,
+      "loss": 0.879,
+      "step": 1275
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9834294785440724e-05,
+      "loss": 0.8969,
+      "step": 1276
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9833897323285952e-05,
+      "loss": 0.9459,
+      "step": 1277
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983349938901568e-05,
+      "loss": 0.8529,
+      "step": 1278
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983310098264901e-05,
+      "loss": 0.8694,
+      "step": 1279
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983270210420507e-05,
+      "loss": 0.8943,
+      "step": 1280
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9832302753703007e-05,
+      "loss": 0.482,
+      "step": 1281
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9831902931161995e-05,
+      "loss": 0.9682,
+      "step": 1282
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983150263660123e-05,
+      "loss": 0.8733,
+      "step": 1283
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983110187003993e-05,
+      "loss": 0.8911,
+      "step": 1284
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9830700631497336e-05,
+      "loss": 0.4679,
+      "step": 1285
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9830298920992707e-05,
+      "loss": 0.5433,
+      "step": 1286
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9829896738545334e-05,
+      "loss": 0.8763,
+      "step": 1287
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982949408417452e-05,
+      "loss": 0.9189,
+      "step": 1288
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9829090957899598e-05,
+      "loss": 0.8395,
+      "step": 1289
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9828687359739927e-05,
+      "loss": 0.8495,
+      "step": 1290
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9828283289714875e-05,
+      "loss": 0.8659,
+      "step": 1291
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9827878747843848e-05,
+      "loss": 0.8568,
+      "step": 1292
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9827473734146263e-05,
+      "loss": 0.8894,
+      "step": 1293
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9827068248641566e-05,
+      "loss": 0.9253,
+      "step": 1294
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9826662291349224e-05,
+      "loss": 0.4877,
+      "step": 1295
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982625586228873e-05,
+      "loss": 0.8988,
+      "step": 1296
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9825848961479584e-05,
+      "loss": 0.846,
+      "step": 1297
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9825441588941334e-05,
+      "loss": 0.5047,
+      "step": 1298
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9825033744693533e-05,
+      "loss": 0.8986,
+      "step": 1299
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982462542875576e-05,
+      "loss": 0.8616,
+      "step": 1300
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9824216641147623e-05,
+      "loss": 0.5106,
+      "step": 1301
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9823807381888742e-05,
+      "loss": 0.4796,
+      "step": 1302
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9823397650998765e-05,
+      "loss": 0.8956,
+      "step": 1303
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9822987448497367e-05,
+      "loss": 0.4581,
+      "step": 1304
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9822576774404236e-05,
+      "loss": 0.4805,
+      "step": 1305
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982216562873909e-05,
+      "loss": 0.4869,
+      "step": 1306
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9821754011521676e-05,
+      "loss": 0.9008,
+      "step": 1307
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9821341922771738e-05,
+      "loss": 0.455,
+      "step": 1308
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9820929362509076e-05,
+      "loss": 0.4318,
+      "step": 1309
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982051633075349e-05,
+      "loss": 0.936,
+      "step": 1310
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982010282752481e-05,
+      "loss": 0.4277,
+      "step": 1311
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9819688852842885e-05,
+      "loss": 0.9069,
+      "step": 1312
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981927440672759e-05,
+      "loss": 0.8896,
+      "step": 1313
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981885948919883e-05,
+      "loss": 0.4497,
+      "step": 1314
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9818444100276517e-05,
+      "loss": 0.8817,
+      "step": 1315
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9818028239980594e-05,
+      "loss": 0.8961,
+      "step": 1316
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981761190833103e-05,
+      "loss": 0.8968,
+      "step": 1317
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981719510534781e-05,
+      "loss": 0.8436,
+      "step": 1318
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9816777831050943e-05,
+      "loss": 0.4829,
+      "step": 1319
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9816360085460464e-05,
+      "loss": 0.8839,
+      "step": 1320
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9815941868596424e-05,
+      "loss": 0.8749,
+      "step": 1321
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981552318047891e-05,
+      "loss": 0.8813,
+      "step": 1322
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9815104021128014e-05,
+      "loss": 0.8837,
+      "step": 1323
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981468439056387e-05,
+      "loss": 0.8712,
+      "step": 1324
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9814264288806613e-05,
+      "loss": 0.4595,
+      "step": 1325
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9813843715876415e-05,
+      "loss": 0.8572,
+      "step": 1326
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981342267179347e-05,
+      "loss": 0.8904,
+      "step": 1327
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9813001156577985e-05,
+      "loss": 0.8855,
+      "step": 1328
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9812579170250207e-05,
+      "loss": 0.8432,
+      "step": 1329
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981215671283039e-05,
+      "loss": 0.9595,
+      "step": 1330
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9811733784338817e-05,
+      "loss": 0.4926,
+      "step": 1331
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9811310384795787e-05,
+      "loss": 0.9442,
+      "step": 1332
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9810886514221632e-05,
+      "loss": 0.9044,
+      "step": 1333
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.98104621726367e-05,
+      "loss": 0.8781,
+      "step": 1334
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9810037360061366e-05,
+      "loss": 0.9257,
+      "step": 1335
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980961207651602e-05,
+      "loss": 0.8922,
+      "step": 1336
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9809186322021084e-05,
+      "loss": 0.9046,
+      "step": 1337
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9808760096596996e-05,
+      "loss": 0.9419,
+      "step": 1338
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9808333400264218e-05,
+      "loss": 0.8518,
+      "step": 1339
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9807906233043233e-05,
+      "loss": 0.4808,
+      "step": 1340
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9807478594954556e-05,
+      "loss": 0.8836,
+      "step": 1341
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980705048601871e-05,
+      "loss": 0.873,
+      "step": 1342
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9806621906256256e-05,
+      "loss": 0.8814,
+      "step": 1343
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9806192855687762e-05,
+      "loss": 0.9102,
+      "step": 1344
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980576333433383e-05,
+      "loss": 0.9392,
+      "step": 1345
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980533334221508e-05,
+      "loss": 0.8651,
+      "step": 1346
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9804902879352156e-05,
+      "loss": 0.9103,
+      "step": 1347
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980447194576572e-05,
+      "loss": 0.8855,
+      "step": 1348
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980404054147647e-05,
+      "loss": 0.875,
+      "step": 1349
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9803608666505113e-05,
+      "loss": 0.8683,
+      "step": 1350
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9803176320872375e-05,
+      "loss": 0.8899,
+      "step": 1351
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980274350459902e-05,
+      "loss": 0.9322,
+      "step": 1352
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980231021770583e-05,
+      "loss": 0.8645,
+      "step": 1353
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9801876460213605e-05,
+      "loss": 0.8993,
+      "step": 1354
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9801442232143164e-05,
+      "loss": 0.4925,
+      "step": 1355
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980100753351536e-05,
+      "loss": 0.8756,
+      "step": 1356
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9800572364351054e-05,
+      "loss": 0.8666,
+      "step": 1357
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9800136724671143e-05,
+      "loss": 0.9319,
+      "step": 1358
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9799700614496548e-05,
+      "loss": 0.8498,
+      "step": 1359
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9799264033848196e-05,
+      "loss": 0.8976,
+      "step": 1360
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979882698274705e-05,
+      "loss": 0.8922,
+      "step": 1361
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.97983894612141e-05,
+      "loss": 0.8648,
+      "step": 1362
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9797951469270336e-05,
+      "loss": 0.8944,
+      "step": 1363
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.97975130069368e-05,
+      "loss": 0.8313,
+      "step": 1364
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9797074074234533e-05,
+      "loss": 0.4841,
+      "step": 1365
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9796634671184614e-05,
+      "loss": 0.925,
+      "step": 1366
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979619479780813e-05,
+      "loss": 0.8399,
+      "step": 1367
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9795754454126207e-05,
+      "loss": 0.8939,
+      "step": 1368
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9795313640159985e-05,
+      "loss": 0.856,
+      "step": 1369
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9794872355930623e-05,
+      "loss": 0.8417,
+      "step": 1370
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979443060145931e-05,
+      "loss": 0.8773,
+      "step": 1371
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979398837676725e-05,
+      "loss": 0.8924,
+      "step": 1372
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979354568187567e-05,
+      "loss": 0.9191,
+      "step": 1373
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979310251680584e-05,
+      "loss": 0.892,
+      "step": 1374
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979265888157902e-05,
+      "loss": 0.8653,
+      "step": 1375
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9792214776216515e-05,
+      "loss": 0.4904,
+      "step": 1376
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9791770200739643e-05,
+      "loss": 0.935,
+      "step": 1377
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979132515516975e-05,
+      "loss": 0.8919,
+      "step": 1378
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.97908796395282e-05,
+      "loss": 0.9232,
+      "step": 1379
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979043365383639e-05,
+      "loss": 0.9176,
+      "step": 1380
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978998719811572e-05,
+      "loss": 0.8971,
+      "step": 1381
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978954027238763e-05,
+      "loss": 0.9236,
+      "step": 1382
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9789092876673576e-05,
+      "loss": 0.5152,
+      "step": 1383
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9788645010995032e-05,
+      "loss": 0.8745,
+      "step": 1384
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9788196675373505e-05,
+      "loss": 0.8951,
+      "step": 1385
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978774786983052e-05,
+      "loss": 0.9114,
+      "step": 1386
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978729859438762e-05,
+      "loss": 0.891,
+      "step": 1387
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9786848849066378e-05,
+      "loss": 0.9243,
+      "step": 1388
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978639863388838e-05,
+      "loss": 0.9368,
+      "step": 1389
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9785947948875247e-05,
+      "loss": 0.9062,
+      "step": 1390
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9785496794048612e-05,
+      "loss": 0.4649,
+      "step": 1391
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9785045169430134e-05,
+      "loss": 0.8954,
+      "step": 1392
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.97845930750415e-05,
+      "loss": 0.4872,
+      "step": 1393
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9784140510904407e-05,
+      "loss": 0.4681,
+      "step": 1394
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978368747704059e-05,
+      "loss": 0.8679,
+      "step": 1395
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9783233973471792e-05,
+      "loss": 0.8836,
+      "step": 1396
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978278000021979e-05,
+      "loss": 0.9399,
+      "step": 1397
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9782325557306375e-05,
+      "loss": 0.9084,
+      "step": 1398
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9781870644753367e-05,
+      "loss": 0.8781,
+      "step": 1399
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9781415262582607e-05,
+      "loss": 0.4807,
+      "step": 1400
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9780959410815954e-05,
+      "loss": 0.4594,
+      "step": 1401
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9780503089475295e-05,
+      "loss": 0.9015,
+      "step": 1402
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9780046298582535e-05,
+      "loss": 0.926,
+      "step": 1403
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9779589038159606e-05,
+      "loss": 0.9475,
+      "step": 1404
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9779131308228466e-05,
+      "loss": 0.5095,
+      "step": 1405
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977867310881108e-05,
+      "loss": 0.894,
+      "step": 1406
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9778214439929453e-05,
+      "loss": 0.5117,
+      "step": 1407
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9777755301605602e-05,
+      "loss": 0.5161,
+      "step": 1408
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977729569386157e-05,
+      "loss": 0.9143,
+      "step": 1409
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9776835616719426e-05,
+      "loss": 0.4647,
+      "step": 1410
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977637507020125e-05,
+      "loss": 0.8433,
+      "step": 1411
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977591405432916e-05,
+      "loss": 0.5157,
+      "step": 1412
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977545256912529e-05,
+      "loss": 0.4635,
+      "step": 1413
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9774990614611788e-05,
+      "loss": 0.8582,
+      "step": 1414
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9774528190810833e-05,
+      "loss": 0.8964,
+      "step": 1415
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9774065297744633e-05,
+      "loss": 0.463,
+      "step": 1416
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97736019354354e-05,
+      "loss": 0.482,
+      "step": 1417
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977313810390539e-05,
+      "loss": 0.8945,
+      "step": 1418
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9772673803176865e-05,
+      "loss": 0.8991,
+      "step": 1419
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977220903327212e-05,
+      "loss": 0.8663,
+      "step": 1420
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9771743794213464e-05,
+      "loss": 0.9222,
+      "step": 1421
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9771278086023228e-05,
+      "loss": 0.9026,
+      "step": 1422
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9770811908723784e-05,
+      "loss": 0.5035,
+      "step": 1423
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9770345262337504e-05,
+      "loss": 0.9695,
+      "step": 1424
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9769878146886787e-05,
+      "loss": 0.8819,
+      "step": 1425
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976941056239407e-05,
+      "loss": 0.9256,
+      "step": 1426
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9768942508881794e-05,
+      "loss": 0.8927,
+      "step": 1427
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9768473986372427e-05,
+      "loss": 0.859,
+      "step": 1428
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976800499488847e-05,
+      "loss": 0.8778,
+      "step": 1429
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9767535534452435e-05,
+      "loss": 0.8738,
+      "step": 1430
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976706560508686e-05,
+      "loss": 0.9081,
+      "step": 1431
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9766595206814303e-05,
+      "loss": 0.4878,
+      "step": 1432
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976612433965735e-05,
+      "loss": 0.9188,
+      "step": 1433
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9765653003638614e-05,
+      "loss": 0.4925,
+      "step": 1434
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9765181198780713e-05,
+      "loss": 0.4918,
+      "step": 1435
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9764708925106304e-05,
+      "loss": 0.942,
+      "step": 1436
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9764236182638055e-05,
+      "loss": 0.8821,
+      "step": 1437
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9763762971398666e-05,
+      "loss": 0.8868,
+      "step": 1438
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9763289291410858e-05,
+      "loss": 0.8934,
+      "step": 1439
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976281514269736e-05,
+      "loss": 0.4904,
+      "step": 1440
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976234052528095e-05,
+      "loss": 0.8704,
+      "step": 1441
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9761865439184407e-05,
+      "loss": 0.9319,
+      "step": 1442
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976138988443054e-05,
+      "loss": 0.908,
+      "step": 1443
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9760913861042176e-05,
+      "loss": 0.8889,
+      "step": 1444
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9760437369042175e-05,
+      "loss": 0.8863,
+      "step": 1445
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975996040845341e-05,
+      "loss": 0.8863,
+      "step": 1446
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9759482979298782e-05,
+      "loss": 0.5002,
+      "step": 1447
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9759005081601205e-05,
+      "loss": 0.9354,
+      "step": 1448
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975852671538363e-05,
+      "loss": 0.8515,
+      "step": 1449
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9758047880669017e-05,
+      "loss": 0.5376,
+      "step": 1450
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975756857748036e-05,
+      "loss": 0.8946,
+      "step": 1451
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9757088805840666e-05,
+      "loss": 0.4859,
+      "step": 1452
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975660856577297e-05,
+      "loss": 0.9085,
+      "step": 1453
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9756127857300324e-05,
+      "loss": 0.9103,
+      "step": 1454
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9755646680445812e-05,
+      "loss": 0.8951,
+      "step": 1455
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9755165035232533e-05,
+      "loss": 0.9129,
+      "step": 1456
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9754682921683607e-05,
+      "loss": 0.8918,
+      "step": 1457
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9754200339822186e-05,
+      "loss": 0.4788,
+      "step": 1458
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975371728967143e-05,
+      "loss": 0.51,
+      "step": 1459
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975323377125454e-05,
+      "loss": 0.8916,
+      "step": 1460
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975274978459472e-05,
+      "loss": 0.8606,
+      "step": 1461
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975226532971521e-05,
+      "loss": 0.5091,
+      "step": 1462
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9751780406639266e-05,
+      "loss": 0.4787,
+      "step": 1463
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975129501539017e-05,
+      "loss": 0.421,
+      "step": 1464
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975080915599123e-05,
+      "loss": 0.8883,
+      "step": 1465
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975032282846576e-05,
+      "loss": 0.865,
+      "step": 1466
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9749836032837123e-05,
+      "loss": 0.8532,
+      "step": 1467
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9749348769128676e-05,
+      "loss": 0.8842,
+      "step": 1468
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974886103736382e-05,
+      "loss": 0.8806,
+      "step": 1469
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9748372837565967e-05,
+      "loss": 0.9115,
+      "step": 1470
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9747884169758556e-05,
+      "loss": 0.8566,
+      "step": 1471
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9747395033965052e-05,
+      "loss": 0.5001,
+      "step": 1472
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9746905430208928e-05,
+      "loss": 0.8983,
+      "step": 1473
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9746415358513697e-05,
+      "loss": 0.8766,
+      "step": 1474
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974592481890288e-05,
+      "loss": 0.9067,
+      "step": 1475
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974543381140004e-05,
+      "loss": 0.8778,
+      "step": 1476
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9744942336028737e-05,
+      "loss": 0.8807,
+      "step": 1477
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9744450392812573e-05,
+      "loss": 0.8844,
+      "step": 1478
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9743957981775162e-05,
+      "loss": 0.9173,
+      "step": 1479
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974346510294015e-05,
+      "loss": 0.8804,
+      "step": 1480
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9742971756331193e-05,
+      "loss": 0.8961,
+      "step": 1481
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9742477941971974e-05,
+      "loss": 0.8337,
+      "step": 1482
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9741983659886213e-05,
+      "loss": 0.8901,
+      "step": 1483
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9741488910097628e-05,
+      "loss": 0.8448,
+      "step": 1484
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9740993692629975e-05,
+      "loss": 0.9175,
+      "step": 1485
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9740498007507032e-05,
+      "loss": 0.528,
+      "step": 1486
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9740001854752592e-05,
+      "loss": 0.871,
+      "step": 1487
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9739505234390476e-05,
+      "loss": 0.9121,
+      "step": 1488
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973900814644453e-05,
+      "loss": 0.8909,
+      "step": 1489
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9738510590938612e-05,
+      "loss": 0.8692,
+      "step": 1490
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9738012567896617e-05,
+      "loss": 0.9171,
+      "step": 1491
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9737514077342447e-05,
+      "loss": 0.925,
+      "step": 1492
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9737015119300036e-05,
+      "loss": 0.9044,
+      "step": 1493
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9736515693793345e-05,
+      "loss": 0.8613,
+      "step": 1494
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973601580084634e-05,
+      "loss": 0.9252,
+      "step": 1495
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9735515440483027e-05,
+      "loss": 0.9284,
+      "step": 1496
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973501461272743e-05,
+      "loss": 0.918,
+      "step": 1497
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9734513317603587e-05,
+      "loss": 0.511,
+      "step": 1498
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973401155513557e-05,
+      "loss": 0.9048,
+      "step": 1499
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9733509325347464e-05,
+      "loss": 0.4755,
+      "step": 1500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9733006628263384e-05,
+      "loss": 0.8968,
+      "step": 1501
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9732503463907463e-05,
+      "loss": 0.8985,
+      "step": 1502
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9731999832303855e-05,
+      "loss": 0.8846,
+      "step": 1503
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9731495733476738e-05,
+      "loss": 0.8658,
+      "step": 1504
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973099116745032e-05,
+      "loss": 0.9341,
+      "step": 1505
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9730486134248822e-05,
+      "loss": 0.4854,
+      "step": 1506
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9729980633896484e-05,
+      "loss": 0.4976,
+      "step": 1507
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972947466641758e-05,
+      "loss": 0.8912,
+      "step": 1508
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9728968231836404e-05,
+      "loss": 0.9196,
+      "step": 1509
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9728461330177264e-05,
+      "loss": 0.8827,
+      "step": 1510
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9727953961464496e-05,
+      "loss": 0.865,
+      "step": 1511
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972744612572246e-05,
+      "loss": 0.8723,
+      "step": 1512
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972693782297554e-05,
+      "loss": 0.8503,
+      "step": 1513
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972642905324813e-05,
+      "loss": 0.8168,
+      "step": 1514
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9725919816564666e-05,
+      "loss": 0.8533,
+      "step": 1515
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972541011294959e-05,
+      "loss": 0.8976,
+      "step": 1516
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9724899942427372e-05,
+      "loss": 0.8802,
+      "step": 1517
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9724389305022507e-05,
+      "loss": 0.8869,
+      "step": 1518
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972387820075951e-05,
+      "loss": 0.9279,
+      "step": 1519
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9723366629662916e-05,
+      "loss": 0.8987,
+      "step": 1520
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972285459175729e-05,
+      "loss": 0.5117,
+      "step": 1521
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972234208706721e-05,
+      "loss": 0.8904,
+      "step": 1522
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972182911561728e-05,
+      "loss": 0.9318,
+      "step": 1523
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9721315677432135e-05,
+      "loss": 0.8446,
+      "step": 1524
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9720801772536412e-05,
+      "loss": 0.4583,
+      "step": 1525
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9720287400954795e-05,
+      "loss": 0.8869,
+      "step": 1526
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9719772562711973e-05,
+      "loss": 0.519,
+      "step": 1527
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9719257257832664e-05,
+      "loss": 0.4781,
+      "step": 1528
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971874148634161e-05,
+      "loss": 0.9553,
+      "step": 1529
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9718225248263563e-05,
+      "loss": 0.8662,
+      "step": 1530
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9717708543623314e-05,
+      "loss": 0.8714,
+      "step": 1531
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9717191372445673e-05,
+      "loss": 0.4542,
+      "step": 1532
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9716673734755464e-05,
+      "loss": 0.8487,
+      "step": 1533
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9716155630577536e-05,
+      "loss": 0.9066,
+      "step": 1534
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971563705993677e-05,
+      "loss": 0.9131,
+      "step": 1535
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9715118022858054e-05,
+      "loss": 0.8459,
+      "step": 1536
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971459851936631e-05,
+      "loss": 0.8177,
+      "step": 1537
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9714078549486483e-05,
+      "loss": 0.9045,
+      "step": 1538
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971355811324353e-05,
+      "loss": 0.863,
+      "step": 1539
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971303721066244e-05,
+      "loss": 0.9403,
+      "step": 1540
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971251584176822e-05,
+      "loss": 0.932,
+      "step": 1541
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97119940065859e-05,
+      "loss": 0.8864,
+      "step": 1542
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9711471705140536e-05,
+      "loss": 0.4834,
+      "step": 1543
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9710948937457197e-05,
+      "loss": 0.874,
+      "step": 1544
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9710425703560984e-05,
+      "loss": 0.8785,
+      "step": 1545
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9709902003477014e-05,
+      "loss": 0.9111,
+      "step": 1546
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9709377837230437e-05,
+      "loss": 0.863,
+      "step": 1547
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9708853204846412e-05,
+      "loss": 0.8957,
+      "step": 1548
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9708328106350125e-05,
+      "loss": 0.8712,
+      "step": 1549
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.970780254176679e-05,
+      "loss": 0.9308,
+      "step": 1550
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.970727651112163e-05,
+      "loss": 0.8788,
+      "step": 1551
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9706750014439912e-05,
+      "loss": 0.4712,
+      "step": 1552
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9706223051746907e-05,
+      "loss": 0.8866,
+      "step": 1553
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9705695623067908e-05,
+      "loss": 0.8851,
+      "step": 1554
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9705167728428242e-05,
+      "loss": 0.9025,
+      "step": 1555
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9704639367853254e-05,
+      "loss": 0.8893,
+      "step": 1556
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9704110541368305e-05,
+      "loss": 0.9122,
+      "step": 1557
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970358124899879e-05,
+      "loss": 0.8416,
+      "step": 1558
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9703051490770114e-05,
+      "loss": 0.9391,
+      "step": 1559
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970252126670771e-05,
+      "loss": 0.9398,
+      "step": 1560
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970199057683704e-05,
+      "loss": 0.8887,
+      "step": 1561
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970145942118358e-05,
+      "loss": 0.841,
+      "step": 1562
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9700927799772826e-05,
+      "loss": 0.8766,
+      "step": 1563
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9700395712630298e-05,
+      "loss": 0.8702,
+      "step": 1564
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969986315978155e-05,
+      "loss": 0.4608,
+      "step": 1565
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9699330141252144e-05,
+      "loss": 0.8797,
+      "step": 1566
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969879665706767e-05,
+      "loss": 0.9129,
+      "step": 1567
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969826270725374e-05,
+      "loss": 0.4963,
+      "step": 1568
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9697728291835992e-05,
+      "loss": 0.875,
+      "step": 1569
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969719341084008e-05,
+      "loss": 0.902,
+      "step": 1570
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9696658064291677e-05,
+      "loss": 0.9329,
+      "step": 1571
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9696122252216497e-05,
+      "loss": 0.8723,
+      "step": 1572
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9695585974640254e-05,
+      "loss": 0.8528,
+      "step": 1573
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9695049231588695e-05,
+      "loss": 0.8864,
+      "step": 1574
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9694512023087593e-05,
+      "loss": 0.4818,
+      "step": 1575
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9693974349162736e-05,
+      "loss": 0.8429,
+      "step": 1576
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969343620983994e-05,
+      "loss": 0.8513,
+      "step": 1577
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9692897605145035e-05,
+      "loss": 0.8552,
+      "step": 1578
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9692358535103886e-05,
+      "loss": 0.9057,
+      "step": 1579
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969181899974237e-05,
+      "loss": 0.8342,
+      "step": 1580
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969127899908638e-05,
+      "loss": 0.4654,
+      "step": 1581
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9690738533161858e-05,
+      "loss": 0.87,
+      "step": 1582
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969019760199474e-05,
+      "loss": 0.9509,
+      "step": 1583
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9689656205611002e-05,
+      "loss": 0.8472,
+      "step": 1584
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968911434403663e-05,
+      "loss": 0.8739,
+      "step": 1585
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968857201729764e-05,
+      "loss": 0.841,
+      "step": 1586
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968802922542007e-05,
+      "loss": 0.9136,
+      "step": 1587
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968748596842998e-05,
+      "loss": 0.4753,
+      "step": 1588
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968694224635345e-05,
+      "loss": 0.457,
+      "step": 1589
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9686398059216578e-05,
+      "loss": 0.8763,
+      "step": 1590
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9685853407045497e-05,
+      "loss": 0.9352,
+      "step": 1591
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9685308289866355e-05,
+      "loss": 0.4586,
+      "step": 1592
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968476270770532e-05,
+      "loss": 0.8853,
+      "step": 1593
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9684216660588582e-05,
+      "loss": 0.4889,
+      "step": 1594
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9683670148542363e-05,
+      "loss": 0.86,
+      "step": 1595
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9683123171592898e-05,
+      "loss": 0.9076,
+      "step": 1596
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9682575729766442e-05,
+      "loss": 0.5021,
+      "step": 1597
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9682027823089284e-05,
+      "loss": 0.474,
+      "step": 1598
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9681479451587725e-05,
+      "loss": 0.8662,
+      "step": 1599
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968093061528809e-05,
+      "loss": 0.4874,
+      "step": 1600
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968038131421673e-05,
+      "loss": 0.4546,
+      "step": 1601
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9679831548400022e-05,
+      "loss": 0.79,
+      "step": 1602
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967928131786435e-05,
+      "loss": 0.4956,
+      "step": 1603
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9678730622636136e-05,
+      "loss": 0.8901,
+      "step": 1604
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9678179462741814e-05,
+      "loss": 0.848,
+      "step": 1605
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9677627838207853e-05,
+      "loss": 0.8487,
+      "step": 1606
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9677075749060726e-05,
+      "loss": 0.4714,
+      "step": 1607
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9676523195326943e-05,
+      "loss": 0.4615,
+      "step": 1608
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9675970177033034e-05,
+      "loss": 0.9098,
+      "step": 1609
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9675416694205542e-05,
+      "loss": 0.8205,
+      "step": 1610
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9674862746871043e-05,
+      "loss": 0.4871,
+      "step": 1611
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9674308335056135e-05,
+      "loss": 0.8549,
+      "step": 1612
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9673753458787426e-05,
+      "loss": 0.898,
+      "step": 1613
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9673198118091566e-05,
+      "loss": 0.8441,
+      "step": 1614
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967264231299521e-05,
+      "loss": 0.8474,
+      "step": 1615
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967208604352504e-05,
+      "loss": 0.934,
+      "step": 1616
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9671529309707767e-05,
+      "loss": 0.4983,
+      "step": 1617
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9670972111570115e-05,
+      "loss": 0.8915,
+      "step": 1618
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9670414449138833e-05,
+      "loss": 0.8804,
+      "step": 1619
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96698563224407e-05,
+      "loss": 0.8949,
+      "step": 1620
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966929773150251e-05,
+      "loss": 0.8256,
+      "step": 1621
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9668738676351073e-05,
+      "loss": 0.9139,
+      "step": 1622
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9668179157013235e-05,
+      "loss": 0.8664,
+      "step": 1623
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966761917351586e-05,
+      "loss": 0.4952,
+      "step": 1624
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966705872588583e-05,
+      "loss": 0.892,
+      "step": 1625
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9666497814150048e-05,
+      "loss": 0.8784,
+      "step": 1626
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9665936438335443e-05,
+      "loss": 0.8482,
+      "step": 1627
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9665374598468973e-05,
+      "loss": 0.8815,
+      "step": 1628
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966481229457761e-05,
+      "loss": 0.8187,
+      "step": 1629
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966424952668834e-05,
+      "loss": 0.4945,
+      "step": 1630
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966368629482819e-05,
+      "loss": 0.4741,
+      "step": 1631
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96631225990242e-05,
+      "loss": 0.9015,
+      "step": 1632
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9662558439303425e-05,
+      "loss": 0.8451,
+      "step": 1633
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966199381569296e-05,
+      "loss": 0.9272,
+      "step": 1634
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9661428728219905e-05,
+      "loss": 0.4909,
+      "step": 1635
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9660863176911392e-05,
+      "loss": 0.7844,
+      "step": 1636
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966029716179457e-05,
+      "loss": 0.9032,
+      "step": 1637
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9659730682896617e-05,
+      "loss": 0.8584,
+      "step": 1638
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9659163740244724e-05,
+      "loss": 0.91,
+      "step": 1639
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9658596333866114e-05,
+      "loss": 0.8893,
+      "step": 1640
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9658028463788025e-05,
+      "loss": 0.9088,
+      "step": 1641
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9657460130037718e-05,
+      "loss": 0.884,
+      "step": 1642
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9656891332642486e-05,
+      "loss": 0.845,
+      "step": 1643
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9656322071629626e-05,
+      "loss": 0.4289,
+      "step": 1644
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9655752347026476e-05,
+      "loss": 0.9074,
+      "step": 1645
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965518215886038e-05,
+      "loss": 0.9092,
+      "step": 1646
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9654611507158724e-05,
+      "loss": 0.8498,
+      "step": 1647
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9654040391948892e-05,
+      "loss": 0.5297,
+      "step": 1648
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9653468813258308e-05,
+      "loss": 0.9095,
+      "step": 1649
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9652896771114416e-05,
+      "loss": 0.8951,
+      "step": 1650
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9652324265544674e-05,
+      "loss": 0.8704,
+      "step": 1651
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9651751296576565e-05,
+      "loss": 0.8372,
+      "step": 1652
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9651177864237606e-05,
+      "loss": 0.8971,
+      "step": 1653
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9650603968555324e-05,
+      "loss": 0.4432,
+      "step": 1654
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9650029609557265e-05,
+      "loss": 0.845,
+      "step": 1655
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9649454787271008e-05,
+      "loss": 0.9179,
+      "step": 1656
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964887950172415e-05,
+      "loss": 0.9071,
+      "step": 1657
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964830375294431e-05,
+      "loss": 0.8818,
+      "step": 1658
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9647727540959125e-05,
+      "loss": 0.8581,
+      "step": 1659
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9647150865796262e-05,
+      "loss": 0.8486,
+      "step": 1660
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9646573727483407e-05,
+      "loss": 0.8652,
+      "step": 1661
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9645996126048267e-05,
+      "loss": 0.8663,
+      "step": 1662
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9645418061518572e-05,
+      "loss": 0.8835,
+      "step": 1663
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9644839533922074e-05,
+      "loss": 0.9092,
+      "step": 1664
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964426054328655e-05,
+      "loss": 0.8798,
+      "step": 1665
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9643681089639794e-05,
+      "loss": 0.8499,
+      "step": 1666
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964310117300962e-05,
+      "loss": 0.465,
+      "step": 1667
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964252079342388e-05,
+      "loss": 0.9016,
+      "step": 1668
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964193995091043e-05,
+      "loss": 0.8544,
+      "step": 1669
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9641358645497158e-05,
+      "loss": 0.8812,
+      "step": 1670
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9640776877211973e-05,
+      "loss": 0.9013,
+      "step": 1671
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9640194646082806e-05,
+      "loss": 0.5024,
+      "step": 1672
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9639611952137604e-05,
+      "loss": 0.8373,
+      "step": 1673
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9639028795404344e-05,
+      "loss": 0.8889,
+      "step": 1674
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963844517591102e-05,
+      "loss": 0.8412,
+      "step": 1675
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9637861093685663e-05,
+      "loss": 0.4548,
+      "step": 1676
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9637276548756302e-05,
+      "loss": 0.9282,
+      "step": 1677
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9636691541151003e-05,
+      "loss": 0.8883,
+      "step": 1678
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9636106070897853e-05,
+      "loss": 0.8941,
+      "step": 1679
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963552013802496e-05,
+      "loss": 0.4731,
+      "step": 1680
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963493374256045e-05,
+      "loss": 0.902,
+      "step": 1681
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9634346884532483e-05,
+      "loss": 0.8978,
+      "step": 1682
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9633759563969227e-05,
+      "loss": 0.9184,
+      "step": 1683
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9633171780898882e-05,
+      "loss": 0.4413,
+      "step": 1684
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963258353534967e-05,
+      "loss": 0.8416,
+      "step": 1685
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9631994827349824e-05,
+      "loss": 0.8673,
+      "step": 1686
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963140565692761e-05,
+      "loss": 0.4299,
+      "step": 1687
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9630816024111317e-05,
+      "loss": 0.4616,
+      "step": 1688
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9630225928929248e-05,
+      "loss": 0.8555,
+      "step": 1689
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9629635371409736e-05,
+      "loss": 0.9069,
+      "step": 1690
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.962904435158113e-05,
+      "loss": 0.8842,
+      "step": 1691
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.962845286947181e-05,
+      "loss": 0.872,
+      "step": 1692
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.962786092511017e-05,
+      "loss": 0.9591,
+      "step": 1693
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9627268518524626e-05,
+      "loss": 0.8324,
+      "step": 1694
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9626675649743617e-05,
+      "loss": 0.9119,
+      "step": 1695
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9626082318795614e-05,
+      "loss": 0.8282,
+      "step": 1696
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96254885257091e-05,
+      "loss": 0.8893,
+      "step": 1697
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9624894270512575e-05,
+      "loss": 0.7897,
+      "step": 1698
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9624299553234577e-05,
+      "loss": 0.8855,
+      "step": 1699
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.962370437390365e-05,
+      "loss": 0.8368,
+      "step": 1700
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9623108732548378e-05,
+      "loss": 0.8481,
+      "step": 1701
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9622512629197347e-05,
+      "loss": 0.5105,
+      "step": 1702
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9621916063879183e-05,
+      "loss": 0.8864,
+      "step": 1703
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9621319036622523e-05,
+      "loss": 0.4791,
+      "step": 1704
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9620721547456026e-05,
+      "loss": 0.9295,
+      "step": 1705
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.962012359640839e-05,
+      "loss": 0.837,
+      "step": 1706
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9619525183508305e-05,
+      "loss": 0.4909,
+      "step": 1707
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9618926308784513e-05,
+      "loss": 0.8841,
+      "step": 1708
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9618326972265755e-05,
+      "loss": 0.8255,
+      "step": 1709
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9617727173980815e-05,
+      "loss": 0.8789,
+      "step": 1710
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961712691395848e-05,
+      "loss": 0.8636,
+      "step": 1711
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9616526192227577e-05,
+      "loss": 0.9211,
+      "step": 1712
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9615925008816937e-05,
+      "loss": 0.8891,
+      "step": 1713
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9615323363755427e-05,
+      "loss": 0.8719,
+      "step": 1714
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961472125707193e-05,
+      "loss": 0.8457,
+      "step": 1715
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9614118688795353e-05,
+      "loss": 0.9014,
+      "step": 1716
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9613515658954627e-05,
+      "loss": 0.8623,
+      "step": 1717
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.96129121675787e-05,
+      "loss": 0.9314,
+      "step": 1718
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9612308214696537e-05,
+      "loss": 0.4732,
+      "step": 1719
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9611703800337148e-05,
+      "loss": 0.4657,
+      "step": 1720
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9611098924529547e-05,
+      "loss": 0.8636,
+      "step": 1721
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9610493587302768e-05,
+      "loss": 0.8732,
+      "step": 1722
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9609887788685873e-05,
+      "loss": 0.9412,
+      "step": 1723
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9609281528707947e-05,
+      "loss": 0.8698,
+      "step": 1724
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9608674807398104e-05,
+      "loss": 0.48,
+      "step": 1725
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9608067624785457e-05,
+      "loss": 0.8622,
+      "step": 1726
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960745998089917e-05,
+      "loss": 0.9239,
+      "step": 1727
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9606851875768404e-05,
+      "loss": 0.8738,
+      "step": 1728
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9606243309422362e-05,
+      "loss": 0.4585,
+      "step": 1729
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9605634281890255e-05,
+      "loss": 0.8488,
+      "step": 1730
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9605024793201325e-05,
+      "loss": 0.8667,
+      "step": 1731
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9604414843384837e-05,
+      "loss": 0.8589,
+      "step": 1732
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9603804432470067e-05,
+      "loss": 0.8867,
+      "step": 1733
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960319356048632e-05,
+      "loss": 0.8842,
+      "step": 1734
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9602582227462927e-05,
+      "loss": 0.8828,
+      "step": 1735
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9601970433429237e-05,
+      "loss": 0.927,
+      "step": 1736
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960135817841462e-05,
+      "loss": 0.4571,
+      "step": 1737
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9600745462448473e-05,
+      "loss": 0.7925,
+      "step": 1738
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960013228556021e-05,
+      "loss": 0.5355,
+      "step": 1739
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9599518647779265e-05,
+      "loss": 0.9034,
+      "step": 1740
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9598904549135106e-05,
+      "loss": 0.8682,
+      "step": 1741
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959828998965721e-05,
+      "loss": 0.87,
+      "step": 1742
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959767496937508e-05,
+      "loss": 0.8338,
+      "step": 1743
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959705948831825e-05,
+      "loss": 0.8743,
+      "step": 1744
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959644354651626e-05,
+      "loss": 0.9042,
+      "step": 1745
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959582714399868e-05,
+      "loss": 0.7727,
+      "step": 1746
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9595210280795114e-05,
+      "loss": 0.8193,
+      "step": 1747
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959459295693517e-05,
+      "loss": 0.9061,
+      "step": 1748
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9593975172448482e-05,
+      "loss": 0.4762,
+      "step": 1749
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9593356927364713e-05,
+      "loss": 0.5153,
+      "step": 1750
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9592738221713544e-05,
+      "loss": 0.8762,
+      "step": 1751
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959211905552468e-05,
+      "loss": 0.9186,
+      "step": 1752
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9591499428827846e-05,
+      "loss": 0.8993,
+      "step": 1753
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9590879341652784e-05,
+      "loss": 0.8382,
+      "step": 1754
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959025879402927e-05,
+      "loss": 0.8897,
+      "step": 1755
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9589637785987094e-05,
+      "loss": 0.8759,
+      "step": 1756
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9589016317556073e-05,
+      "loss": 0.895,
+      "step": 1757
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9588394388766036e-05,
+      "loss": 0.8681,
+      "step": 1758
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9587771999646847e-05,
+      "loss": 0.5013,
+      "step": 1759
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9587149150228386e-05,
+      "loss": 0.9332,
+      "step": 1760
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9586525840540553e-05,
+      "loss": 0.8582,
+      "step": 1761
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958590207061327e-05,
+      "loss": 0.8828,
+      "step": 1762
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958527784047649e-05,
+      "loss": 0.9171,
+      "step": 1763
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958465315016018e-05,
+      "loss": 0.4697,
+      "step": 1764
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9584027999694327e-05,
+      "loss": 0.8591,
+      "step": 1765
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9583402389108947e-05,
+      "loss": 0.8758,
+      "step": 1766
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9582776318434077e-05,
+      "loss": 0.9028,
+      "step": 1767
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9582149787699767e-05,
+      "loss": 0.5199,
+      "step": 1768
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9581522796936104e-05,
+      "loss": 0.8728,
+      "step": 1769
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958089534617318e-05,
+      "loss": 0.8915,
+      "step": 1770
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9580267435441128e-05,
+      "loss": 0.8543,
+      "step": 1771
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957963906477009e-05,
+      "loss": 0.8918,
+      "step": 1772
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957901023419023e-05,
+      "loss": 0.8626,
+      "step": 1773
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957838094373174e-05,
+      "loss": 0.4698,
+      "step": 1774
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9577751193424832e-05,
+      "loss": 0.8792,
+      "step": 1775
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957712098329974e-05,
+      "loss": 0.8675,
+      "step": 1776
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957649031338672e-05,
+      "loss": 0.8779,
+      "step": 1777
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9575859183716047e-05,
+      "loss": 0.911,
+      "step": 1778
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9575227594318024e-05,
+      "loss": 0.8708,
+      "step": 1779
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9574595545222972e-05,
+      "loss": 0.8734,
+      "step": 1780
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9573963036461236e-05,
+      "loss": 0.8525,
+      "step": 1781
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957333006806318e-05,
+      "loss": 0.5139,
+      "step": 1782
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9572696640059192e-05,
+      "loss": 0.8212,
+      "step": 1783
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9572062752479684e-05,
+      "loss": 0.8609,
+      "step": 1784
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9571428405355085e-05,
+      "loss": 0.8639,
+      "step": 1785
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9570793598715857e-05,
+      "loss": 0.4753,
+      "step": 1786
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957015833259247e-05,
+      "loss": 0.8266,
+      "step": 1787
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9569522607015424e-05,
+      "loss": 0.5015,
+      "step": 1788
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956888642201524e-05,
+      "loss": 0.4939,
+      "step": 1789
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9568249777622457e-05,
+      "loss": 0.8816,
+      "step": 1790
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9567612673867647e-05,
+      "loss": 0.8592,
+      "step": 1791
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956697511078139e-05,
+      "loss": 0.841,
+      "step": 1792
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9566337088394296e-05,
+      "loss": 0.4479,
+      "step": 1793
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9565698606737e-05,
+      "loss": 0.8406,
+      "step": 1794
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9565059665840153e-05,
+      "loss": 0.8716,
+      "step": 1795
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9564420265734422e-05,
+      "loss": 0.9086,
+      "step": 1796
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9563780406450517e-05,
+      "loss": 0.8048,
+      "step": 1797
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956314008801915e-05,
+      "loss": 0.4929,
+      "step": 1798
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956249931047106e-05,
+      "loss": 0.8683,
+      "step": 1799
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9561858073837016e-05,
+      "loss": 0.8818,
+      "step": 1800
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.95612163781478e-05,
+      "loss": 0.8205,
+      "step": 1801
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9560574223434216e-05,
+      "loss": 0.8164,
+      "step": 1802
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.95599316097271e-05,
+      "loss": 0.8826,
+      "step": 1803
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9559288537057294e-05,
+      "loss": 0.8667,
+      "step": 1804
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9558645005455684e-05,
+      "loss": 0.4554,
+      "step": 1805
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9558001014953155e-05,
+      "loss": 0.8773,
+      "step": 1806
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9557356565580626e-05,
+      "loss": 0.4741,
+      "step": 1807
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9556711657369042e-05,
+      "loss": 0.87,
+      "step": 1808
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955606629034936e-05,
+      "loss": 0.8736,
+      "step": 1809
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955542046455256e-05,
+      "loss": 0.9102,
+      "step": 1810
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9554774180009653e-05,
+      "loss": 0.4585,
+      "step": 1811
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9554127436751665e-05,
+      "loss": 0.8728,
+      "step": 1812
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9553480234809643e-05,
+      "loss": 0.9029,
+      "step": 1813
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955283257421466e-05,
+      "loss": 0.8352,
+      "step": 1814
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9552184454997817e-05,
+      "loss": 0.4793,
+      "step": 1815
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9551535877190217e-05,
+      "loss": 0.8582,
+      "step": 1816
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9550886840823e-05,
+      "loss": 0.8879,
+      "step": 1817
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9550237345927337e-05,
+      "loss": 0.8684,
+      "step": 1818
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9549587392534398e-05,
+      "loss": 0.88,
+      "step": 1819
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9548936980675387e-05,
+      "loss": 0.8878,
+      "step": 1820
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9548286110381538e-05,
+      "loss": 0.8473,
+      "step": 1821
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954763478168409e-05,
+      "loss": 0.4473,
+      "step": 1822
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954698299461432e-05,
+      "loss": 0.4585,
+      "step": 1823
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954633074920351e-05,
+      "loss": 0.474,
+      "step": 1824
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9545678045482984e-05,
+      "loss": 0.9329,
+      "step": 1825
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954502488348407e-05,
+      "loss": 0.8778,
+      "step": 1826
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9544371263238127e-05,
+      "loss": 0.8842,
+      "step": 1827
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954371718477654e-05,
+      "loss": 0.8342,
+      "step": 1828
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9543062648130708e-05,
+      "loss": 0.8462,
+      "step": 1829
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954240765333205e-05,
+      "loss": 0.8567,
+      "step": 1830
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9541752200412016e-05,
+      "loss": 0.8965,
+      "step": 1831
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954109628940207e-05,
+      "loss": 0.4728,
+      "step": 1832
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9540439920333708e-05,
+      "loss": 0.89,
+      "step": 1833
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9539783093238438e-05,
+      "loss": 0.5304,
+      "step": 1834
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953912580814779e-05,
+      "loss": 0.8967,
+      "step": 1835
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953846806509333e-05,
+      "loss": 0.8967,
+      "step": 1836
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9537809864106623e-05,
+      "loss": 0.872,
+      "step": 1837
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9537151205219277e-05,
+      "loss": 0.9132,
+      "step": 1838
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9536492088462908e-05,
+      "loss": 0.901,
+      "step": 1839
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9535832513869164e-05,
+      "loss": 0.8429,
+      "step": 1840
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953517248146971e-05,
+      "loss": 0.5085,
+      "step": 1841
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9534511991296232e-05,
+      "loss": 0.8418,
+      "step": 1842
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9533851043380442e-05,
+      "loss": 0.8574,
+      "step": 1843
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9533189637754065e-05,
+      "loss": 0.908,
+      "step": 1844
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9532527774448858e-05,
+      "loss": 0.8509,
+      "step": 1845
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9531865453496604e-05,
+      "loss": 0.893,
+      "step": 1846
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953120267492909e-05,
+      "loss": 0.866,
+      "step": 1847
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9530539438778136e-05,
+      "loss": 0.9033,
+      "step": 1848
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9529875745075587e-05,
+      "loss": 0.7997,
+      "step": 1849
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9529211593853308e-05,
+      "loss": 0.8342,
+      "step": 1850
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952854698514318e-05,
+      "loss": 0.5017,
+      "step": 1851
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9527881918977115e-05,
+      "loss": 0.4747,
+      "step": 1852
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9527216395387037e-05,
+      "loss": 0.8801,
+      "step": 1853
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.95265504144049e-05,
+      "loss": 0.8704,
+      "step": 1854
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952588397606267e-05,
+      "loss": 0.8799,
+      "step": 1855
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9525217080392355e-05,
+      "loss": 0.8497,
+      "step": 1856
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9524549727425962e-05,
+      "loss": 0.8268,
+      "step": 1857
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952388191719554e-05,
+      "loss": 0.8208,
+      "step": 1858
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9523213649733135e-05,
+      "loss": 0.4798,
+      "step": 1859
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952254492507084e-05,
+      "loss": 0.883,
+      "step": 1860
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9521875743240757e-05,
+      "loss": 0.4372,
+      "step": 1861
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9521206104275016e-05,
+      "loss": 0.8623,
+      "step": 1862
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952053600820576e-05,
+      "loss": 0.4652,
+      "step": 1863
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9519865455065165e-05,
+      "loss": 0.8669,
+      "step": 1864
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9519194444885422e-05,
+      "loss": 0.8427,
+      "step": 1865
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9518522977698743e-05,
+      "loss": 0.8669,
+      "step": 1866
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9517851053537365e-05,
+      "loss": 0.8828,
+      "step": 1867
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.951717867243355e-05,
+      "loss": 0.8582,
+      "step": 1868
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9516505834419575e-05,
+      "loss": 0.8549,
+      "step": 1869
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9515832539527747e-05,
+      "loss": 0.4782,
+      "step": 1870
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9515158787790383e-05,
+      "loss": 0.5045,
+      "step": 1871
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9514484579239832e-05,
+      "loss": 0.4567,
+      "step": 1872
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9513809913908466e-05,
+      "loss": 0.4657,
+      "step": 1873
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.951313479182867e-05,
+      "loss": 0.8306,
+      "step": 1874
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9512459213032857e-05,
+      "loss": 0.4776,
+      "step": 1875
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9511783177553463e-05,
+      "loss": 0.8212,
+      "step": 1876
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.951110668542294e-05,
+      "loss": 0.8594,
+      "step": 1877
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.951042973667377e-05,
+      "loss": 0.5065,
+      "step": 1878
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950975233133845e-05,
+      "loss": 0.8862,
+      "step": 1879
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9509074469449507e-05,
+      "loss": 0.8315,
+      "step": 1880
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9508396151039473e-05,
+      "loss": 0.9073,
+      "step": 1881
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950771737614092e-05,
+      "loss": 0.4953,
+      "step": 1882
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950703814478644e-05,
+      "loss": 0.8606,
+      "step": 1883
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9506358457008633e-05,
+      "loss": 0.9062,
+      "step": 1884
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950567831284014e-05,
+      "loss": 0.5264,
+      "step": 1885
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9504997712313606e-05,
+      "loss": 0.8454,
+      "step": 1886
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9504316655461707e-05,
+      "loss": 0.8935,
+      "step": 1887
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9503635142317144e-05,
+      "loss": 0.4993,
+      "step": 1888
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9502953172912633e-05,
+      "loss": 0.8998,
+      "step": 1889
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9502270747280912e-05,
+      "loss": 0.8485,
+      "step": 1890
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9501587865454753e-05,
+      "loss": 0.866,
+      "step": 1891
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9500904527466928e-05,
+      "loss": 0.8427,
+      "step": 1892
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950022073335025e-05,
+      "loss": 0.9055,
+      "step": 1893
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949953648313755e-05,
+      "loss": 0.4767,
+      "step": 1894
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9498851776861675e-05,
+      "loss": 0.4723,
+      "step": 1895
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9498166614555495e-05,
+      "loss": 0.8538,
+      "step": 1896
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9497480996251905e-05,
+      "loss": 0.8409,
+      "step": 1897
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9496794921983823e-05,
+      "loss": 0.4729,
+      "step": 1898
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9496108391784183e-05,
+      "loss": 0.5553,
+      "step": 1899
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949542140568595e-05,
+      "loss": 0.8629,
+      "step": 1900
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9494733963722102e-05,
+      "loss": 0.8876,
+      "step": 1901
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949404606592564e-05,
+      "loss": 0.8306,
+      "step": 1902
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9493357712329594e-05,
+      "loss": 0.8511,
+      "step": 1903
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949266890296701e-05,
+      "loss": 0.8428,
+      "step": 1904
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9491979637870956e-05,
+      "loss": 0.8791,
+      "step": 1905
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9491289917074523e-05,
+      "loss": 0.4563,
+      "step": 1906
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9490599740610826e-05,
+      "loss": 0.8761,
+      "step": 1907
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9489909108512997e-05,
+      "loss": 0.885,
+      "step": 1908
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.948921802081419e-05,
+      "loss": 0.8697,
+      "step": 1909
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9488526477547593e-05,
+      "loss": 0.5065,
+      "step": 1910
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9487834478746396e-05,
+      "loss": 0.4833,
+      "step": 1911
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9487142024443826e-05,
+      "loss": 0.8742,
+      "step": 1912
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9486449114673127e-05,
+      "loss": 0.8587,
+      "step": 1913
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.948575574946756e-05,
+      "loss": 0.8451,
+      "step": 1914
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9485061928860424e-05,
+      "loss": 0.8733,
+      "step": 1915
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9484367652885016e-05,
+      "loss": 0.8591,
+      "step": 1916
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9483672921574677e-05,
+      "loss": 0.4806,
+      "step": 1917
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9482977734962753e-05,
+      "loss": 0.8568,
+      "step": 1918
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9482282093082628e-05,
+      "loss": 0.8558,
+      "step": 1919
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9481585995967688e-05,
+      "loss": 0.8846,
+      "step": 1920
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.948088944365136e-05,
+      "loss": 0.8745,
+      "step": 1921
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9480192436167082e-05,
+      "loss": 0.8864,
+      "step": 1922
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9479494973548315e-05,
+      "loss": 0.8787,
+      "step": 1923
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.947879705582855e-05,
+      "loss": 0.4555,
+      "step": 1924
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9478098683041287e-05,
+      "loss": 0.889,
+      "step": 1925
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9477399855220054e-05,
+      "loss": 0.4508,
+      "step": 1926
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9476700572398404e-05,
+      "loss": 0.887,
+      "step": 1927
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9476000834609906e-05,
+      "loss": 0.8486,
+      "step": 1928
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.947530064188816e-05,
+      "loss": 0.883,
+      "step": 1929
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9474599994266774e-05,
+      "loss": 0.4728,
+      "step": 1930
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9473898891779392e-05,
+      "loss": 0.4642,
+      "step": 1931
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9473197334459667e-05,
+      "loss": 0.8668,
+      "step": 1932
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.947249532234128e-05,
+      "loss": 0.8305,
+      "step": 1933
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.947179285545794e-05,
+      "loss": 0.4609,
+      "step": 1934
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.947108993384337e-05,
+      "loss": 0.8704,
+      "step": 1935
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.947038655753131e-05,
+      "loss": 0.8651,
+      "step": 1936
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946968272655554e-05,
+      "loss": 0.4817,
+      "step": 1937
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9468978440949837e-05,
+      "loss": 0.8938,
+      "step": 1938
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9468273700748024e-05,
+      "loss": 0.8644,
+      "step": 1939
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946756850598393e-05,
+      "loss": 0.8466,
+      "step": 1940
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946686285669141e-05,
+      "loss": 0.7975,
+      "step": 1941
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9466156752904344e-05,
+      "loss": 0.467,
+      "step": 1942
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9465450194656626e-05,
+      "loss": 0.8629,
+      "step": 1943
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9464743181982187e-05,
+      "loss": 0.7985,
+      "step": 1944
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946403571491496e-05,
+      "loss": 0.5016,
+      "step": 1945
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9463327793488916e-05,
+      "loss": 0.4249,
+      "step": 1946
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946261941773804e-05,
+      "loss": 0.4455,
+      "step": 1947
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946191058769634e-05,
+      "loss": 0.8082,
+      "step": 1948
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9461201303397842e-05,
+      "loss": 0.9247,
+      "step": 1949
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9460491564876606e-05,
+      "loss": 0.8597,
+      "step": 1950
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.94597813721667e-05,
+      "loss": 0.8729,
+      "step": 1951
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9459070725302223e-05,
+      "loss": 0.5324,
+      "step": 1952
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9458359624317293e-05,
+      "loss": 0.8363,
+      "step": 1953
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9457648069246044e-05,
+      "loss": 0.8457,
+      "step": 1954
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945693606012264e-05,
+      "loss": 0.4297,
+      "step": 1955
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9456223596981267e-05,
+      "loss": 0.9117,
+      "step": 1956
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9455510679856126e-05,
+      "loss": 0.923,
+      "step": 1957
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9454797308781447e-05,
+      "loss": 0.8696,
+      "step": 1958
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945408348379147e-05,
+      "loss": 0.8945,
+      "step": 1959
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9453369204920473e-05,
+      "loss": 0.8797,
+      "step": 1960
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9452654472202747e-05,
+      "loss": 0.904,
+      "step": 1961
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9451939285672598e-05,
+      "loss": 0.8987,
+      "step": 1962
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945122364536437e-05,
+      "loss": 0.8772,
+      "step": 1963
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9450507551312423e-05,
+      "loss": 0.4906,
+      "step": 1964
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9449791003551125e-05,
+      "loss": 0.5295,
+      "step": 1965
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9449074002114884e-05,
+      "loss": 0.8069,
+      "step": 1966
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.944835654703812e-05,
+      "loss": 0.8897,
+      "step": 1967
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.944763863835528e-05,
+      "loss": 0.871,
+      "step": 1968
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9446920276100828e-05,
+      "loss": 0.8754,
+      "step": 1969
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.944620146030925e-05,
+      "loss": 0.4739,
+      "step": 1970
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9445482191015053e-05,
+      "loss": 0.4864,
+      "step": 1971
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.944476246825278e-05,
+      "loss": 0.5213,
+      "step": 1972
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9444042292056974e-05,
+      "loss": 0.8544,
+      "step": 1973
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9443321662462216e-05,
+      "loss": 0.4624,
+      "step": 1974
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9442600579503096e-05,
+      "loss": 0.8874,
+      "step": 1975
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.944187904321424e-05,
+      "loss": 0.8567,
+      "step": 1976
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.944115705363028e-05,
+      "loss": 0.5154,
+      "step": 1977
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9440434610785884e-05,
+      "loss": 0.8502,
+      "step": 1978
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9439711714715735e-05,
+      "loss": 0.8775,
+      "step": 1979
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9438988365454535e-05,
+      "loss": 0.9887,
+      "step": 1980
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9438264563037018e-05,
+      "loss": 0.8191,
+      "step": 1981
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9437540307497928e-05,
+      "loss": 0.4699,
+      "step": 1982
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9436815598872037e-05,
+      "loss": 0.5128,
+      "step": 1983
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9436090437194136e-05,
+      "loss": 0.5057,
+      "step": 1984
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.943536482249904e-05,
+      "loss": 0.8512,
+      "step": 1985
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9434638754821586e-05,
+      "loss": 0.8698,
+      "step": 1986
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.943391223419663e-05,
+      "loss": 0.8875,
+      "step": 1987
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9433185260659056e-05,
+      "loss": 0.8641,
+      "step": 1988
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9432457834243763e-05,
+      "loss": 0.8612,
+      "step": 1989
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.943172995498567e-05,
+      "loss": 0.4811,
+      "step": 1990
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9431001622919725e-05,
+      "loss": 0.8969,
+      "step": 1991
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9430272838080896e-05,
+      "loss": 0.4866,
+      "step": 1992
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9429543600504166e-05,
+      "loss": 0.4515,
+      "step": 1993
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9428813910224554e-05,
+      "loss": 0.4346,
+      "step": 1994
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9428083767277082e-05,
+      "loss": 0.8539,
+      "step": 1995
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.942735317169681e-05,
+      "loss": 0.8851,
+      "step": 1996
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.942662212351881e-05,
+      "loss": 0.4801,
+      "step": 1997
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9425890622778182e-05,
+      "loss": 0.8678,
+      "step": 1998
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9425158669510037e-05,
+      "loss": 0.8299,
+      "step": 1999
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9424426263749527e-05,
+      "loss": 0.8603,
+      "step": 2000
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9423693405531802e-05,
+      "loss": 0.8464,
+      "step": 2001
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.942296009489206e-05,
+      "loss": 0.4911,
+      "step": 2002
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.942222633186549e-05,
+      "loss": 0.8083,
+      "step": 2003
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.942149211648733e-05,
+      "loss": 0.8583,
+      "step": 2004
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9420757448792826e-05,
+      "loss": 0.8191,
+      "step": 2005
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.942002232881725e-05,
+      "loss": 0.8511,
+      "step": 2006
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.941928675659589e-05,
+      "loss": 0.8348,
+      "step": 2007
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.941855073216407e-05,
+      "loss": 0.4826,
+      "step": 2008
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9417814255557112e-05,
+      "loss": 0.831,
+      "step": 2009
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9417077326810384e-05,
+      "loss": 0.8915,
+      "step": 2010
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.941633994595926e-05,
+      "loss": 0.8565,
+      "step": 2011
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9415602113039142e-05,
+      "loss": 0.8385,
+      "step": 2012
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.941486382808546e-05,
+      "loss": 0.4857,
+      "step": 2013
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9414125091133643e-05,
+      "loss": 0.8563,
+      "step": 2014
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.941338590221917e-05,
+      "loss": 0.8964,
+      "step": 2015
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9412646261377522e-05,
+      "loss": 0.8709,
+      "step": 2016
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9411906168644215e-05,
+      "loss": 0.4951,
+      "step": 2017
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9411165624054772e-05,
+      "loss": 0.8587,
+      "step": 2018
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.941042462764475e-05,
+      "loss": 0.8688,
+      "step": 2019
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.940968317944973e-05,
+      "loss": 0.4496,
+      "step": 2020
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9408941279505293e-05,
+      "loss": 0.8588,
+      "step": 2021
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.940819892784707e-05,
+      "loss": 0.8349,
+      "step": 2022
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9407456124510692e-05,
+      "loss": 0.8574,
+      "step": 2023
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9406712869531827e-05,
+      "loss": 0.9116,
+      "step": 2024
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9405969162946156e-05,
+      "loss": 0.883,
+      "step": 2025
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.940522500478938e-05,
+      "loss": 0.8156,
+      "step": 2026
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9404480395097226e-05,
+      "loss": 0.8533,
+      "step": 2027
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.940373533390545e-05,
+      "loss": 0.8565,
+      "step": 2028
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.940298982124981e-05,
+      "loss": 0.901,
+      "step": 2029
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9402243857166104e-05,
+      "loss": 0.8346,
+      "step": 2030
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9401497441690144e-05,
+      "loss": 0.826,
+      "step": 2031
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9400750574857763e-05,
+      "loss": 0.886,
+      "step": 2032
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9400003256704823e-05,
+      "loss": 0.4487,
+      "step": 2033
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9399255487267195e-05,
+      "loss": 0.8904,
+      "step": 2034
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.939850726658078e-05,
+      "loss": 0.8579,
+      "step": 2035
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9397758594681504e-05,
+      "loss": 0.8664,
+      "step": 2036
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9397009471605304e-05,
+      "loss": 0.8633,
+      "step": 2037
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.939625989738815e-05,
+      "loss": 0.8184,
+      "step": 2038
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9395509872066026e-05,
+      "loss": 0.5093,
+      "step": 2039
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9394759395674935e-05,
+      "loss": 0.857,
+      "step": 2040
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9394008468250916e-05,
+      "loss": 0.7861,
+      "step": 2041
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9393257089830014e-05,
+      "loss": 0.8284,
+      "step": 2042
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.93925052604483e-05,
+      "loss": 0.8438,
+      "step": 2043
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.939175298014188e-05,
+      "loss": 0.8518,
+      "step": 2044
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9391000248946854e-05,
+      "loss": 0.8942,
+      "step": 2045
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.939024706689937e-05,
+      "loss": 0.8515,
+      "step": 2046
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9389493434035588e-05,
+      "loss": 0.4422,
+      "step": 2047
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9388739350391685e-05,
+      "loss": 0.8789,
+      "step": 2048
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9387984816003868e-05,
+      "loss": 0.4589,
+      "step": 2049
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9387229830908353e-05,
+      "loss": 0.4876,
+      "step": 2050
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9386474395141395e-05,
+      "loss": 0.8965,
+      "step": 2051
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9385718508739263e-05,
+      "loss": 0.8862,
+      "step": 2052
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.938496217173824e-05,
+      "loss": 0.8701,
+      "step": 2053
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9384205384174636e-05,
+      "loss": 0.7863,
+      "step": 2054
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.938344814608479e-05,
+      "loss": 0.8732,
+      "step": 2055
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.938269045750505e-05,
+      "loss": 0.841,
+      "step": 2056
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9381932318471794e-05,
+      "loss": 0.4613,
+      "step": 2057
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9381173729021423e-05,
+      "loss": 0.8981,
+      "step": 2058
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9380414689190353e-05,
+      "loss": 0.5307,
+      "step": 2059
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9379655199015026e-05,
+      "loss": 0.5021,
+      "step": 2060
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9378895258531902e-05,
+      "loss": 0.8689,
+      "step": 2061
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9378134867777468e-05,
+      "loss": 0.4798,
+      "step": 2062
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9377374026788227e-05,
+      "loss": 0.8395,
+      "step": 2063
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9376612735600708e-05,
+      "loss": 0.9022,
+      "step": 2064
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9375850994251456e-05,
+      "loss": 0.8865,
+      "step": 2065
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9375088802777045e-05,
+      "loss": 0.8211,
+      "step": 2066
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.937432616121407e-05,
+      "loss": 0.8552,
+      "step": 2067
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9373563069599138e-05,
+      "loss": 0.8718,
+      "step": 2068
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.937279952796889e-05,
+      "loss": 0.8858,
+      "step": 2069
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9372035536359977e-05,
+      "loss": 0.4222,
+      "step": 2070
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.937127109480909e-05,
+      "loss": 0.8105,
+      "step": 2071
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.937050620335291e-05,
+      "loss": 0.8939,
+      "step": 2072
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9369740862028173e-05,
+      "loss": 0.8631,
+      "step": 2073
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9368975070871615e-05,
+      "loss": 0.8912,
+      "step": 2074
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9368208829920006e-05,
+      "loss": 0.8626,
+      "step": 2075
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9367442139210128e-05,
+      "loss": 0.8364,
+      "step": 2076
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9366674998778792e-05,
+      "loss": 0.9599,
+      "step": 2077
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.936590740866283e-05,
+      "loss": 0.4516,
+      "step": 2078
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9365139368899083e-05,
+      "loss": 0.8102,
+      "step": 2079
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9364370879524438e-05,
+      "loss": 0.4875,
+      "step": 2080
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9363601940575778e-05,
+      "loss": 0.8351,
+      "step": 2081
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9362832552090025e-05,
+      "loss": 0.8497,
+      "step": 2082
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9362062714104114e-05,
+      "loss": 0.4867,
+      "step": 2083
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9361292426655004e-05,
+      "loss": 0.8339,
+      "step": 2084
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9360521689779677e-05,
+      "loss": 0.8737,
+      "step": 2085
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9359750503515138e-05,
+      "loss": 0.8703,
+      "step": 2086
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9358978867898407e-05,
+      "loss": 0.4761,
+      "step": 2087
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.935820678296653e-05,
+      "loss": 0.8672,
+      "step": 2088
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.935743424875657e-05,
+      "loss": 0.8249,
+      "step": 2089
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9356661265305624e-05,
+      "loss": 0.8581,
+      "step": 2090
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.93558878326508e-05,
+      "loss": 0.8204,
+      "step": 2091
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9355113950829224e-05,
+      "loss": 0.5106,
+      "step": 2092
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9354339619878052e-05,
+      "loss": 0.8937,
+      "step": 2093
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9353564839834464e-05,
+      "loss": 0.9269,
+      "step": 2094
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9352789610735653e-05,
+      "loss": 0.4837,
+      "step": 2095
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9352013932618833e-05,
+      "loss": 0.4788,
+      "step": 2096
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9351237805521248e-05,
+      "loss": 0.8687,
+      "step": 2097
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.935046122948016e-05,
+      "loss": 0.4881,
+      "step": 2098
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9349684204532844e-05,
+      "loss": 0.8755,
+      "step": 2099
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9348906730716614e-05,
+      "loss": 0.8837,
+      "step": 2100
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.934812880806879e-05,
+      "loss": 0.8263,
+      "step": 2101
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.934735043662672e-05,
+      "loss": 0.8268,
+      "step": 2102
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9346571616427776e-05,
+      "loss": 0.8853,
+      "step": 2103
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9345792347509343e-05,
+      "loss": 0.4681,
+      "step": 2104
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.934501262990884e-05,
+      "loss": 0.8654,
+      "step": 2105
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9344232463663692e-05,
+      "loss": 0.436,
+      "step": 2106
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.934345184881136e-05,
+      "loss": 0.8421,
+      "step": 2107
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.934267078538932e-05,
+      "loss": 0.8163,
+      "step": 2108
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9341889273435068e-05,
+      "loss": 0.4779,
+      "step": 2109
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9341107312986124e-05,
+      "loss": 0.8646,
+      "step": 2110
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.934032490408003e-05,
+      "loss": 0.8479,
+      "step": 2111
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.933954204675435e-05,
+      "loss": 0.5364,
+      "step": 2112
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9338758741046663e-05,
+      "loss": 0.8371,
+      "step": 2113
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.933797498699458e-05,
+      "loss": 0.9318,
+      "step": 2114
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9337190784635728e-05,
+      "loss": 0.5022,
+      "step": 2115
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.933640613400775e-05,
+      "loss": 0.8808,
+      "step": 2116
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9335621035148326e-05,
+      "loss": 0.8822,
+      "step": 2117
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.933483548809514e-05,
+      "loss": 0.8461,
+      "step": 2118
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9334049492885906e-05,
+      "loss": 0.8565,
+      "step": 2119
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9333263049558362e-05,
+      "loss": 0.8774,
+      "step": 2120
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9332476158150263e-05,
+      "loss": 0.8061,
+      "step": 2121
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9331688818699387e-05,
+      "loss": 0.8365,
+      "step": 2122
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9330901031243533e-05,
+      "loss": 0.8759,
+      "step": 2123
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9330112795820518e-05,
+      "loss": 0.4621,
+      "step": 2124
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.932932411246819e-05,
+      "loss": 0.514,
+      "step": 2125
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9328534981224416e-05,
+      "loss": 0.9002,
+      "step": 2126
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9327745402127073e-05,
+      "loss": 0.5103,
+      "step": 2127
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9326955375214074e-05,
+      "loss": 0.8238,
+      "step": 2128
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9326164900523342e-05,
+      "loss": 0.8331,
+      "step": 2129
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.932537397809283e-05,
+      "loss": 0.8658,
+      "step": 2130
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.932458260796051e-05,
+      "loss": 0.8682,
+      "step": 2131
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9323790790164374e-05,
+      "loss": 0.9186,
+      "step": 2132
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.932299852474244e-05,
+      "loss": 0.8567,
+      "step": 2133
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9322205811732734e-05,
+      "loss": 0.8656,
+      "step": 2134
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9321412651173323e-05,
+      "loss": 0.847,
+      "step": 2135
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9320619043102282e-05,
+      "loss": 0.9126,
+      "step": 2136
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9319824987557712e-05,
+      "loss": 0.8247,
+      "step": 2137
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.931903048457773e-05,
+      "loss": 0.8533,
+      "step": 2138
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.931823553420049e-05,
+      "loss": 0.4437,
+      "step": 2139
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.931744013646415e-05,
+      "loss": 0.8772,
+      "step": 2140
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9316644291406897e-05,
+      "loss": 0.8742,
+      "step": 2141
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.931584799906694e-05,
+      "loss": 0.8595,
+      "step": 2142
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9315051259482502e-05,
+      "loss": 0.4992,
+      "step": 2143
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9314254072691842e-05,
+      "loss": 0.8657,
+      "step": 2144
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9313456438733228e-05,
+      "loss": 0.8379,
+      "step": 2145
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9312658357644955e-05,
+      "loss": 0.8316,
+      "step": 2146
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9311859829465333e-05,
+      "loss": 0.8515,
+      "step": 2147
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.931106085423271e-05,
+      "loss": 0.4979,
+      "step": 2148
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.931026143198543e-05,
+      "loss": 0.8901,
+      "step": 2149
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9309461562761885e-05,
+      "loss": 0.4607,
+      "step": 2150
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9308661246600466e-05,
+      "loss": 0.4562,
+      "step": 2151
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9307860483539602e-05,
+      "loss": 0.8514,
+      "step": 2152
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9307059273617735e-05,
+      "loss": 0.4791,
+      "step": 2153
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9306257616873327e-05,
+      "loss": 0.9153,
+      "step": 2154
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.930545551334487e-05,
+      "loss": 0.8779,
+      "step": 2155
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9304652963070868e-05,
+      "loss": 0.8863,
+      "step": 2156
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9303849966089852e-05,
+      "loss": 0.834,
+      "step": 2157
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9303046522440375e-05,
+      "loss": 0.4649,
+      "step": 2158
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9302242632161005e-05,
+      "loss": 0.8005,
+      "step": 2159
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.930143829529034e-05,
+      "loss": 0.8654,
+      "step": 2160
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9300633511866997e-05,
+      "loss": 0.8397,
+      "step": 2161
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.929982828192961e-05,
+      "loss": 0.9146,
+      "step": 2162
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9299022605516833e-05,
+      "loss": 0.5167,
+      "step": 2163
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9298216482667354e-05,
+      "loss": 0.4909,
+      "step": 2164
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9297409913419867e-05,
+      "loss": 0.8621,
+      "step": 2165
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.92966028978131e-05,
+      "loss": 0.881,
+      "step": 2166
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9295795435885796e-05,
+      "loss": 0.845,
+      "step": 2167
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9294987527676717e-05,
+      "loss": 0.4579,
+      "step": 2168
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9294179173224653e-05,
+      "loss": 0.8919,
+      "step": 2169
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.929337037256841e-05,
+      "loss": 0.8698,
+      "step": 2170
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9292561125746817e-05,
+      "loss": 0.4616,
+      "step": 2171
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9291751432798735e-05,
+      "loss": 0.8703,
+      "step": 2172
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.929094129376302e-05,
+      "loss": 0.8068,
+      "step": 2173
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.929013070867858e-05,
+      "loss": 0.8501,
+      "step": 2174
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9289319677584325e-05,
+      "loss": 0.8623,
+      "step": 2175
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.928850820051919e-05,
+      "loss": 0.8903,
+      "step": 2176
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9287696277522133e-05,
+      "loss": 0.8792,
+      "step": 2177
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.928688390863214e-05,
+      "loss": 0.5063,
+      "step": 2178
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9286071093888207e-05,
+      "loss": 0.8973,
+      "step": 2179
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9285257833329352e-05,
+      "loss": 0.4925,
+      "step": 2180
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.928444412699463e-05,
+      "loss": 0.8124,
+      "step": 2181
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9283629974923097e-05,
+      "loss": 0.8669,
+      "step": 2182
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9282815377153846e-05,
+      "loss": 0.8701,
+      "step": 2183
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9282000333725976e-05,
+      "loss": 0.4949,
+      "step": 2184
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.928118484467863e-05,
+      "loss": 0.8397,
+      "step": 2185
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9280368910050943e-05,
+      "loss": 0.8483,
+      "step": 2186
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.92795525298821e-05,
+      "loss": 0.4515,
+      "step": 2187
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9278735704211288e-05,
+      "loss": 0.9261,
+      "step": 2188
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9277918433077724e-05,
+      "loss": 0.8379,
+      "step": 2189
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9277100716520646e-05,
+      "loss": 0.8718,
+      "step": 2190
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9276282554579306e-05,
+      "loss": 0.8809,
+      "step": 2191
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.927546394729299e-05,
+      "loss": 0.4667,
+      "step": 2192
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9274644894700994e-05,
+      "loss": 0.8432,
+      "step": 2193
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9273825396842644e-05,
+      "loss": 0.9284,
+      "step": 2194
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9273005453757275e-05,
+      "loss": 0.8496,
+      "step": 2195
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9272185065484263e-05,
+      "loss": 0.8877,
+      "step": 2196
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9271364232062987e-05,
+      "loss": 0.877,
+      "step": 2197
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9270542953532852e-05,
+      "loss": 0.8617,
+      "step": 2198
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9269721229933297e-05,
+      "loss": 0.8488,
+      "step": 2199
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.926889906130376e-05,
+      "loss": 0.8534,
+      "step": 2200
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.926807644768372e-05,
+      "loss": 0.8584,
+      "step": 2201
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.926725338911267e-05,
+      "loss": 0.4974,
+      "step": 2202
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.926642988563012e-05,
+      "loss": 0.8715,
+      "step": 2203
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9265605937275608e-05,
+      "loss": 0.8478,
+      "step": 2204
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.926478154408869e-05,
+      "loss": 0.898,
+      "step": 2205
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9263956706108945e-05,
+      "loss": 0.8781,
+      "step": 2206
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.926313142337597e-05,
+      "loss": 0.8621,
+      "step": 2207
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9262305695929392e-05,
+      "loss": 0.8741,
+      "step": 2208
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9261479523808852e-05,
+      "loss": 0.5039,
+      "step": 2209
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9260652907054006e-05,
+      "loss": 0.8614,
+      "step": 2210
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.925982584570455e-05,
+      "loss": 0.8612,
+      "step": 2211
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.925899833980018e-05,
+      "loss": 0.8331,
+      "step": 2212
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9258170389380632e-05,
+      "loss": 0.8344,
+      "step": 2213
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.925734199448565e-05,
+      "loss": 0.4688,
+      "step": 2214
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.925651315515501e-05,
+      "loss": 0.8326,
+      "step": 2215
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9255683871428495e-05,
+      "loss": 0.8614,
+      "step": 2216
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9254854143345925e-05,
+      "loss": 0.8547,
+      "step": 2217
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9254023970947134e-05,
+      "loss": 0.4474,
+      "step": 2218
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9253193354271972e-05,
+      "loss": 0.8313,
+      "step": 2219
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9252362293360325e-05,
+      "loss": 0.5025,
+      "step": 2220
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9251530788252085e-05,
+      "loss": 0.8204,
+      "step": 2221
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.925069883898717e-05,
+      "loss": 0.8792,
+      "step": 2222
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9249866445605528e-05,
+      "loss": 0.8626,
+      "step": 2223
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.924903360814712e-05,
+      "loss": 0.8535,
+      "step": 2224
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9248200326651923e-05,
+      "loss": 0.4615,
+      "step": 2225
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9247366601159945e-05,
+      "loss": 0.8466,
+      "step": 2226
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.924653243171122e-05,
+      "loss": 0.8396,
+      "step": 2227
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9245697818345784e-05,
+      "loss": 0.7973,
+      "step": 2228
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9244862761103717e-05,
+      "loss": 0.8822,
+      "step": 2229
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.92440272600251e-05,
+      "loss": 0.8249,
+      "step": 2230
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9243191315150044e-05,
+      "loss": 0.8489,
+      "step": 2231
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.924235492651869e-05,
+      "loss": 0.4324,
+      "step": 2232
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.924151809417119e-05,
+      "loss": 0.5013,
+      "step": 2233
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9240680818147717e-05,
+      "loss": 0.8174,
+      "step": 2234
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9239843098488467e-05,
+      "loss": 0.8456,
+      "step": 2235
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.923900493523366e-05,
+      "loss": 0.473,
+      "step": 2236
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.923816632842354e-05,
+      "loss": 0.8424,
+      "step": 2237
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9237327278098355e-05,
+      "loss": 0.8184,
+      "step": 2238
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9236487784298396e-05,
+      "loss": 0.8459,
+      "step": 2239
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9235647847063967e-05,
+      "loss": 0.8491,
+      "step": 2240
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.923480746643539e-05,
+      "loss": 0.867,
+      "step": 2241
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.923396664245301e-05,
+      "loss": 0.8714,
+      "step": 2242
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9233125375157197e-05,
+      "loss": 0.8525,
+      "step": 2243
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9232283664588334e-05,
+      "loss": 0.8788,
+      "step": 2244
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.923144151078684e-05,
+      "loss": 0.8683,
+      "step": 2245
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9230598913793136e-05,
+      "loss": 0.8353,
+      "step": 2246
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.922975587364768e-05,
+      "loss": 0.8788,
+      "step": 2247
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9228912390390944e-05,
+      "loss": 0.4642,
+      "step": 2248
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.922806846406342e-05,
+      "loss": 0.8064,
+      "step": 2249
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9227224094705633e-05,
+      "loss": 0.8856,
+      "step": 2250
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.922637928235811e-05,
+      "loss": 0.8403,
+      "step": 2251
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9225534027061416e-05,
+      "loss": 0.8117,
+      "step": 2252
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9224688328856127e-05,
+      "loss": 0.8311,
+      "step": 2253
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9223842187782846e-05,
+      "loss": 0.8571,
+      "step": 2254
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9222995603882194e-05,
+      "loss": 0.8175,
+      "step": 2255
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9222148577194818e-05,
+      "loss": 0.4551,
+      "step": 2256
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.922130110776138e-05,
+      "loss": 0.8254,
+      "step": 2257
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9220453195622566e-05,
+      "loss": 0.8676,
+      "step": 2258
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9219604840819085e-05,
+      "loss": 0.4575,
+      "step": 2259
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9218756043391664e-05,
+      "loss": 0.8393,
+      "step": 2260
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9217906803381058e-05,
+      "loss": 0.8365,
+      "step": 2261
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.921705712082803e-05,
+      "loss": 0.9045,
+      "step": 2262
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9216206995773373e-05,
+      "loss": 0.8555,
+      "step": 2263
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.921535642825791e-05,
+      "loss": 0.8595,
+      "step": 2264
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.921450541832247e-05,
+      "loss": 0.8488,
+      "step": 2265
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9213653966007907e-05,
+      "loss": 0.9185,
+      "step": 2266
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9212802071355098e-05,
+      "loss": 0.8392,
+      "step": 2267
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.921194973440495e-05,
+      "loss": 0.5171,
+      "step": 2268
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.921109695519837e-05,
+      "loss": 0.4617,
+      "step": 2269
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.921024373377631e-05,
+      "loss": 0.8398,
+      "step": 2270
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9209390070179728e-05,
+      "loss": 0.8734,
+      "step": 2271
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.920853596444961e-05,
+      "loss": 0.8809,
+      "step": 2272
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9207681416626952e-05,
+      "loss": 0.8398,
+      "step": 2273
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.920682642675279e-05,
+      "loss": 0.8253,
+      "step": 2274
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.920597099486817e-05,
+      "loss": 0.4953,
+      "step": 2275
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9205115121014155e-05,
+      "loss": 0.4223,
+      "step": 2276
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9204258805231838e-05,
+      "loss": 0.8873,
+      "step": 2277
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9203402047562332e-05,
+      "loss": 0.9103,
+      "step": 2278
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9202544848046766e-05,
+      "loss": 0.4799,
+      "step": 2279
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9201687206726292e-05,
+      "loss": 0.8359,
+      "step": 2280
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9200829123642087e-05,
+      "loss": 0.9293,
+      "step": 2281
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9199970598835347e-05,
+      "loss": 0.8729,
+      "step": 2282
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.919911163234729e-05,
+      "loss": 0.8298,
+      "step": 2283
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9198252224219146e-05,
+      "loss": 0.8753,
+      "step": 2284
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9197392374492184e-05,
+      "loss": 0.826,
+      "step": 2285
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9196532083207684e-05,
+      "loss": 0.508,
+      "step": 2286
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9195671350406943e-05,
+      "loss": 0.5382,
+      "step": 2287
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9194810176131288e-05,
+      "loss": 0.9121,
+      "step": 2288
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9193948560422054e-05,
+      "loss": 0.439,
+      "step": 2289
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9193086503320617e-05,
+      "loss": 0.4481,
+      "step": 2290
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.919222400486836e-05,
+      "loss": 0.8575,
+      "step": 2291
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.919136106510669e-05,
+      "loss": 0.8732,
+      "step": 2292
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9190497684077036e-05,
+      "loss": 0.8809,
+      "step": 2293
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.918963386182085e-05,
+      "loss": 0.4665,
+      "step": 2294
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.91887695983796e-05,
+      "loss": 0.859,
+      "step": 2295
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9187904893794777e-05,
+      "loss": 0.8504,
+      "step": 2296
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.91870397481079e-05,
+      "loss": 0.8857,
+      "step": 2297
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.91861741613605e-05,
+      "loss": 0.4764,
+      "step": 2298
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9185308133594137e-05,
+      "loss": 0.8835,
+      "step": 2299
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9184441664850385e-05,
+      "loss": 0.4899,
+      "step": 2300
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.918357475517084e-05,
+      "loss": 0.8574,
+      "step": 2301
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9182707404597126e-05,
+      "loss": 0.84,
+      "step": 2302
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9181839613170877e-05,
+      "loss": 0.8601,
+      "step": 2303
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9180971380933765e-05,
+      "loss": 0.8747,
+      "step": 2304
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9180102707927466e-05,
+      "loss": 0.79,
+      "step": 2305
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9179233594193685e-05,
+      "loss": 0.8608,
+      "step": 2306
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9178364039774147e-05,
+      "loss": 0.4724,
+      "step": 2307
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.91774940447106e-05,
+      "loss": 0.8743,
+      "step": 2308
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9176623609044812e-05,
+      "loss": 0.8379,
+      "step": 2309
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.917575273281857e-05,
+      "loss": 0.5039,
+      "step": 2310
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9174881416073683e-05,
+      "loss": 0.8292,
+      "step": 2311
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9174009658851985e-05,
+      "loss": 0.85,
+      "step": 2312
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9173137461195327e-05,
+      "loss": 0.4518,
+      "step": 2313
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9172264823145582e-05,
+      "loss": 0.8386,
+      "step": 2314
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.917139174474464e-05,
+      "loss": 0.8511,
+      "step": 2315
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9170518226034426e-05,
+      "loss": 0.4725,
+      "step": 2316
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.916964426705687e-05,
+      "loss": 0.8439,
+      "step": 2317
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9168769867853932e-05,
+      "loss": 0.8838,
+      "step": 2318
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.916789502846759e-05,
+      "loss": 0.8962,
+      "step": 2319
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9167019748939847e-05,
+      "loss": 0.8679,
+      "step": 2320
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.916614402931272e-05,
+      "loss": 0.8132,
+      "step": 2321
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9165267869628253e-05,
+      "loss": 0.8774,
+      "step": 2322
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9164391269928514e-05,
+      "loss": 0.8082,
+      "step": 2323
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.916351423025558e-05,
+      "loss": 0.4622,
+      "step": 2324
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9162636750651566e-05,
+      "loss": 0.8799,
+      "step": 2325
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.916175883115859e-05,
+      "loss": 0.8722,
+      "step": 2326
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9160880471818802e-05,
+      "loss": 0.4785,
+      "step": 2327
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9160001672674372e-05,
+      "loss": 0.4461,
+      "step": 2328
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.915912243376749e-05,
+      "loss": 0.8661,
+      "step": 2329
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9158242755140376e-05,
+      "loss": 0.88,
+      "step": 2330
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9157362636835247e-05,
+      "loss": 0.8259,
+      "step": 2331
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.915648207889437e-05,
+      "loss": 0.9031,
+      "step": 2332
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.915560108136001e-05,
+      "loss": 0.8169,
+      "step": 2333
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.915471964427447e-05,
+      "loss": 0.846,
+      "step": 2334
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9153837767680062e-05,
+      "loss": 0.8039,
+      "step": 2335
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9152955451619127e-05,
+      "loss": 0.8843,
+      "step": 2336
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.915207269613402e-05,
+      "loss": 0.817,
+      "step": 2337
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9151189501267126e-05,
+      "loss": 0.8259,
+      "step": 2338
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9150305867060847e-05,
+      "loss": 0.859,
+      "step": 2339
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.91494217935576e-05,
+      "loss": 0.8589,
+      "step": 2340
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.914853728079983e-05,
+      "loss": 0.8236,
+      "step": 2341
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9147652328830006e-05,
+      "loss": 0.8407,
+      "step": 2342
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.914676693769061e-05,
+      "loss": 0.8589,
+      "step": 2343
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9145881107424148e-05,
+      "loss": 0.498,
+      "step": 2344
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.914499483807315e-05,
+      "loss": 0.4766,
+      "step": 2345
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9144108129680162e-05,
+      "loss": 0.8712,
+      "step": 2346
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.914322098228776e-05,
+      "loss": 0.8676,
+      "step": 2347
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9142333395938526e-05,
+      "loss": 0.8023,
+      "step": 2348
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.914144537067508e-05,
+      "loss": 0.8928,
+      "step": 2349
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.914055690654005e-05,
+      "loss": 0.8433,
+      "step": 2350
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9139668003576097e-05,
+      "loss": 0.8333,
+      "step": 2351
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.913877866182589e-05,
+      "loss": 0.4499,
+      "step": 2352
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9137888881332126e-05,
+      "loss": 0.8569,
+      "step": 2353
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9136998662137525e-05,
+      "loss": 0.8267,
+      "step": 2354
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9136108004284824e-05,
+      "loss": 0.8766,
+      "step": 2355
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9135216907816784e-05,
+      "loss": 0.4603,
+      "step": 2356
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9134325372776184e-05,
+      "loss": 0.8338,
+      "step": 2357
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9133433399205824e-05,
+      "loss": 0.8575,
+      "step": 2358
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.913254098714853e-05,
+      "loss": 0.4796,
+      "step": 2359
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9131648136647146e-05,
+      "loss": 0.4521,
+      "step": 2360
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9130754847744534e-05,
+      "loss": 0.5273,
+      "step": 2361
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9129861120483584e-05,
+      "loss": 0.7891,
+      "step": 2362
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9128966954907197e-05,
+      "loss": 0.8169,
+      "step": 2363
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9128072351058308e-05,
+      "loss": 0.9012,
+      "step": 2364
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.912717730897986e-05,
+      "loss": 0.8437,
+      "step": 2365
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9126281828714828e-05,
+      "loss": 0.8755,
+      "step": 2366
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.91253859103062e-05,
+      "loss": 0.9289,
+      "step": 2367
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.912448955379699e-05,
+      "loss": 0.8383,
+      "step": 2368
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9123592759230226e-05,
+      "loss": 0.8155,
+      "step": 2369
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9122695526648968e-05,
+      "loss": 0.4283,
+      "step": 2370
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.912179785609629e-05,
+      "loss": 0.8196,
+      "step": 2371
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.912089974761529e-05,
+      "loss": 0.846,
+      "step": 2372
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9120001201249078e-05,
+      "loss": 0.8924,
+      "step": 2373
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9119102217040802e-05,
+      "loss": 0.8763,
+      "step": 2374
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9118202795033616e-05,
+      "loss": 0.4798,
+      "step": 2375
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.91173029352707e-05,
+      "loss": 0.8363,
+      "step": 2376
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.911640263779525e-05,
+      "loss": 0.8816,
+      "step": 2377
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9115501902650502e-05,
+      "loss": 0.892,
+      "step": 2378
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9114600729879693e-05,
+      "loss": 0.4431,
+      "step": 2379
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9113699119526083e-05,
+      "loss": 0.8669,
+      "step": 2380
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9112797071632966e-05,
+      "loss": 0.4757,
+      "step": 2381
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.911189458624364e-05,
+      "loss": 0.883,
+      "step": 2382
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.911099166340143e-05,
+      "loss": 0.4559,
+      "step": 2383
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9110088303149696e-05,
+      "loss": 0.8271,
+      "step": 2384
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.91091845055318e-05,
+      "loss": 0.8508,
+      "step": 2385
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9108280270591136e-05,
+      "loss": 0.8803,
+      "step": 2386
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9107375598371112e-05,
+      "loss": 0.8779,
+      "step": 2387
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9106470488915165e-05,
+      "loss": 0.9192,
+      "step": 2388
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9105564942266742e-05,
+      "loss": 0.8973,
+      "step": 2389
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.910465895846932e-05,
+      "loss": 0.8456,
+      "step": 2390
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9103752537566394e-05,
+      "loss": 0.8211,
+      "step": 2391
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9102845679601482e-05,
+      "loss": 0.4671,
+      "step": 2392
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.910193838461812e-05,
+      "loss": 0.8587,
+      "step": 2393
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.910103065265987e-05,
+      "loss": 0.8742,
+      "step": 2394
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.910012248377031e-05,
+      "loss": 0.4484,
+      "step": 2395
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9099213877993032e-05,
+      "loss": 0.5157,
+      "step": 2396
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9098304835371666e-05,
+      "loss": 0.8454,
+      "step": 2397
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9097395355949852e-05,
+      "loss": 0.8482,
+      "step": 2398
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9096485439771254e-05,
+      "loss": 0.8796,
+      "step": 2399
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9095575086879555e-05,
+      "loss": 0.807,
+      "step": 2400
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.909466429731846e-05,
+      "loss": 0.8864,
+      "step": 2401
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9093753071131697e-05,
+      "loss": 0.8532,
+      "step": 2402
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.909284140836301e-05,
+      "loss": 0.7294,
+      "step": 2403
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.909192930905617e-05,
+      "loss": 0.8338,
+      "step": 2404
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9091016773254966e-05,
+      "loss": 0.4673,
+      "step": 2405
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9090103801003202e-05,
+      "loss": 0.8716,
+      "step": 2406
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9089190392344715e-05,
+      "loss": 0.8398,
+      "step": 2407
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9088276547323355e-05,
+      "loss": 0.8909,
+      "step": 2408
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9087362265982995e-05,
+      "loss": 0.8619,
+      "step": 2409
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9086447548367534e-05,
+      "loss": 0.8874,
+      "step": 2410
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9085532394520875e-05,
+      "loss": 0.9006,
+      "step": 2411
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.908461680448696e-05,
+      "loss": 0.8441,
+      "step": 2412
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9083700778309747e-05,
+      "loss": 0.8842,
+      "step": 2413
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9082784316033213e-05,
+      "loss": 0.8799,
+      "step": 2414
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9081867417701356e-05,
+      "loss": 0.8451,
+      "step": 2415
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9080950083358195e-05,
+      "loss": 0.4929,
+      "step": 2416
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9080032313047766e-05,
+      "loss": 0.87,
+      "step": 2417
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9079114106814138e-05,
+      "loss": 0.8565,
+      "step": 2418
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.907819546470139e-05,
+      "loss": 0.8348,
+      "step": 2419
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9077276386753623e-05,
+      "loss": 0.8744,
+      "step": 2420
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9076356873014963e-05,
+      "loss": 0.8306,
+      "step": 2421
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9075436923529555e-05,
+      "loss": 0.9183,
+      "step": 2422
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9074516538341563e-05,
+      "loss": 0.8555,
+      "step": 2423
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9073595717495178e-05,
+      "loss": 0.4536,
+      "step": 2424
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9072674461034602e-05,
+      "loss": 0.482,
+      "step": 2425
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9071752769004067e-05,
+      "loss": 0.8342,
+      "step": 2426
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.907083064144782e-05,
+      "loss": 0.8203,
+      "step": 2427
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9069908078410135e-05,
+      "loss": 0.8862,
+      "step": 2428
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.90689850799353e-05,
+      "loss": 0.8792,
+      "step": 2429
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9068061646067636e-05,
+      "loss": 0.8758,
+      "step": 2430
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.906713777685146e-05,
+      "loss": 0.8577,
+      "step": 2431
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9066213472331138e-05,
+      "loss": 0.8431,
+      "step": 2432
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9065288732551046e-05,
+      "loss": 0.856,
+      "step": 2433
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.906436355755557e-05,
+      "loss": 0.8487,
+      "step": 2434
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9063437947389135e-05,
+      "loss": 0.8782,
+      "step": 2435
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9062511902096177e-05,
+      "loss": 0.8472,
+      "step": 2436
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9061585421721153e-05,
+      "loss": 0.4768,
+      "step": 2437
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9060658506308544e-05,
+      "loss": 0.8741,
+      "step": 2438
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.905973115590285e-05,
+      "loss": 0.8152,
+      "step": 2439
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9058803370548586e-05,
+      "loss": 0.4925,
+      "step": 2440
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9057875150290304e-05,
+      "loss": 0.4601,
+      "step": 2441
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.905694649517256e-05,
+      "loss": 0.91,
+      "step": 2442
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9056017405239944e-05,
+      "loss": 0.859,
+      "step": 2443
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9055087880537052e-05,
+      "loss": 0.7977,
+      "step": 2444
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9054157921108517e-05,
+      "loss": 0.4424,
+      "step": 2445
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9053227526998986e-05,
+      "loss": 0.8243,
+      "step": 2446
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.905229669825312e-05,
+      "loss": 0.86,
+      "step": 2447
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.905136543491561e-05,
+      "loss": 0.8939,
+      "step": 2448
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9050433737031167e-05,
+      "loss": 0.502,
+      "step": 2449
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9049501604644516e-05,
+      "loss": 0.4704,
+      "step": 2450
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.904856903780041e-05,
+      "loss": 0.8328,
+      "step": 2451
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9047636036543626e-05,
+      "loss": 0.4905,
+      "step": 2452
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9046702600918953e-05,
+      "loss": 0.4439,
+      "step": 2453
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9045768730971198e-05,
+      "loss": 0.4452,
+      "step": 2454
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.90448344267452e-05,
+      "loss": 0.4429,
+      "step": 2455
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.904389968828582e-05,
+      "loss": 0.918,
+      "step": 2456
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9042964515637926e-05,
+      "loss": 0.8923,
+      "step": 2457
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9042028908846415e-05,
+      "loss": 0.8682,
+      "step": 2458
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9041092867956207e-05,
+      "loss": 0.828,
+      "step": 2459
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9040156393012243e-05,
+      "loss": 0.7674,
+      "step": 2460
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9039219484059476e-05,
+      "loss": 0.8747,
+      "step": 2461
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.903828214114289e-05,
+      "loss": 0.4506,
+      "step": 2462
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9037344364307483e-05,
+      "loss": 0.8564,
+      "step": 2463
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.903640615359828e-05,
+      "loss": 0.83,
+      "step": 2464
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9035467509060327e-05,
+      "loss": 0.4896,
+      "step": 2465
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9034528430738675e-05,
+      "loss": 0.8557,
+      "step": 2466
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.903358891867842e-05,
+      "loss": 0.8547,
+      "step": 2467
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9032648972924662e-05,
+      "loss": 0.813,
+      "step": 2468
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.903170859352253e-05,
+      "loss": 0.8158,
+      "step": 2469
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9030767780517165e-05,
+      "loss": 0.4834,
+      "step": 2470
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.902982653395374e-05,
+      "loss": 0.4858,
+      "step": 2471
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.902888485387744e-05,
+      "loss": 0.8422,
+      "step": 2472
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9027942740333477e-05,
+      "loss": 0.8319,
+      "step": 2473
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9027000193367077e-05,
+      "loss": 0.9006,
+      "step": 2474
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9026057213023494e-05,
+      "loss": 0.8425,
+      "step": 2475
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9025113799348002e-05,
+      "loss": 0.7984,
+      "step": 2476
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9024169952385887e-05,
+      "loss": 0.8819,
+      "step": 2477
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9023225672182463e-05,
+      "loss": 0.8897,
+      "step": 2478
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.902228095878307e-05,
+      "loss": 0.4819,
+      "step": 2479
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9021335812233052e-05,
+      "loss": 0.4512,
+      "step": 2480
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9020390232577796e-05,
+      "loss": 0.8792,
+      "step": 2481
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9019444219862695e-05,
+      "loss": 0.827,
+      "step": 2482
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.901849777413316e-05,
+      "loss": 0.4908,
+      "step": 2483
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.901755089543464e-05,
+      "loss": 0.9058,
+      "step": 2484
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9016603583812582e-05,
+      "loss": 0.8575,
+      "step": 2485
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.901565583931247e-05,
+      "loss": 0.9052,
+      "step": 2486
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.901470766197981e-05,
+      "loss": 0.4492,
+      "step": 2487
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9013759051860116e-05,
+      "loss": 0.8305,
+      "step": 2488
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9012810008998928e-05,
+      "loss": 0.8696,
+      "step": 2489
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9011860533441818e-05,
+      "loss": 0.8623,
+      "step": 2490
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.901091062523436e-05,
+      "loss": 0.8908,
+      "step": 2491
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9009960284422164e-05,
+      "loss": 0.474,
+      "step": 2492
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9009009511050852e-05,
+      "loss": 0.8351,
+      "step": 2493
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.900805830516607e-05,
+      "loss": 0.8283,
+      "step": 2494
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9007106666813487e-05,
+      "loss": 0.8795,
+      "step": 2495
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9006154596038788e-05,
+      "loss": 0.8652,
+      "step": 2496
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9005202092887677e-05,
+      "loss": 0.831,
+      "step": 2497
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.900424915740589e-05,
+      "loss": 0.4522,
+      "step": 2498
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9003295789639175e-05,
+      "loss": 0.5106,
+      "step": 2499
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9002341989633297e-05,
+      "loss": 0.8163,
+      "step": 2500
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9001387757434055e-05,
+      "loss": 0.4701,
+      "step": 2501
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.900043309308725e-05,
+      "loss": 0.853,
+      "step": 2502
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8999477996638722e-05,
+      "loss": 0.8817,
+      "step": 2503
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8998522468134325e-05,
+      "loss": 0.4613,
+      "step": 2504
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8997566507619933e-05,
+      "loss": 0.9025,
+      "step": 2505
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8996610115141435e-05,
+      "loss": 0.475,
+      "step": 2506
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.899565329074475e-05,
+      "loss": 0.5074,
+      "step": 2507
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.899469603447582e-05,
+      "loss": 0.8327,
+      "step": 2508
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.899373834638059e-05,
+      "loss": 0.4773,
+      "step": 2509
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8992780226505044e-05,
+      "loss": 0.8165,
+      "step": 2510
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8991821674895182e-05,
+      "loss": 0.8422,
+      "step": 2511
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.899086269159702e-05,
+      "loss": 0.8917,
+      "step": 2512
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.89899032766566e-05,
+      "loss": 0.4581,
+      "step": 2513
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8988943430119978e-05,
+      "loss": 0.8263,
+      "step": 2514
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8987983152033244e-05,
+      "loss": 0.8318,
+      "step": 2515
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8987022442442493e-05,
+      "loss": 0.4515,
+      "step": 2516
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.898606130139385e-05,
+      "loss": 0.9013,
+      "step": 2517
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8985099728933457e-05,
+      "loss": 0.8607,
+      "step": 2518
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8984137725107482e-05,
+      "loss": 0.8603,
+      "step": 2519
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8983175289962102e-05,
+      "loss": 0.7898,
+      "step": 2520
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.898221242354353e-05,
+      "loss": 0.8718,
+      "step": 2521
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8981249125897988e-05,
+      "loss": 0.8714,
+      "step": 2522
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8980285397071726e-05,
+      "loss": 0.8589,
+      "step": 2523
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.897932123711101e-05,
+      "loss": 0.4357,
+      "step": 2524
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.897835664606213e-05,
+      "loss": 0.5396,
+      "step": 2525
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8977391623971392e-05,
+      "loss": 0.8866,
+      "step": 2526
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8976426170885127e-05,
+      "loss": 0.8456,
+      "step": 2527
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.897546028684969e-05,
+      "loss": 0.8016,
+      "step": 2528
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8974493971911442e-05,
+      "loss": 0.8784,
+      "step": 2529
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8973527226116785e-05,
+      "loss": 0.5023,
+      "step": 2530
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8972560049512124e-05,
+      "loss": 0.8383,
+      "step": 2531
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.89715924421439e-05,
+      "loss": 0.4863,
+      "step": 2532
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8970624404058558e-05,
+      "loss": 0.483,
+      "step": 2533
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.896965593530258e-05,
+      "loss": 0.8573,
+      "step": 2534
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8968687035922455e-05,
+      "loss": 0.8597,
+      "step": 2535
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8967717705964703e-05,
+      "loss": 0.8468,
+      "step": 2536
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.896674794547586e-05,
+      "loss": 0.8276,
+      "step": 2537
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8965777754502483e-05,
+      "loss": 0.832,
+      "step": 2538
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.896480713309115e-05,
+      "loss": 0.8261,
+      "step": 2539
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.896383608128846e-05,
+      "loss": 0.5203,
+      "step": 2540
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8962864599141026e-05,
+      "loss": 0.465,
+      "step": 2541
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8961892686695496e-05,
+      "loss": 0.874,
+      "step": 2542
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.896092034399853e-05,
+      "loss": 0.8654,
+      "step": 2543
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8959947571096803e-05,
+      "loss": 0.4209,
+      "step": 2544
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8958974368037023e-05,
+      "loss": 0.8835,
+      "step": 2545
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8958000734865914e-05,
+      "loss": 0.7988,
+      "step": 2546
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.895702667163021e-05,
+      "loss": 0.892,
+      "step": 2547
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8956052178376684e-05,
+      "loss": 0.4753,
+      "step": 2548
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8955077255152115e-05,
+      "loss": 0.8509,
+      "step": 2549
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.895410190200331e-05,
+      "loss": 0.4512,
+      "step": 2550
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8953126118977095e-05,
+      "loss": 0.48,
+      "step": 2551
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8952149906120314e-05,
+      "loss": 0.4897,
+      "step": 2552
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.895117326347984e-05,
+      "loss": 0.8235,
+      "step": 2553
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.895019619110255e-05,
+      "loss": 0.4437,
+      "step": 2554
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8949218689035364e-05,
+      "loss": 0.8941,
+      "step": 2555
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.89482407573252e-05,
+      "loss": 0.4933,
+      "step": 2556
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.894726239601902e-05,
+      "loss": 0.8238,
+      "step": 2557
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8946283605163786e-05,
+      "loss": 0.8364,
+      "step": 2558
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.894530438480649e-05,
+      "loss": 0.4768,
+      "step": 2559
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8944324734994137e-05,
+      "loss": 0.4831,
+      "step": 2560
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.894334465577377e-05,
+      "loss": 0.8774,
+      "step": 2561
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8942364147192437e-05,
+      "loss": 0.4802,
+      "step": 2562
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8941383209297215e-05,
+      "loss": 0.8862,
+      "step": 2563
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.894040184213519e-05,
+      "loss": 0.8588,
+      "step": 2564
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.893942004575348e-05,
+      "loss": 0.8394,
+      "step": 2565
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8938437820199224e-05,
+      "loss": 0.4661,
+      "step": 2566
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8937455165519572e-05,
+      "loss": 0.8233,
+      "step": 2567
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8936472081761703e-05,
+      "loss": 0.8653,
+      "step": 2568
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8935488568972813e-05,
+      "loss": 0.8288,
+      "step": 2569
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.893450462720012e-05,
+      "loss": 0.9027,
+      "step": 2570
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8933520256490862e-05,
+      "loss": 0.8517,
+      "step": 2571
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8932535456892297e-05,
+      "loss": 0.9129,
+      "step": 2572
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8931550228451704e-05,
+      "loss": 0.8275,
+      "step": 2573
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8930564571216386e-05,
+      "loss": 0.8862,
+      "step": 2574
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8929578485233658e-05,
+      "loss": 0.4879,
+      "step": 2575
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.892859197055087e-05,
+      "loss": 0.8605,
+      "step": 2576
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.892760502721537e-05,
+      "loss": 0.8438,
+      "step": 2577
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8926617655274553e-05,
+      "loss": 0.8246,
+      "step": 2578
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8925629854775814e-05,
+      "loss": 0.8657,
+      "step": 2579
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.892464162576658e-05,
+      "loss": 0.8235,
+      "step": 2580
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8923652968294294e-05,
+      "loss": 0.8772,
+      "step": 2581
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.892266388240642e-05,
+      "loss": 0.4622,
+      "step": 2582
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.892167436815044e-05,
+      "loss": 0.8451,
+      "step": 2583
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8920684425573865e-05,
+      "loss": 0.8443,
+      "step": 2584
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8919694054724222e-05,
+      "loss": 0.4361,
+      "step": 2585
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.891870325564905e-05,
+      "loss": 0.8431,
+      "step": 2586
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8917712028395922e-05,
+      "loss": 0.4429,
+      "step": 2587
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8916720373012425e-05,
+      "loss": 0.872,
+      "step": 2588
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.891572828954617e-05,
+      "loss": 0.8523,
+      "step": 2589
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.891473577804478e-05,
+      "loss": 0.8892,
+      "step": 2590
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8913742838555906e-05,
+      "loss": 0.8804,
+      "step": 2591
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8912749471127223e-05,
+      "loss": 0.4701,
+      "step": 2592
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8911755675806417e-05,
+      "loss": 0.8599,
+      "step": 2593
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.89107614526412e-05,
+      "loss": 0.8876,
+      "step": 2594
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.89097668016793e-05,
+      "loss": 0.8555,
+      "step": 2595
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8908771722968477e-05,
+      "loss": 0.8641,
+      "step": 2596
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.89077762165565e-05,
+      "loss": 0.801,
+      "step": 2597
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8906780282491158e-05,
+      "loss": 0.8155,
+      "step": 2598
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8905783920820272e-05,
+      "loss": 0.8114,
+      "step": 2599
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8904787131591674e-05,
+      "loss": 0.8146,
+      "step": 2600
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8903789914853215e-05,
+      "loss": 0.4764,
+      "step": 2601
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8902792270652776e-05,
+      "loss": 0.9046,
+      "step": 2602
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.890179419903825e-05,
+      "loss": 0.8484,
+      "step": 2603
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8900795700057552e-05,
+      "loss": 0.8427,
+      "step": 2604
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.889979677375862e-05,
+      "loss": 0.4474,
+      "step": 2605
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8898797420189417e-05,
+      "loss": 0.4563,
+      "step": 2606
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.889779763939791e-05,
+      "loss": 0.8711,
+      "step": 2607
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.889679743143211e-05,
+      "loss": 0.8545,
+      "step": 2608
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.889579679634002e-05,
+      "loss": 0.8609,
+      "step": 2609
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8894795734169694e-05,
+      "loss": 0.8233,
+      "step": 2610
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8893794244969188e-05,
+      "loss": 0.4603,
+      "step": 2611
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.889279232878658e-05,
+      "loss": 0.8144,
+      "step": 2612
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.889178998566997e-05,
+      "loss": 0.8394,
+      "step": 2613
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8890787215667487e-05,
+      "loss": 0.8142,
+      "step": 2614
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8889784018827265e-05,
+      "loss": 0.463,
+      "step": 2615
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.888878039519747e-05,
+      "loss": 0.4828,
+      "step": 2616
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8887776344826283e-05,
+      "loss": 0.7751,
+      "step": 2617
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.888677186776191e-05,
+      "loss": 0.8251,
+      "step": 2618
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8885766964052573e-05,
+      "loss": 0.8479,
+      "step": 2619
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8884761633746516e-05,
+      "loss": 0.4445,
+      "step": 2620
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8883755876892004e-05,
+      "loss": 0.8705,
+      "step": 2621
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8882749693537328e-05,
+      "loss": 0.812,
+      "step": 2622
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8881743083730786e-05,
+      "loss": 0.8686,
+      "step": 2623
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8880736047520705e-05,
+      "loss": 0.8452,
+      "step": 2624
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.887972858495544e-05,
+      "loss": 0.8739,
+      "step": 2625
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.887872069608335e-05,
+      "loss": 0.8673,
+      "step": 2626
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.887771238095282e-05,
+      "loss": 0.8446,
+      "step": 2627
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.887670363961227e-05,
+      "loss": 0.9023,
+      "step": 2628
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8875694472110123e-05,
+      "loss": 0.8778,
+      "step": 2629
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8874684878494824e-05,
+      "loss": 0.8972,
+      "step": 2630
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8873674858814844e-05,
+      "loss": 0.8901,
+      "step": 2631
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.887266441311868e-05,
+      "loss": 0.8743,
+      "step": 2632
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8871653541454833e-05,
+      "loss": 0.8662,
+      "step": 2633
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.887064224387184e-05,
+      "loss": 0.4304,
+      "step": 2634
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.886963052041825e-05,
+      "loss": 0.9148,
+      "step": 2635
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8868618371142634e-05,
+      "loss": 0.8289,
+      "step": 2636
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8867605796093588e-05,
+      "loss": 0.8493,
+      "step": 2637
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8866592795319724e-05,
+      "loss": 0.8766,
+      "step": 2638
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.886557936886967e-05,
+      "loss": 0.8361,
+      "step": 2639
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8864565516792084e-05,
+      "loss": 0.8336,
+      "step": 2640
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.886355123913564e-05,
+      "loss": 0.4357,
+      "step": 2641
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.886253653594903e-05,
+      "loss": 0.9187,
+      "step": 2642
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.886152140728097e-05,
+      "loss": 0.4561,
+      "step": 2643
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.88605058531802e-05,
+      "loss": 0.4733,
+      "step": 2644
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8859489873695465e-05,
+      "loss": 0.8484,
+      "step": 2645
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8858473468875554e-05,
+      "loss": 0.4458,
+      "step": 2646
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8857456638769253e-05,
+      "loss": 0.8429,
+      "step": 2647
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8856439383425386e-05,
+      "loss": 0.837,
+      "step": 2648
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8855421702892788e-05,
+      "loss": 0.8269,
+      "step": 2649
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8854403597220314e-05,
+      "loss": 0.8831,
+      "step": 2650
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8853385066456847e-05,
+      "loss": 0.465,
+      "step": 2651
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8852366110651278e-05,
+      "loss": 0.8359,
+      "step": 2652
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.885134672985254e-05,
+      "loss": 0.8547,
+      "step": 2653
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8850326924109557e-05,
+      "loss": 0.8335,
+      "step": 2654
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8849306693471297e-05,
+      "loss": 0.5068,
+      "step": 2655
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.884828603798674e-05,
+      "loss": 0.4628,
+      "step": 2656
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8847264957704886e-05,
+      "loss": 0.5076,
+      "step": 2657
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8846243452674753e-05,
+      "loss": 0.8003,
+      "step": 2658
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8845221522945387e-05,
+      "loss": 0.8733,
+      "step": 2659
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8844199168565844e-05,
+      "loss": 0.4361,
+      "step": 2660
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8843176389585215e-05,
+      "loss": 0.4847,
+      "step": 2661
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8842153186052595e-05,
+      "loss": 0.8441,
+      "step": 2662
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8841129558017114e-05,
+      "loss": 0.4643,
+      "step": 2663
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8840105505527904e-05,
+      "loss": 0.8295,
+      "step": 2664
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8839081028634137e-05,
+      "loss": 0.4913,
+      "step": 2665
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8838056127385002e-05,
+      "loss": 0.8486,
+      "step": 2666
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.883703080182969e-05,
+      "loss": 0.8529,
+      "step": 2667
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8836005052017436e-05,
+      "loss": 0.4927,
+      "step": 2668
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.883497887799748e-05,
+      "loss": 0.8244,
+      "step": 2669
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8833952279819098e-05,
+      "loss": 0.8949,
+      "step": 2670
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8832925257531557e-05,
+      "loss": 0.4951,
+      "step": 2671
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8831897811184182e-05,
+      "loss": 0.8098,
+      "step": 2672
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.883086994082629e-05,
+      "loss": 0.8442,
+      "step": 2673
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8829841646507224e-05,
+      "loss": 0.4709,
+      "step": 2674
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8828812928276364e-05,
+      "loss": 0.9032,
+      "step": 2675
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.882778378618309e-05,
+      "loss": 0.903,
+      "step": 2676
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8826754220276808e-05,
+      "loss": 0.476,
+      "step": 2677
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8825724230606953e-05,
+      "loss": 0.8669,
+      "step": 2678
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8824693817222965e-05,
+      "loss": 0.5149,
+      "step": 2679
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8823662980174323e-05,
+      "loss": 0.8733,
+      "step": 2680
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.882263171951051e-05,
+      "loss": 0.4162,
+      "step": 2681
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8821600035281036e-05,
+      "loss": 0.8994,
+      "step": 2682
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8820567927535437e-05,
+      "loss": 0.458,
+      "step": 2683
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8819535396323254e-05,
+      "loss": 0.9471,
+      "step": 2684
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8818502441694067e-05,
+      "loss": 0.8486,
+      "step": 2685
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.881746906369746e-05,
+      "loss": 0.4644,
+      "step": 2686
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.881643526238305e-05,
+      "loss": 0.8783,
+      "step": 2687
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8815401037800464e-05,
+      "loss": 0.4286,
+      "step": 2688
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8814366389999357e-05,
+      "loss": 0.8712,
+      "step": 2689
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.88133313190294e-05,
+      "loss": 0.4393,
+      "step": 2690
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8812295824940284e-05,
+      "loss": 0.8853,
+      "step": 2691
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.881125990778173e-05,
+      "loss": 0.8413,
+      "step": 2692
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8810223567603464e-05,
+      "loss": 0.4581,
+      "step": 2693
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8809186804455238e-05,
+      "loss": 0.8421,
+      "step": 2694
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.880814961838683e-05,
+      "loss": 0.8679,
+      "step": 2695
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8807112009448037e-05,
+      "loss": 0.4797,
+      "step": 2696
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.880607397768867e-05,
+      "loss": 0.4751,
+      "step": 2697
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8805035523158563e-05,
+      "loss": 0.8569,
+      "step": 2698
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.880399664590757e-05,
+      "loss": 0.8406,
+      "step": 2699
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.880295734598557e-05,
+      "loss": 0.4494,
+      "step": 2700
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8801917623442457e-05,
+      "loss": 0.8738,
+      "step": 2701
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8800877478328147e-05,
+      "loss": 0.7958,
+      "step": 2702
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.879983691069258e-05,
+      "loss": 0.4681,
+      "step": 2703
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8798795920585707e-05,
+      "loss": 0.4664,
+      "step": 2704
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8797754508057506e-05,
+      "loss": 0.473,
+      "step": 2705
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8796712673157976e-05,
+      "loss": 0.8377,
+      "step": 2706
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8795670415937133e-05,
+      "loss": 0.8401,
+      "step": 2707
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8794627736445015e-05,
+      "loss": 0.9276,
+      "step": 2708
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8793584634731686e-05,
+      "loss": 0.4658,
+      "step": 2709
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8792541110847213e-05,
+      "loss": 0.4442,
+      "step": 2710
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.87914971648417e-05,
+      "loss": 0.4782,
+      "step": 2711
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.879045279676527e-05,
+      "loss": 0.858,
+      "step": 2712
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.878940800666806e-05,
+      "loss": 0.8784,
+      "step": 2713
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8788362794600223e-05,
+      "loss": 0.8468,
+      "step": 2714
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8787317160611943e-05,
+      "loss": 0.8693,
+      "step": 2715
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8786271104753422e-05,
+      "loss": 0.8901,
+      "step": 2716
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8785224627074876e-05,
+      "loss": 0.8458,
+      "step": 2717
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8784177727626546e-05,
+      "loss": 0.8058,
+      "step": 2718
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.87831304064587e-05,
+      "loss": 0.8158,
+      "step": 2719
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8782082663621608e-05,
+      "loss": 0.8321,
+      "step": 2720
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8781034499165577e-05,
+      "loss": 0.8436,
+      "step": 2721
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8779985913140927e-05,
+      "loss": 0.8359,
+      "step": 2722
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8778936905597997e-05,
+      "loss": 0.8861,
+      "step": 2723
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8777887476587153e-05,
+      "loss": 0.8358,
+      "step": 2724
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.877683762615878e-05,
+      "loss": 0.4892,
+      "step": 2725
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.877578735436327e-05,
+      "loss": 0.8864,
+      "step": 2726
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8774736661251053e-05,
+      "loss": 0.8167,
+      "step": 2727
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.877368554687257e-05,
+      "loss": 0.854,
+      "step": 2728
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8772634011278284e-05,
+      "loss": 0.5189,
+      "step": 2729
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.877158205451868e-05,
+      "loss": 0.8357,
+      "step": 2730
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.877052967664426e-05,
+      "loss": 0.4174,
+      "step": 2731
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8769476877705543e-05,
+      "loss": 0.8473,
+      "step": 2732
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.876842365775308e-05,
+      "loss": 0.8732,
+      "step": 2733
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.876737001683743e-05,
+      "loss": 0.779,
+      "step": 2734
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.876631595500918e-05,
+      "loss": 0.8542,
+      "step": 2735
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8765261472318936e-05,
+      "loss": 0.8733,
+      "step": 2736
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8764206568817317e-05,
+      "loss": 0.8525,
+      "step": 2737
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8763151244554973e-05,
+      "loss": 0.946,
+      "step": 2738
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8762095499582564e-05,
+      "loss": 0.8467,
+      "step": 2739
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.876103933395078e-05,
+      "loss": 0.8614,
+      "step": 2740
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.875998274771033e-05,
+      "loss": 0.8433,
+      "step": 2741
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8758925740911926e-05,
+      "loss": 0.4312,
+      "step": 2742
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8757868313606327e-05,
+      "loss": 0.8072,
+      "step": 2743
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8756810465844296e-05,
+      "loss": 0.8592,
+      "step": 2744
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8755752197676612e-05,
+      "loss": 0.4663,
+      "step": 2745
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8754693509154092e-05,
+      "loss": 0.8229,
+      "step": 2746
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8753634400327553e-05,
+      "loss": 0.8792,
+      "step": 2747
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.875257487124785e-05,
+      "loss": 0.8486,
+      "step": 2748
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.875151492196585e-05,
+      "loss": 0.8128,
+      "step": 2749
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.875045455253243e-05,
+      "loss": 0.8239,
+      "step": 2750
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.87493937629985e-05,
+      "loss": 0.8554,
+      "step": 2751
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8748332553415e-05,
+      "loss": 0.4644,
+      "step": 2752
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8747270923832863e-05,
+      "loss": 0.8225,
+      "step": 2753
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8746208874303063e-05,
+      "loss": 0.8637,
+      "step": 2754
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8745146404876592e-05,
+      "loss": 0.8389,
+      "step": 2755
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8744083515604448e-05,
+      "loss": 0.8595,
+      "step": 2756
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8743020206537666e-05,
+      "loss": 0.8845,
+      "step": 2757
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.874195647772729e-05,
+      "loss": 0.469,
+      "step": 2758
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8740892329224397e-05,
+      "loss": 0.4697,
+      "step": 2759
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8739827761080066e-05,
+      "loss": 0.8501,
+      "step": 2760
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8738762773345415e-05,
+      "loss": 0.8312,
+      "step": 2761
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8737697366071567e-05,
+      "loss": 0.8638,
+      "step": 2762
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.873663153930967e-05,
+      "loss": 0.8803,
+      "step": 2763
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8735565293110894e-05,
+      "loss": 0.872,
+      "step": 2764
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8734498627526432e-05,
+      "loss": 0.8483,
+      "step": 2765
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.873343154260749e-05,
+      "loss": 0.8149,
+      "step": 2766
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8732364038405303e-05,
+      "loss": 0.8439,
+      "step": 2767
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8731296114971113e-05,
+      "loss": 0.8704,
+      "step": 2768
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8730227772356198e-05,
+      "loss": 0.4411,
+      "step": 2769
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.872915901061184e-05,
+      "loss": 0.4907,
+      "step": 2770
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8728089829789353e-05,
+      "loss": 0.8992,
+      "step": 2771
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.872702022994007e-05,
+      "loss": 0.8239,
+      "step": 2772
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8725950211115335e-05,
+      "loss": 0.8231,
+      "step": 2773
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8724879773366526e-05,
+      "loss": 0.8565,
+      "step": 2774
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.872380891674503e-05,
+      "loss": 0.8532,
+      "step": 2775
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8722737641302255e-05,
+      "loss": 0.843,
+      "step": 2776
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8721665947089634e-05,
+      "loss": 0.4656,
+      "step": 2777
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8720593834158625e-05,
+      "loss": 0.8664,
+      "step": 2778
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8719521302560687e-05,
+      "loss": 0.4712,
+      "step": 2779
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8718448352347317e-05,
+      "loss": 0.8648,
+      "step": 2780
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8717374983570024e-05,
+      "loss": 0.4392,
+      "step": 2781
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8716301196280348e-05,
+      "loss": 0.8373,
+      "step": 2782
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.871522699052983e-05,
+      "loss": 0.8037,
+      "step": 2783
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.871415236637004e-05,
+      "loss": 0.8442,
+      "step": 2784
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.871307732385258e-05,
+      "loss": 0.8572,
+      "step": 2785
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8712001863029062e-05,
+      "loss": 0.4606,
+      "step": 2786
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8710925983951106e-05,
+      "loss": 0.8101,
+      "step": 2787
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8709849686670368e-05,
+      "loss": 0.842,
+      "step": 2788
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8708772971238528e-05,
+      "loss": 0.8283,
+      "step": 2789
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.870769583770727e-05,
+      "loss": 0.8353,
+      "step": 2790
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.870661828612831e-05,
+      "loss": 0.8507,
+      "step": 2791
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8705540316553375e-05,
+      "loss": 0.8258,
+      "step": 2792
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.870446192903422e-05,
+      "loss": 0.856,
+      "step": 2793
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8703383123622624e-05,
+      "loss": 0.8619,
+      "step": 2794
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.870230390037037e-05,
+      "loss": 0.8443,
+      "step": 2795
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8701224259329273e-05,
+      "loss": 0.8346,
+      "step": 2796
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.870014420055117e-05,
+      "loss": 0.8482,
+      "step": 2797
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8699063724087905e-05,
+      "loss": 0.9113,
+      "step": 2798
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8697982829991356e-05,
+      "loss": 0.8343,
+      "step": 2799
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8696901518313416e-05,
+      "loss": 0.4818,
+      "step": 2800
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8695819789106e-05,
+      "loss": 0.8346,
+      "step": 2801
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8694737642421036e-05,
+      "loss": 0.4917,
+      "step": 2802
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8693655078310477e-05,
+      "loss": 0.8251,
+      "step": 2803
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.86925720968263e-05,
+      "loss": 0.8428,
+      "step": 2804
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8691488698020488e-05,
+      "loss": 0.8221,
+      "step": 2805
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8690404881945064e-05,
+      "loss": 0.8313,
+      "step": 2806
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.868932064865206e-05,
+      "loss": 0.8621,
+      "step": 2807
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8688235998193526e-05,
+      "loss": 0.4867,
+      "step": 2808
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8687150930621536e-05,
+      "loss": 0.901,
+      "step": 2809
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.868606544598818e-05,
+      "loss": 0.8266,
+      "step": 2810
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.868497954434558e-05,
+      "loss": 0.8267,
+      "step": 2811
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8683893225745854e-05,
+      "loss": 0.9044,
+      "step": 2812
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.868280649024117e-05,
+      "loss": 0.4283,
+      "step": 2813
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8681719337883692e-05,
+      "loss": 0.8395,
+      "step": 2814
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8680631768725615e-05,
+      "loss": 0.8194,
+      "step": 2815
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.867954378281916e-05,
+      "loss": 0.4578,
+      "step": 2816
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8678455380216543e-05,
+      "loss": 0.8856,
+      "step": 2817
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8677366560970035e-05,
+      "loss": 0.8639,
+      "step": 2818
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.86762773251319e-05,
+      "loss": 0.8504,
+      "step": 2819
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.867518767275443e-05,
+      "loss": 0.4897,
+      "step": 2820
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.867409760388994e-05,
+      "loss": 0.8333,
+      "step": 2821
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.867300711859077e-05,
+      "loss": 0.8282,
+      "step": 2822
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.867191621690926e-05,
+      "loss": 0.8419,
+      "step": 2823
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.867082489889779e-05,
+      "loss": 0.4578,
+      "step": 2824
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8669733164608756e-05,
+      "loss": 0.8755,
+      "step": 2825
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8668641014094572e-05,
+      "loss": 0.8467,
+      "step": 2826
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.866754844740766e-05,
+      "loss": 0.8723,
+      "step": 2827
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.866645546460048e-05,
+      "loss": 0.8727,
+      "step": 2828
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.866536206572551e-05,
+      "loss": 0.8537,
+      "step": 2829
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8664268250835237e-05,
+      "loss": 0.4482,
+      "step": 2830
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8663174019982175e-05,
+      "loss": 0.8284,
+      "step": 2831
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8662079373218852e-05,
+      "loss": 0.8576,
+      "step": 2832
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8660984310597833e-05,
+      "loss": 0.8338,
+      "step": 2833
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.865988883217168e-05,
+      "loss": 0.8248,
+      "step": 2834
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.865879293799299e-05,
+      "loss": 0.8755,
+      "step": 2835
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8657696628114377e-05,
+      "loss": 0.8277,
+      "step": 2836
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.865659990258847e-05,
+      "loss": 0.8191,
+      "step": 2837
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8655502761467922e-05,
+      "loss": 0.811,
+      "step": 2838
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.865440520480541e-05,
+      "loss": 0.8036,
+      "step": 2839
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8653307232653627e-05,
+      "loss": 0.8801,
+      "step": 2840
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8652208845065275e-05,
+      "loss": 0.8438,
+      "step": 2841
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.86511100420931e-05,
+      "loss": 0.8768,
+      "step": 2842
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8650010823789845e-05,
+      "loss": 0.8446,
+      "step": 2843
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.864891119020829e-05,
+      "loss": 0.8309,
+      "step": 2844
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.864781114140122e-05,
+      "loss": 0.4703,
+      "step": 2845
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.864671067742145e-05,
+      "loss": 0.8052,
+      "step": 2846
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.864560979832181e-05,
+      "loss": 0.8332,
+      "step": 2847
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.864450850415516e-05,
+      "loss": 0.8376,
+      "step": 2848
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8643406794974365e-05,
+      "loss": 0.8426,
+      "step": 2849
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8642304670832317e-05,
+      "loss": 0.8641,
+      "step": 2850
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8641202131781934e-05,
+      "loss": 0.8384,
+      "step": 2851
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8640099177876135e-05,
+      "loss": 0.7873,
+      "step": 2852
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8638995809167886e-05,
+      "loss": 0.8238,
+      "step": 2853
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8637892025710148e-05,
+      "loss": 0.8413,
+      "step": 2854
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.863678782755592e-05,
+      "loss": 0.5003,
+      "step": 2855
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8635683214758213e-05,
+      "loss": 0.4562,
+      "step": 2856
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8634578187370053e-05,
+      "loss": 0.4727,
+      "step": 2857
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8633472745444495e-05,
+      "loss": 0.8411,
+      "step": 2858
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.863236688903461e-05,
+      "loss": 0.8333,
+      "step": 2859
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8631260618193488e-05,
+      "loss": 0.862,
+      "step": 2860
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8630153932974236e-05,
+      "loss": 0.8502,
+      "step": 2861
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8629046833429994e-05,
+      "loss": 0.8453,
+      "step": 2862
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8627939319613905e-05,
+      "loss": 0.8261,
+      "step": 2863
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8626831391579145e-05,
+      "loss": 0.7971,
+      "step": 2864
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.86257230493789e-05,
+      "loss": 0.8001,
+      "step": 2865
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8624614293066382e-05,
+      "loss": 0.8872,
+      "step": 2866
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8623505122694822e-05,
+      "loss": 0.4544,
+      "step": 2867
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.862239553831747e-05,
+      "loss": 0.8292,
+      "step": 2868
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8621285539987596e-05,
+      "loss": 0.856,
+      "step": 2869
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.862017512775849e-05,
+      "loss": 0.4973,
+      "step": 2870
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8619064301683463e-05,
+      "loss": 0.8731,
+      "step": 2871
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.861795306181584e-05,
+      "loss": 0.8315,
+      "step": 2872
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8616841408208975e-05,
+      "loss": 0.4509,
+      "step": 2873
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8615729340916234e-05,
+      "loss": 0.8618,
+      "step": 2874
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8614616859991013e-05,
+      "loss": 0.8651,
+      "step": 2875
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8613503965486712e-05,
+      "loss": 0.832,
+      "step": 2876
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8612390657456766e-05,
+      "loss": 0.8818,
+      "step": 2877
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8611276935954622e-05,
+      "loss": 0.4896,
+      "step": 2878
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8610162801033752e-05,
+      "loss": 0.8712,
+      "step": 2879
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8609048252747634e-05,
+      "loss": 0.8208,
+      "step": 2880
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8607933291149787e-05,
+      "loss": 0.8497,
+      "step": 2881
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8606817916293736e-05,
+      "loss": 0.8648,
+      "step": 2882
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8605702128233032e-05,
+      "loss": 0.4582,
+      "step": 2883
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8604585927021233e-05,
+      "loss": 0.4549,
+      "step": 2884
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.860346931271194e-05,
+      "loss": 0.4171,
+      "step": 2885
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8602352285358748e-05,
+      "loss": 0.47,
+      "step": 2886
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8601234845015295e-05,
+      "loss": 0.8531,
+      "step": 2887
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.860011699173522e-05,
+      "loss": 0.8331,
+      "step": 2888
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8598998725572194e-05,
+      "loss": 0.4833,
+      "step": 2889
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8597880046579903e-05,
+      "loss": 0.8255,
+      "step": 2890
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8596760954812057e-05,
+      "loss": 0.7985,
+      "step": 2891
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8595641450322374e-05,
+      "loss": 0.4489,
+      "step": 2892
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.859452153316461e-05,
+      "loss": 0.4921,
+      "step": 2893
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8593401203392526e-05,
+      "loss": 0.8422,
+      "step": 2894
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8592280461059905e-05,
+      "loss": 0.7923,
+      "step": 2895
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.859115930622056e-05,
+      "loss": 0.8467,
+      "step": 2896
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8590037738928312e-05,
+      "loss": 0.4746,
+      "step": 2897
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.858891575923701e-05,
+      "loss": 0.8304,
+      "step": 2898
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.858779336720051e-05,
+      "loss": 0.8443,
+      "step": 2899
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8586670562872707e-05,
+      "loss": 0.8641,
+      "step": 2900
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.85855473463075e-05,
+      "loss": 0.8278,
+      "step": 2901
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.858442371755882e-05,
+      "loss": 0.4482,
+      "step": 2902
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.85832996766806e-05,
+      "loss": 0.8372,
+      "step": 2903
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8582175223726814e-05,
+      "loss": 0.4855,
+      "step": 2904
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8581050358751444e-05,
+      "loss": 0.8458,
+      "step": 2905
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8579925081808488e-05,
+      "loss": 0.8155,
+      "step": 2906
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8578799392951977e-05,
+      "loss": 0.8907,
+      "step": 2907
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.857767329223595e-05,
+      "loss": 0.9212,
+      "step": 2908
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8576546779714475e-05,
+      "loss": 0.8121,
+      "step": 2909
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8575419855441627e-05,
+      "loss": 0.8491,
+      "step": 2910
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.857429251947151e-05,
+      "loss": 0.8849,
+      "step": 2911
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8573164771858253e-05,
+      "loss": 0.7706,
+      "step": 2912
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.857203661265599e-05,
+      "loss": 0.7853,
+      "step": 2913
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8570908041918882e-05,
+      "loss": 0.8672,
+      "step": 2914
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.856977905970112e-05,
+      "loss": 0.793,
+      "step": 2915
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.85686496660569e-05,
+      "loss": 0.4805,
+      "step": 2916
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8567519861040443e-05,
+      "loss": 0.7987,
+      "step": 2917
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8566389644705987e-05,
+      "loss": 0.8887,
+      "step": 2918
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8565259017107797e-05,
+      "loss": 0.489,
+      "step": 2919
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.856412797830015e-05,
+      "loss": 0.8433,
+      "step": 2920
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.856299652833735e-05,
+      "loss": 0.8419,
+      "step": 2921
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8561864667273715e-05,
+      "loss": 0.4769,
+      "step": 2922
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8560732395163585e-05,
+      "loss": 0.8839,
+      "step": 2923
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.855959971206131e-05,
+      "loss": 0.8428,
+      "step": 2924
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8558466618021285e-05,
+      "loss": 0.8762,
+      "step": 2925
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.85573331130979e-05,
+      "loss": 0.8235,
+      "step": 2926
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8556199197345576e-05,
+      "loss": 0.8893,
+      "step": 2927
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8555064870818743e-05,
+      "loss": 0.8234,
+      "step": 2928
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.855393013357187e-05,
+      "loss": 0.862,
+      "step": 2929
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8552794985659427e-05,
+      "loss": 0.4702,
+      "step": 2930
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.855165942713592e-05,
+      "loss": 0.4826,
+      "step": 2931
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8550523458055852e-05,
+      "loss": 0.8555,
+      "step": 2932
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8549387078473774e-05,
+      "loss": 0.8653,
+      "step": 2933
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8548250288444232e-05,
+      "loss": 0.4654,
+      "step": 2934
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.854711308802181e-05,
+      "loss": 0.4278,
+      "step": 2935
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.85459754772611e-05,
+      "loss": 0.8076,
+      "step": 2936
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8544837456216716e-05,
+      "loss": 0.8429,
+      "step": 2937
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8543699024943298e-05,
+      "loss": 0.8163,
+      "step": 2938
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8542560183495497e-05,
+      "loss": 0.851,
+      "step": 2939
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.854142093192799e-05,
+      "loss": 0.8826,
+      "step": 2940
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8540281270295465e-05,
+      "loss": 0.4524,
+      "step": 2941
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8539141198652647e-05,
+      "loss": 0.8936,
+      "step": 2942
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8538000717054262e-05,
+      "loss": 0.8269,
+      "step": 2943
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8536859825555065e-05,
+      "loss": 0.8775,
+      "step": 2944
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8535718524209828e-05,
+      "loss": 0.8451,
+      "step": 2945
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8534576813073345e-05,
+      "loss": 0.8944,
+      "step": 2946
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8533434692200428e-05,
+      "loss": 0.8728,
+      "step": 2947
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.853229216164591e-05,
+      "loss": 0.8208,
+      "step": 2948
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8531149221464645e-05,
+      "loss": 0.8974,
+      "step": 2949
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8530005871711497e-05,
+      "loss": 0.8056,
+      "step": 2950
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8528862112441365e-05,
+      "loss": 0.8616,
+      "step": 2951
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8527717943709153e-05,
+      "loss": 0.7798,
+      "step": 2952
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8526573365569797e-05,
+      "loss": 0.457,
+      "step": 2953
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8525428378078245e-05,
+      "loss": 0.8337,
+      "step": 2954
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8524282981289464e-05,
+      "loss": 0.8079,
+      "step": 2955
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.852313717525845e-05,
+      "loss": 0.4707,
+      "step": 2956
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8521990960040206e-05,
+      "loss": 0.4754,
+      "step": 2957
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.852084433568976e-05,
+      "loss": 0.4522,
+      "step": 2958
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8519697302262166e-05,
+      "loss": 0.8549,
+      "step": 2959
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8518549859812485e-05,
+      "loss": 0.8933,
+      "step": 2960
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8517402008395813e-05,
+      "loss": 0.8353,
+      "step": 2961
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8516253748067246e-05,
+      "loss": 0.8175,
+      "step": 2962
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8515105078881924e-05,
+      "loss": 0.8442,
+      "step": 2963
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8513956000894984e-05,
+      "loss": 0.8387,
+      "step": 2964
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8512806514161596e-05,
+      "loss": 0.8601,
+      "step": 2965
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.851165661873694e-05,
+      "loss": 0.8306,
+      "step": 2966
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.851050631467623e-05,
+      "loss": 0.8304,
+      "step": 2967
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8509355602034688e-05,
+      "loss": 0.8511,
+      "step": 2968
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.850820448086755e-05,
+      "loss": 0.4551,
+      "step": 2969
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8507052951230092e-05,
+      "loss": 0.8769,
+      "step": 2970
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8505901013177598e-05,
+      "loss": 0.84,
+      "step": 2971
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.850474866676536e-05,
+      "loss": 0.7928,
+      "step": 2972
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.850359591204871e-05,
+      "loss": 0.8217,
+      "step": 2973
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8502442749082988e-05,
+      "loss": 0.8299,
+      "step": 2974
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8501289177923555e-05,
+      "loss": 0.787,
+      "step": 2975
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8500135198625796e-05,
+      "loss": 0.7863,
+      "step": 2976
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.849898081124511e-05,
+      "loss": 0.8523,
+      "step": 2977
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8497826015836916e-05,
+      "loss": 0.4383,
+      "step": 2978
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8496670812456662e-05,
+      "loss": 0.8171,
+      "step": 2979
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.84955152011598e-05,
+      "loss": 0.4746,
+      "step": 2980
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.849435918200181e-05,
+      "loss": 0.8731,
+      "step": 2981
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.84932027550382e-05,
+      "loss": 0.4486,
+      "step": 2982
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.849204592032448e-05,
+      "loss": 0.4695,
+      "step": 2983
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8490888677916192e-05,
+      "loss": 0.8341,
+      "step": 2984
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8489731027868893e-05,
+      "loss": 0.8604,
+      "step": 2985
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8488572970238163e-05,
+      "loss": 0.8567,
+      "step": 2986
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8487414505079597e-05,
+      "loss": 0.7844,
+      "step": 2987
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8486255632448812e-05,
+      "loss": 0.7968,
+      "step": 2988
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8485096352401447e-05,
+      "loss": 0.488,
+      "step": 2989
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8483936664993152e-05,
+      "loss": 0.8062,
+      "step": 2990
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8482776570279605e-05,
+      "loss": 0.8545,
+      "step": 2991
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8481616068316504e-05,
+      "loss": 0.8347,
+      "step": 2992
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.848045515915956e-05,
+      "loss": 0.4345,
+      "step": 2993
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.847929384286451e-05,
+      "loss": 0.475,
+      "step": 2994
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8478132119487105e-05,
+      "loss": 0.4498,
+      "step": 2995
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8476969989083118e-05,
+      "loss": 0.4726,
+      "step": 2996
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.847580745170835e-05,
+      "loss": 0.8876,
+      "step": 2997
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8474644507418595e-05,
+      "loss": 0.8682,
+      "step": 2998
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.84734811562697e-05,
+      "loss": 0.8318,
+      "step": 2999
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8472317398317517e-05,
+      "loss": 0.89,
+      "step": 3000
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.847115323361791e-05,
+      "loss": 0.4585,
+      "step": 3001
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.846998866222677e-05,
+      "loss": 0.7845,
+      "step": 3002
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.846882368420001e-05,
+      "loss": 0.8595,
+      "step": 3003
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8467658299593552e-05,
+      "loss": 0.7907,
+      "step": 3004
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8466492508463356e-05,
+      "loss": 0.4481,
+      "step": 3005
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8465326310865386e-05,
+      "loss": 0.4841,
+      "step": 3006
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8464159706855628e-05,
+      "loss": 0.4625,
+      "step": 3007
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8462992696490087e-05,
+      "loss": 0.8405,
+      "step": 3008
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.84618252798248e-05,
+      "loss": 0.8281,
+      "step": 3009
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.84606574569158e-05,
+      "loss": 0.8347,
+      "step": 3010
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8459489227819166e-05,
+      "loss": 0.5064,
+      "step": 3011
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8458320592590976e-05,
+      "loss": 0.835,
+      "step": 3012
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8457151551287334e-05,
+      "loss": 0.8526,
+      "step": 3013
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.845598210396437e-05,
+      "loss": 0.8602,
+      "step": 3014
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8454812250678226e-05,
+      "loss": 0.8262,
+      "step": 3015
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8453641991485063e-05,
+      "loss": 0.8253,
+      "step": 3016
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8452471326441067e-05,
+      "loss": 0.821,
+      "step": 3017
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8451300255602435e-05,
+      "loss": 0.8381,
+      "step": 3018
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.84501287790254e-05,
+      "loss": 0.8677,
+      "step": 3019
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8448956896766187e-05,
+      "loss": 0.7713,
+      "step": 3020
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.844778460888107e-05,
+      "loss": 0.8451,
+      "step": 3021
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.844661191542633e-05,
+      "loss": 0.8034,
+      "step": 3022
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8445438816458258e-05,
+      "loss": 0.4963,
+      "step": 3023
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.844426531203318e-05,
+      "loss": 0.8626,
+      "step": 3024
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.844309140220743e-05,
+      "loss": 0.4377,
+      "step": 3025
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8441917087037372e-05,
+      "loss": 0.853,
+      "step": 3026
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.844074236657938e-05,
+      "loss": 0.4418,
+      "step": 3027
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.843956724088985e-05,
+      "loss": 0.8613,
+      "step": 3028
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.84383917100252e-05,
+      "loss": 0.8438,
+      "step": 3029
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8437215774041863e-05,
+      "loss": 0.8532,
+      "step": 3030
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.84360394329963e-05,
+      "loss": 0.827,
+      "step": 3031
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.843486268694499e-05,
+      "loss": 0.4196,
+      "step": 3032
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8433685535944412e-05,
+      "loss": 0.8914,
+      "step": 3033
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8432507980051092e-05,
+      "loss": 0.4777,
+      "step": 3034
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8431330019321563e-05,
+      "loss": 0.8332,
+      "step": 3035
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.843015165381237e-05,
+      "loss": 0.8486,
+      "step": 3036
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8428972883580094e-05,
+      "loss": 0.8394,
+      "step": 3037
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.842779370868132e-05,
+      "loss": 0.9224,
+      "step": 3038
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8426614129172666e-05,
+      "loss": 0.839,
+      "step": 3039
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8425434145110755e-05,
+      "loss": 0.8741,
+      "step": 3040
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8424253756552238e-05,
+      "loss": 0.5129,
+      "step": 3041
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.842307296355379e-05,
+      "loss": 0.4315,
+      "step": 3042
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.842189176617209e-05,
+      "loss": 0.8484,
+      "step": 3043
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8420710164463856e-05,
+      "loss": 0.9365,
+      "step": 3044
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8419528158485813e-05,
+      "loss": 0.8382,
+      "step": 3045
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8418345748294702e-05,
+      "loss": 0.8549,
+      "step": 3046
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8417162933947296e-05,
+      "loss": 0.7992,
+      "step": 3047
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.841597971550038e-05,
+      "loss": 0.4716,
+      "step": 3048
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8414796093010758e-05,
+      "loss": 0.8118,
+      "step": 3049
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8413612066535247e-05,
+      "loss": 0.7693,
+      "step": 3050
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8412427636130706e-05,
+      "loss": 0.837,
+      "step": 3051
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8411242801853984e-05,
+      "loss": 0.9315,
+      "step": 3052
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8410057563761973e-05,
+      "loss": 0.8501,
+      "step": 3053
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8408871921911574e-05,
+      "loss": 0.8374,
+      "step": 3054
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8407685876359704e-05,
+      "loss": 0.8761,
+      "step": 3055
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8406499427163304e-05,
+      "loss": 0.477,
+      "step": 3056
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.840531257437934e-05,
+      "loss": 0.8036,
+      "step": 3057
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8404125318064786e-05,
+      "loss": 0.8666,
+      "step": 3058
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8402937658276647e-05,
+      "loss": 0.8683,
+      "step": 3059
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8401749595071933e-05,
+      "loss": 0.8145,
+      "step": 3060
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.840056112850769e-05,
+      "loss": 0.8731,
+      "step": 3061
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8399372258640967e-05,
+      "loss": 0.8135,
+      "step": 3062
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8398182985528848e-05,
+      "loss": 0.873,
+      "step": 3063
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8396993309228427e-05,
+      "loss": 0.4712,
+      "step": 3064
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8395803229796812e-05,
+      "loss": 0.8011,
+      "step": 3065
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8394612747291146e-05,
+      "loss": 0.8724,
+      "step": 3066
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8393421861768583e-05,
+      "loss": 0.8007,
+      "step": 3067
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.839223057328629e-05,
+      "loss": 0.8524,
+      "step": 3068
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8391038881901466e-05,
+      "loss": 0.8541,
+      "step": 3069
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.838984678767132e-05,
+      "loss": 0.8568,
+      "step": 3070
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8388654290653077e-05,
+      "loss": 0.8106,
+      "step": 3071
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8387461390903997e-05,
+      "loss": 0.4696,
+      "step": 3072
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.838626808848135e-05,
+      "loss": 0.8438,
+      "step": 3073
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.838507438344242e-05,
+      "loss": 0.8337,
+      "step": 3074
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8383880275844513e-05,
+      "loss": 0.824,
+      "step": 3075
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8382685765744966e-05,
+      "loss": 0.7956,
+      "step": 3076
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8381490853201122e-05,
+      "loss": 0.5129,
+      "step": 3077
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8380295538270343e-05,
+      "loss": 0.8821,
+      "step": 3078
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8379099821010017e-05,
+      "loss": 0.4737,
+      "step": 3079
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8377903701477556e-05,
+      "loss": 0.8217,
+      "step": 3080
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.837670717973038e-05,
+      "loss": 0.8811,
+      "step": 3081
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.837551025582593e-05,
+      "loss": 0.8523,
+      "step": 3082
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.837431292982167e-05,
+      "loss": 0.871,
+      "step": 3083
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.837311520177509e-05,
+      "loss": 0.8302,
+      "step": 3084
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.837191707174368e-05,
+      "loss": 0.7974,
+      "step": 3085
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.837071853978497e-05,
+      "loss": 0.4856,
+      "step": 3086
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8369519605956494e-05,
+      "loss": 0.8113,
+      "step": 3087
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8368320270315813e-05,
+      "loss": 0.8173,
+      "step": 3088
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.836712053292051e-05,
+      "loss": 0.8243,
+      "step": 3089
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.836592039382818e-05,
+      "loss": 0.885,
+      "step": 3090
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8364719853096442e-05,
+      "loss": 0.8985,
+      "step": 3091
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.836351891078293e-05,
+      "loss": 0.8565,
+      "step": 3092
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8362317566945303e-05,
+      "loss": 0.8475,
+      "step": 3093
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.836111582164124e-05,
+      "loss": 0.8235,
+      "step": 3094
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8359913674928425e-05,
+      "loss": 0.8501,
+      "step": 3095
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8358711126864578e-05,
+      "loss": 0.4837,
+      "step": 3096
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8357508177507433e-05,
+      "loss": 0.825,
+      "step": 3097
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8356304826914742e-05,
+      "loss": 0.8269,
+      "step": 3098
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8355101075144276e-05,
+      "loss": 0.8221,
+      "step": 3099
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8353896922253823e-05,
+      "loss": 0.8119,
+      "step": 3100
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8352692368301194e-05,
+      "loss": 0.8344,
+      "step": 3101
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8351487413344228e-05,
+      "loss": 0.8956,
+      "step": 3102
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8350282057440757e-05,
+      "loss": 0.4687,
+      "step": 3103
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8349076300648664e-05,
+      "loss": 0.8056,
+      "step": 3104
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.834787014302583e-05,
+      "loss": 0.8377,
+      "step": 3105
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8346663584630157e-05,
+      "loss": 0.8435,
+      "step": 3106
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8345456625519576e-05,
+      "loss": 0.8145,
+      "step": 3107
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.834424926575203e-05,
+      "loss": 0.7857,
+      "step": 3108
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.834304150538549e-05,
+      "loss": 0.8068,
+      "step": 3109
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.834183334447793e-05,
+      "loss": 0.8315,
+      "step": 3110
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8340624783087358e-05,
+      "loss": 0.8378,
+      "step": 3111
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8339415821271792e-05,
+      "loss": 0.8171,
+      "step": 3112
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8338206459089272e-05,
+      "loss": 0.8434,
+      "step": 3113
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8336996696597866e-05,
+      "loss": 0.8575,
+      "step": 3114
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8335786533855647e-05,
+      "loss": 0.8538,
+      "step": 3115
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8334575970920715e-05,
+      "loss": 0.4559,
+      "step": 3116
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8333365007851186e-05,
+      "loss": 0.4929,
+      "step": 3117
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.83321536447052e-05,
+      "loss": 0.8256,
+      "step": 3118
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8330941881540917e-05,
+      "loss": 0.8594,
+      "step": 3119
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8329729718416508e-05,
+      "loss": 0.4793,
+      "step": 3120
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8328517155390165e-05,
+      "loss": 0.8892,
+      "step": 3121
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8327304192520103e-05,
+      "loss": 0.8651,
+      "step": 3122
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.832609082986456e-05,
+      "loss": 0.8751,
+      "step": 3123
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8324877067481782e-05,
+      "loss": 0.8435,
+      "step": 3124
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.832366290543005e-05,
+      "loss": 0.4898,
+      "step": 3125
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8322448343767645e-05,
+      "loss": 0.8887,
+      "step": 3126
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.832123338255288e-05,
+      "loss": 0.8807,
+      "step": 3127
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8320018021844085e-05,
+      "loss": 0.8258,
+      "step": 3128
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8318802261699608e-05,
+      "loss": 0.4566,
+      "step": 3129
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8317586102177812e-05,
+      "loss": 0.813,
+      "step": 3130
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8316369543337092e-05,
+      "loss": 0.4463,
+      "step": 3131
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.831515258523585e-05,
+      "loss": 0.8313,
+      "step": 3132
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.831393522793251e-05,
+      "loss": 0.8489,
+      "step": 3133
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8312717471485516e-05,
+      "loss": 0.8377,
+      "step": 3134
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8311499315953328e-05,
+      "loss": 0.8572,
+      "step": 3135
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8310280761394435e-05,
+      "loss": 0.8221,
+      "step": 3136
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.830906180786734e-05,
+      "loss": 0.881,
+      "step": 3137
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8307842455430557e-05,
+      "loss": 0.8307,
+      "step": 3138
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8306622704142625e-05,
+      "loss": 0.8768,
+      "step": 3139
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.830540255406211e-05,
+      "loss": 0.819,
+      "step": 3140
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8304182005247583e-05,
+      "loss": 0.8403,
+      "step": 3141
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.830296105775765e-05,
+      "loss": 0.832,
+      "step": 3142
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.830173971165092e-05,
+      "loss": 0.47,
+      "step": 3143
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8300517966986028e-05,
+      "loss": 0.847,
+      "step": 3144
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8299295823821635e-05,
+      "loss": 0.8619,
+      "step": 3145
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8298073282216412e-05,
+      "loss": 0.8523,
+      "step": 3146
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.829685034222905e-05,
+      "loss": 0.8117,
+      "step": 3147
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.829562700391826e-05,
+      "loss": 0.5192,
+      "step": 3148
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.829440326734278e-05,
+      "loss": 0.8466,
+      "step": 3149
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.829317913256136e-05,
+      "loss": 0.8165,
+      "step": 3150
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.829195459963276e-05,
+      "loss": 0.8535,
+      "step": 3151
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.829072966861577e-05,
+      "loss": 0.4413,
+      "step": 3152
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.828950433956921e-05,
+      "loss": 0.4533,
+      "step": 3153
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8288278612551895e-05,
+      "loss": 0.8714,
+      "step": 3154
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8287052487622677e-05,
+      "loss": 0.4758,
+      "step": 3155
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.828582596484042e-05,
+      "loss": 0.8301,
+      "step": 3156
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8284599044264002e-05,
+      "loss": 0.8693,
+      "step": 3157
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8283371725952334e-05,
+      "loss": 0.8143,
+      "step": 3158
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8282144009964336e-05,
+      "loss": 0.8474,
+      "step": 3159
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.828091589635895e-05,
+      "loss": 0.8216,
+      "step": 3160
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.827968738519513e-05,
+      "loss": 0.8229,
+      "step": 3161
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8278458476531866e-05,
+      "loss": 0.8464,
+      "step": 3162
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.827722917042815e-05,
+      "loss": 0.8613,
+      "step": 3163
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8275999466943002e-05,
+      "loss": 0.855,
+      "step": 3164
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8274769366135455e-05,
+      "loss": 0.8018,
+      "step": 3165
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8273538868064573e-05,
+      "loss": 0.8706,
+      "step": 3166
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8272307972789422e-05,
+      "loss": 0.8247,
+      "step": 3167
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.82710766803691e-05,
+      "loss": 0.7857,
+      "step": 3168
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8269844990862722e-05,
+      "loss": 0.4607,
+      "step": 3169
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.826861290432942e-05,
+      "loss": 0.8406,
+      "step": 3170
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8267380420828343e-05,
+      "loss": 0.5016,
+      "step": 3171
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.826614754041866e-05,
+      "loss": 0.8717,
+      "step": 3172
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8264914263159564e-05,
+      "loss": 0.8321,
+      "step": 3173
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.826368058911026e-05,
+      "loss": 0.8459,
+      "step": 3174
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8262446518329976e-05,
+      "loss": 0.827,
+      "step": 3175
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8261212050877962e-05,
+      "loss": 0.4821,
+      "step": 3176
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.825997718681348e-05,
+      "loss": 0.8601,
+      "step": 3177
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8258741926195817e-05,
+      "loss": 0.4728,
+      "step": 3178
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8257506269084274e-05,
+      "loss": 0.8724,
+      "step": 3179
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8256270215538177e-05,
+      "loss": 0.4733,
+      "step": 3180
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8255033765616863e-05,
+      "loss": 0.4361,
+      "step": 3181
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8253796919379695e-05,
+      "loss": 0.8232,
+      "step": 3182
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8252559676886058e-05,
+      "loss": 0.8717,
+      "step": 3183
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.825132203819534e-05,
+      "loss": 0.8241,
+      "step": 3184
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.825008400336697e-05,
+      "loss": 0.8921,
+      "step": 3185
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8248845572460374e-05,
+      "loss": 0.7948,
+      "step": 3186
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8247606745535016e-05,
+      "loss": 0.8026,
+      "step": 3187
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.824636752265037e-05,
+      "loss": 0.7982,
+      "step": 3188
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8245127903865924e-05,
+      "loss": 0.8258,
+      "step": 3189
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8243887889241196e-05,
+      "loss": 0.8024,
+      "step": 3190
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8242647478835717e-05,
+      "loss": 0.8631,
+      "step": 3191
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8241406672709038e-05,
+      "loss": 0.8306,
+      "step": 3192
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.824016547092073e-05,
+      "loss": 0.796,
+      "step": 3193
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.823892387353038e-05,
+      "loss": 0.9198,
+      "step": 3194
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8237681880597595e-05,
+      "loss": 0.4828,
+      "step": 3195
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8236439492182e-05,
+      "loss": 0.8028,
+      "step": 3196
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8235196708343246e-05,
+      "loss": 0.818,
+      "step": 3197
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8233953529140994e-05,
+      "loss": 0.8129,
+      "step": 3198
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8232709954634934e-05,
+      "loss": 0.4585,
+      "step": 3199
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8231465984884758e-05,
+      "loss": 0.7831,
+      "step": 3200
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.82302216199502e-05,
+      "loss": 0.8263,
+      "step": 3201
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.822897685989099e-05,
+      "loss": 0.8178,
+      "step": 3202
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8227731704766892e-05,
+      "loss": 0.4496,
+      "step": 3203
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8226486154637686e-05,
+      "loss": 0.7997,
+      "step": 3204
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.822524020956317e-05,
+      "loss": 0.8267,
+      "step": 3205
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8223993869603158e-05,
+      "loss": 0.4552,
+      "step": 3206
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8222747134817487e-05,
+      "loss": 0.8525,
+      "step": 3207
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.822150000526601e-05,
+      "loss": 0.4796,
+      "step": 3208
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.82202524810086e-05,
+      "loss": 0.832,
+      "step": 3209
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8219004562105152e-05,
+      "loss": 0.8521,
+      "step": 3210
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8217756248615578e-05,
+      "loss": 0.437,
+      "step": 3211
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8216507540599805e-05,
+      "loss": 0.8628,
+      "step": 3212
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8215258438117783e-05,
+      "loss": 0.8519,
+      "step": 3213
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8214008941229483e-05,
+      "loss": 0.8923,
+      "step": 3214
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8212759049994887e-05,
+      "loss": 0.8518,
+      "step": 3215
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8211508764474003e-05,
+      "loss": 0.8725,
+      "step": 3216
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.821025808472686e-05,
+      "loss": 0.8274,
+      "step": 3217
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.82090070108135e-05,
+      "loss": 0.8593,
+      "step": 3218
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.820775554279398e-05,
+      "loss": 0.8447,
+      "step": 3219
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.820650368072839e-05,
+      "loss": 0.8245,
+      "step": 3220
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8205251424676823e-05,
+      "loss": 0.8574,
+      "step": 3221
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8203998774699405e-05,
+      "loss": 0.8244,
+      "step": 3222
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.820274573085627e-05,
+      "loss": 0.8496,
+      "step": 3223
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.820149229320758e-05,
+      "loss": 0.8293,
+      "step": 3224
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8200238461813506e-05,
+      "loss": 0.8007,
+      "step": 3225
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8198984236734246e-05,
+      "loss": 0.8475,
+      "step": 3226
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8197729618030017e-05,
+      "loss": 0.8657,
+      "step": 3227
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8196474605761047e-05,
+      "loss": 0.8375,
+      "step": 3228
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8195219199987586e-05,
+      "loss": 0.47,
+      "step": 3229
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8193963400769913e-05,
+      "loss": 0.8419,
+      "step": 3230
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8192707208168313e-05,
+      "loss": 0.8152,
+      "step": 3231
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8191450622243092e-05,
+      "loss": 0.8507,
+      "step": 3232
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8190193643054578e-05,
+      "loss": 0.7656,
+      "step": 3233
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8188936270663124e-05,
+      "loss": 0.8196,
+      "step": 3234
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.818767850512909e-05,
+      "loss": 0.4856,
+      "step": 3235
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8186420346512858e-05,
+      "loss": 0.4485,
+      "step": 3236
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8185161794874833e-05,
+      "loss": 0.8099,
+      "step": 3237
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8183902850275443e-05,
+      "loss": 0.4642,
+      "step": 3238
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8182643512775117e-05,
+      "loss": 0.8321,
+      "step": 3239
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8181383782434324e-05,
+      "loss": 0.8067,
+      "step": 3240
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8180123659313538e-05,
+      "loss": 0.8368,
+      "step": 3241
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8178863143473255e-05,
+      "loss": 0.8411,
+      "step": 3242
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8177602234973996e-05,
+      "loss": 0.8202,
+      "step": 3243
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8176340933876295e-05,
+      "loss": 0.8441,
+      "step": 3244
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.81750792402407e-05,
+      "loss": 0.4474,
+      "step": 3245
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8173817154127787e-05,
+      "loss": 0.8405,
+      "step": 3246
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8172554675598148e-05,
+      "loss": 0.4216,
+      "step": 3247
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8171291804712395e-05,
+      "loss": 0.4457,
+      "step": 3248
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8170028541531155e-05,
+      "loss": 0.83,
+      "step": 3249
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8168764886115077e-05,
+      "loss": 0.8575,
+      "step": 3250
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.816750083852483e-05,
+      "loss": 0.8663,
+      "step": 3251
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.816623639882109e-05,
+      "loss": 0.8862,
+      "step": 3252
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8164971567064574e-05,
+      "loss": 0.4305,
+      "step": 3253
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8163706343316e-05,
+      "loss": 0.9006,
+      "step": 3254
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8162440727636108e-05,
+      "loss": 0.833,
+      "step": 3255
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.816117472008566e-05,
+      "loss": 0.8592,
+      "step": 3256
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8159908320725438e-05,
+      "loss": 0.8641,
+      "step": 3257
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.815864152961624e-05,
+      "loss": 0.8489,
+      "step": 3258
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8157374346818883e-05,
+      "loss": 0.7935,
+      "step": 3259
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.81561067723942e-05,
+      "loss": 0.8597,
+      "step": 3260
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.815483880640305e-05,
+      "loss": 0.8749,
+      "step": 3261
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8153570448906306e-05,
+      "loss": 0.4841,
+      "step": 3262
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8152301699964858e-05,
+      "loss": 0.7952,
+      "step": 3263
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8151032559639622e-05,
+      "loss": 0.8461,
+      "step": 3264
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8149763027991524e-05,
+      "loss": 0.4575,
+      "step": 3265
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8148493105081513e-05,
+      "loss": 0.8511,
+      "step": 3266
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.814722279097056e-05,
+      "loss": 0.4612,
+      "step": 3267
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.814595208571965e-05,
+      "loss": 0.8383,
+      "step": 3268
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8144680989389788e-05,
+      "loss": 0.8654,
+      "step": 3269
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8143409502041996e-05,
+      "loss": 0.8368,
+      "step": 3270
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.814213762373732e-05,
+      "loss": 0.842,
+      "step": 3271
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.814086535453682e-05,
+      "loss": 0.4577,
+      "step": 3272
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8139592694501575e-05,
+      "loss": 0.7855,
+      "step": 3273
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.813831964369269e-05,
+      "loss": 0.4522,
+      "step": 3274
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8137046202171276e-05,
+      "loss": 0.7991,
+      "step": 3275
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8135772369998476e-05,
+      "loss": 0.4329,
+      "step": 3276
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8134498147235436e-05,
+      "loss": 0.8603,
+      "step": 3277
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.813322353394334e-05,
+      "loss": 0.8008,
+      "step": 3278
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8131948530183373e-05,
+      "loss": 0.8781,
+      "step": 3279
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8130673136016756e-05,
+      "loss": 0.8907,
+      "step": 3280
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.812939735150471e-05,
+      "loss": 0.8203,
+      "step": 3281
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.812812117670849e-05,
+      "loss": 0.8571,
+      "step": 3282
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.812684461168936e-05,
+      "loss": 0.8283,
+      "step": 3283
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8125567656508608e-05,
+      "loss": 0.8291,
+      "step": 3284
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8124290311227543e-05,
+      "loss": 0.8212,
+      "step": 3285
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8123012575907484e-05,
+      "loss": 0.8577,
+      "step": 3286
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8121734450609777e-05,
+      "loss": 0.8617,
+      "step": 3287
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.812045593539578e-05,
+      "loss": 0.8476,
+      "step": 3288
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8119177030326878e-05,
+      "loss": 0.8534,
+      "step": 3289
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8117897735464464e-05,
+      "loss": 0.8039,
+      "step": 3290
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8116618050869964e-05,
+      "loss": 0.4482,
+      "step": 3291
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8115337976604804e-05,
+      "loss": 0.8069,
+      "step": 3292
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8114057512730448e-05,
+      "loss": 0.8153,
+      "step": 3293
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8112776659308364e-05,
+      "loss": 0.8232,
+      "step": 3294
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.811149541640005e-05,
+      "loss": 0.8479,
+      "step": 3295
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.811021378406701e-05,
+      "loss": 0.847,
+      "step": 3296
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8108931762370778e-05,
+      "loss": 0.4808,
+      "step": 3297
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8107649351372905e-05,
+      "loss": 0.4771,
+      "step": 3298
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8106366551134954e-05,
+      "loss": 0.4625,
+      "step": 3299
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8105083361718512e-05,
+      "loss": 0.8566,
+      "step": 3300
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8103799783185184e-05,
+      "loss": 0.842,
+      "step": 3301
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8102515815596596e-05,
+      "loss": 0.8286,
+      "step": 3302
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8101231459014385e-05,
+      "loss": 0.428,
+      "step": 3303
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8099946713500217e-05,
+      "loss": 0.7967,
+      "step": 3304
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8098661579115765e-05,
+      "loss": 0.8368,
+      "step": 3305
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8097376055922733e-05,
+      "loss": 0.7705,
+      "step": 3306
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8096090143982835e-05,
+      "loss": 0.4562,
+      "step": 3307
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8094803843357803e-05,
+      "loss": 0.4809,
+      "step": 3308
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.80935171541094e-05,
+      "loss": 0.8447,
+      "step": 3309
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8092230076299388e-05,
+      "loss": 0.8332,
+      "step": 3310
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.809094260998957e-05,
+      "loss": 0.4597,
+      "step": 3311
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.808965475524174e-05,
+      "loss": 0.8676,
+      "step": 3312
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8088366512117744e-05,
+      "loss": 0.8432,
+      "step": 3313
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.808707788067942e-05,
+      "loss": 0.8379,
+      "step": 3314
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.808578886098863e-05,
+      "loss": 0.8301,
+      "step": 3315
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8084499453107268e-05,
+      "loss": 0.8214,
+      "step": 3316
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8083209657097235e-05,
+      "loss": 0.8912,
+      "step": 3317
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.808191947302045e-05,
+      "loss": 0.8189,
+      "step": 3318
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.808062890093885e-05,
+      "loss": 0.8444,
+      "step": 3319
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8079337940914402e-05,
+      "loss": 0.4683,
+      "step": 3320
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.807804659300908e-05,
+      "loss": 0.7983,
+      "step": 3321
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8076754857284875e-05,
+      "loss": 0.8405,
+      "step": 3322
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.807546273380381e-05,
+      "loss": 0.4611,
+      "step": 3323
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8074170222627914e-05,
+      "loss": 0.811,
+      "step": 3324
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8072877323819247e-05,
+      "loss": 0.8162,
+      "step": 3325
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8071584037439867e-05,
+      "loss": 0.8556,
+      "step": 3326
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8070290363551875e-05,
+      "loss": 0.8171,
+      "step": 3327
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8068996302217373e-05,
+      "loss": 0.454,
+      "step": 3328
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8067701853498484e-05,
+      "loss": 0.8373,
+      "step": 3329
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8066407017457362e-05,
+      "loss": 0.8136,
+      "step": 3330
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8065111794156164e-05,
+      "loss": 0.8289,
+      "step": 3331
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8063816183657077e-05,
+      "loss": 0.8454,
+      "step": 3332
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.80625201860223e-05,
+      "loss": 0.8449,
+      "step": 3333
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.806122380131405e-05,
+      "loss": 0.8505,
+      "step": 3334
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.805992702959457e-05,
+      "loss": 0.8573,
+      "step": 3335
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8058629870926113e-05,
+      "loss": 0.8489,
+      "step": 3336
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8057332325370954e-05,
+      "loss": 0.8291,
+      "step": 3337
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.805603439299139e-05,
+      "loss": 0.8655,
+      "step": 3338
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8054736073849733e-05,
+      "loss": 0.8136,
+      "step": 3339
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8053437368008314e-05,
+      "loss": 0.4762,
+      "step": 3340
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8052138275529477e-05,
+      "loss": 0.4487,
+      "step": 3341
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8050838796475594e-05,
+      "loss": 0.8225,
+      "step": 3342
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8049538930909058e-05,
+      "loss": 0.866,
+      "step": 3343
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8048238678892264e-05,
+      "loss": 0.8294,
+      "step": 3344
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8046938040487642e-05,
+      "loss": 0.7841,
+      "step": 3345
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.804563701575763e-05,
+      "loss": 0.4322,
+      "step": 3346
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8044335604764694e-05,
+      "loss": 0.4635,
+      "step": 3347
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.804303380757131e-05,
+      "loss": 0.4703,
+      "step": 3348
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8041731624239974e-05,
+      "loss": 0.8358,
+      "step": 3349
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8040429054833207e-05,
+      "loss": 0.9011,
+      "step": 3350
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8039126099413542e-05,
+      "loss": 0.8343,
+      "step": 3351
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8037822758043532e-05,
+      "loss": 0.4651,
+      "step": 3352
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8036519030785752e-05,
+      "loss": 0.8493,
+      "step": 3353
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.803521491770279e-05,
+      "loss": 0.8995,
+      "step": 3354
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8033910418857258e-05,
+      "loss": 0.8106,
+      "step": 3355
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.803260553431178e-05,
+      "loss": 0.4373,
+      "step": 3356
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8031300264129e-05,
+      "loss": 0.4821,
+      "step": 3357
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.802999460837159e-05,
+      "loss": 0.872,
+      "step": 3358
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.802868856710223e-05,
+      "loss": 0.7671,
+      "step": 3359
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.802738214038362e-05,
+      "loss": 0.8712,
+      "step": 3360
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8026075328278483e-05,
+      "loss": 0.8071,
+      "step": 3361
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8024768130849555e-05,
+      "loss": 0.4185,
+      "step": 3362
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8023460548159596e-05,
+      "loss": 0.8655,
+      "step": 3363
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8022152580271376e-05,
+      "loss": 0.8725,
+      "step": 3364
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8020844227247697e-05,
+      "loss": 0.8384,
+      "step": 3365
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8019535489151368e-05,
+      "loss": 0.4746,
+      "step": 3366
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8018226366045218e-05,
+      "loss": 0.8342,
+      "step": 3367
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.80169168579921e-05,
+      "loss": 0.8488,
+      "step": 3368
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8015606965054883e-05,
+      "loss": 0.8629,
+      "step": 3369
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8014296687296452e-05,
+      "loss": 0.476,
+      "step": 3370
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8012986024779712e-05,
+      "loss": 0.4829,
+      "step": 3371
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.801167497756758e-05,
+      "loss": 0.8562,
+      "step": 3372
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8010363545723014e-05,
+      "loss": 0.4346,
+      "step": 3373
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.800905172930896e-05,
+      "loss": 0.8103,
+      "step": 3374
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8007739528388403e-05,
+      "loss": 0.8366,
+      "step": 3375
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.800642694302434e-05,
+      "loss": 0.4436,
+      "step": 3376
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8005113973279786e-05,
+      "loss": 0.7992,
+      "step": 3377
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8003800619217777e-05,
+      "loss": 0.7985,
+      "step": 3378
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.800248688090136e-05,
+      "loss": 0.7475,
+      "step": 3379
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8001172758393616e-05,
+      "loss": 0.8167,
+      "step": 3380
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7999858251757628e-05,
+      "loss": 0.8275,
+      "step": 3381
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7998543361056503e-05,
+      "loss": 0.8436,
+      "step": 3382
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7997228086353377e-05,
+      "loss": 0.4353,
+      "step": 3383
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7995912427711382e-05,
+      "loss": 0.83,
+      "step": 3384
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.799459638519369e-05,
+      "loss": 0.8173,
+      "step": 3385
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.799327995886348e-05,
+      "loss": 0.8566,
+      "step": 3386
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.799196314878395e-05,
+      "loss": 0.8656,
+      "step": 3387
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7990645955018328e-05,
+      "loss": 0.4974,
+      "step": 3388
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7989328377629843e-05,
+      "loss": 0.8439,
+      "step": 3389
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.798801041668175e-05,
+      "loss": 0.8131,
+      "step": 3390
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7986692072237328e-05,
+      "loss": 0.8621,
+      "step": 3391
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.798537334435986e-05,
+      "loss": 0.8032,
+      "step": 3392
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.798405423311267e-05,
+      "loss": 0.8456,
+      "step": 3393
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.798273473855908e-05,
+      "loss": 0.8999,
+      "step": 3394
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.798141486076244e-05,
+      "loss": 0.8535,
+      "step": 3395
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7980094599786108e-05,
+      "loss": 0.891,
+      "step": 3396
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.797877395569348e-05,
+      "loss": 0.8436,
+      "step": 3397
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7977452928547953e-05,
+      "loss": 0.8409,
+      "step": 3398
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7976131518412946e-05,
+      "loss": 0.8086,
+      "step": 3399
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.79748097253519e-05,
+      "loss": 0.8454,
+      "step": 3400
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.797348754942828e-05,
+      "loss": 0.8542,
+      "step": 3401
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7972164990705553e-05,
+      "loss": 0.7967,
+      "step": 3402
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.797084204924722e-05,
+      "loss": 0.7969,
+      "step": 3403
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.796951872511679e-05,
+      "loss": 0.8569,
+      "step": 3404
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.796819501837779e-05,
+      "loss": 0.4477,
+      "step": 3405
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7966870929093784e-05,
+      "loss": 0.823,
+      "step": 3406
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7965546457328327e-05,
+      "loss": 0.8373,
+      "step": 3407
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.796422160314501e-05,
+      "loss": 0.859,
+      "step": 3408
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7962896366607443e-05,
+      "loss": 0.8037,
+      "step": 3409
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.796157074777924e-05,
+      "loss": 0.4517,
+      "step": 3410
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.796024474672405e-05,
+      "loss": 0.44,
+      "step": 3411
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7958918363505532e-05,
+      "loss": 0.8251,
+      "step": 3412
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7957591598187357e-05,
+      "loss": 0.4866,
+      "step": 3413
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.795626445083323e-05,
+      "loss": 0.4437,
+      "step": 3414
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7954936921506864e-05,
+      "loss": 0.8023,
+      "step": 3415
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7953609010271987e-05,
+      "loss": 0.851,
+      "step": 3416
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7952280717192357e-05,
+      "loss": 0.8064,
+      "step": 3417
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7950952042331746e-05,
+      "loss": 0.8664,
+      "step": 3418
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7949622985753937e-05,
+      "loss": 0.8105,
+      "step": 3419
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7948293547522737e-05,
+      "loss": 0.4984,
+      "step": 3420
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.794696372770197e-05,
+      "loss": 0.7317,
+      "step": 3421
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7945633526355486e-05,
+      "loss": 0.8055,
+      "step": 3422
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.794430294354714e-05,
+      "loss": 0.4676,
+      "step": 3423
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7942971979340817e-05,
+      "loss": 0.8313,
+      "step": 3424
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.794164063380041e-05,
+      "loss": 0.4378,
+      "step": 3425
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7940308906989832e-05,
+      "loss": 0.4342,
+      "step": 3426
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7938976798973033e-05,
+      "loss": 0.8269,
+      "step": 3427
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7937644309813952e-05,
+      "loss": 0.8248,
+      "step": 3428
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7936311439576567e-05,
+      "loss": 0.844,
+      "step": 3429
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7934978188324864e-05,
+      "loss": 0.8343,
+      "step": 3430
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7933644556122854e-05,
+      "loss": 0.8341,
+      "step": 3431
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7932310543034562e-05,
+      "loss": 0.828,
+      "step": 3432
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7930976149124036e-05,
+      "loss": 0.4238,
+      "step": 3433
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7929641374455333e-05,
+      "loss": 0.5163,
+      "step": 3434
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7928306219092538e-05,
+      "loss": 0.8237,
+      "step": 3435
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.792697068309975e-05,
+      "loss": 0.8258,
+      "step": 3436
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7925634766541086e-05,
+      "loss": 0.85,
+      "step": 3437
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7924298469480684e-05,
+      "loss": 0.4889,
+      "step": 3438
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7922961791982695e-05,
+      "loss": 0.8208,
+      "step": 3439
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7921624734111292e-05,
+      "loss": 0.8251,
+      "step": 3440
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7920287295930672e-05,
+      "loss": 0.7871,
+      "step": 3441
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7918949477505035e-05,
+      "loss": 0.8949,
+      "step": 3442
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7917611278898615e-05,
+      "loss": 0.4652,
+      "step": 3443
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7916272700175656e-05,
+      "loss": 0.4233,
+      "step": 3444
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7914933741400414e-05,
+      "loss": 0.8063,
+      "step": 3445
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7913594402637184e-05,
+      "loss": 0.4636,
+      "step": 3446
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7912254683950257e-05,
+      "loss": 0.4898,
+      "step": 3447
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7910914585403958e-05,
+      "loss": 0.8224,
+      "step": 3448
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7909574107062618e-05,
+      "loss": 0.8545,
+      "step": 3449
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7908233248990593e-05,
+      "loss": 0.8793,
+      "step": 3450
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7906892011252258e-05,
+      "loss": 0.8163,
+      "step": 3451
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.790555039391201e-05,
+      "loss": 0.8851,
+      "step": 3452
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.790420839703425e-05,
+      "loss": 0.5275,
+      "step": 3453
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7902866020683404e-05,
+      "loss": 0.4485,
+      "step": 3454
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7901523264923928e-05,
+      "loss": 0.8581,
+      "step": 3455
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7900180129820278e-05,
+      "loss": 0.4272,
+      "step": 3456
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7898836615436942e-05,
+      "loss": 0.8492,
+      "step": 3457
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7897492721838422e-05,
+      "loss": 0.8204,
+      "step": 3458
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.789614844908923e-05,
+      "loss": 0.8203,
+      "step": 3459
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7894803797253907e-05,
+      "loss": 0.4678,
+      "step": 3460
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7893458766397006e-05,
+      "loss": 0.821,
+      "step": 3461
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7892113356583107e-05,
+      "loss": 0.8576,
+      "step": 3462
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7890767567876795e-05,
+      "loss": 0.8956,
+      "step": 3463
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7889421400342685e-05,
+      "loss": 0.801,
+      "step": 3464
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7888074854045395e-05,
+      "loss": 0.8303,
+      "step": 3465
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.788672792904959e-05,
+      "loss": 0.4999,
+      "step": 3466
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7885380625419918e-05,
+      "loss": 0.8619,
+      "step": 3467
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7884032943221065e-05,
+      "loss": 0.8448,
+      "step": 3468
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.788268488251774e-05,
+      "loss": 0.5088,
+      "step": 3469
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.788133644337465e-05,
+      "loss": 0.8732,
+      "step": 3470
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7879987625856537e-05,
+      "loss": 0.468,
+      "step": 3471
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7878638430028164e-05,
+      "loss": 0.8118,
+      "step": 3472
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7877288855954294e-05,
+      "loss": 0.4343,
+      "step": 3473
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.787593890369973e-05,
+      "loss": 0.4567,
+      "step": 3474
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7874588573329267e-05,
+      "loss": 0.8444,
+      "step": 3475
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7873237864907743e-05,
+      "loss": 0.8416,
+      "step": 3476
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7871886778500002e-05,
+      "loss": 0.8713,
+      "step": 3477
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7870535314170907e-05,
+      "loss": 0.7734,
+      "step": 3478
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7869183471985344e-05,
+      "loss": 0.8535,
+      "step": 3479
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.786783125200821e-05,
+      "loss": 0.4436,
+      "step": 3480
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7866478654304426e-05,
+      "loss": 0.8442,
+      "step": 3481
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7865125678938925e-05,
+      "loss": 0.8174,
+      "step": 3482
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7863772325976666e-05,
+      "loss": 0.4475,
+      "step": 3483
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.786241859548262e-05,
+      "loss": 0.4654,
+      "step": 3484
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.786106448752178e-05,
+      "loss": 0.8656,
+      "step": 3485
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7859710002159153e-05,
+      "loss": 0.8094,
+      "step": 3486
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7858355139459772e-05,
+      "loss": 0.8388,
+      "step": 3487
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7856999899488677e-05,
+      "loss": 0.8575,
+      "step": 3488
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.785564428231093e-05,
+      "loss": 0.4572,
+      "step": 3489
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.785428828799162e-05,
+      "loss": 0.8395,
+      "step": 3490
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.785293191659584e-05,
+      "loss": 0.8302,
+      "step": 3491
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7851575168188714e-05,
+      "loss": 0.4649,
+      "step": 3492
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.785021804283537e-05,
+      "loss": 0.4191,
+      "step": 3493
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7848860540600976e-05,
+      "loss": 0.7956,
+      "step": 3494
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.784750266155069e-05,
+      "loss": 0.8296,
+      "step": 3495
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.784614440574971e-05,
+      "loss": 0.4258,
+      "step": 3496
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7844785773263236e-05,
+      "loss": 0.4437,
+      "step": 3497
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.784342676415651e-05,
+      "loss": 0.8917,
+      "step": 3498
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7842067378494765e-05,
+      "loss": 0.7837,
+      "step": 3499
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7840707616343266e-05,
+      "loss": 0.8016,
+      "step": 3500
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7839347477767293e-05,
+      "loss": 0.8883,
+      "step": 3501
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.783798696283215e-05,
+      "loss": 0.4941,
+      "step": 3502
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7836626071603143e-05,
+      "loss": 0.4865,
+      "step": 3503
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.783526480414562e-05,
+      "loss": 0.8228,
+      "step": 3504
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7833903160524924e-05,
+      "loss": 0.4543,
+      "step": 3505
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7832541140806435e-05,
+      "loss": 0.8384,
+      "step": 3506
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7831178745055536e-05,
+      "loss": 0.8247,
+      "step": 3507
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7829815973337636e-05,
+      "loss": 0.8489,
+      "step": 3508
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.782845282571816e-05,
+      "loss": 0.777,
+      "step": 3509
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7827089302262553e-05,
+      "loss": 0.4401,
+      "step": 3510
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7825725403036274e-05,
+      "loss": 0.8751,
+      "step": 3511
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.78243611281048e-05,
+      "loss": 0.807,
+      "step": 3512
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7822996477533636e-05,
+      "loss": 0.7925,
+      "step": 3513
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.782163145138829e-05,
+      "loss": 0.8107,
+      "step": 3514
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.78202660497343e-05,
+      "loss": 0.8045,
+      "step": 3515
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.781890027263722e-05,
+      "loss": 0.865,
+      "step": 3516
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.781753412016261e-05,
+      "loss": 0.8064,
+      "step": 3517
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7816167592376066e-05,
+      "loss": 0.8437,
+      "step": 3518
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7814800689343193e-05,
+      "loss": 0.446,
+      "step": 3519
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.781343341112961e-05,
+      "loss": 0.8276,
+      "step": 3520
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7812065757800963e-05,
+      "loss": 0.8278,
+      "step": 3521
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.781069772942291e-05,
+      "loss": 0.7976,
+      "step": 3522
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.780932932606113e-05,
+      "loss": 0.8045,
+      "step": 3523
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7807960547781313e-05,
+      "loss": 0.4684,
+      "step": 3524
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.780659139464918e-05,
+      "loss": 0.8233,
+      "step": 3525
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.780522186673046e-05,
+      "loss": 0.7927,
+      "step": 3526
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7803851964090905e-05,
+      "loss": 0.8047,
+      "step": 3527
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7802481686796275e-05,
+      "loss": 0.4249,
+      "step": 3528
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7801111034912365e-05,
+      "loss": 0.8965,
+      "step": 3529
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.779974000850497e-05,
+      "loss": 0.4662,
+      "step": 3530
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7798368607639916e-05,
+      "loss": 0.8203,
+      "step": 3531
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7796996832383046e-05,
+      "loss": 0.854,
+      "step": 3532
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7795624682800213e-05,
+      "loss": 0.834,
+      "step": 3533
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7794252158957295e-05,
+      "loss": 0.8516,
+      "step": 3534
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7792879260920184e-05,
+      "loss": 0.8139,
+      "step": 3535
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7791505988754788e-05,
+      "loss": 0.8199,
+      "step": 3536
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.779013234252704e-05,
+      "loss": 0.8557,
+      "step": 3537
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.778875832230289e-05,
+      "loss": 0.8272,
+      "step": 3538
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.77873839281483e-05,
+      "loss": 0.8433,
+      "step": 3539
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7786009160129255e-05,
+      "loss": 0.864,
+      "step": 3540
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7784634018311753e-05,
+      "loss": 0.7881,
+      "step": 3541
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7783258502761815e-05,
+      "loss": 0.4445,
+      "step": 3542
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.778188261354548e-05,
+      "loss": 0.8302,
+      "step": 3543
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.77805063507288e-05,
+      "loss": 0.4441,
+      "step": 3544
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7779129714377854e-05,
+      "loss": 0.8419,
+      "step": 3545
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7777752704558722e-05,
+      "loss": 0.9066,
+      "step": 3546
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7776375321337523e-05,
+      "loss": 0.4821,
+      "step": 3547
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.777499756478038e-05,
+      "loss": 0.8298,
+      "step": 3548
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.777361943495343e-05,
+      "loss": 0.7823,
+      "step": 3549
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7772240931922855e-05,
+      "loss": 0.8149,
+      "step": 3550
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7770862055754818e-05,
+      "loss": 0.8379,
+      "step": 3551
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7769482806515523e-05,
+      "loss": 0.8481,
+      "step": 3552
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.776810318427119e-05,
+      "loss": 0.4748,
+      "step": 3553
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7766723189088047e-05,
+      "loss": 0.7879,
+      "step": 3554
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.776534282103235e-05,
+      "loss": 0.7975,
+      "step": 3555
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7763962080170368e-05,
+      "loss": 0.8347,
+      "step": 3556
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.776258096656839e-05,
+      "loss": 0.4761,
+      "step": 3557
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.776119948029272e-05,
+      "loss": 0.8712,
+      "step": 3558
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.775981762140968e-05,
+      "loss": 0.8618,
+      "step": 3559
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.775843538998562e-05,
+      "loss": 0.8038,
+      "step": 3560
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.775705278608689e-05,
+      "loss": 0.4778,
+      "step": 3561
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7755669809779874e-05,
+      "loss": 0.4836,
+      "step": 3562
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7754286461130966e-05,
+      "loss": 0.8283,
+      "step": 3563
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7752902740206574e-05,
+      "loss": 0.7934,
+      "step": 3564
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7751518647073138e-05,
+      "loss": 0.8626,
+      "step": 3565
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7750134181797097e-05,
+      "loss": 0.8745,
+      "step": 3566
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7748749344444924e-05,
+      "loss": 0.8177,
+      "step": 3567
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7747364135083104e-05,
+      "loss": 0.8062,
+      "step": 3568
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7745978553778136e-05,
+      "loss": 0.8848,
+      "step": 3569
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7744592600596542e-05,
+      "loss": 0.8063,
+      "step": 3570
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.774320627560486e-05,
+      "loss": 0.8718,
+      "step": 3571
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.774181957886965e-05,
+      "loss": 0.4829,
+      "step": 3572
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7740432510457478e-05,
+      "loss": 0.817,
+      "step": 3573
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7739045070434944e-05,
+      "loss": 0.8198,
+      "step": 3574
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.773765725886865e-05,
+      "loss": 0.7973,
+      "step": 3575
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7736269075825227e-05,
+      "loss": 0.4538,
+      "step": 3576
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7734880521371324e-05,
+      "loss": 0.8428,
+      "step": 3577
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.77334915955736e-05,
+      "loss": 0.8132,
+      "step": 3578
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7732102298498735e-05,
+      "loss": 0.4681,
+      "step": 3579
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.773071263021343e-05,
+      "loss": 0.8312,
+      "step": 3580
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.77293225907844e-05,
+      "loss": 0.4393,
+      "step": 3581
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7727932180278384e-05,
+      "loss": 0.44,
+      "step": 3582
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7726541398762125e-05,
+      "loss": 0.8484,
+      "step": 3583
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.77251502463024e-05,
+      "loss": 0.834,
+      "step": 3584
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7723758722965995e-05,
+      "loss": 0.8576,
+      "step": 3585
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7722366828819718e-05,
+      "loss": 0.7386,
+      "step": 3586
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.772097456393039e-05,
+      "loss": 0.8266,
+      "step": 3587
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.771958192836485e-05,
+      "loss": 0.8209,
+      "step": 3588
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7718188922189964e-05,
+      "loss": 0.7987,
+      "step": 3589
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.77167955454726e-05,
+      "loss": 0.7806,
+      "step": 3590
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7715401798279658e-05,
+      "loss": 0.8344,
+      "step": 3591
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.771400768067805e-05,
+      "loss": 0.7827,
+      "step": 3592
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7712613192734703e-05,
+      "loss": 0.4538,
+      "step": 3593
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.771121833451657e-05,
+      "loss": 0.8423,
+      "step": 3594
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7709823106090616e-05,
+      "loss": 0.8131,
+      "step": 3595
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.770842750752382e-05,
+      "loss": 0.8483,
+      "step": 3596
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7707031538883188e-05,
+      "loss": 0.8872,
+      "step": 3597
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7705635200235737e-05,
+      "loss": 0.8721,
+      "step": 3598
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7704238491648503e-05,
+      "loss": 0.7756,
+      "step": 3599
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.770284141318854e-05,
+      "loss": 0.8153,
+      "step": 3600
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7701443964922926e-05,
+      "loss": 0.4709,
+      "step": 3601
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7700046146918745e-05,
+      "loss": 0.7985,
+      "step": 3602
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7698647959243107e-05,
+      "loss": 0.8464,
+      "step": 3603
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7697249401963135e-05,
+      "loss": 0.4366,
+      "step": 3604
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7695850475145975e-05,
+      "loss": 0.8215,
+      "step": 3605
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.769445117885879e-05,
+      "loss": 0.8371,
+      "step": 3606
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7693051513168756e-05,
+      "loss": 0.8003,
+      "step": 3607
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.769165147814307e-05,
+      "loss": 0.8476,
+      "step": 3608
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7690251073848945e-05,
+      "loss": 0.8288,
+      "step": 3609
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7688850300353613e-05,
+      "loss": 0.8078,
+      "step": 3610
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.768744915772433e-05,
+      "loss": 0.8367,
+      "step": 3611
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7686047646028358e-05,
+      "loss": 0.8539,
+      "step": 3612
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.768464576533298e-05,
+      "loss": 0.8068,
+      "step": 3613
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.76832435157055e-05,
+      "loss": 0.8243,
+      "step": 3614
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7681840897213246e-05,
+      "loss": 0.7959,
+      "step": 3615
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7680437909923543e-05,
+      "loss": 0.8606,
+      "step": 3616
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.767903455390376e-05,
+      "loss": 0.7761,
+      "step": 3617
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7677630829221264e-05,
+      "loss": 0.8294,
+      "step": 3618
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7676226735943448e-05,
+      "loss": 0.8312,
+      "step": 3619
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7674822274137716e-05,
+      "loss": 0.4257,
+      "step": 3620
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7673417443871504e-05,
+      "loss": 0.8054,
+      "step": 3621
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7672012245212253e-05,
+      "loss": 0.8501,
+      "step": 3622
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7670606678227422e-05,
+      "loss": 0.8226,
+      "step": 3623
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7669200742984494e-05,
+      "loss": 0.8672,
+      "step": 3624
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7667794439550964e-05,
+      "loss": 0.7982,
+      "step": 3625
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.766638776799435e-05,
+      "loss": 0.8181,
+      "step": 3626
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7664980728382184e-05,
+      "loss": 0.8528,
+      "step": 3627
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7663573320782013e-05,
+      "loss": 0.8932,
+      "step": 3628
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7662165545261413e-05,
+      "loss": 0.8258,
+      "step": 3629
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7660757401887963e-05,
+      "loss": 0.8206,
+      "step": 3630
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7659348890729266e-05,
+      "loss": 0.8152,
+      "step": 3631
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.765794001185295e-05,
+      "loss": 0.4863,
+      "step": 3632
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7656530765326653e-05,
+      "loss": 0.4563,
+      "step": 3633
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.765512115121802e-05,
+      "loss": 0.8446,
+      "step": 3634
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.765371116959474e-05,
+      "loss": 0.8195,
+      "step": 3635
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7652300820524496e-05,
+      "loss": 0.4297,
+      "step": 3636
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7650890104075e-05,
+      "loss": 0.8748,
+      "step": 3637
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7649479020313978e-05,
+      "loss": 0.4594,
+      "step": 3638
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7648067569309175e-05,
+      "loss": 0.4605,
+      "step": 3639
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7646655751128355e-05,
+      "loss": 0.7992,
+      "step": 3640
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.76452435658393e-05,
+      "loss": 0.8128,
+      "step": 3641
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7643831013509796e-05,
+      "loss": 0.8164,
+      "step": 3642
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7642418094207677e-05,
+      "loss": 0.8228,
+      "step": 3643
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7641004808000758e-05,
+      "loss": 0.8321,
+      "step": 3644
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.76395911549569e-05,
+      "loss": 0.8311,
+      "step": 3645
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7638177135143968e-05,
+      "loss": 0.8114,
+      "step": 3646
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7636762748629846e-05,
+      "loss": 0.8097,
+      "step": 3647
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7635347995482443e-05,
+      "loss": 0.8564,
+      "step": 3648
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.763393287576967e-05,
+      "loss": 0.8115,
+      "step": 3649
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7632517389559476e-05,
+      "loss": 0.4234,
+      "step": 3650
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.763110153691981e-05,
+      "loss": 0.8396,
+      "step": 3651
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.762968531791865e-05,
+      "loss": 0.4731,
+      "step": 3652
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7628268732623985e-05,
+      "loss": 0.8459,
+      "step": 3653
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.762685178110382e-05,
+      "loss": 0.8562,
+      "step": 3654
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.762543446342619e-05,
+      "loss": 0.7959,
+      "step": 3655
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7624016779659133e-05,
+      "loss": 0.8305,
+      "step": 3656
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7622598729870714e-05,
+      "loss": 0.8519,
+      "step": 3657
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7621180314129004e-05,
+      "loss": 0.8406,
+      "step": 3658
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7619761532502113e-05,
+      "loss": 0.8581,
+      "step": 3659
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7618342385058147e-05,
+      "loss": 0.8464,
+      "step": 3660
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7616922871865237e-05,
+      "loss": 0.8263,
+      "step": 3661
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7615502992991533e-05,
+      "loss": 0.8542,
+      "step": 3662
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7614082748505206e-05,
+      "loss": 0.8112,
+      "step": 3663
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7612662138474438e-05,
+      "loss": 0.8128,
+      "step": 3664
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7611241162967432e-05,
+      "loss": 0.8009,
+      "step": 3665
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7609819822052405e-05,
+      "loss": 0.8382,
+      "step": 3666
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7608398115797593e-05,
+      "loss": 0.8316,
+      "step": 3667
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7606976044271258e-05,
+      "loss": 0.4509,
+      "step": 3668
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7605553607541664e-05,
+      "loss": 0.8468,
+      "step": 3669
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7604130805677108e-05,
+      "loss": 0.8528,
+      "step": 3670
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.760270763874589e-05,
+      "loss": 0.4744,
+      "step": 3671
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.760128410681634e-05,
+      "loss": 0.7974,
+      "step": 3672
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.75998602099568e-05,
+      "loss": 0.8577,
+      "step": 3673
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7598435948235627e-05,
+      "loss": 0.8552,
+      "step": 3674
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.75970113217212e-05,
+      "loss": 0.8522,
+      "step": 3675
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7595586330481915e-05,
+      "loss": 0.8239,
+      "step": 3676
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7594160974586184e-05,
+      "loss": 0.8608,
+      "step": 3677
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7592735254102435e-05,
+      "loss": 0.8323,
+      "step": 3678
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7591309169099117e-05,
+      "loss": 0.8337,
+      "step": 3679
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7589882719644694e-05,
+      "loss": 0.8393,
+      "step": 3680
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.758845590580765e-05,
+      "loss": 0.8401,
+      "step": 3681
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.758702872765648e-05,
+      "loss": 0.8242,
+      "step": 3682
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7585601185259705e-05,
+      "loss": 0.8173,
+      "step": 3683
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7584173278685867e-05,
+      "loss": 0.4864,
+      "step": 3684
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.758274500800351e-05,
+      "loss": 0.8576,
+      "step": 3685
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.75813163732812e-05,
+      "loss": 0.8513,
+      "step": 3686
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.757988737458753e-05,
+      "loss": 0.8402,
+      "step": 3687
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.757845801199111e-05,
+      "loss": 0.8477,
+      "step": 3688
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.757702828556055e-05,
+      "loss": 0.8164,
+      "step": 3689
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.75755981953645e-05,
+      "loss": 0.8199,
+      "step": 3690
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7574167741471614e-05,
+      "loss": 0.456,
+      "step": 3691
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7572736923950566e-05,
+      "loss": 0.8477,
+      "step": 3692
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7571305742870046e-05,
+      "loss": 0.7786,
+      "step": 3693
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.756987419829877e-05,
+      "loss": 0.8394,
+      "step": 3694
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7568442290305456e-05,
+      "loss": 0.8532,
+      "step": 3695
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7567010018958858e-05,
+      "loss": 0.8269,
+      "step": 3696
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.756557738432773e-05,
+      "loss": 0.8385,
+      "step": 3697
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7564144386480856e-05,
+      "loss": 0.8516,
+      "step": 3698
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.756271102548703e-05,
+      "loss": 0.4719,
+      "step": 3699
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7561277301415066e-05,
+      "loss": 0.8107,
+      "step": 3700
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.75598432143338e-05,
+      "loss": 0.8559,
+      "step": 3701
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.755840876431208e-05,
+      "loss": 0.7938,
+      "step": 3702
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7556973951418768e-05,
+      "loss": 0.8506,
+      "step": 3703
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.755553877572275e-05,
+      "loss": 0.4637,
+      "step": 3704
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7554103237292926e-05,
+      "loss": 0.4078,
+      "step": 3705
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7552667336198224e-05,
+      "loss": 0.8093,
+      "step": 3706
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.755123107250757e-05,
+      "loss": 0.8037,
+      "step": 3707
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.754979444628992e-05,
+      "loss": 0.9049,
+      "step": 3708
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.754835745761424e-05,
+      "loss": 0.4783,
+      "step": 3709
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7546920106549532e-05,
+      "loss": 0.841,
+      "step": 3710
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.754548239316479e-05,
+      "loss": 0.8289,
+      "step": 3711
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7544044317529044e-05,
+      "loss": 0.8543,
+      "step": 3712
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7542605879711327e-05,
+      "loss": 0.7875,
+      "step": 3713
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7541167079780702e-05,
+      "loss": 0.9302,
+      "step": 3714
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7539727917806248e-05,
+      "loss": 0.8607,
+      "step": 3715
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.753828839385705e-05,
+      "loss": 0.8675,
+      "step": 3716
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7536848508002222e-05,
+      "loss": 0.8729,
+      "step": 3717
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.753540826031089e-05,
+      "loss": 0.4608,
+      "step": 3718
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.75339676508522e-05,
+      "loss": 0.7987,
+      "step": 3719
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7532526679695317e-05,
+      "loss": 0.8341,
+      "step": 3720
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7531085346909416e-05,
+      "loss": 0.782,
+      "step": 3721
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7529643652563693e-05,
+      "loss": 0.8413,
+      "step": 3722
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7528201596727368e-05,
+      "loss": 0.837,
+      "step": 3723
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7526759179469667e-05,
+      "loss": 0.792,
+      "step": 3724
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7525316400859843e-05,
+      "loss": 0.8116,
+      "step": 3725
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.752387326096716e-05,
+      "loss": 0.7961,
+      "step": 3726
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7522429759860903e-05,
+      "loss": 0.7876,
+      "step": 3727
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.752098589761037e-05,
+      "loss": 0.8338,
+      "step": 3728
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7519541674284887e-05,
+      "loss": 0.4928,
+      "step": 3729
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7518097089953784e-05,
+      "loss": 0.8081,
+      "step": 3730
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7516652144686414e-05,
+      "loss": 0.85,
+      "step": 3731
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7515206838552143e-05,
+      "loss": 0.7891,
+      "step": 3732
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7513761171620366e-05,
+      "loss": 0.8393,
+      "step": 3733
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.751231514396049e-05,
+      "loss": 0.8682,
+      "step": 3734
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.751086875564193e-05,
+      "loss": 0.8538,
+      "step": 3735
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7509422006734128e-05,
+      "loss": 0.84,
+      "step": 3736
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7507974897306543e-05,
+      "loss": 0.4731,
+      "step": 3737
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7506527427428647e-05,
+      "loss": 0.5004,
+      "step": 3738
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7505079597169933e-05,
+      "loss": 0.8525,
+      "step": 3739
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.750363140659991e-05,
+      "loss": 0.4912,
+      "step": 3740
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7502182855788104e-05,
+      "loss": 0.838,
+      "step": 3741
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7500733944804057e-05,
+      "loss": 0.8407,
+      "step": 3742
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7499284673717328e-05,
+      "loss": 0.4682,
+      "step": 3743
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7497835042597503e-05,
+      "loss": 0.8249,
+      "step": 3744
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.749638505151417e-05,
+      "loss": 0.7898,
+      "step": 3745
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7494934700536943e-05,
+      "loss": 0.8454,
+      "step": 3746
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7493483989735452e-05,
+      "loss": 0.8059,
+      "step": 3747
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7492032919179347e-05,
+      "loss": 0.7878,
+      "step": 3748
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7490581488938286e-05,
+      "loss": 0.8949,
+      "step": 3749
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.748912969908196e-05,
+      "loss": 0.8807,
+      "step": 3750
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7487677549680058e-05,
+      "loss": 0.8429,
+      "step": 3751
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7486225040802306e-05,
+      "loss": 0.8427,
+      "step": 3752
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7484772172518432e-05,
+      "loss": 0.4454,
+      "step": 3753
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7483318944898187e-05,
+      "loss": 0.8357,
+      "step": 3754
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7481865358011337e-05,
+      "loss": 0.8322,
+      "step": 3755
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.748041141192767e-05,
+      "loss": 0.4658,
+      "step": 3756
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7478957106716992e-05,
+      "loss": 0.4956,
+      "step": 3757
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7477502442449116e-05,
+      "loss": 0.8289,
+      "step": 3758
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7476047419193884e-05,
+      "loss": 0.8674,
+      "step": 3759
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7474592037021142e-05,
+      "loss": 0.8091,
+      "step": 3760
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.747313629600077e-05,
+      "loss": 0.8421,
+      "step": 3761
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.747168019620266e-05,
+      "loss": 0.7952,
+      "step": 3762
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7470223737696705e-05,
+      "loss": 0.8279,
+      "step": 3763
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7468766920552837e-05,
+      "loss": 0.8189,
+      "step": 3764
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7467309744840993e-05,
+      "loss": 0.4684,
+      "step": 3765
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7465852210631134e-05,
+      "loss": 0.4585,
+      "step": 3766
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7464394317993233e-05,
+      "loss": 0.818,
+      "step": 3767
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.746293606699728e-05,
+      "loss": 0.8208,
+      "step": 3768
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.746147745771329e-05,
+      "loss": 0.8313,
+      "step": 3769
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.746001849021128e-05,
+      "loss": 0.804,
+      "step": 3770
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.74585591645613e-05,
+      "loss": 0.7967,
+      "step": 3771
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.745709948083341e-05,
+      "loss": 0.8876,
+      "step": 3772
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7455639439097688e-05,
+      "loss": 0.8157,
+      "step": 3773
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.745417903942423e-05,
+      "loss": 0.8274,
+      "step": 3774
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.745271828188314e-05,
+      "loss": 0.5329,
+      "step": 3775
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7451257166544563e-05,
+      "loss": 0.4454,
+      "step": 3776
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7449795693478633e-05,
+      "loss": 0.8058,
+      "step": 3777
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.744833386275552e-05,
+      "loss": 0.8541,
+      "step": 3778
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.74468716744454e-05,
+      "loss": 0.8387,
+      "step": 3779
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7445409128618472e-05,
+      "loss": 0.839,
+      "step": 3780
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.744394622534496e-05,
+      "loss": 0.8112,
+      "step": 3781
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7442482964695085e-05,
+      "loss": 0.7664,
+      "step": 3782
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7441019346739107e-05,
+      "loss": 0.8322,
+      "step": 3783
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.743955537154728e-05,
+      "loss": 0.5068,
+      "step": 3784
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.74380910391899e-05,
+      "loss": 0.7904,
+      "step": 3785
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.743662634973726e-05,
+      "loss": 0.8271,
+      "step": 3786
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7435161303259688e-05,
+      "loss": 0.8137,
+      "step": 3787
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7433695899827506e-05,
+      "loss": 0.838,
+      "step": 3788
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.743223013951108e-05,
+      "loss": 0.4379,
+      "step": 3789
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.743076402238077e-05,
+      "loss": 0.4782,
+      "step": 3790
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.742929754850697e-05,
+      "loss": 0.8176,
+      "step": 3791
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7427830717960076e-05,
+      "loss": 0.8538,
+      "step": 3792
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.742636353081051e-05,
+      "loss": 0.8181,
+      "step": 3793
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7424895987128723e-05,
+      "loss": 0.762,
+      "step": 3794
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7423428086985154e-05,
+      "loss": 0.854,
+      "step": 3795
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7421959830450283e-05,
+      "loss": 0.8267,
+      "step": 3796
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.74204912175946e-05,
+      "loss": 0.8401,
+      "step": 3797
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.741902224848861e-05,
+      "loss": 0.7833,
+      "step": 3798
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7417552923202835e-05,
+      "loss": 0.8196,
+      "step": 3799
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7416083241807823e-05,
+      "loss": 0.4347,
+      "step": 3800
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.741461320437412e-05,
+      "loss": 0.8424,
+      "step": 3801
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.741314281097231e-05,
+      "loss": 0.8272,
+      "step": 3802
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7411672061672985e-05,
+      "loss": 0.768,
+      "step": 3803
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7410200956546753e-05,
+      "loss": 0.8246,
+      "step": 3804
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7408729495664237e-05,
+      "loss": 0.8225,
+      "step": 3805
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7407257679096083e-05,
+      "loss": 0.8324,
+      "step": 3806
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.740578550691295e-05,
+      "loss": 0.8624,
+      "step": 3807
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7404312979185518e-05,
+      "loss": 0.5157,
+      "step": 3808
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7402840095984478e-05,
+      "loss": 0.7555,
+      "step": 3809
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7401366857380545e-05,
+      "loss": 0.8591,
+      "step": 3810
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7399893263444448e-05,
+      "loss": 0.8596,
+      "step": 3811
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.739841931424693e-05,
+      "loss": 0.4422,
+      "step": 3812
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7396945009858758e-05,
+      "loss": 0.7943,
+      "step": 3813
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7395470350350704e-05,
+      "loss": 0.8722,
+      "step": 3814
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7393995335793574e-05,
+      "loss": 0.78,
+      "step": 3815
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7392519966258176e-05,
+      "loss": 0.822,
+      "step": 3816
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.739104424181534e-05,
+      "loss": 0.4525,
+      "step": 3817
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.738956816253592e-05,
+      "loss": 0.7842,
+      "step": 3818
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7388091728490777e-05,
+      "loss": 0.7876,
+      "step": 3819
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7386614939750795e-05,
+      "loss": 0.4275,
+      "step": 3820
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7385137796386874e-05,
+      "loss": 0.856,
+      "step": 3821
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7383660298469923e-05,
+      "loss": 0.4655,
+      "step": 3822
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7382182446070886e-05,
+      "loss": 0.8338,
+      "step": 3823
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7380704239260704e-05,
+      "loss": 0.83,
+      "step": 3824
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.737922567811035e-05,
+      "loss": 0.4901,
+      "step": 3825
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7377746762690808e-05,
+      "loss": 0.8243,
+      "step": 3826
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7376267493073072e-05,
+      "loss": 0.8013,
+      "step": 3827
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7374787869328172e-05,
+      "loss": 0.8111,
+      "step": 3828
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.737330789152713e-05,
+      "loss": 0.827,
+      "step": 3829
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7371827559741014e-05,
+      "loss": 0.8437,
+      "step": 3830
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.737034687404088e-05,
+      "loss": 0.873,
+      "step": 3831
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7368865834497815e-05,
+      "loss": 0.4503,
+      "step": 3832
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.736738444118293e-05,
+      "loss": 0.8169,
+      "step": 3833
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.736590269416734e-05,
+      "loss": 0.8492,
+      "step": 3834
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7364420593522184e-05,
+      "loss": 0.7878,
+      "step": 3835
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7362938139318614e-05,
+      "loss": 0.8338,
+      "step": 3836
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7361455331627802e-05,
+      "loss": 0.8046,
+      "step": 3837
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.735997217052094e-05,
+      "loss": 0.4772,
+      "step": 3838
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7358488656069227e-05,
+      "loss": 0.4876,
+      "step": 3839
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.735700478834389e-05,
+      "loss": 0.4505,
+      "step": 3840
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7355520567416165e-05,
+      "loss": 0.7774,
+      "step": 3841
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.735403599335731e-05,
+      "loss": 0.8898,
+      "step": 3842
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7352551066238593e-05,
+      "loss": 0.7984,
+      "step": 3843
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.735106578613131e-05,
+      "loss": 0.4233,
+      "step": 3844
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.734958015310676e-05,
+      "loss": 0.886,
+      "step": 3845
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.734809416723628e-05,
+      "loss": 0.7758,
+      "step": 3846
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7346607828591197e-05,
+      "loss": 0.7959,
+      "step": 3847
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7345121137242877e-05,
+      "loss": 0.8376,
+      "step": 3848
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7343634093262692e-05,
+      "loss": 0.4591,
+      "step": 3849
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.734214669672203e-05,
+      "loss": 0.4428,
+      "step": 3850
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7340658947692306e-05,
+      "loss": 0.8182,
+      "step": 3851
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7339170846244942e-05,
+      "loss": 0.8604,
+      "step": 3852
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.733768239245138e-05,
+      "loss": 0.8229,
+      "step": 3853
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.733619358638308e-05,
+      "loss": 0.7715,
+      "step": 3854
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7334704428111515e-05,
+      "loss": 0.8391,
+      "step": 3855
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7333214917708184e-05,
+      "loss": 0.7856,
+      "step": 3856
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.733172505524459e-05,
+      "loss": 0.745,
+      "step": 3857
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7330234840792268e-05,
+      "loss": 0.4249,
+      "step": 3858
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7328744274422753e-05,
+      "loss": 0.8198,
+      "step": 3859
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.732725335620761e-05,
+      "loss": 0.4574,
+      "step": 3860
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7325762086218415e-05,
+      "loss": 0.455,
+      "step": 3861
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.732427046452677e-05,
+      "loss": 0.5002,
+      "step": 3862
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7322778491204272e-05,
+      "loss": 0.7848,
+      "step": 3863
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.732128616632256e-05,
+      "loss": 0.8419,
+      "step": 3864
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7319793489953273e-05,
+      "loss": 0.8599,
+      "step": 3865
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.731830046216808e-05,
+      "loss": 0.7896,
+      "step": 3866
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7316807083038654e-05,
+      "loss": 0.7982,
+      "step": 3867
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.731531335263669e-05,
+      "loss": 0.4479,
+      "step": 3868
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7313819271033906e-05,
+      "loss": 0.8136,
+      "step": 3869
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7312324838302026e-05,
+      "loss": 0.7963,
+      "step": 3870
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.73108300545128e-05,
+      "loss": 0.414,
+      "step": 3871
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7309334919737987e-05,
+      "loss": 0.7766,
+      "step": 3872
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7307839434049366e-05,
+      "loss": 0.8098,
+      "step": 3873
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7306343597518744e-05,
+      "loss": 0.4744,
+      "step": 3874
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7304847410217924e-05,
+      "loss": 0.8715,
+      "step": 3875
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7303350872218738e-05,
+      "loss": 0.8579,
+      "step": 3876
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7301853983593035e-05,
+      "loss": 0.8055,
+      "step": 3877
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7300356744412684e-05,
+      "loss": 0.4431,
+      "step": 3878
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7298859154749558e-05,
+      "loss": 0.7611,
+      "step": 3879
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.729736121467556e-05,
+      "loss": 0.8542,
+      "step": 3880
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.72958629242626e-05,
+      "loss": 0.8104,
+      "step": 3881
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7294364283582612e-05,
+      "loss": 0.781,
+      "step": 3882
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7292865292707545e-05,
+      "loss": 0.8056,
+      "step": 3883
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7291365951709366e-05,
+      "loss": 0.8864,
+      "step": 3884
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.728986626066005e-05,
+      "loss": 0.8045,
+      "step": 3885
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7288366219631604e-05,
+      "loss": 0.7916,
+      "step": 3886
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7286865828696036e-05,
+      "loss": 0.7707,
+      "step": 3887
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7285365087925384e-05,
+      "loss": 0.8125,
+      "step": 3888
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.728386399739169e-05,
+      "loss": 0.4344,
+      "step": 3889
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.728236255716703e-05,
+      "loss": 0.4521,
+      "step": 3890
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.728086076732348e-05,
+      "loss": 0.8448,
+      "step": 3891
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7279358627933138e-05,
+      "loss": 0.8262,
+      "step": 3892
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7277856139068125e-05,
+      "loss": 0.8576,
+      "step": 3893
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.727635330080057e-05,
+      "loss": 0.8253,
+      "step": 3894
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7274850113202622e-05,
+      "loss": 0.8627,
+      "step": 3895
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7273346576346454e-05,
+      "loss": 0.8313,
+      "step": 3896
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7271842690304247e-05,
+      "loss": 0.8245,
+      "step": 3897
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7270338455148198e-05,
+      "loss": 0.7781,
+      "step": 3898
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7268833870950522e-05,
+      "loss": 0.4687,
+      "step": 3899
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7267328937783454e-05,
+      "loss": 0.8788,
+      "step": 3900
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.726582365571925e-05,
+      "loss": 0.481,
+      "step": 3901
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7264318024830174e-05,
+      "loss": 0.4081,
+      "step": 3902
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7262812045188506e-05,
+      "loss": 0.8366,
+      "step": 3903
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.726130571686655e-05,
+      "loss": 0.4286,
+      "step": 3904
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7259799039936625e-05,
+      "loss": 0.8485,
+      "step": 3905
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.725829201447106e-05,
+      "loss": 0.4851,
+      "step": 3906
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.725678464054221e-05,
+      "loss": 0.4493,
+      "step": 3907
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.725527691822244e-05,
+      "loss": 0.8544,
+      "step": 3908
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7253768847584135e-05,
+      "loss": 0.4599,
+      "step": 3909
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7252260428699697e-05,
+      "loss": 0.8548,
+      "step": 3910
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7250751661641545e-05,
+      "loss": 0.4401,
+      "step": 3911
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.724924254648211e-05,
+      "loss": 0.7775,
+      "step": 3912
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7247733083293843e-05,
+      "loss": 0.4874,
+      "step": 3913
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7246223272149213e-05,
+      "loss": 0.8194,
+      "step": 3914
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7244713113120705e-05,
+      "loss": 0.8529,
+      "step": 3915
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7243202606280823e-05,
+      "loss": 0.8287,
+      "step": 3916
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.724169175170208e-05,
+      "loss": 0.8378,
+      "step": 3917
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7240180549457012e-05,
+      "loss": 0.8558,
+      "step": 3918
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7238668999618168e-05,
+      "loss": 0.7956,
+      "step": 3919
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7237157102258124e-05,
+      "loss": 0.7999,
+      "step": 3920
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7235644857449455e-05,
+      "loss": 0.8265,
+      "step": 3921
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.723413226526477e-05,
+      "loss": 0.4291,
+      "step": 3922
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.723261932577668e-05,
+      "loss": 0.8425,
+      "step": 3923
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.723110603905783e-05,
+      "loss": 0.8076,
+      "step": 3924
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.722959240518086e-05,
+      "loss": 0.8415,
+      "step": 3925
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7228078424218442e-05,
+      "loss": 0.8262,
+      "step": 3926
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7226564096243265e-05,
+      "loss": 0.8151,
+      "step": 3927
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7225049421328024e-05,
+      "loss": 0.8365,
+      "step": 3928
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.722353439954544e-05,
+      "loss": 0.4774,
+      "step": 3929
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.722201903096825e-05,
+      "loss": 0.7922,
+      "step": 3930
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.72205033156692e-05,
+      "loss": 0.4826,
+      "step": 3931
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7218987253721066e-05,
+      "loss": 0.8685,
+      "step": 3932
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.721747084519662e-05,
+      "loss": 0.774,
+      "step": 3933
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.721595409016868e-05,
+      "loss": 0.8551,
+      "step": 3934
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7214436988710047e-05,
+      "loss": 0.8476,
+      "step": 3935
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7212919540893564e-05,
+      "loss": 0.8262,
+      "step": 3936
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7211401746792085e-05,
+      "loss": 0.7971,
+      "step": 3937
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7209883606478473e-05,
+      "loss": 0.5066,
+      "step": 3938
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7208365120025613e-05,
+      "loss": 0.4492,
+      "step": 3939
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7206846287506406e-05,
+      "loss": 0.4426,
+      "step": 3940
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7205327108993767e-05,
+      "loss": 0.4536,
+      "step": 3941
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7203807584560637e-05,
+      "loss": 0.4321,
+      "step": 3942
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7202287714279964e-05,
+      "loss": 0.8091,
+      "step": 3943
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7200767498224715e-05,
+      "loss": 0.4044,
+      "step": 3944
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7199246936467872e-05,
+      "loss": 0.8111,
+      "step": 3945
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7197726029082437e-05,
+      "loss": 0.4629,
+      "step": 3946
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7196204776141427e-05,
+      "loss": 0.8587,
+      "step": 3947
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7194683177717877e-05,
+      "loss": 0.836,
+      "step": 3948
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7193161233884838e-05,
+      "loss": 0.8051,
+      "step": 3949
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7191638944715376e-05,
+      "loss": 0.8478,
+      "step": 3950
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.719011631028257e-05,
+      "loss": 0.4913,
+      "step": 3951
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7188593330659525e-05,
+      "loss": 0.7655,
+      "step": 3952
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7187070005919362e-05,
+      "loss": 0.4595,
+      "step": 3953
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7185546336135206e-05,
+      "loss": 0.8174,
+      "step": 3954
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.718402232138021e-05,
+      "loss": 0.8387,
+      "step": 3955
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7182497961727543e-05,
+      "loss": 0.863,
+      "step": 3956
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7180973257250384e-05,
+      "loss": 0.814,
+      "step": 3957
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7179448208021935e-05,
+      "loss": 0.808,
+      "step": 3958
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.717792281411541e-05,
+      "loss": 0.8688,
+      "step": 3959
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7176397075604042e-05,
+      "loss": 0.8552,
+      "step": 3960
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7174870992561087e-05,
+      "loss": 0.7862,
+      "step": 3961
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.71733445650598e-05,
+      "loss": 0.4708,
+      "step": 3962
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7171817793173467e-05,
+      "loss": 0.8533,
+      "step": 3963
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.717029067697539e-05,
+      "loss": 0.4782,
+      "step": 3964
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7168763216538883e-05,
+      "loss": 0.8437,
+      "step": 3965
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7167235411937272e-05,
+      "loss": 0.8356,
+      "step": 3966
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7165707263243915e-05,
+      "loss": 0.4473,
+      "step": 3967
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.716417877053217e-05,
+      "loss": 0.8373,
+      "step": 3968
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.716264993387542e-05,
+      "loss": 0.8122,
+      "step": 3969
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7161120753347067e-05,
+      "loss": 0.8117,
+      "step": 3970
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.715959122902052e-05,
+      "loss": 0.7949,
+      "step": 3971
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.715806136096921e-05,
+      "loss": 0.851,
+      "step": 3972
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.715653114926659e-05,
+      "loss": 0.7981,
+      "step": 3973
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7155000593986117e-05,
+      "loss": 0.8077,
+      "step": 3974
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7153469695201278e-05,
+      "loss": 0.7882,
+      "step": 3975
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7151938452985565e-05,
+      "loss": 0.8566,
+      "step": 3976
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.715040686741249e-05,
+      "loss": 0.859,
+      "step": 3977
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.714887493855559e-05,
+      "loss": 0.8248,
+      "step": 3978
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7147342666488406e-05,
+      "loss": 0.8294,
+      "step": 3979
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7145810051284496e-05,
+      "loss": 0.7729,
+      "step": 3980
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7144277093017453e-05,
+      "loss": 0.8524,
+      "step": 3981
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7142743791760865e-05,
+      "loss": 0.7988,
+      "step": 3982
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7141210147588337e-05,
+      "loss": 0.8724,
+      "step": 3983
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7139676160573512e-05,
+      "loss": 0.7946,
+      "step": 3984
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7138141830790027e-05,
+      "loss": 0.4437,
+      "step": 3985
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7136607158311542e-05,
+      "loss": 0.8331,
+      "step": 3986
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7135072143211738e-05,
+      "loss": 0.821,
+      "step": 3987
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.713353678556431e-05,
+      "loss": 0.8088,
+      "step": 3988
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7132001085442972e-05,
+      "loss": 0.4049,
+      "step": 3989
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7130465042921445e-05,
+      "loss": 0.8353,
+      "step": 3990
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7128928658073478e-05,
+      "loss": 0.4617,
+      "step": 3991
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7127391930972825e-05,
+      "loss": 0.7892,
+      "step": 3992
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.712585486169327e-05,
+      "loss": 0.78,
+      "step": 3993
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.71243174503086e-05,
+      "loss": 0.8386,
+      "step": 3994
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7122779696892628e-05,
+      "loss": 0.7789,
+      "step": 3995
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7121241601519184e-05,
+      "loss": 0.8182,
+      "step": 3996
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.71197031642621e-05,
+      "loss": 0.8567,
+      "step": 3997
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7118164385195243e-05,
+      "loss": 0.8408,
+      "step": 3998
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7116625264392484e-05,
+      "loss": 0.4906,
+      "step": 3999
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.711508580192772e-05,
+      "loss": 0.8184,
+      "step": 4000
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7113545997874852e-05,
+      "loss": 0.823,
+      "step": 4001
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7112005852307813e-05,
+      "loss": 0.8516,
+      "step": 4002
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7110465365300532e-05,
+      "loss": 0.8069,
+      "step": 4003
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7108924536926976e-05,
+      "loss": 0.5318,
+      "step": 4004
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7107383367261118e-05,
+      "loss": 0.7989,
+      "step": 4005
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.710584185637694e-05,
+      "loss": 0.8471,
+      "step": 4006
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7104300004348462e-05,
+      "loss": 0.8082,
+      "step": 4007
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7102757811249692e-05,
+      "loss": 0.7692,
+      "step": 4008
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.710121527715468e-05,
+      "loss": 0.8185,
+      "step": 4009
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7099672402137476e-05,
+      "loss": 0.825,
+      "step": 4010
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.709812918627215e-05,
+      "loss": 0.788,
+      "step": 4011
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.70965856296328e-05,
+      "loss": 0.502,
+      "step": 4012
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7095041732293522e-05,
+      "loss": 0.8537,
+      "step": 4013
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7093497494328438e-05,
+      "loss": 0.869,
+      "step": 4014
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7091952915811684e-05,
+      "loss": 0.8341,
+      "step": 4015
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7090407996817423e-05,
+      "loss": 0.4507,
+      "step": 4016
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.708886273741981e-05,
+      "loss": 0.8641,
+      "step": 4017
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7087317137693044e-05,
+      "loss": 0.8184,
+      "step": 4018
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7085771197711323e-05,
+      "loss": 0.8253,
+      "step": 4019
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7084224917548868e-05,
+      "loss": 0.8436,
+      "step": 4020
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.708267829727991e-05,
+      "loss": 0.8113,
+      "step": 4021
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.70811313369787e-05,
+      "loss": 0.7961,
+      "step": 4022
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7079584036719513e-05,
+      "loss": 0.4667,
+      "step": 4023
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.707803639657663e-05,
+      "loss": 0.8029,
+      "step": 4024
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.707648841662435e-05,
+      "loss": 0.8181,
+      "step": 4025
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7074940096936992e-05,
+      "loss": 0.785,
+      "step": 4026
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7073391437588886e-05,
+      "loss": 0.8393,
+      "step": 4027
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7071842438654385e-05,
+      "loss": 0.8199,
+      "step": 4028
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7070293100207856e-05,
+      "loss": 0.8134,
+      "step": 4029
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.706874342232368e-05,
+      "loss": 0.8198,
+      "step": 4030
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.706719340507625e-05,
+      "loss": 0.8053,
+      "step": 4031
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7065643048539984e-05,
+      "loss": 0.8683,
+      "step": 4032
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.706409235278932e-05,
+      "loss": 0.7641,
+      "step": 4033
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.70625413178987e-05,
+      "loss": 0.4923,
+      "step": 4034
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.706098994394258e-05,
+      "loss": 0.7865,
+      "step": 4035
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7059438230995455e-05,
+      "loss": 0.853,
+      "step": 4036
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7057886179131808e-05,
+      "loss": 0.8535,
+      "step": 4037
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7056333788426156e-05,
+      "loss": 0.8922,
+      "step": 4038
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7054781058953035e-05,
+      "loss": 0.4899,
+      "step": 4039
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.705322799078698e-05,
+      "loss": 0.7788,
+      "step": 4040
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7051674584002556e-05,
+      "loss": 0.795,
+      "step": 4041
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.705012083867434e-05,
+      "loss": 0.8587,
+      "step": 4042
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7048566754876926e-05,
+      "loss": 0.7801,
+      "step": 4043
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7047012332684923e-05,
+      "loss": 0.8338,
+      "step": 4044
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.704545757217296e-05,
+      "loss": 0.8374,
+      "step": 4045
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7043902473415676e-05,
+      "loss": 0.828,
+      "step": 4046
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7042347036487733e-05,
+      "loss": 0.8353,
+      "step": 4047
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7040791261463804e-05,
+      "loss": 0.8123,
+      "step": 4048
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7039235148418577e-05,
+      "loss": 0.8147,
+      "step": 4049
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7037678697426767e-05,
+      "loss": 0.8531,
+      "step": 4050
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7036121908563093e-05,
+      "loss": 0.8431,
+      "step": 4051
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7034564781902293e-05,
+      "loss": 0.8561,
+      "step": 4052
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.703300731751913e-05,
+      "loss": 0.4107,
+      "step": 4053
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7031449515488366e-05,
+      "loss": 0.8339,
+      "step": 4054
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7029891375884802e-05,
+      "loss": 0.8201,
+      "step": 4055
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7028332898783233e-05,
+      "loss": 0.8521,
+      "step": 4056
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7026774084258484e-05,
+      "loss": 0.8177,
+      "step": 4057
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.702521493238539e-05,
+      "loss": 0.7727,
+      "step": 4058
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7023655443238804e-05,
+      "loss": 0.8422,
+      "step": 4059
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7022095616893598e-05,
+      "loss": 0.8338,
+      "step": 4060
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7020535453424658e-05,
+      "loss": 0.8603,
+      "step": 4061
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7018974952906885e-05,
+      "loss": 0.8174,
+      "step": 4062
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7017414115415194e-05,
+      "loss": 0.8457,
+      "step": 4063
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7015852941024524e-05,
+      "loss": 0.8002,
+      "step": 4064
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7014291429809823e-05,
+      "loss": 0.7635,
+      "step": 4065
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7012729581846054e-05,
+      "loss": 0.8436,
+      "step": 4066
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7011167397208207e-05,
+      "loss": 0.8221,
+      "step": 4067
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7009604875971277e-05,
+      "loss": 0.7818,
+      "step": 4068
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.700804201821028e-05,
+      "loss": 0.4402,
+      "step": 4069
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7006478824000248e-05,
+      "loss": 0.4406,
+      "step": 4070
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7004915293416223e-05,
+      "loss": 0.8255,
+      "step": 4071
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7003351426533276e-05,
+      "loss": 0.8163,
+      "step": 4072
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7001787223426483e-05,
+      "loss": 0.4562,
+      "step": 4073
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7000222684170938e-05,
+      "loss": 0.8172,
+      "step": 4074
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6998657808841756e-05,
+      "loss": 0.847,
+      "step": 4075
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6997092597514068e-05,
+      "loss": 0.863,
+      "step": 4076
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.699552705026301e-05,
+      "loss": 0.41,
+      "step": 4077
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6993961167163748e-05,
+      "loss": 0.8552,
+      "step": 4078
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6992394948291457e-05,
+      "loss": 0.7922,
+      "step": 4079
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.699082839372133e-05,
+      "loss": 0.4391,
+      "step": 4080
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6989261503528575e-05,
+      "loss": 0.8495,
+      "step": 4081
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.698769427778842e-05,
+      "loss": 0.4619,
+      "step": 4082
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6986126716576098e-05,
+      "loss": 0.7986,
+      "step": 4083
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6984558819966875e-05,
+      "loss": 0.8724,
+      "step": 4084
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.698299058803602e-05,
+      "loss": 0.8865,
+      "step": 4085
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6981422020858823e-05,
+      "loss": 0.8736,
+      "step": 4086
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6979853118510587e-05,
+      "loss": 0.8085,
+      "step": 4087
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6978283881066636e-05,
+      "loss": 0.4874,
+      "step": 4088
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.697671430860231e-05,
+      "loss": 0.4686,
+      "step": 4089
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6975144401192957e-05,
+      "loss": 0.8076,
+      "step": 4090
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.697357415891395e-05,
+      "loss": 0.8077,
+      "step": 4091
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6972003581840675e-05,
+      "loss": 0.8175,
+      "step": 4092
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.697043267004853e-05,
+      "loss": 0.8132,
+      "step": 4093
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6968861423612937e-05,
+      "loss": 0.4387,
+      "step": 4094
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6967289842609327e-05,
+      "loss": 0.4934,
+      "step": 4095
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6965717927113156e-05,
+      "loss": 0.4983,
+      "step": 4096
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.696414567719988e-05,
+      "loss": 0.8092,
+      "step": 4097
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.696257309294499e-05,
+      "loss": 0.5102,
+      "step": 4098
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6961000174423978e-05,
+      "loss": 0.4187,
+      "step": 4099
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6959426921712366e-05,
+      "loss": 0.822,
+      "step": 4100
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6957853334885677e-05,
+      "loss": 0.7135,
+      "step": 4101
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.695627941401946e-05,
+      "loss": 0.8046,
+      "step": 4102
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.695470515918928e-05,
+      "loss": 0.4067,
+      "step": 4103
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6953130570470707e-05,
+      "loss": 0.4432,
+      "step": 4104
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6951555647939346e-05,
+      "loss": 0.796,
+      "step": 4105
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.69499803916708e-05,
+      "loss": 0.8326,
+      "step": 4106
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6948404801740703e-05,
+      "loss": 0.8328,
+      "step": 4107
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6946828878224683e-05,
+      "loss": 0.8732,
+      "step": 4108
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6945252621198416e-05,
+      "loss": 0.4049,
+      "step": 4109
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6943676030737565e-05,
+      "loss": 0.4167,
+      "step": 4110
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6942099106917824e-05,
+      "loss": 0.8493,
+      "step": 4111
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.69405218498149e-05,
+      "loss": 0.51,
+      "step": 4112
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6938944259504513e-05,
+      "loss": 0.8071,
+      "step": 4113
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6937366336062404e-05,
+      "loss": 0.7707,
+      "step": 4114
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.693578807956433e-05,
+      "loss": 0.8485,
+      "step": 4115
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.693420949008606e-05,
+      "loss": 0.7893,
+      "step": 4116
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6932630567703375e-05,
+      "loss": 0.4613,
+      "step": 4117
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6931051312492078e-05,
+      "loss": 0.8257,
+      "step": 4118
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6929471724527997e-05,
+      "loss": 0.8046,
+      "step": 4119
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.692789180388696e-05,
+      "loss": 0.4492,
+      "step": 4120
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6926311550644814e-05,
+      "loss": 0.7992,
+      "step": 4121
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6924730964877436e-05,
+      "loss": 0.8127,
+      "step": 4122
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6923150046660694e-05,
+      "loss": 0.4815,
+      "step": 4123
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.69215687960705e-05,
+      "loss": 0.8397,
+      "step": 4124
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6919987213182755e-05,
+      "loss": 0.4815,
+      "step": 4125
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6918405298073402e-05,
+      "loss": 0.4525,
+      "step": 4126
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.691682305081838e-05,
+      "loss": 0.7969,
+      "step": 4127
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.691524047149365e-05,
+      "loss": 0.846,
+      "step": 4128
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6913657560175197e-05,
+      "loss": 0.8294,
+      "step": 4129
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6912074316939006e-05,
+      "loss": 0.8514,
+      "step": 4130
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6910490741861094e-05,
+      "loss": 0.4582,
+      "step": 4131
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6908906835017483e-05,
+      "loss": 0.8349,
+      "step": 4132
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6907322596484217e-05,
+      "loss": 0.832,
+      "step": 4133
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.690573802633735e-05,
+      "loss": 0.4679,
+      "step": 4134
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.690415312465296e-05,
+      "loss": 0.8585,
+      "step": 4135
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6902567891507137e-05,
+      "loss": 0.4514,
+      "step": 4136
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6900982326975982e-05,
+      "loss": 0.8179,
+      "step": 4137
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6899396431135618e-05,
+      "loss": 0.4307,
+      "step": 4138
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6897810204062187e-05,
+      "loss": 0.8409,
+      "step": 4139
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6896223645831837e-05,
+      "loss": 0.8265,
+      "step": 4140
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6894636756520735e-05,
+      "loss": 0.4683,
+      "step": 4141
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.689304953620507e-05,
+      "loss": 0.8357,
+      "step": 4142
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.689146198496104e-05,
+      "loss": 0.8158,
+      "step": 4143
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6889874102864868e-05,
+      "loss": 0.5159,
+      "step": 4144
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.688828588999278e-05,
+      "loss": 0.7922,
+      "step": 4145
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.688669734642103e-05,
+      "loss": 0.4659,
+      "step": 4146
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6885108472225876e-05,
+      "loss": 0.8459,
+      "step": 4147
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.68835192674836e-05,
+      "loss": 0.816,
+      "step": 4148
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.68819297322705e-05,
+      "loss": 0.8437,
+      "step": 4149
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.688033986666289e-05,
+      "loss": 0.8205,
+      "step": 4150
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6878749670737092e-05,
+      "loss": 0.8371,
+      "step": 4151
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.687715914456946e-05,
+      "loss": 0.8459,
+      "step": 4152
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.687556828823634e-05,
+      "loss": 0.8058,
+      "step": 4153
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6873977101814118e-05,
+      "loss": 0.802,
+      "step": 4154
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6872385585379178e-05,
+      "loss": 0.8521,
+      "step": 4155
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.687079373900793e-05,
+      "loss": 0.4473,
+      "step": 4156
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6869201562776802e-05,
+      "loss": 0.7902,
+      "step": 4157
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.686760905676222e-05,
+      "loss": 0.8051,
+      "step": 4158
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.686601622104066e-05,
+      "loss": 0.8128,
+      "step": 4159
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6864423055688564e-05,
+      "loss": 0.8928,
+      "step": 4160
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6862829560782442e-05,
+      "loss": 0.8038,
+      "step": 4161
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6861235736398785e-05,
+      "loss": 0.4616,
+      "step": 4162
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6859641582614116e-05,
+      "loss": 0.7985,
+      "step": 4163
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6858047099504964e-05,
+      "loss": 0.7745,
+      "step": 4164
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.685645228714788e-05,
+      "loss": 0.443,
+      "step": 4165
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6854857145619434e-05,
+      "loss": 0.8275,
+      "step": 4166
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.68532616749962e-05,
+      "loss": 0.7955,
+      "step": 4167
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.685166587535478e-05,
+      "loss": 0.764,
+      "step": 4168
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6850069746771783e-05,
+      "loss": 0.8264,
+      "step": 4169
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6848473289323838e-05,
+      "loss": 0.8251,
+      "step": 4170
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6846876503087593e-05,
+      "loss": 0.863,
+      "step": 4171
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6845279388139704e-05,
+      "loss": 0.8097,
+      "step": 4172
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.684368194455685e-05,
+      "loss": 0.4666,
+      "step": 4173
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.684208417241572e-05,
+      "loss": 0.7666,
+      "step": 4174
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6840486071793023e-05,
+      "loss": 0.8494,
+      "step": 4175
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6838887642765484e-05,
+      "loss": 0.7966,
+      "step": 4176
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6837288885409838e-05,
+      "loss": 0.8788,
+      "step": 4177
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6835689799802843e-05,
+      "loss": 0.468,
+      "step": 4178
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6834090386021267e-05,
+      "loss": 0.8778,
+      "step": 4179
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6832490644141896e-05,
+      "loss": 0.7769,
+      "step": 4180
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6830890574241536e-05,
+      "loss": 0.4387,
+      "step": 4181
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6829290176397e-05,
+      "loss": 0.8252,
+      "step": 4182
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6827689450685125e-05,
+      "loss": 0.8893,
+      "step": 4183
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6826088397182757e-05,
+      "loss": 0.8084,
+      "step": 4184
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6824487015966763e-05,
+      "loss": 0.8334,
+      "step": 4185
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6822885307114025e-05,
+      "loss": 0.8378,
+      "step": 4186
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6821283270701435e-05,
+      "loss": 0.8129,
+      "step": 4187
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.681968090680591e-05,
+      "loss": 0.4444,
+      "step": 4188
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.681807821550438e-05,
+      "loss": 0.7834,
+      "step": 4189
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.681647519687378e-05,
+      "loss": 0.8264,
+      "step": 4190
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6814871850991076e-05,
+      "loss": 0.8194,
+      "step": 4191
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6813268177933236e-05,
+      "loss": 0.8588,
+      "step": 4192
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.681166417777726e-05,
+      "loss": 0.7633,
+      "step": 4193
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.681005985060015e-05,
+      "loss": 0.7922,
+      "step": 4194
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6808455196478925e-05,
+      "loss": 0.7955,
+      "step": 4195
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.680685021549063e-05,
+      "loss": 0.8793,
+      "step": 4196
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.680524490771231e-05,
+      "loss": 0.4793,
+      "step": 4197
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6803639273221044e-05,
+      "loss": 0.8416,
+      "step": 4198
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6802033312093907e-05,
+      "loss": 0.4344,
+      "step": 4199
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6800427024408006e-05,
+      "loss": 0.8674,
+      "step": 4200
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6798820410240456e-05,
+      "loss": 0.4591,
+      "step": 4201
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6797213469668386e-05,
+      "loss": 0.8361,
+      "step": 4202
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.679560620276895e-05,
+      "loss": 0.7692,
+      "step": 4203
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6793998609619306e-05,
+      "loss": 0.4514,
+      "step": 4204
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.679239069029663e-05,
+      "loss": 0.7781,
+      "step": 4205
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6790782444878125e-05,
+      "loss": 0.8354,
+      "step": 4206
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6789173873440993e-05,
+      "loss": 0.8131,
+      "step": 4207
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.678756497606247e-05,
+      "loss": 0.4956,
+      "step": 4208
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.678595575281979e-05,
+      "loss": 0.8264,
+      "step": 4209
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6784346203790205e-05,
+      "loss": 0.8056,
+      "step": 4210
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6782736329051e-05,
+      "loss": 0.8308,
+      "step": 4211
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6781126128679455e-05,
+      "loss": 0.8287,
+      "step": 4212
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.677951560275288e-05,
+      "loss": 0.8042,
+      "step": 4213
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.677790475134859e-05,
+      "loss": 0.4759,
+      "step": 4214
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6776293574543923e-05,
+      "loss": 0.4779,
+      "step": 4215
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.677468207241623e-05,
+      "loss": 0.7795,
+      "step": 4216
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6773070245042872e-05,
+      "loss": 0.8141,
+      "step": 4217
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6771458092501238e-05,
+      "loss": 0.815,
+      "step": 4218
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6769845614868728e-05,
+      "loss": 0.8313,
+      "step": 4219
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6768232812222747e-05,
+      "loss": 0.8251,
+      "step": 4220
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6766619684640727e-05,
+      "loss": 0.8271,
+      "step": 4221
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6765006232200116e-05,
+      "loss": 0.8504,
+      "step": 4222
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.676339245497837e-05,
+      "loss": 0.4516,
+      "step": 4223
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6761778353052967e-05,
+      "loss": 0.7702,
+      "step": 4224
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6760163926501404e-05,
+      "loss": 0.8134,
+      "step": 4225
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6758549175401177e-05,
+      "loss": 0.8371,
+      "step": 4226
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6756934099829814e-05,
+      "loss": 0.8485,
+      "step": 4227
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.675531869986485e-05,
+      "loss": 0.783,
+      "step": 4228
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.675370297558385e-05,
+      "loss": 0.8459,
+      "step": 4229
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6752086927064368e-05,
+      "loss": 0.7905,
+      "step": 4230
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6750470554383995e-05,
+      "loss": 0.772,
+      "step": 4231
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6748853857620334e-05,
+      "loss": 0.8533,
+      "step": 4232
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6747236836851e-05,
+      "loss": 0.8395,
+      "step": 4233
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6745619492153627e-05,
+      "loss": 0.4827,
+      "step": 4234
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6744001823605853e-05,
+      "loss": 0.8182,
+      "step": 4235
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6742383831285345e-05,
+      "loss": 0.8426,
+      "step": 4236
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.674076551526979e-05,
+      "loss": 0.857,
+      "step": 4237
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.673914687563687e-05,
+      "loss": 0.8456,
+      "step": 4238
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.67375279124643e-05,
+      "loss": 0.7992,
+      "step": 4239
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.67359086258298e-05,
+      "loss": 0.4387,
+      "step": 4240
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.673428901581112e-05,
+      "loss": 0.8619,
+      "step": 4241
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6732669082486005e-05,
+      "loss": 0.4215,
+      "step": 4242
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6731048825932236e-05,
+      "loss": 0.8339,
+      "step": 4243
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6729428246227585e-05,
+      "loss": 0.7886,
+      "step": 4244
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.672780734344987e-05,
+      "loss": 0.8276,
+      "step": 4245
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6726186117676906e-05,
+      "loss": 0.7982,
+      "step": 4246
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.672456456898652e-05,
+      "loss": 0.8339,
+      "step": 4247
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6722942697456565e-05,
+      "loss": 0.7982,
+      "step": 4248
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6721320503164906e-05,
+      "loss": 0.8206,
+      "step": 4249
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.671969798618942e-05,
+      "loss": 0.8352,
+      "step": 4250
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6718075146608007e-05,
+      "loss": 0.8062,
+      "step": 4251
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6716451984498572e-05,
+      "loss": 0.7865,
+      "step": 4252
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6714828499939043e-05,
+      "loss": 0.8027,
+      "step": 4253
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6713204693007367e-05,
+      "loss": 0.4334,
+      "step": 4254
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6711580563781497e-05,
+      "loss": 0.8189,
+      "step": 4255
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.670995611233941e-05,
+      "loss": 0.9021,
+      "step": 4256
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6708331338759083e-05,
+      "loss": 0.8424,
+      "step": 4257
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6706706243118534e-05,
+      "loss": 0.8104,
+      "step": 4258
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6705080825495768e-05,
+      "loss": 0.4776,
+      "step": 4259
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6703455085968837e-05,
+      "loss": 0.4629,
+      "step": 4260
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6701829024615773e-05,
+      "loss": 0.8066,
+      "step": 4261
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6700202641514655e-05,
+      "loss": 0.8148,
+      "step": 4262
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.669857593674356e-05,
+      "loss": 0.8041,
+      "step": 4263
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.669694891038058e-05,
+      "loss": 0.7784,
+      "step": 4264
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6695321562503833e-05,
+      "loss": 0.8477,
+      "step": 4265
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6693693893191442e-05,
+      "loss": 0.8183,
+      "step": 4266
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6692065902521554e-05,
+      "loss": 0.473,
+      "step": 4267
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.669043759057232e-05,
+      "loss": 0.4816,
+      "step": 4268
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.668880895742192e-05,
+      "loss": 0.8353,
+      "step": 4269
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.668718000314854e-05,
+      "loss": 0.8273,
+      "step": 4270
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6685550727830392e-05,
+      "loss": 0.8225,
+      "step": 4271
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6683921131545682e-05,
+      "loss": 0.828,
+      "step": 4272
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.668229121437266e-05,
+      "loss": 0.8049,
+      "step": 4273
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6680660976389568e-05,
+      "loss": 0.9177,
+      "step": 4274
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.667903041767467e-05,
+      "loss": 0.8532,
+      "step": 4275
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6677399538306255e-05,
+      "loss": 0.8721,
+      "step": 4276
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.667576833836261e-05,
+      "loss": 0.8483,
+      "step": 4277
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6674136817922063e-05,
+      "loss": 0.7713,
+      "step": 4278
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6672504977062924e-05,
+      "loss": 0.7897,
+      "step": 4279
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6670872815863548e-05,
+      "loss": 0.8093,
+      "step": 4280
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6669240334402288e-05,
+      "loss": 0.7857,
+      "step": 4281
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.666760753275752e-05,
+      "loss": 0.8643,
+      "step": 4282
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.666597441100763e-05,
+      "loss": 0.4915,
+      "step": 4283
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6664340969231028e-05,
+      "loss": 0.8137,
+      "step": 4284
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.666270720750613e-05,
+      "loss": 0.8318,
+      "step": 4285
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6661073125911368e-05,
+      "loss": 0.7827,
+      "step": 4286
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6659438724525202e-05,
+      "loss": 0.425,
+      "step": 4287
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.665780400342609e-05,
+      "loss": 0.4397,
+      "step": 4288
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.665616896269252e-05,
+      "loss": 0.8229,
+      "step": 4289
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.665453360240298e-05,
+      "loss": 0.7664,
+      "step": 4290
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6652897922635987e-05,
+      "loss": 0.8246,
+      "step": 4291
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.665126192347007e-05,
+      "loss": 0.7998,
+      "step": 4292
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6649625604983768e-05,
+      "loss": 0.7993,
+      "step": 4293
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6647988967255638e-05,
+      "loss": 0.4702,
+      "step": 4294
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.664635201036426e-05,
+      "loss": 0.8003,
+      "step": 4295
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.664471473438822e-05,
+      "loss": 0.4472,
+      "step": 4296
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6643077139406116e-05,
+      "loss": 0.7588,
+      "step": 4297
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6641439225496576e-05,
+      "loss": 0.8379,
+      "step": 4298
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6639800992738226e-05,
+      "loss": 0.7941,
+      "step": 4299
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6638162441209727e-05,
+      "loss": 0.851,
+      "step": 4300
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.663652357098973e-05,
+      "loss": 0.7902,
+      "step": 4301
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6634884382156927e-05,
+      "loss": 0.8035,
+      "step": 4302
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6633244874790013e-05,
+      "loss": 0.7887,
+      "step": 4303
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.663160504896769e-05,
+      "loss": 0.4605,
+      "step": 4304
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.662996490476869e-05,
+      "loss": 0.8318,
+      "step": 4305
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.662832444227176e-05,
+      "loss": 0.7979,
+      "step": 4306
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6626683661555647e-05,
+      "loss": 0.4513,
+      "step": 4307
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6625042562699134e-05,
+      "loss": 0.416,
+      "step": 4308
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6623401145781e-05,
+      "loss": 0.806,
+      "step": 4309
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.662175941088005e-05,
+      "loss": 0.8247,
+      "step": 4310
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.66201173580751e-05,
+      "loss": 0.8338,
+      "step": 4311
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.661847498744499e-05,
+      "loss": 0.4707,
+      "step": 4312
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.661683229906856e-05,
+      "loss": 0.7907,
+      "step": 4313
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.661518929302468e-05,
+      "loss": 0.8161,
+      "step": 4314
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6613545969392233e-05,
+      "loss": 0.8121,
+      "step": 4315
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.66119023282501e-05,
+      "loss": 0.7561,
+      "step": 4316
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.66102583696772e-05,
+      "loss": 0.4539,
+      "step": 4317
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.660861409375246e-05,
+      "loss": 0.7998,
+      "step": 4318
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6606969500554812e-05,
+      "loss": 0.8449,
+      "step": 4319
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6605324590163214e-05,
+      "loss": 0.7991,
+      "step": 4320
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6603679362656642e-05,
+      "loss": 0.854,
+      "step": 4321
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6602033818114075e-05,
+      "loss": 0.8492,
+      "step": 4322
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6600387956614517e-05,
+      "loss": 0.8155,
+      "step": 4323
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6598741778236982e-05,
+      "loss": 0.7725,
+      "step": 4324
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6597095283060505e-05,
+      "loss": 0.7742,
+      "step": 4325
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6595448471164125e-05,
+      "loss": 0.8086,
+      "step": 4326
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6593801342626914e-05,
+      "loss": 0.7995,
+      "step": 4327
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6592153897527944e-05,
+      "loss": 0.7861,
+      "step": 4328
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.659050613594631e-05,
+      "loss": 0.7993,
+      "step": 4329
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.658885805796111e-05,
+      "loss": 0.7689,
+      "step": 4330
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.658720966365148e-05,
+      "loss": 0.5053,
+      "step": 4331
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.658556095309655e-05,
+      "loss": 0.7556,
+      "step": 4332
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6583911926375468e-05,
+      "loss": 0.8085,
+      "step": 4333
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6582262583567413e-05,
+      "loss": 0.8482,
+      "step": 4334
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6580612924751562e-05,
+      "loss": 0.8652,
+      "step": 4335
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6578962950007116e-05,
+      "loss": 0.428,
+      "step": 4336
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6577312659413287e-05,
+      "loss": 0.796,
+      "step": 4337
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6575662053049305e-05,
+      "loss": 0.8041,
+      "step": 4338
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6574011130994415e-05,
+      "loss": 0.4616,
+      "step": 4339
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.657235989332787e-05,
+      "loss": 0.8169,
+      "step": 4340
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6570708340128952e-05,
+      "loss": 0.8067,
+      "step": 4341
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6569056471476943e-05,
+      "loss": 0.776,
+      "step": 4342
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6567404287451155e-05,
+      "loss": 0.838,
+      "step": 4343
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6565751788130905e-05,
+      "loss": 0.481,
+      "step": 4344
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6564098973595523e-05,
+      "loss": 0.5048,
+      "step": 4345
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6562445843924366e-05,
+      "loss": 0.8085,
+      "step": 4346
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6560792399196793e-05,
+      "loss": 0.4564,
+      "step": 4347
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6559138639492195e-05,
+      "loss": 0.3974,
+      "step": 4348
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.655748456488995e-05,
+      "loss": 0.8454,
+      "step": 4349
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6555830175469486e-05,
+      "loss": 0.8516,
+      "step": 4350
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6554175471310216e-05,
+      "loss": 0.757,
+      "step": 4351
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6552520452491586e-05,
+      "loss": 0.8399,
+      "step": 4352
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.655086511909305e-05,
+      "loss": 0.7615,
+      "step": 4353
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6549209471194083e-05,
+      "loss": 0.4414,
+      "step": 4354
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6547553508874168e-05,
+      "loss": 0.8662,
+      "step": 4355
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6545897232212804e-05,
+      "loss": 0.7907,
+      "step": 4356
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6544240641289508e-05,
+      "loss": 0.8049,
+      "step": 4357
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6542583736183816e-05,
+      "loss": 0.803,
+      "step": 4358
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6540926516975267e-05,
+      "loss": 0.7859,
+      "step": 4359
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6539268983743433e-05,
+      "loss": 0.4193,
+      "step": 4360
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.653761113656788e-05,
+      "loss": 0.8445,
+      "step": 4361
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.65359529755282e-05,
+      "loss": 0.8489,
+      "step": 4362
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6534294500704007e-05,
+      "loss": 0.7951,
+      "step": 4363
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6532635712174918e-05,
+      "loss": 0.438,
+      "step": 4364
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.653097661002057e-05,
+      "loss": 0.4748,
+      "step": 4365
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6529317194320614e-05,
+      "loss": 0.4495,
+      "step": 4366
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6527657465154723e-05,
+      "loss": 0.866,
+      "step": 4367
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6525997422602568e-05,
+      "loss": 0.8112,
+      "step": 4368
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6524337066743857e-05,
+      "loss": 0.4596,
+      "step": 4369
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6522676397658294e-05,
+      "loss": 0.7716,
+      "step": 4370
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6521015415425612e-05,
+      "loss": 0.8211,
+      "step": 4371
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.651935412012555e-05,
+      "loss": 0.4806,
+      "step": 4372
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.651769251183786e-05,
+      "loss": 0.8587,
+      "step": 4373
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6516030590642323e-05,
+      "loss": 0.8404,
+      "step": 4374
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6514368356618728e-05,
+      "loss": 0.82,
+      "step": 4375
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6512705809846865e-05,
+      "loss": 0.4685,
+      "step": 4376
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6511042950406556e-05,
+      "loss": 0.7708,
+      "step": 4377
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.650937977837764e-05,
+      "loss": 0.8588,
+      "step": 4378
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.650771629383996e-05,
+      "loss": 0.8415,
+      "step": 4379
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6506052496873376e-05,
+      "loss": 0.7563,
+      "step": 4380
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6504388387557768e-05,
+      "loss": 0.4412,
+      "step": 4381
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6502723965973026e-05,
+      "loss": 0.8084,
+      "step": 4382
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6501059232199058e-05,
+      "loss": 0.789,
+      "step": 4383
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6499394186315786e-05,
+      "loss": 0.8118,
+      "step": 4384
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6497728828403148e-05,
+      "loss": 0.8468,
+      "step": 4385
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6496063158541096e-05,
+      "loss": 0.4255,
+      "step": 4386
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6494397176809597e-05,
+      "loss": 0.8123,
+      "step": 4387
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6492730883288635e-05,
+      "loss": 0.784,
+      "step": 4388
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6491064278058203e-05,
+      "loss": 0.4683,
+      "step": 4389
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.648939736119832e-05,
+      "loss": 0.8564,
+      "step": 4390
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6487730132789003e-05,
+      "loss": 0.8345,
+      "step": 4391
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6486062592910297e-05,
+      "loss": 0.8556,
+      "step": 4392
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6484394741642263e-05,
+      "loss": 0.4282,
+      "step": 4393
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6482726579064972e-05,
+      "loss": 0.7648,
+      "step": 4394
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.648105810525851e-05,
+      "loss": 0.8161,
+      "step": 4395
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6479389320302978e-05,
+      "loss": 0.8023,
+      "step": 4396
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6477720224278493e-05,
+      "loss": 0.7918,
+      "step": 4397
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6476050817265187e-05,
+      "loss": 0.8055,
+      "step": 4398
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6474381099343206e-05,
+      "loss": 0.8728,
+      "step": 4399
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6472711070592714e-05,
+      "loss": 0.4578,
+      "step": 4400
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.647104073109388e-05,
+      "loss": 0.4941,
+      "step": 4401
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6469370080926907e-05,
+      "loss": 0.8604,
+      "step": 4402
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.646769912017199e-05,
+      "loss": 0.4395,
+      "step": 4403
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6466027848909354e-05,
+      "loss": 0.8072,
+      "step": 4404
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6464356267219238e-05,
+      "loss": 0.8644,
+      "step": 4405
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.646268437518189e-05,
+      "loss": 0.8313,
+      "step": 4406
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6461012172877578e-05,
+      "loss": 0.8274,
+      "step": 4407
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.645933966038658e-05,
+      "loss": 0.7839,
+      "step": 4408
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6457666837789196e-05,
+      "loss": 0.8512,
+      "step": 4409
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.645599370516573e-05,
+      "loss": 0.4461,
+      "step": 4410
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.645432026259651e-05,
+      "loss": 0.7992,
+      "step": 4411
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6452646510161884e-05,
+      "loss": 0.9009,
+      "step": 4412
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6450972447942196e-05,
+      "loss": 0.4567,
+      "step": 4413
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6449298076017822e-05,
+      "loss": 0.8591,
+      "step": 4414
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6447623394469143e-05,
+      "loss": 0.7729,
+      "step": 4415
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6445948403376564e-05,
+      "loss": 0.8104,
+      "step": 4416
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6444273102820493e-05,
+      "loss": 0.7508,
+      "step": 4417
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6442597492881366e-05,
+      "loss": 0.4508,
+      "step": 4418
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6440921573639624e-05,
+      "loss": 0.4955,
+      "step": 4419
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.643924534517573e-05,
+      "loss": 0.7653,
+      "step": 4420
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6437568807570152e-05,
+      "loss": 0.786,
+      "step": 4421
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6435891960903383e-05,
+      "loss": 0.8431,
+      "step": 4422
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6434214805255925e-05,
+      "loss": 0.4426,
+      "step": 4423
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6432537340708297e-05,
+      "loss": 0.8043,
+      "step": 4424
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6430859567341028e-05,
+      "loss": 0.7942,
+      "step": 4425
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.642918148523468e-05,
+      "loss": 0.8379,
+      "step": 4426
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.64275030944698e-05,
+      "loss": 0.8452,
+      "step": 4427
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6425824395126975e-05,
+      "loss": 0.8123,
+      "step": 4428
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.642414538728679e-05,
+      "loss": 0.8028,
+      "step": 4429
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6422466071029863e-05,
+      "loss": 0.804,
+      "step": 4430
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.642078644643681e-05,
+      "loss": 0.8089,
+      "step": 4431
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.641910651358827e-05,
+      "loss": 0.7994,
+      "step": 4432
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6417426272564893e-05,
+      "loss": 0.7919,
+      "step": 4433
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6415745723447346e-05,
+      "loss": 0.4406,
+      "step": 4434
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6414064866316314e-05,
+      "loss": 0.7725,
+      "step": 4435
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6412383701252488e-05,
+      "loss": 0.8173,
+      "step": 4436
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.641070222833658e-05,
+      "loss": 0.4448,
+      "step": 4437
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6409020447649316e-05,
+      "loss": 0.8379,
+      "step": 4438
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6407338359271442e-05,
+      "loss": 0.8452,
+      "step": 4439
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6405655963283708e-05,
+      "loss": 0.8444,
+      "step": 4440
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6403973259766882e-05,
+      "loss": 0.8138,
+      "step": 4441
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.640229024880176e-05,
+      "loss": 0.8569,
+      "step": 4442
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6400606930469125e-05,
+      "loss": 0.8057,
+      "step": 4443
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6398923304849807e-05,
+      "loss": 0.8029,
+      "step": 4444
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6397239372024625e-05,
+      "loss": 0.7843,
+      "step": 4445
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6395555132074425e-05,
+      "loss": 0.4715,
+      "step": 4446
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.639387058508007e-05,
+      "loss": 0.8154,
+      "step": 4447
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.639218573112243e-05,
+      "loss": 0.7948,
+      "step": 4448
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6390500570282393e-05,
+      "loss": 0.8187,
+      "step": 4449
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6388815102640863e-05,
+      "loss": 0.8371,
+      "step": 4450
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6387129328278756e-05,
+      "loss": 0.8194,
+      "step": 4451
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6385443247277003e-05,
+      "loss": 0.8207,
+      "step": 4452
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6383756859716558e-05,
+      "loss": 0.8145,
+      "step": 4453
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6382070165678375e-05,
+      "loss": 0.8288,
+      "step": 4454
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6380383165243436e-05,
+      "loss": 0.4528,
+      "step": 4455
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.637869585849273e-05,
+      "loss": 0.8233,
+      "step": 4456
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6377008245507263e-05,
+      "loss": 0.7937,
+      "step": 4457
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6375320326368053e-05,
+      "loss": 0.7786,
+      "step": 4458
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6373632101156137e-05,
+      "loss": 0.8049,
+      "step": 4459
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.637194356995257e-05,
+      "loss": 0.7704,
+      "step": 4460
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6370254732838407e-05,
+      "loss": 0.8196,
+      "step": 4461
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6368565589894736e-05,
+      "loss": 0.8112,
+      "step": 4462
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6366876141202647e-05,
+      "loss": 0.7829,
+      "step": 4463
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.636518638684325e-05,
+      "loss": 0.4277,
+      "step": 4464
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.636349632689767e-05,
+      "loss": 0.8458,
+      "step": 4465
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.636180596144704e-05,
+      "loss": 0.7678,
+      "step": 4466
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6360115290572515e-05,
+      "loss": 0.8137,
+      "step": 4467
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6358424314355263e-05,
+      "loss": 0.8517,
+      "step": 4468
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6356733032876468e-05,
+      "loss": 0.8344,
+      "step": 4469
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6355041446217327e-05,
+      "loss": 0.8194,
+      "step": 4470
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6353349554459045e-05,
+      "loss": 0.4638,
+      "step": 4471
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6351657357682854e-05,
+      "loss": 0.8161,
+      "step": 4472
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6349964855969994e-05,
+      "loss": 0.805,
+      "step": 4473
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6348272049401717e-05,
+      "loss": 0.8022,
+      "step": 4474
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6346578938059296e-05,
+      "loss": 0.4227,
+      "step": 4475
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6344885522024016e-05,
+      "loss": 0.8282,
+      "step": 4476
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6343191801377175e-05,
+      "loss": 0.8198,
+      "step": 4477
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6341497776200082e-05,
+      "loss": 0.464,
+      "step": 4478
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6339803446574076e-05,
+      "loss": 0.7997,
+      "step": 4479
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6338108812580493e-05,
+      "loss": 0.8185,
+      "step": 4480
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.633641387430069e-05,
+      "loss": 0.8276,
+      "step": 4481
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6334718631816042e-05,
+      "loss": 0.8156,
+      "step": 4482
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6333023085207937e-05,
+      "loss": 0.8354,
+      "step": 4483
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6331327234557773e-05,
+      "loss": 0.8708,
+      "step": 4484
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6329631079946966e-05,
+      "loss": 0.8013,
+      "step": 4485
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6327934621456953e-05,
+      "loss": 0.8321,
+      "step": 4486
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6326237859169172e-05,
+      "loss": 0.8336,
+      "step": 4487
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6324540793165086e-05,
+      "loss": 0.8356,
+      "step": 4488
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6322843423526167e-05,
+      "loss": 0.8385,
+      "step": 4489
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6321145750333906e-05,
+      "loss": 0.4553,
+      "step": 4490
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.631944777366981e-05,
+      "loss": 0.8092,
+      "step": 4491
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6317749493615388e-05,
+      "loss": 0.7807,
+      "step": 4492
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.631605091025218e-05,
+      "loss": 0.7889,
+      "step": 4493
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6314352023661736e-05,
+      "loss": 0.8303,
+      "step": 4494
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.631265283392561e-05,
+      "loss": 0.752,
+      "step": 4495
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6310953341125384e-05,
+      "loss": 0.7939,
+      "step": 4496
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6309253545342645e-05,
+      "loss": 0.8338,
+      "step": 4497
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6307553446659e-05,
+      "loss": 0.822,
+      "step": 4498
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.630585304515607e-05,
+      "loss": 0.4616,
+      "step": 4499
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6304152340915485e-05,
+      "loss": 0.7815,
+      "step": 4500
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6302451334018906e-05,
+      "loss": 0.7888,
+      "step": 4501
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6300750024547985e-05,
+      "loss": 0.7816,
+      "step": 4502
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.62990484125844e-05,
+      "loss": 0.8801,
+      "step": 4503
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.629734649820985e-05,
+      "loss": 0.4213,
+      "step": 4504
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.629564428150604e-05,
+      "loss": 0.4761,
+      "step": 4505
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.629394176255469e-05,
+      "loss": 0.7881,
+      "step": 4506
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6292238941437538e-05,
+      "loss": 0.4776,
+      "step": 4507
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6290535818236336e-05,
+      "loss": 0.7951,
+      "step": 4508
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.628883239303285e-05,
+      "loss": 0.8338,
+      "step": 4509
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.628712866590885e-05,
+      "loss": 0.8037,
+      "step": 4510
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.628542463694614e-05,
+      "loss": 0.4947,
+      "step": 4511
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6283720306226526e-05,
+      "loss": 0.4372,
+      "step": 4512
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6282015673831835e-05,
+      "loss": 0.5008,
+      "step": 4513
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6280310739843897e-05,
+      "loss": 0.4835,
+      "step": 4514
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.627860550434457e-05,
+      "loss": 0.8164,
+      "step": 4515
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.627689996741572e-05,
+      "loss": 0.8186,
+      "step": 4516
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6275194129139226e-05,
+      "loss": 0.4451,
+      "step": 4517
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6273487989596982e-05,
+      "loss": 0.8246,
+      "step": 4518
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6271781548870903e-05,
+      "loss": 0.7745,
+      "step": 4519
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.627007480704291e-05,
+      "loss": 0.8114,
+      "step": 4520
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6268367764194942e-05,
+      "loss": 0.436,
+      "step": 4521
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6266660420408956e-05,
+      "loss": 0.8147,
+      "step": 4522
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6264952775766915e-05,
+      "loss": 0.8429,
+      "step": 4523
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6263244830350807e-05,
+      "loss": 0.8655,
+      "step": 4524
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6261536584242624e-05,
+      "loss": 0.8006,
+      "step": 4525
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6259828037524376e-05,
+      "loss": 0.8298,
+      "step": 4526
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6258119190278093e-05,
+      "loss": 0.8048,
+      "step": 4527
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.625641004258581e-05,
+      "loss": 0.8277,
+      "step": 4528
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6254700594529585e-05,
+      "loss": 0.8105,
+      "step": 4529
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6252990846191492e-05,
+      "loss": 0.8482,
+      "step": 4530
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6251280797653606e-05,
+      "loss": 0.808,
+      "step": 4531
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.624957044899803e-05,
+      "loss": 0.7725,
+      "step": 4532
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6247859800306868e-05,
+      "loss": 0.8202,
+      "step": 4533
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6246148851662255e-05,
+      "loss": 0.8431,
+      "step": 4534
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6244437603146325e-05,
+      "loss": 0.7988,
+      "step": 4535
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6242726054841244e-05,
+      "loss": 0.4498,
+      "step": 4536
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.624101420682917e-05,
+      "loss": 0.4669,
+      "step": 4537
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6239302059192293e-05,
+      "loss": 0.8445,
+      "step": 4538
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6237589612012812e-05,
+      "loss": 0.8428,
+      "step": 4539
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6235876865372938e-05,
+      "loss": 0.8002,
+      "step": 4540
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6234163819354902e-05,
+      "loss": 0.8059,
+      "step": 4541
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.623245047404094e-05,
+      "loss": 0.8297,
+      "step": 4542
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6230736829513306e-05,
+      "loss": 0.8486,
+      "step": 4543
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.622902288585428e-05,
+      "loss": 0.8315,
+      "step": 4544
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6227308643146142e-05,
+      "loss": 0.8193,
+      "step": 4545
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6225594101471186e-05,
+      "loss": 0.4533,
+      "step": 4546
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6223879260911733e-05,
+      "loss": 0.8592,
+      "step": 4547
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6222164121550106e-05,
+      "loss": 0.8142,
+      "step": 4548
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6220448683468652e-05,
+      "loss": 0.8223,
+      "step": 4549
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6218732946749717e-05,
+      "loss": 0.4438,
+      "step": 4550
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6217016911475686e-05,
+      "loss": 0.8267,
+      "step": 4551
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6215300577728932e-05,
+      "loss": 0.8966,
+      "step": 4552
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6213583945591865e-05,
+      "loss": 0.8814,
+      "step": 4553
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.621186701514689e-05,
+      "loss": 0.8337,
+      "step": 4554
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6210149786476437e-05,
+      "loss": 0.7665,
+      "step": 4555
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6208432259662954e-05,
+      "loss": 0.8535,
+      "step": 4556
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.620671443478889e-05,
+      "loss": 0.8415,
+      "step": 4557
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.620499631193672e-05,
+      "loss": 0.7757,
+      "step": 4558
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6203277891188932e-05,
+      "loss": 0.4372,
+      "step": 4559
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6201559172628017e-05,
+      "loss": 0.8141,
+      "step": 4560
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.61998401563365e-05,
+      "loss": 0.8233,
+      "step": 4561
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6198120842396904e-05,
+      "loss": 0.7745,
+      "step": 4562
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6196401230891768e-05,
+      "loss": 0.8711,
+      "step": 4563
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6194681321903654e-05,
+      "loss": 0.4975,
+      "step": 4564
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.619296111551513e-05,
+      "loss": 0.5076,
+      "step": 4565
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6191240611808785e-05,
+      "loss": 0.4758,
+      "step": 4566
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6189519810867212e-05,
+      "loss": 0.7601,
+      "step": 4567
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6187798712773037e-05,
+      "loss": 0.407,
+      "step": 4568
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6186077317608872e-05,
+      "loss": 0.8564,
+      "step": 4569
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6184355625457372e-05,
+      "loss": 0.794,
+      "step": 4570
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.618263363640119e-05,
+      "loss": 0.8584,
+      "step": 4571
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6180911350522996e-05,
+      "loss": 0.4658,
+      "step": 4572
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6179188767905474e-05,
+      "loss": 0.4522,
+      "step": 4573
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6177465888631325e-05,
+      "loss": 0.8207,
+      "step": 4574
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6175742712783263e-05,
+      "loss": 0.4405,
+      "step": 4575
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6174019240444017e-05,
+      "loss": 0.8398,
+      "step": 4576
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.617229547169633e-05,
+      "loss": 0.81,
+      "step": 4577
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6170571406622952e-05,
+      "loss": 0.4333,
+      "step": 4578
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.616884704530666e-05,
+      "loss": 0.8104,
+      "step": 4579
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6167122387830237e-05,
+      "loss": 0.4671,
+      "step": 4580
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6165397434276483e-05,
+      "loss": 0.4702,
+      "step": 4581
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.616367218472821e-05,
+      "loss": 0.7858,
+      "step": 4582
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6161946639268245e-05,
+      "loss": 0.4804,
+      "step": 4583
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.616022079797943e-05,
+      "loss": 0.8285,
+      "step": 4584
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6158494660944618e-05,
+      "loss": 0.4251,
+      "step": 4585
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6156768228246685e-05,
+      "loss": 0.468,
+      "step": 4586
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6155041499968515e-05,
+      "loss": 0.8358,
+      "step": 4587
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6153314476193004e-05,
+      "loss": 0.8153,
+      "step": 4588
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6151587157003063e-05,
+      "loss": 0.8204,
+      "step": 4589
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.614985954248162e-05,
+      "loss": 0.8183,
+      "step": 4590
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.614813163271162e-05,
+      "loss": 0.8183,
+      "step": 4591
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6146403427776013e-05,
+      "loss": 0.8181,
+      "step": 4592
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.614467492775777e-05,
+      "loss": 0.7803,
+      "step": 4593
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6142946132739875e-05,
+      "loss": 0.7942,
+      "step": 4594
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6141217042805325e-05,
+      "loss": 0.8188,
+      "step": 4595
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6139487658037133e-05,
+      "loss": 0.841,
+      "step": 4596
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6137757978518325e-05,
+      "loss": 0.8128,
+      "step": 4597
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.613602800433194e-05,
+      "loss": 0.8186,
+      "step": 4598
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6134297735561034e-05,
+      "loss": 0.8209,
+      "step": 4599
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6132567172288675e-05,
+      "loss": 0.4751,
+      "step": 4600
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6130836314597943e-05,
+      "loss": 0.8026,
+      "step": 4601
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6129105162571933e-05,
+      "loss": 0.7839,
+      "step": 4602
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6127373716293764e-05,
+      "loss": 0.8484,
+      "step": 4603
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.612564197584656e-05,
+      "loss": 0.8376,
+      "step": 4604
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6123909941313455e-05,
+      "loss": 0.4453,
+      "step": 4605
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.61221776127776e-05,
+      "loss": 0.7913,
+      "step": 4606
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.612044499032217e-05,
+      "loss": 0.7817,
+      "step": 4607
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.611871207403034e-05,
+      "loss": 0.7739,
+      "step": 4608
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.611697886398531e-05,
+      "loss": 0.436,
+      "step": 4609
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6115245360270283e-05,
+      "loss": 0.852,
+      "step": 4610
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6113511562968493e-05,
+      "loss": 0.8042,
+      "step": 4611
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6111777472163172e-05,
+      "loss": 0.8178,
+      "step": 4612
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6110043087937576e-05,
+      "loss": 0.8385,
+      "step": 4613
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6108308410374962e-05,
+      "loss": 0.8324,
+      "step": 4614
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.610657343955862e-05,
+      "loss": 0.4996,
+      "step": 4615
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6104838175571837e-05,
+      "loss": 0.4882,
+      "step": 4616
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6103102618497922e-05,
+      "loss": 0.8092,
+      "step": 4617
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.61013667684202e-05,
+      "loss": 0.8261,
+      "step": 4618
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6099630625422013e-05,
+      "loss": 0.8074,
+      "step": 4619
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6097894189586703e-05,
+      "loss": 0.4332,
+      "step": 4620
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6096157460997633e-05,
+      "loss": 0.7979,
+      "step": 4621
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.609442043973819e-05,
+      "loss": 0.4655,
+      "step": 4622
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.609268312589176e-05,
+      "loss": 0.4305,
+      "step": 4623
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.609094551954175e-05,
+      "loss": 0.8231,
+      "step": 4624
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6089207620771586e-05,
+      "loss": 0.8284,
+      "step": 4625
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6087469429664696e-05,
+      "loss": 0.8008,
+      "step": 4626
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6085730946304537e-05,
+      "loss": 0.8384,
+      "step": 4627
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6083992170774567e-05,
+      "loss": 0.8057,
+      "step": 4628
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6082253103158257e-05,
+      "loss": 0.8025,
+      "step": 4629
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.608051374353911e-05,
+      "loss": 0.8382,
+      "step": 4630
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.607877409200062e-05,
+      "loss": 0.4572,
+      "step": 4631
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.607703414862631e-05,
+      "loss": 0.4684,
+      "step": 4632
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6075293913499717e-05,
+      "loss": 0.4465,
+      "step": 4633
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6073553386704384e-05,
+      "loss": 0.8364,
+      "step": 4634
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.607181256832387e-05,
+      "loss": 0.8347,
+      "step": 4635
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6070071458441757e-05,
+      "loss": 0.4219,
+      "step": 4636
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6068330057141624e-05,
+      "loss": 0.4123,
+      "step": 4637
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6066588364507083e-05,
+      "loss": 0.4439,
+      "step": 4638
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6064846380621744e-05,
+      "loss": 0.7817,
+      "step": 4639
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6063104105569243e-05,
+      "loss": 0.4416,
+      "step": 4640
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6061361539433222e-05,
+      "loss": 0.7801,
+      "step": 4641
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6059618682297342e-05,
+      "loss": 0.8099,
+      "step": 4642
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6057875534245275e-05,
+      "loss": 0.8462,
+      "step": 4643
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6056132095360705e-05,
+      "loss": 0.7448,
+      "step": 4644
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6054388365727333e-05,
+      "loss": 0.7897,
+      "step": 4645
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6052644345428882e-05,
+      "loss": 0.7682,
+      "step": 4646
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6050900034549067e-05,
+      "loss": 0.8034,
+      "step": 4647
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6049155433171645e-05,
+      "loss": 0.8265,
+      "step": 4648
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6047410541380362e-05,
+      "loss": 0.4597,
+      "step": 4649
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6045665359258994e-05,
+      "loss": 0.7861,
+      "step": 4650
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6043919886891322e-05,
+      "loss": 0.7671,
+      "step": 4651
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6042174124361146e-05,
+      "loss": 0.8254,
+      "step": 4652
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6040428071752276e-05,
+      "loss": 0.8303,
+      "step": 4653
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6038681729148544e-05,
+      "loss": 0.7562,
+      "step": 4654
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6036935096633782e-05,
+      "loss": 0.8022,
+      "step": 4655
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6035188174291854e-05,
+      "loss": 0.8003,
+      "step": 4656
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.603344096220662e-05,
+      "loss": 0.7987,
+      "step": 4657
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6031693460461966e-05,
+      "loss": 0.8125,
+      "step": 4658
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6029945669141785e-05,
+      "loss": 0.8018,
+      "step": 4659
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6028197588329985e-05,
+      "loss": 0.8118,
+      "step": 4660
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6026449218110496e-05,
+      "loss": 0.8086,
+      "step": 4661
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6024700558567253e-05,
+      "loss": 0.5102,
+      "step": 4662
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6022951609784204e-05,
+      "loss": 0.7593,
+      "step": 4663
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.602120237184532e-05,
+      "loss": 0.8835,
+      "step": 4664
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.601945284483457e-05,
+      "loss": 0.8226,
+      "step": 4665
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6017703028835962e-05,
+      "loss": 0.8281,
+      "step": 4666
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.601595292393349e-05,
+      "loss": 0.7813,
+      "step": 4667
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6014202530211182e-05,
+      "loss": 0.814,
+      "step": 4668
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6012451847753066e-05,
+      "loss": 0.7996,
+      "step": 4669
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.60107008766432e-05,
+      "loss": 0.8034,
+      "step": 4670
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6008949616965635e-05,
+      "loss": 0.7841,
+      "step": 4671
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.600719806880446e-05,
+      "loss": 0.8278,
+      "step": 4672
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6005446232243756e-05,
+      "loss": 0.8661,
+      "step": 4673
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6003694107367627e-05,
+      "loss": 0.4495,
+      "step": 4674
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6001941694260197e-05,
+      "loss": 0.7631,
+      "step": 4675
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.600018899300559e-05,
+      "loss": 0.8262,
+      "step": 4676
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5998436003687955e-05,
+      "loss": 0.78,
+      "step": 4677
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5996682726391454e-05,
+      "loss": 0.45,
+      "step": 4678
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5994929161200254e-05,
+      "loss": 0.8233,
+      "step": 4679
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5993175308198548e-05,
+      "loss": 0.8344,
+      "step": 4680
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5991421167470535e-05,
+      "loss": 0.7978,
+      "step": 4681
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5989666739100425e-05,
+      "loss": 0.4777,
+      "step": 4682
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5987912023172452e-05,
+      "loss": 0.4315,
+      "step": 4683
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5986157019770855e-05,
+      "loss": 0.8053,
+      "step": 4684
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.598440172897989e-05,
+      "loss": 0.7825,
+      "step": 4685
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.598264615088383e-05,
+      "loss": 0.433,
+      "step": 4686
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5980890285566957e-05,
+      "loss": 0.8021,
+      "step": 4687
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5979134133113563e-05,
+      "loss": 0.4485,
+      "step": 4688
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5977377693607966e-05,
+      "loss": 0.4354,
+      "step": 4689
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.597562096713449e-05,
+      "loss": 0.81,
+      "step": 4690
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5973863953777467e-05,
+      "loss": 0.8277,
+      "step": 4691
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.597210665362126e-05,
+      "loss": 0.8397,
+      "step": 4692
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.597034906675023e-05,
+      "loss": 0.438,
+      "step": 4693
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.596859119324875e-05,
+      "loss": 0.8136,
+      "step": 4694
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.596683303320123e-05,
+      "loss": 0.8342,
+      "step": 4695
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5965074586692065e-05,
+      "loss": 0.797,
+      "step": 4696
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5963315853805678e-05,
+      "loss": 0.7949,
+      "step": 4697
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5961556834626503e-05,
+      "loss": 0.7824,
+      "step": 4698
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5959797529238994e-05,
+      "loss": 0.7708,
+      "step": 4699
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.595803793772761e-05,
+      "loss": 0.7726,
+      "step": 4700
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5956278060176827e-05,
+      "loss": 0.8426,
+      "step": 4701
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.595451789667114e-05,
+      "loss": 0.826,
+      "step": 4702
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.595275744729504e-05,
+      "loss": 0.8094,
+      "step": 4703
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5950996712133056e-05,
+      "loss": 0.801,
+      "step": 4704
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5949235691269716e-05,
+      "loss": 0.4095,
+      "step": 4705
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5947474384789566e-05,
+      "loss": 0.8291,
+      "step": 4706
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5945712792777165e-05,
+      "loss": 0.8399,
+      "step": 4707
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5943950915317083e-05,
+      "loss": 0.8339,
+      "step": 4708
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5942188752493904e-05,
+      "loss": 0.4355,
+      "step": 4709
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5940426304392232e-05,
+      "loss": 0.8452,
+      "step": 4710
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5938663571096677e-05,
+      "loss": 0.8598,
+      "step": 4711
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5936900552691868e-05,
+      "loss": 0.7877,
+      "step": 4712
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5935137249262444e-05,
+      "loss": 0.8007,
+      "step": 4713
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.593337366089306e-05,
+      "loss": 0.8434,
+      "step": 4714
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.593160978766839e-05,
+      "loss": 0.4346,
+      "step": 4715
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5929845629673103e-05,
+      "loss": 0.858,
+      "step": 4716
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5928081186991907e-05,
+      "loss": 0.8937,
+      "step": 4717
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5926316459709504e-05,
+      "loss": 0.8193,
+      "step": 4718
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5924551447910622e-05,
+      "loss": 0.808,
+      "step": 4719
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5922786151679993e-05,
+      "loss": 0.4421,
+      "step": 4720
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5921020571102366e-05,
+      "loss": 0.4695,
+      "step": 4721
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.591925470626251e-05,
+      "loss": 0.42,
+      "step": 4722
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.59174885572452e-05,
+      "loss": 0.8466,
+      "step": 4723
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5915722124135227e-05,
+      "loss": 0.7643,
+      "step": 4724
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5913955407017394e-05,
+      "loss": 0.8087,
+      "step": 4725
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.591218840597652e-05,
+      "loss": 0.7655,
+      "step": 4726
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5910421121097438e-05,
+      "loss": 0.4373,
+      "step": 4727
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.590865355246499e-05,
+      "loss": 0.8371,
+      "step": 4728
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5906885700164042e-05,
+      "loss": 0.837,
+      "step": 4729
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5905117564279463e-05,
+      "loss": 0.7831,
+      "step": 4730
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5903349144896137e-05,
+      "loss": 0.8151,
+      "step": 4731
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.590158044209897e-05,
+      "loss": 0.4787,
+      "step": 4732
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5899811455972872e-05,
+      "loss": 0.4826,
+      "step": 4733
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5898042186602765e-05,
+      "loss": 0.4862,
+      "step": 4734
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5896272634073603e-05,
+      "loss": 0.7795,
+      "step": 4735
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5894502798470326e-05,
+      "loss": 0.8123,
+      "step": 4736
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5892732679877912e-05,
+      "loss": 0.8155,
+      "step": 4737
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.589096227838134e-05,
+      "loss": 0.8,
+      "step": 4738
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.58891915940656e-05,
+      "loss": 0.8128,
+      "step": 4739
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.588742062701571e-05,
+      "loss": 0.7534,
+      "step": 4740
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5885649377316688e-05,
+      "loss": 0.8125,
+      "step": 4741
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.588387784505357e-05,
+      "loss": 0.7815,
+      "step": 4742
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.58821060303114e-05,
+      "loss": 0.8311,
+      "step": 4743
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5880333933175247e-05,
+      "loss": 0.47,
+      "step": 4744
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.587856155373019e-05,
+      "loss": 0.8387,
+      "step": 4745
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5876788892061314e-05,
+      "loss": 0.8629,
+      "step": 4746
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5875015948253727e-05,
+      "loss": 0.8586,
+      "step": 4747
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.587324272239254e-05,
+      "loss": 0.7728,
+      "step": 4748
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.587146921456289e-05,
+      "loss": 0.8253,
+      "step": 4749
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5869695424849918e-05,
+      "loss": 0.7591,
+      "step": 4750
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5867921353338783e-05,
+      "loss": 0.8117,
+      "step": 4751
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5866147000114657e-05,
+      "loss": 0.429,
+      "step": 4752
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.586437236526272e-05,
+      "loss": 0.8318,
+      "step": 4753
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.586259744886818e-05,
+      "loss": 0.8668,
+      "step": 4754
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.586082225101624e-05,
+      "loss": 0.8402,
+      "step": 4755
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.585904677179213e-05,
+      "loss": 0.4272,
+      "step": 4756
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.585727101128109e-05,
+      "loss": 0.4636,
+      "step": 4757
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5855494969568367e-05,
+      "loss": 0.803,
+      "step": 4758
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.585371864673923e-05,
+      "loss": 0.4428,
+      "step": 4759
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.585194204287896e-05,
+      "loss": 0.8145,
+      "step": 4760
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.585016515807285e-05,
+      "loss": 0.8097,
+      "step": 4761
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5848387992406203e-05,
+      "loss": 0.7834,
+      "step": 4762
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.584661054596434e-05,
+      "loss": 0.7826,
+      "step": 4763
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.58448328188326e-05,
+      "loss": 0.7909,
+      "step": 4764
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5843054811096323e-05,
+      "loss": 0.8294,
+      "step": 4765
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5841276522840873e-05,
+      "loss": 0.8266,
+      "step": 4766
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.583949795415162e-05,
+      "loss": 0.4034,
+      "step": 4767
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5837719105113954e-05,
+      "loss": 0.8466,
+      "step": 4768
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.583593997581328e-05,
+      "loss": 0.7763,
+      "step": 4769
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5834160566335003e-05,
+      "loss": 0.8012,
+      "step": 4770
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5832380876764556e-05,
+      "loss": 0.4847,
+      "step": 4771
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.583060090718738e-05,
+      "loss": 0.8071,
+      "step": 4772
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.582882065768893e-05,
+      "loss": 0.812,
+      "step": 4773
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5827040128354672e-05,
+      "loss": 0.4774,
+      "step": 4774
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.582525931927009e-05,
+      "loss": 0.7409,
+      "step": 4775
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.582347823052068e-05,
+      "loss": 0.7362,
+      "step": 4776
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5821696862191944e-05,
+      "loss": 0.8158,
+      "step": 4777
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5819915214369407e-05,
+      "loss": 0.8674,
+      "step": 4778
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5818133287138606e-05,
+      "loss": 0.779,
+      "step": 4779
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5816351080585087e-05,
+      "loss": 0.4436,
+      "step": 4780
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.581456859479441e-05,
+      "loss": 0.4481,
+      "step": 4781
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.581278582985216e-05,
+      "loss": 0.7769,
+      "step": 4782
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5811002785843914e-05,
+      "loss": 0.8464,
+      "step": 4783
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5809219462855282e-05,
+      "loss": 0.8301,
+      "step": 4784
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5807435860971877e-05,
+      "loss": 0.8141,
+      "step": 4785
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5805651980279324e-05,
+      "loss": 0.8177,
+      "step": 4786
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.580386782086327e-05,
+      "loss": 0.4626,
+      "step": 4787
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5802083382809374e-05,
+      "loss": 0.813,
+      "step": 4788
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5800298666203295e-05,
+      "loss": 0.7786,
+      "step": 4789
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5798513671130727e-05,
+      "loss": 0.804,
+      "step": 4790
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5796728397677355e-05,
+      "loss": 0.8434,
+      "step": 4791
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.57949428459289e-05,
+      "loss": 0.8315,
+      "step": 4792
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.579315701597107e-05,
+      "loss": 0.4472,
+      "step": 4793
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5791370907889612e-05,
+      "loss": 0.7412,
+      "step": 4794
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5789584521770268e-05,
+      "loss": 0.4175,
+      "step": 4795
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5787797857698806e-05,
+      "loss": 0.8137,
+      "step": 4796
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5786010915761006e-05,
+      "loss": 0.4886,
+      "step": 4797
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.578422369604265e-05,
+      "loss": 0.8569,
+      "step": 4798
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.578243619862954e-05,
+      "loss": 0.4605,
+      "step": 4799
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5780648423607493e-05,
+      "loss": 0.7837,
+      "step": 4800
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5778860371062342e-05,
+      "loss": 0.8203,
+      "step": 4801
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5777072041079928e-05,
+      "loss": 0.7771,
+      "step": 4802
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.57752834337461e-05,
+      "loss": 0.4343,
+      "step": 4803
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.577349454914674e-05,
+      "loss": 0.4408,
+      "step": 4804
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.577170538736772e-05,
+      "loss": 0.476,
+      "step": 4805
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5769915948494942e-05,
+      "loss": 0.7607,
+      "step": 4806
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5768126232614312e-05,
+      "loss": 0.8475,
+      "step": 4807
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5766336239811755e-05,
+      "loss": 0.8331,
+      "step": 4808
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5764545970173202e-05,
+      "loss": 0.776,
+      "step": 4809
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5762755423784605e-05,
+      "loss": 0.7851,
+      "step": 4810
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5760964600731926e-05,
+      "loss": 0.8118,
+      "step": 4811
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5759173501101144e-05,
+      "loss": 0.4358,
+      "step": 4812
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5757382124978244e-05,
+      "loss": 0.8722,
+      "step": 4813
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.575559047244923e-05,
+      "loss": 0.8204,
+      "step": 4814
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5753798543600117e-05,
+      "loss": 0.4614,
+      "step": 4815
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.575200633851693e-05,
+      "loss": 0.8259,
+      "step": 4816
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5750213857285718e-05,
+      "loss": 0.8664,
+      "step": 4817
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5748421099992528e-05,
+      "loss": 0.7926,
+      "step": 4818
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5746628066723437e-05,
+      "loss": 0.8022,
+      "step": 4819
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.574483475756452e-05,
+      "loss": 0.4166,
+      "step": 4820
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5743041172601876e-05,
+      "loss": 0.8153,
+      "step": 4821
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.574124731192161e-05,
+      "loss": 0.4511,
+      "step": 4822
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5739453175609848e-05,
+      "loss": 0.4456,
+      "step": 4823
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.573765876375272e-05,
+      "loss": 0.7834,
+      "step": 4824
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5735864076436376e-05,
+      "loss": 0.8023,
+      "step": 4825
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5734069113746978e-05,
+      "loss": 0.4621,
+      "step": 4826
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5732273875770696e-05,
+      "loss": 0.4731,
+      "step": 4827
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5730478362593725e-05,
+      "loss": 0.8338,
+      "step": 4828
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5728682574302258e-05,
+      "loss": 0.7583,
+      "step": 4829
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5726886510982514e-05,
+      "loss": 0.4319,
+      "step": 4830
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.572509017272072e-05,
+      "loss": 0.8026,
+      "step": 4831
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5723293559603114e-05,
+      "loss": 0.4261,
+      "step": 4832
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5721496671715953e-05,
+      "loss": 0.7852,
+      "step": 4833
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.57196995091455e-05,
+      "loss": 0.4568,
+      "step": 4834
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5717902071978034e-05,
+      "loss": 0.7947,
+      "step": 4835
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5716104360299855e-05,
+      "loss": 0.432,
+      "step": 4836
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5714306374197265e-05,
+      "loss": 0.7732,
+      "step": 4837
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.571250811375658e-05,
+      "loss": 0.8246,
+      "step": 4838
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5710709579064137e-05,
+      "loss": 0.8519,
+      "step": 4839
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5708910770206282e-05,
+      "loss": 0.7881,
+      "step": 4840
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.570711168726937e-05,
+      "loss": 0.7701,
+      "step": 4841
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.570531233033978e-05,
+      "loss": 0.8171,
+      "step": 4842
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5703512699503893e-05,
+      "loss": 0.7508,
+      "step": 4843
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5701712794848107e-05,
+      "loss": 0.7985,
+      "step": 4844
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5699912616458834e-05,
+      "loss": 0.7531,
+      "step": 4845
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.56981121644225e-05,
+      "loss": 0.8079,
+      "step": 4846
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.569631143882554e-05,
+      "loss": 0.8155,
+      "step": 4847
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.569451043975441e-05,
+      "loss": 0.808,
+      "step": 4848
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.569270916729557e-05,
+      "loss": 0.8308,
+      "step": 4849
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5690907621535502e-05,
+      "loss": 0.889,
+      "step": 4850
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.568910580256069e-05,
+      "loss": 0.7874,
+      "step": 4851
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5687303710457642e-05,
+      "loss": 0.7822,
+      "step": 4852
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5685501345312873e-05,
+      "loss": 0.8011,
+      "step": 4853
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5683698707212912e-05,
+      "loss": 0.7976,
+      "step": 4854
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5681895796244302e-05,
+      "loss": 0.8456,
+      "step": 4855
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.56800926124936e-05,
+      "loss": 0.8826,
+      "step": 4856
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5678289156047374e-05,
+      "loss": 0.811,
+      "step": 4857
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.567648542699221e-05,
+      "loss": 0.4563,
+      "step": 4858
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5674681425414697e-05,
+      "loss": 0.8032,
+      "step": 4859
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5672877151401448e-05,
+      "loss": 0.7989,
+      "step": 4860
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5671072605039083e-05,
+      "loss": 0.399,
+      "step": 4861
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5669267786414234e-05,
+      "loss": 0.4515,
+      "step": 4862
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5667462695613554e-05,
+      "loss": 0.7875,
+      "step": 4863
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5665657332723697e-05,
+      "loss": 0.7622,
+      "step": 4864
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5663851697831343e-05,
+      "loss": 0.7381,
+      "step": 4865
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.566204579102317e-05,
+      "loss": 0.8197,
+      "step": 4866
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.566023961238589e-05,
+      "loss": 0.786,
+      "step": 4867
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5658433162006204e-05,
+      "loss": 0.8418,
+      "step": 4868
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5656626439970847e-05,
+      "loss": 0.8071,
+      "step": 4869
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5654819446366553e-05,
+      "loss": 0.7965,
+      "step": 4870
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5653012181280077e-05,
+      "loss": 0.8243,
+      "step": 4871
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.565120464479818e-05,
+      "loss": 0.7713,
+      "step": 4872
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.564939683700764e-05,
+      "loss": 0.8333,
+      "step": 4873
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5647588757995252e-05,
+      "loss": 0.4516,
+      "step": 4874
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5645780407847818e-05,
+      "loss": 0.8052,
+      "step": 4875
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5643971786652158e-05,
+      "loss": 0.7973,
+      "step": 4876
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5642162894495094e-05,
+      "loss": 0.7977,
+      "step": 4877
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5640353731463472e-05,
+      "loss": 0.4642,
+      "step": 4878
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.563854429764416e-05,
+      "loss": 0.4789,
+      "step": 4879
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.563673459312401e-05,
+      "loss": 0.7535,
+      "step": 4880
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5634924617989913e-05,
+      "loss": 0.7709,
+      "step": 4881
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5633114372328764e-05,
+      "loss": 0.723,
+      "step": 4882
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.563130385622747e-05,
+      "loss": 0.8132,
+      "step": 4883
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.562949306977295e-05,
+      "loss": 0.783,
+      "step": 4884
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5627682013052138e-05,
+      "loss": 0.815,
+      "step": 4885
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.562587068615199e-05,
+      "loss": 0.4825,
+      "step": 4886
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.562405908915945e-05,
+      "loss": 0.8125,
+      "step": 4887
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5622247222161505e-05,
+      "loss": 0.4312,
+      "step": 4888
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5620435085245134e-05,
+      "loss": 0.4752,
+      "step": 4889
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5618622678497338e-05,
+      "loss": 0.4632,
+      "step": 4890
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5616810002005127e-05,
+      "loss": 0.7949,
+      "step": 4891
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5614997055855528e-05,
+      "loss": 0.4769,
+      "step": 4892
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.561318384013558e-05,
+      "loss": 0.8308,
+      "step": 4893
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.561137035493233e-05,
+      "loss": 0.4149,
+      "step": 4894
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5609556600332843e-05,
+      "loss": 0.8512,
+      "step": 4895
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5607742576424194e-05,
+      "loss": 0.817,
+      "step": 4896
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5605928283293475e-05,
+      "loss": 0.4378,
+      "step": 4897
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5604113721027788e-05,
+      "loss": 0.8031,
+      "step": 4898
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5602298889714245e-05,
+      "loss": 0.7575,
+      "step": 4899
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5600483789439983e-05,
+      "loss": 0.7792,
+      "step": 4900
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.559866842029213e-05,
+      "loss": 0.4791,
+      "step": 4901
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5596852782357853e-05,
+      "loss": 0.7885,
+      "step": 4902
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.559503687572431e-05,
+      "loss": 0.4483,
+      "step": 4903
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5593220700478683e-05,
+      "loss": 0.7761,
+      "step": 4904
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.559140425670817e-05,
+      "loss": 0.7215,
+      "step": 4905
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.558958754449997e-05,
+      "loss": 0.8133,
+      "step": 4906
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5587770563941304e-05,
+      "loss": 0.7659,
+      "step": 4907
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.55859533151194e-05,
+      "loss": 0.4356,
+      "step": 4908
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.558413579812151e-05,
+      "loss": 0.779,
+      "step": 4909
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5582318013034886e-05,
+      "loss": 0.8033,
+      "step": 4910
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5580499959946798e-05,
+      "loss": 0.8472,
+      "step": 4911
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.557868163894453e-05,
+      "loss": 0.4482,
+      "step": 4912
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5576863050115374e-05,
+      "loss": 0.7917,
+      "step": 4913
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5575044193546645e-05,
+      "loss": 0.8045,
+      "step": 4914
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5573225069325663e-05,
+      "loss": 0.7982,
+      "step": 4915
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5571405677539755e-05,
+      "loss": 0.7918,
+      "step": 4916
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.556958601827628e-05,
+      "loss": 0.8186,
+      "step": 4917
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5567766091622592e-05,
+      "loss": 0.8052,
+      "step": 4918
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.556594589766606e-05,
+      "loss": 0.7762,
+      "step": 4919
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5564125436494076e-05,
+      "loss": 0.797,
+      "step": 4920
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5562304708194035e-05,
+      "loss": 0.7206,
+      "step": 4921
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5560483712853348e-05,
+      "loss": 0.7907,
+      "step": 4922
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5558662450559444e-05,
+      "loss": 0.4507,
+      "step": 4923
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5556840921399753e-05,
+      "loss": 0.8067,
+      "step": 4924
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.555501912546173e-05,
+      "loss": 0.764,
+      "step": 4925
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5553197062832837e-05,
+      "loss": 0.7958,
+      "step": 4926
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5551374733600545e-05,
+      "loss": 0.4285,
+      "step": 4927
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.554955213785235e-05,
+      "loss": 0.8433,
+      "step": 4928
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5547729275675742e-05,
+      "loss": 0.7898,
+      "step": 4929
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5545906147158245e-05,
+      "loss": 0.7869,
+      "step": 4930
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5544082752387383e-05,
+      "loss": 0.7728,
+      "step": 4931
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5542259091450695e-05,
+      "loss": 0.7981,
+      "step": 4932
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5540435164435726e-05,
+      "loss": 0.8195,
+      "step": 4933
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5538610971430054e-05,
+      "loss": 0.4408,
+      "step": 4934
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5536786512521244e-05,
+      "loss": 0.7874,
+      "step": 4935
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5534961787796894e-05,
+      "loss": 0.7889,
+      "step": 4936
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5533136797344606e-05,
+      "loss": 0.8359,
+      "step": 4937
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5531311541251995e-05,
+      "loss": 0.7946,
+      "step": 4938
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.552948601960669e-05,
+      "loss": 0.7984,
+      "step": 4939
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5527660232496335e-05,
+      "loss": 0.8172,
+      "step": 4940
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5525834180008578e-05,
+      "loss": 0.8153,
+      "step": 4941
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5524007862231095e-05,
+      "loss": 0.776,
+      "step": 4942
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5522181279251553e-05,
+      "loss": 0.7971,
+      "step": 4943
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5520354431157655e-05,
+      "loss": 0.8362,
+      "step": 4944
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5518527318037105e-05,
+      "loss": 0.775,
+      "step": 4945
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5516699939977616e-05,
+      "loss": 0.8047,
+      "step": 4946
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5514872297066922e-05,
+      "loss": 0.4645,
+      "step": 4947
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5513044389392768e-05,
+      "loss": 0.8042,
+      "step": 4948
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5511216217042904e-05,
+      "loss": 0.8824,
+      "step": 4949
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5509387780105102e-05,
+      "loss": 0.8345,
+      "step": 4950
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5507559078667146e-05,
+      "loss": 0.7988,
+      "step": 4951
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5505730112816828e-05,
+      "loss": 0.8135,
+      "step": 4952
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5503900882641954e-05,
+      "loss": 0.7898,
+      "step": 4953
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5502071388230344e-05,
+      "loss": 0.8119,
+      "step": 4954
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5500241629669832e-05,
+      "loss": 0.7634,
+      "step": 4955
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.549841160704826e-05,
+      "loss": 0.8113,
+      "step": 4956
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5496581320453487e-05,
+      "loss": 0.76,
+      "step": 4957
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5494750769973382e-05,
+      "loss": 0.4946,
+      "step": 4958
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5492919955695827e-05,
+      "loss": 0.8242,
+      "step": 4959
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.549108887770872e-05,
+      "loss": 0.8039,
+      "step": 4960
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.548925753609997e-05,
+      "loss": 0.8242,
+      "step": 4961
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.54874259309575e-05,
+      "loss": 0.807,
+      "step": 4962
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5485594062369235e-05,
+      "loss": 0.7824,
+      "step": 4963
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5483761930423126e-05,
+      "loss": 0.4347,
+      "step": 4964
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5481929535207133e-05,
+      "loss": 0.4367,
+      "step": 4965
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.548009687680923e-05,
+      "loss": 0.8669,
+      "step": 4966
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.547826395531739e-05,
+      "loss": 0.7947,
+      "step": 4967
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5476430770819625e-05,
+      "loss": 0.8175,
+      "step": 4968
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5474597323403933e-05,
+      "loss": 0.4207,
+      "step": 4969
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.547276361315834e-05,
+      "loss": 0.8165,
+      "step": 4970
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.547092964017088e-05,
+      "loss": 0.7657,
+      "step": 4971
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.54690954045296e-05,
+      "loss": 0.4538,
+      "step": 4972
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.546726090632256e-05,
+      "loss": 0.8499,
+      "step": 4973
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5465426145637832e-05,
+      "loss": 0.4368,
+      "step": 4974
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5463591122563503e-05,
+      "loss": 0.4356,
+      "step": 4975
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5461755837187665e-05,
+      "loss": 0.7747,
+      "step": 4976
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.545992028959844e-05,
+      "loss": 0.8125,
+      "step": 4977
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5458084479883938e-05,
+      "loss": 0.8371,
+      "step": 4978
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.54562484081323e-05,
+      "loss": 0.836,
+      "step": 4979
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5454412074431674e-05,
+      "loss": 0.8478,
+      "step": 4980
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.545257547887022e-05,
+      "loss": 0.4603,
+      "step": 4981
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.545073862153611e-05,
+      "loss": 0.8073,
+      "step": 4982
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.544890150251753e-05,
+      "loss": 0.8167,
+      "step": 4983
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.544706412190268e-05,
+      "loss": 0.796,
+      "step": 4984
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5445226479779773e-05,
+      "loss": 0.4375,
+      "step": 4985
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5443388576237025e-05,
+      "loss": 0.4209,
+      "step": 4986
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5441550411362677e-05,
+      "loss": 0.7933,
+      "step": 4987
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5439711985244978e-05,
+      "loss": 0.8323,
+      "step": 4988
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5437873297972184e-05,
+      "loss": 0.882,
+      "step": 4989
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5436034349632577e-05,
+      "loss": 0.4766,
+      "step": 4990
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.543419514031443e-05,
+      "loss": 0.808,
+      "step": 4991
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.543235567010606e-05,
+      "loss": 0.7572,
+      "step": 4992
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.543051593909576e-05,
+      "loss": 0.8426,
+      "step": 4993
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5428675947371866e-05,
+      "loss": 0.814,
+      "step": 4994
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5426835695022705e-05,
+      "loss": 0.7692,
+      "step": 4995
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.542499518213663e-05,
+      "loss": 0.764,
+      "step": 4996
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5423154408802006e-05,
+      "loss": 0.8018,
+      "step": 4997
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5421313375107202e-05,
+      "loss": 0.4232,
+      "step": 4998
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5419472081140607e-05,
+      "loss": 0.833,
+      "step": 4999
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5417630526990613e-05,
+      "loss": 0.8198,
+      "step": 5000
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.541578871274564e-05,
+      "loss": 0.8136,
+      "step": 5001
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5413946638494107e-05,
+      "loss": 0.3914,
+      "step": 5002
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.541210430432445e-05,
+      "loss": 0.8433,
+      "step": 5003
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.541026171032512e-05,
+      "loss": 0.8193,
+      "step": 5004
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5408418856584574e-05,
+      "loss": 0.7902,
+      "step": 5005
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5406575743191286e-05,
+      "loss": 0.8543,
+      "step": 5006
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5404732370233747e-05,
+      "loss": 0.7911,
+      "step": 5007
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5402888737800458e-05,
+      "loss": 0.8018,
+      "step": 5008
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5401044845979917e-05,
+      "loss": 0.4395,
+      "step": 5009
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5399200694860656e-05,
+      "loss": 0.8055,
+      "step": 5010
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5397356284531207e-05,
+      "loss": 0.4245,
+      "step": 5011
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.539551161508013e-05,
+      "loss": 0.8102,
+      "step": 5012
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.539366668659597e-05,
+      "loss": 0.771,
+      "step": 5013
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5391821499167305e-05,
+      "loss": 0.4884,
+      "step": 5014
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5389976052882728e-05,
+      "loss": 0.7335,
+      "step": 5015
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.538813034783083e-05,
+      "loss": 0.8024,
+      "step": 5016
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5386284384100217e-05,
+      "loss": 0.4368,
+      "step": 5017
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5384438161779524e-05,
+      "loss": 0.7997,
+      "step": 5018
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.538259168095738e-05,
+      "loss": 0.8165,
+      "step": 5019
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.538074494172243e-05,
+      "loss": 0.8353,
+      "step": 5020
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5378897944163335e-05,
+      "loss": 0.7975,
+      "step": 5021
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5377050688368774e-05,
+      "loss": 0.808,
+      "step": 5022
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5375203174427428e-05,
+      "loss": 0.4274,
+      "step": 5023
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.537335540242799e-05,
+      "loss": 0.7925,
+      "step": 5024
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5371507372459173e-05,
+      "loss": 0.8451,
+      "step": 5025
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5369659084609702e-05,
+      "loss": 0.8832,
+      "step": 5026
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.53678105389683e-05,
+      "loss": 0.7681,
+      "step": 5027
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5365961735623732e-05,
+      "loss": 0.7604,
+      "step": 5028
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5364112674664745e-05,
+      "loss": 0.8274,
+      "step": 5029
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5362263356180115e-05,
+      "loss": 0.7959,
+      "step": 5030
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5360413780258625e-05,
+      "loss": 0.7878,
+      "step": 5031
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5358563946989064e-05,
+      "loss": 0.4589,
+      "step": 5032
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.535671385646025e-05,
+      "loss": 0.7501,
+      "step": 5033
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5354863508761e-05,
+      "loss": 0.7797,
+      "step": 5034
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.535301290398015e-05,
+      "loss": 0.8499,
+      "step": 5035
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5351162042206543e-05,
+      "loss": 0.4207,
+      "step": 5036
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5349310923529038e-05,
+      "loss": 0.8363,
+      "step": 5037
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.534745954803651e-05,
+      "loss": 0.817,
+      "step": 5038
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.534560791581783e-05,
+      "loss": 0.4313,
+      "step": 5039
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5343756026961906e-05,
+      "loss": 0.8018,
+      "step": 5040
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5341903881557635e-05,
+      "loss": 0.7236,
+      "step": 5041
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.534005147969394e-05,
+      "loss": 0.8233,
+      "step": 5042
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5338198821459755e-05,
+      "loss": 0.4505,
+      "step": 5043
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5336345906944025e-05,
+      "loss": 0.7882,
+      "step": 5044
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5334492736235703e-05,
+      "loss": 0.8278,
+      "step": 5045
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.533263930942376e-05,
+      "loss": 0.4321,
+      "step": 5046
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.533078562659718e-05,
+      "loss": 0.8105,
+      "step": 5047
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.532893168784495e-05,
+      "loss": 0.447,
+      "step": 5048
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.532707749325608e-05,
+      "loss": 0.4477,
+      "step": 5049
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5325223042919583e-05,
+      "loss": 0.4651,
+      "step": 5050
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5323368336924498e-05,
+      "loss": 0.7915,
+      "step": 5051
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5321513375359863e-05,
+      "loss": 0.809,
+      "step": 5052
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.531965815831473e-05,
+      "loss": 0.8031,
+      "step": 5053
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.531780268587817e-05,
+      "loss": 0.4358,
+      "step": 5054
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5315946958139258e-05,
+      "loss": 0.7936,
+      "step": 5055
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5314090975187093e-05,
+      "loss": 0.4542,
+      "step": 5056
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5312234737110773e-05,
+      "loss": 0.784,
+      "step": 5057
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5310378243999412e-05,
+      "loss": 0.5076,
+      "step": 5058
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.530852149594215e-05,
+      "loss": 0.7943,
+      "step": 5059
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5306664493028112e-05,
+      "loss": 0.8286,
+      "step": 5060
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5304807235346463e-05,
+      "loss": 0.7891,
+      "step": 5061
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.530294972298636e-05,
+      "loss": 0.7958,
+      "step": 5062
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5301091956036987e-05,
+      "loss": 0.7783,
+      "step": 5063
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.529923393458753e-05,
+      "loss": 0.8262,
+      "step": 5064
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5297375658727188e-05,
+      "loss": 0.4304,
+      "step": 5065
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5295517128545184e-05,
+      "loss": 0.7867,
+      "step": 5066
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5293658344130734e-05,
+      "loss": 0.8161,
+      "step": 5067
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.529179930557308e-05,
+      "loss": 0.8568,
+      "step": 5068
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5289940012961478e-05,
+      "loss": 0.8382,
+      "step": 5069
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.528808046638518e-05,
+      "loss": 0.501,
+      "step": 5070
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.528622066593347e-05,
+      "loss": 0.4329,
+      "step": 5071
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.528436061169563e-05,
+      "loss": 0.8072,
+      "step": 5072
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5282500303760965e-05,
+      "loss": 0.8002,
+      "step": 5073
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.528063974221878e-05,
+      "loss": 0.7719,
+      "step": 5074
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.52787789271584e-05,
+      "loss": 0.818,
+      "step": 5075
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5276917858669163e-05,
+      "loss": 0.4818,
+      "step": 5076
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5275056536840417e-05,
+      "loss": 0.7957,
+      "step": 5077
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5273194961761525e-05,
+      "loss": 0.8058,
+      "step": 5078
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5271333133521853e-05,
+      "loss": 0.8149,
+      "step": 5079
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.526947105221079e-05,
+      "loss": 0.4323,
+      "step": 5080
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.526760871791773e-05,
+      "loss": 0.8313,
+      "step": 5081
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5265746130732083e-05,
+      "loss": 0.8373,
+      "step": 5082
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5263883290743268e-05,
+      "loss": 0.8073,
+      "step": 5083
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5262020198040723e-05,
+      "loss": 0.7714,
+      "step": 5084
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.526015685271389e-05,
+      "loss": 0.4404,
+      "step": 5085
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5258293254852224e-05,
+      "loss": 0.7857,
+      "step": 5086
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5256429404545197e-05,
+      "loss": 0.4353,
+      "step": 5087
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5254565301882292e-05,
+      "loss": 0.4368,
+      "step": 5088
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5252700946953001e-05,
+      "loss": 0.4372,
+      "step": 5089
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.525083633984683e-05,
+      "loss": 0.8082,
+      "step": 5090
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5248971480653296e-05,
+      "loss": 0.8306,
+      "step": 5091
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5247106369461931e-05,
+      "loss": 0.7876,
+      "step": 5092
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5245241006362274e-05,
+      "loss": 0.8107,
+      "step": 5093
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5243375391443885e-05,
+      "loss": 0.8143,
+      "step": 5094
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5241509524796323e-05,
+      "loss": 0.8092,
+      "step": 5095
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5239643406509172e-05,
+      "loss": 0.4347,
+      "step": 5096
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.523777703667202e-05,
+      "loss": 0.4186,
+      "step": 5097
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5235910415374466e-05,
+      "loss": 0.4398,
+      "step": 5098
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5234043542706131e-05,
+      "loss": 0.7793,
+      "step": 5099
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.523217641875664e-05,
+      "loss": 0.8106,
+      "step": 5100
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5230309043615632e-05,
+      "loss": 0.4806,
+      "step": 5101
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5228441417372754e-05,
+      "loss": 0.768,
+      "step": 5102
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5226573540117673e-05,
+      "loss": 0.8492,
+      "step": 5103
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5224705411940065e-05,
+      "loss": 0.8193,
+      "step": 5104
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5222837032929611e-05,
+      "loss": 0.7597,
+      "step": 5105
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5220968403176014e-05,
+      "loss": 0.7833,
+      "step": 5106
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5219099522768983e-05,
+      "loss": 0.8618,
+      "step": 5107
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5217230391798245e-05,
+      "loss": 0.4467,
+      "step": 5108
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.521536101035353e-05,
+      "loss": 0.8357,
+      "step": 5109
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5213491378524587e-05,
+      "loss": 0.7964,
+      "step": 5110
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5211621496401181e-05,
+      "loss": 0.8403,
+      "step": 5111
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5209751364073076e-05,
+      "loss": 0.8633,
+      "step": 5112
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5207880981630054e-05,
+      "loss": 0.8075,
+      "step": 5113
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5206010349161915e-05,
+      "loss": 0.8268,
+      "step": 5114
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5204139466758467e-05,
+      "loss": 0.4766,
+      "step": 5115
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5202268334509522e-05,
+      "loss": 0.8523,
+      "step": 5116
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.520039695250492e-05,
+      "loss": 0.842,
+      "step": 5117
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5198525320834498e-05,
+      "loss": 0.8272,
+      "step": 5118
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5196653439588115e-05,
+      "loss": 0.7596,
+      "step": 5119
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5194781308855633e-05,
+      "loss": 0.8226,
+      "step": 5120
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5192908928726939e-05,
+      "loss": 0.4492,
+      "step": 5121
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5191036299291917e-05,
+      "loss": 0.7912,
+      "step": 5122
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5189163420640472e-05,
+      "loss": 0.4128,
+      "step": 5123
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.518729029286252e-05,
+      "loss": 0.463,
+      "step": 5124
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5185416916047986e-05,
+      "loss": 0.7759,
+      "step": 5125
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5183543290286814e-05,
+      "loss": 0.7764,
+      "step": 5126
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.518166941566895e-05,
+      "loss": 0.8289,
+      "step": 5127
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5179795292284358e-05,
+      "loss": 0.7758,
+      "step": 5128
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5177920920223013e-05,
+      "loss": 0.8072,
+      "step": 5129
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.51760462995749e-05,
+      "loss": 0.8249,
+      "step": 5130
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5174171430430021e-05,
+      "loss": 0.4682,
+      "step": 5131
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5172296312878385e-05,
+      "loss": 0.4664,
+      "step": 5132
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5170420947010014e-05,
+      "loss": 0.8067,
+      "step": 5133
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5168545332914942e-05,
+      "loss": 0.7928,
+      "step": 5134
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5166669470683219e-05,
+      "loss": 0.9152,
+      "step": 5135
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5164793360404898e-05,
+      "loss": 0.7968,
+      "step": 5136
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5162917002170053e-05,
+      "loss": 0.7994,
+      "step": 5137
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5161040396068764e-05,
+      "loss": 0.7864,
+      "step": 5138
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5159163542191122e-05,
+      "loss": 0.4542,
+      "step": 5139
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5157286440627244e-05,
+      "loss": 0.4538,
+      "step": 5140
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5155409091467234e-05,
+      "loss": 0.4532,
+      "step": 5141
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5153531494801233e-05,
+      "loss": 0.8309,
+      "step": 5142
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5151653650719373e-05,
+      "loss": 0.8191,
+      "step": 5143
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5149775559311812e-05,
+      "loss": 0.8257,
+      "step": 5144
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.514789722066872e-05,
+      "loss": 0.8237,
+      "step": 5145
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5146018634880264e-05,
+      "loss": 0.8108,
+      "step": 5146
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5144139802036639e-05,
+      "loss": 0.8436,
+      "step": 5147
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5142260722228048e-05,
+      "loss": 0.4522,
+      "step": 5148
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5140381395544701e-05,
+      "loss": 0.8549,
+      "step": 5149
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5138501822076819e-05,
+      "loss": 0.8002,
+      "step": 5150
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5136622001914645e-05,
+      "loss": 0.8219,
+      "step": 5151
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.513474193514842e-05,
+      "loss": 0.7818,
+      "step": 5152
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5132861621868414e-05,
+      "loss": 0.7889,
+      "step": 5153
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5130981062164889e-05,
+      "loss": 0.832,
+      "step": 5154
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5129100256128134e-05,
+      "loss": 0.8034,
+      "step": 5155
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5127219203848445e-05,
+      "loss": 0.8139,
+      "step": 5156
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5125337905416127e-05,
+      "loss": 0.8327,
+      "step": 5157
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5123456360921501e-05,
+      "loss": 0.4613,
+      "step": 5158
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5121574570454894e-05,
+      "loss": 0.7711,
+      "step": 5159
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5119692534106656e-05,
+      "loss": 0.4973,
+      "step": 5160
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5117810251967135e-05,
+      "loss": 0.832,
+      "step": 5161
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5115927724126702e-05,
+      "loss": 0.8065,
+      "step": 5162
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5114044950675733e-05,
+      "loss": 0.8415,
+      "step": 5163
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5112161931704618e-05,
+      "loss": 0.7959,
+      "step": 5164
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5110278667303762e-05,
+      "loss": 0.8294,
+      "step": 5165
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5108395157563574e-05,
+      "loss": 0.8202,
+      "step": 5166
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5106511402574482e-05,
+      "loss": 0.7847,
+      "step": 5167
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5104627402426922e-05,
+      "loss": 0.8375,
+      "step": 5168
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5102743157211344e-05,
+      "loss": 0.7985,
+      "step": 5169
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5100858667018208e-05,
+      "loss": 0.7817,
+      "step": 5170
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.509897393193799e-05,
+      "loss": 0.8396,
+      "step": 5171
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5097088952061167e-05,
+      "loss": 0.8096,
+      "step": 5172
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.509520372747824e-05,
+      "loss": 0.7917,
+      "step": 5173
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5093318258279718e-05,
+      "loss": 0.8179,
+      "step": 5174
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5091432544556115e-05,
+      "loss": 0.8205,
+      "step": 5175
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5089546586397968e-05,
+      "loss": 0.8182,
+      "step": 5176
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5087660383895815e-05,
+      "loss": 0.7596,
+      "step": 5177
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5085773937140217e-05,
+      "loss": 0.77,
+      "step": 5178
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5083887246221733e-05,
+      "loss": 0.4405,
+      "step": 5179
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5082000311230946e-05,
+      "loss": 0.7984,
+      "step": 5180
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5080113132258447e-05,
+      "loss": 0.8293,
+      "step": 5181
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5078225709394829e-05,
+      "loss": 0.4617,
+      "step": 5182
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5076338042730716e-05,
+      "loss": 0.7816,
+      "step": 5183
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5074450132356724e-05,
+      "loss": 0.4368,
+      "step": 5184
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.50725619783635e-05,
+      "loss": 0.4431,
+      "step": 5185
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5070673580841684e-05,
+      "loss": 0.7799,
+      "step": 5186
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5068784939881938e-05,
+      "loss": 0.4422,
+      "step": 5187
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5066896055574933e-05,
+      "loss": 0.7771,
+      "step": 5188
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5065006928011352e-05,
+      "loss": 0.7881,
+      "step": 5189
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5063117557281897e-05,
+      "loss": 0.8087,
+      "step": 5190
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5061227943477263e-05,
+      "loss": 0.8527,
+      "step": 5191
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.505933808668818e-05,
+      "loss": 0.8247,
+      "step": 5192
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5057447987005373e-05,
+      "loss": 0.5005,
+      "step": 5193
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5055557644519583e-05,
+      "loss": 0.8064,
+      "step": 5194
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5053667059321562e-05,
+      "loss": 0.8037,
+      "step": 5195
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.505177623150208e-05,
+      "loss": 0.8155,
+      "step": 5196
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5049885161151912e-05,
+      "loss": 0.7847,
+      "step": 5197
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5047993848361842e-05,
+      "loss": 0.793,
+      "step": 5198
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5046102293222676e-05,
+      "loss": 0.4535,
+      "step": 5199
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.504421049582522e-05,
+      "loss": 0.7863,
+      "step": 5200
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5042318456260305e-05,
+      "loss": 0.8204,
+      "step": 5201
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5040426174618756e-05,
+      "loss": 0.4326,
+      "step": 5202
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.503853365099143e-05,
+      "loss": 0.8128,
+      "step": 5203
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.503664088546918e-05,
+      "loss": 0.8153,
+      "step": 5204
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.503474787814287e-05,
+      "loss": 0.4454,
+      "step": 5205
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5032854629103393e-05,
+      "loss": 0.8115,
+      "step": 5206
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5030961138441634e-05,
+      "loss": 0.8336,
+      "step": 5207
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5029067406248504e-05,
+      "loss": 0.8079,
+      "step": 5208
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5027173432614907e-05,
+      "loss": 0.8218,
+      "step": 5209
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5025279217631783e-05,
+      "loss": 0.8513,
+      "step": 5210
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5023384761390065e-05,
+      "loss": 0.7778,
+      "step": 5211
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5021490063980708e-05,
+      "loss": 0.7523,
+      "step": 5212
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5019595125494672e-05,
+      "loss": 0.4388,
+      "step": 5213
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5017699946022932e-05,
+      "loss": 0.4377,
+      "step": 5214
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5015804525656474e-05,
+      "loss": 0.7742,
+      "step": 5215
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5013908864486295e-05,
+      "loss": 0.7754,
+      "step": 5216
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5012012962603399e-05,
+      "loss": 0.7996,
+      "step": 5217
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5010116820098812e-05,
+      "loss": 0.8177,
+      "step": 5218
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5008220437063567e-05,
+      "loss": 0.8443,
+      "step": 5219
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5006323813588704e-05,
+      "loss": 0.7996,
+      "step": 5220
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5004426949765278e-05,
+      "loss": 0.783,
+      "step": 5221
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5002529845684359e-05,
+      "loss": 0.83,
+      "step": 5222
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.500063250143702e-05,
+      "loss": 0.8352,
+      "step": 5223
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4998734917114353e-05,
+      "loss": 0.7437,
+      "step": 5224
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4996837092807458e-05,
+      "loss": 0.7666,
+      "step": 5225
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4994939028607451e-05,
+      "loss": 0.7586,
+      "step": 5226
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4993040724605453e-05,
+      "loss": 0.8122,
+      "step": 5227
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4991142180892601e-05,
+      "loss": 0.8209,
+      "step": 5228
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4989243397560043e-05,
+      "loss": 0.808,
+      "step": 5229
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4987344374698936e-05,
+      "loss": 0.4753,
+      "step": 5230
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4985445112400449e-05,
+      "loss": 0.858,
+      "step": 5231
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4983545610755766e-05,
+      "loss": 0.7916,
+      "step": 5232
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4981645869856078e-05,
+      "loss": 0.8466,
+      "step": 5233
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4979745889792595e-05,
+      "loss": 0.4836,
+      "step": 5234
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4977845670656526e-05,
+      "loss": 0.8283,
+      "step": 5235
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4975945212539103e-05,
+      "loss": 0.8133,
+      "step": 5236
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4974044515531565e-05,
+      "loss": 0.7835,
+      "step": 5237
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4972143579725158e-05,
+      "loss": 0.8138,
+      "step": 5238
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.497024240521115e-05,
+      "loss": 0.7857,
+      "step": 5239
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4968340992080812e-05,
+      "loss": 0.4384,
+      "step": 5240
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4966439340425428e-05,
+      "loss": 0.7775,
+      "step": 5241
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4964537450336296e-05,
+      "loss": 0.4782,
+      "step": 5242
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4962635321904721e-05,
+      "loss": 0.4397,
+      "step": 5243
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4960732955222028e-05,
+      "loss": 0.4652,
+      "step": 5244
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4958830350379541e-05,
+      "loss": 0.8097,
+      "step": 5245
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4956927507468609e-05,
+      "loss": 0.8447,
+      "step": 5246
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.495502442658058e-05,
+      "loss": 0.7815,
+      "step": 5247
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4953121107806818e-05,
+      "loss": 0.7963,
+      "step": 5248
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4951217551238703e-05,
+      "loss": 0.4409,
+      "step": 5249
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4949313756967622e-05,
+      "loss": 0.8219,
+      "step": 5250
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4947409725084977e-05,
+      "loss": 0.8063,
+      "step": 5251
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4945505455682174e-05,
+      "loss": 0.8363,
+      "step": 5252
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4943600948850638e-05,
+      "loss": 0.826,
+      "step": 5253
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4941696204681799e-05,
+      "loss": 0.7641,
+      "step": 5254
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4939791223267108e-05,
+      "loss": 0.7794,
+      "step": 5255
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4937886004698017e-05,
+      "loss": 0.8006,
+      "step": 5256
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4935980549065992e-05,
+      "loss": 0.4388,
+      "step": 5257
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4934074856462515e-05,
+      "loss": 0.8164,
+      "step": 5258
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4932168926979074e-05,
+      "loss": 0.7911,
+      "step": 5259
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4930262760707176e-05,
+      "loss": 0.4837,
+      "step": 5260
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4928356357738326e-05,
+      "loss": 0.4393,
+      "step": 5261
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4926449718164056e-05,
+      "loss": 0.8174,
+      "step": 5262
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4924542842075898e-05,
+      "loss": 0.7758,
+      "step": 5263
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4922635729565397e-05,
+      "loss": 0.4318,
+      "step": 5264
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.492072838072412e-05,
+      "loss": 0.4463,
+      "step": 5265
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4918820795643628e-05,
+      "loss": 0.8192,
+      "step": 5266
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4916912974415507e-05,
+      "loss": 0.815,
+      "step": 5267
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4915004917131345e-05,
+      "loss": 0.7714,
+      "step": 5268
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4913096623882753e-05,
+      "loss": 0.4884,
+      "step": 5269
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.491118809476134e-05,
+      "loss": 0.7944,
+      "step": 5270
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4909279329858737e-05,
+      "loss": 0.7551,
+      "step": 5271
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4907370329266577e-05,
+      "loss": 0.7755,
+      "step": 5272
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4905461093076512e-05,
+      "loss": 0.7971,
+      "step": 5273
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4903551621380206e-05,
+      "loss": 0.8141,
+      "step": 5274
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4901641914269325e-05,
+      "loss": 0.8257,
+      "step": 5275
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4899731971835556e-05,
+      "loss": 0.7554,
+      "step": 5276
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4897821794170589e-05,
+      "loss": 0.7141,
+      "step": 5277
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4895911381366133e-05,
+      "loss": 0.7709,
+      "step": 5278
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4894000733513907e-05,
+      "loss": 0.4229,
+      "step": 5279
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4892089850705634e-05,
+      "loss": 0.8091,
+      "step": 5280
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4890178733033058e-05,
+      "loss": 0.4749,
+      "step": 5281
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4888267380587928e-05,
+      "loss": 0.7847,
+      "step": 5282
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4886355793462007e-05,
+      "loss": 0.4317,
+      "step": 5283
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4884443971747065e-05,
+      "loss": 0.7452,
+      "step": 5284
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4882531915534892e-05,
+      "loss": 0.8033,
+      "step": 5285
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4880619624917278e-05,
+      "loss": 0.8284,
+      "step": 5286
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4878707099986037e-05,
+      "loss": 0.499,
+      "step": 5287
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4876794340832983e-05,
+      "loss": 0.8212,
+      "step": 5288
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4874881347549943e-05,
+      "loss": 0.7849,
+      "step": 5289
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4872968120228768e-05,
+      "loss": 0.8242,
+      "step": 5290
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4871054658961297e-05,
+      "loss": 0.7932,
+      "step": 5291
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4869140963839403e-05,
+      "loss": 0.8209,
+      "step": 5292
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4867227034954954e-05,
+      "loss": 0.825,
+      "step": 5293
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4865312872399842e-05,
+      "loss": 0.7983,
+      "step": 5294
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4863398476265962e-05,
+      "loss": 0.4289,
+      "step": 5295
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4861483846645216e-05,
+      "loss": 0.8207,
+      "step": 5296
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4859568983629536e-05,
+      "loss": 0.7746,
+      "step": 5297
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4857653887310841e-05,
+      "loss": 0.7456,
+      "step": 5298
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.485573855778108e-05,
+      "loss": 0.7273,
+      "step": 5299
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4853822995132201e-05,
+      "loss": 0.8145,
+      "step": 5300
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4851907199456171e-05,
+      "loss": 0.7784,
+      "step": 5301
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4849991170844964e-05,
+      "loss": 0.8151,
+      "step": 5302
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4848074909390569e-05,
+      "loss": 0.7701,
+      "step": 5303
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4846158415184982e-05,
+      "loss": 0.4195,
+      "step": 5304
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4844241688320213e-05,
+      "loss": 0.816,
+      "step": 5305
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4842324728888283e-05,
+      "loss": 0.447,
+      "step": 5306
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4840407536981216e-05,
+      "loss": 0.7865,
+      "step": 5307
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4838490112691064e-05,
+      "loss": 0.4425,
+      "step": 5308
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4836572456109874e-05,
+      "loss": 0.7708,
+      "step": 5309
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4834654567329718e-05,
+      "loss": 0.7952,
+      "step": 5310
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4832736446442667e-05,
+      "loss": 0.8436,
+      "step": 5311
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4830818093540806e-05,
+      "loss": 0.8037,
+      "step": 5312
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4828899508716235e-05,
+      "loss": 0.8011,
+      "step": 5313
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4826980692061065e-05,
+      "loss": 0.8075,
+      "step": 5314
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4825061643667416e-05,
+      "loss": 0.4378,
+      "step": 5315
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.482314236362742e-05,
+      "loss": 0.8098,
+      "step": 5316
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4821222852033216e-05,
+      "loss": 0.832,
+      "step": 5317
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.481930310897696e-05,
+      "loss": 0.8317,
+      "step": 5318
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4817383134550824e-05,
+      "loss": 0.809,
+      "step": 5319
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.481546292884697e-05,
+      "loss": 0.4148,
+      "step": 5320
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4813542491957598e-05,
+      "loss": 0.4249,
+      "step": 5321
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.48116218239749e-05,
+      "loss": 0.4264,
+      "step": 5322
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4809700924991081e-05,
+      "loss": 0.7894,
+      "step": 5323
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4807779795098372e-05,
+      "loss": 0.7825,
+      "step": 5324
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4805858434388997e-05,
+      "loss": 0.7315,
+      "step": 5325
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4803936842955203e-05,
+      "loss": 0.8045,
+      "step": 5326
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.480201502088924e-05,
+      "loss": 0.7562,
+      "step": 5327
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4800092968283375e-05,
+      "loss": 0.7976,
+      "step": 5328
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.479817068522988e-05,
+      "loss": 0.8022,
+      "step": 5329
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4796248171821049e-05,
+      "loss": 0.7742,
+      "step": 5330
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4794325428149173e-05,
+      "loss": 0.8229,
+      "step": 5331
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4792402454306563e-05,
+      "loss": 0.8218,
+      "step": 5332
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4790479250385544e-05,
+      "loss": 0.7772,
+      "step": 5333
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.478855581647844e-05,
+      "loss": 0.7839,
+      "step": 5334
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4786632152677596e-05,
+      "loss": 0.8132,
+      "step": 5335
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4784708259075364e-05,
+      "loss": 0.8189,
+      "step": 5336
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4782784135764113e-05,
+      "loss": 0.8094,
+      "step": 5337
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4780859782836212e-05,
+      "loss": 0.4612,
+      "step": 5338
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4778935200384049e-05,
+      "loss": 0.8088,
+      "step": 5339
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4777010388500023e-05,
+      "loss": 0.8184,
+      "step": 5340
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4775085347276542e-05,
+      "loss": 0.4927,
+      "step": 5341
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4773160076806025e-05,
+      "loss": 0.7979,
+      "step": 5342
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4771234577180901e-05,
+      "loss": 0.8742,
+      "step": 5343
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4769308848493613e-05,
+      "loss": 0.4606,
+      "step": 5344
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4767382890836612e-05,
+      "loss": 0.7946,
+      "step": 5345
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4765456704302362e-05,
+      "loss": 0.8527,
+      "step": 5346
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4763530288983339e-05,
+      "loss": 0.4427,
+      "step": 5347
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4761603644972022e-05,
+      "loss": 0.7853,
+      "step": 5348
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4759676772360918e-05,
+      "loss": 0.7719,
+      "step": 5349
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4757749671242523e-05,
+      "loss": 0.7577,
+      "step": 5350
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4755822341709361e-05,
+      "loss": 0.434,
+      "step": 5351
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4753894783853959e-05,
+      "loss": 0.4444,
+      "step": 5352
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4751966997768858e-05,
+      "loss": 0.7632,
+      "step": 5353
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4750038983546613e-05,
+      "loss": 0.8166,
+      "step": 5354
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.474811074127978e-05,
+      "loss": 0.7801,
+      "step": 5355
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4746182271060933e-05,
+      "loss": 0.8303,
+      "step": 5356
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4744253572982661e-05,
+      "loss": 0.8058,
+      "step": 5357
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4742324647137554e-05,
+      "loss": 0.7449,
+      "step": 5358
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4740395493618213e-05,
+      "loss": 0.4237,
+      "step": 5359
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4738466112517266e-05,
+      "loss": 0.8264,
+      "step": 5360
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4736536503927334e-05,
+      "loss": 0.7633,
+      "step": 5361
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4734606667941059e-05,
+      "loss": 0.7771,
+      "step": 5362
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4732676604651088e-05,
+      "loss": 0.7737,
+      "step": 5363
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.473074631415008e-05,
+      "loss": 0.8145,
+      "step": 5364
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4728815796530708e-05,
+      "loss": 0.781,
+      "step": 5365
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4726885051885654e-05,
+      "loss": 0.7758,
+      "step": 5366
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4724954080307612e-05,
+      "loss": 0.7696,
+      "step": 5367
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4723022881889284e-05,
+      "loss": 0.8154,
+      "step": 5368
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4721091456723388e-05,
+      "loss": 0.865,
+      "step": 5369
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4719159804902647e-05,
+      "loss": 0.769,
+      "step": 5370
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4717227926519799e-05,
+      "loss": 0.7572,
+      "step": 5371
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.471529582166759e-05,
+      "loss": 0.7906,
+      "step": 5372
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.471336349043878e-05,
+      "loss": 0.4231,
+      "step": 5373
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.471143093292614e-05,
+      "loss": 0.7992,
+      "step": 5374
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4709498149222445e-05,
+      "loss": 0.7732,
+      "step": 5375
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4707565139420492e-05,
+      "loss": 0.7931,
+      "step": 5376
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4705631903613076e-05,
+      "loss": 0.7933,
+      "step": 5377
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4703698441893015e-05,
+      "loss": 0.4525,
+      "step": 5378
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4701764754353133e-05,
+      "loss": 0.4525,
+      "step": 5379
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4699830841086262e-05,
+      "loss": 0.7849,
+      "step": 5380
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4697896702185247e-05,
+      "loss": 0.813,
+      "step": 5381
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4695962337742943e-05,
+      "loss": 0.7745,
+      "step": 5382
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4694027747852222e-05,
+      "loss": 0.7463,
+      "step": 5383
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4692092932605954e-05,
+      "loss": 0.788,
+      "step": 5384
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4690157892097037e-05,
+      "loss": 0.8058,
+      "step": 5385
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4688222626418362e-05,
+      "loss": 0.7678,
+      "step": 5386
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4686287135662845e-05,
+      "loss": 0.4513,
+      "step": 5387
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.46843514199234e-05,
+      "loss": 0.4598,
+      "step": 5388
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4682415479292966e-05,
+      "loss": 0.7833,
+      "step": 5389
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4680479313864481e-05,
+      "loss": 0.8031,
+      "step": 5390
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4678542923730901e-05,
+      "loss": 0.4355,
+      "step": 5391
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.467660630898519e-05,
+      "loss": 0.7777,
+      "step": 5392
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4674669469720319e-05,
+      "loss": 0.8394,
+      "step": 5393
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4672732406029284e-05,
+      "loss": 0.471,
+      "step": 5394
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4670795118005067e-05,
+      "loss": 0.4452,
+      "step": 5395
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4668857605740684e-05,
+      "loss": 0.7928,
+      "step": 5396
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4666919869329155e-05,
+      "loss": 0.4566,
+      "step": 5397
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.46649819088635e-05,
+      "loss": 0.8312,
+      "step": 5398
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4663043724436769e-05,
+      "loss": 0.8138,
+      "step": 5399
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4661105316142003e-05,
+      "loss": 0.8086,
+      "step": 5400
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.465916668407227e-05,
+      "loss": 0.8292,
+      "step": 5401
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4657227828320637e-05,
+      "loss": 0.7595,
+      "step": 5402
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.465528874898019e-05,
+      "loss": 0.7769,
+      "step": 5403
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4653349446144018e-05,
+      "loss": 0.4032,
+      "step": 5404
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4651409919905231e-05,
+      "loss": 0.8358,
+      "step": 5405
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4649470170356939e-05,
+      "loss": 0.4195,
+      "step": 5406
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4647530197592268e-05,
+      "loss": 0.758,
+      "step": 5407
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4645590001704356e-05,
+      "loss": 0.4583,
+      "step": 5408
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4643649582786348e-05,
+      "loss": 0.468,
+      "step": 5409
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4641708940931402e-05,
+      "loss": 0.8119,
+      "step": 5410
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4639768076232685e-05,
+      "loss": 0.8237,
+      "step": 5411
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4637826988783381e-05,
+      "loss": 0.8264,
+      "step": 5412
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4635885678676676e-05,
+      "loss": 0.4257,
+      "step": 5413
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4633944146005767e-05,
+      "loss": 0.445,
+      "step": 5414
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.463200239086387e-05,
+      "loss": 0.4519,
+      "step": 5415
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4630060413344207e-05,
+      "loss": 0.4594,
+      "step": 5416
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4628118213540009e-05,
+      "loss": 0.8661,
+      "step": 5417
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4626175791544514e-05,
+      "loss": 0.4066,
+      "step": 5418
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4624233147450985e-05,
+      "loss": 0.4811,
+      "step": 5419
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.462229028135268e-05,
+      "loss": 0.4433,
+      "step": 5420
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4620347193342876e-05,
+      "loss": 0.7686,
+      "step": 5421
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.461840388351486e-05,
+      "loss": 0.7929,
+      "step": 5422
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4616460351961927e-05,
+      "loss": 0.82,
+      "step": 5423
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4614516598777381e-05,
+      "loss": 0.7808,
+      "step": 5424
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4612572624054544e-05,
+      "loss": 0.4829,
+      "step": 5425
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4610628427886744e-05,
+      "loss": 0.8353,
+      "step": 5426
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4608684010367315e-05,
+      "loss": 0.8145,
+      "step": 5427
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4606739371589614e-05,
+      "loss": 0.7579,
+      "step": 5428
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4604794511646996e-05,
+      "loss": 0.4572,
+      "step": 5429
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4602849430632832e-05,
+      "loss": 0.4519,
+      "step": 5430
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4600904128640507e-05,
+      "loss": 0.445,
+      "step": 5431
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4598958605763409e-05,
+      "loss": 0.8523,
+      "step": 5432
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4597012862094944e-05,
+      "loss": 0.8157,
+      "step": 5433
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4595066897728519e-05,
+      "loss": 0.8011,
+      "step": 5434
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4593120712757565e-05,
+      "loss": 0.8187,
+      "step": 5435
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.459117430727551e-05,
+      "loss": 0.4555,
+      "step": 5436
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4589227681375806e-05,
+      "loss": 0.8122,
+      "step": 5437
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4587280835151906e-05,
+      "loss": 0.8162,
+      "step": 5438
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4585333768697275e-05,
+      "loss": 0.796,
+      "step": 5439
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4583386482105389e-05,
+      "loss": 0.8129,
+      "step": 5440
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4581438975469733e-05,
+      "loss": 0.7677,
+      "step": 5441
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.457949124888381e-05,
+      "loss": 0.7943,
+      "step": 5442
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4577543302441127e-05,
+      "loss": 0.4381,
+      "step": 5443
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4575595136235202e-05,
+      "loss": 0.4426,
+      "step": 5444
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4573646750359562e-05,
+      "loss": 0.4355,
+      "step": 5445
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4571698144907753e-05,
+      "loss": 0.8077,
+      "step": 5446
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.456974931997332e-05,
+      "loss": 0.8196,
+      "step": 5447
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.456780027564983e-05,
+      "loss": 0.7689,
+      "step": 5448
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4565851012030848e-05,
+      "loss": 0.8086,
+      "step": 5449
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4563901529209959e-05,
+      "loss": 0.7829,
+      "step": 5450
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4561951827280759e-05,
+      "loss": 0.7814,
+      "step": 5451
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4560001906336845e-05,
+      "loss": 0.8078,
+      "step": 5452
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4558051766471836e-05,
+      "loss": 0.4368,
+      "step": 5453
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4556101407779352e-05,
+      "loss": 0.7645,
+      "step": 5454
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4554150830353032e-05,
+      "loss": 0.7731,
+      "step": 5455
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4552200034286521e-05,
+      "loss": 0.7825,
+      "step": 5456
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4550249019673471e-05,
+      "loss": 0.7792,
+      "step": 5457
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4548297786607547e-05,
+      "loss": 0.7675,
+      "step": 5458
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4546346335182432e-05,
+      "loss": 0.8117,
+      "step": 5459
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.454439466549181e-05,
+      "loss": 0.478,
+      "step": 5460
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.454244277762938e-05,
+      "loss": 0.7856,
+      "step": 5461
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.454049067168885e-05,
+      "loss": 0.8064,
+      "step": 5462
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4538538347763932e-05,
+      "loss": 0.4247,
+      "step": 5463
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4536585805948366e-05,
+      "loss": 0.775,
+      "step": 5464
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4534633046335884e-05,
+      "loss": 0.8273,
+      "step": 5465
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.453268006902024e-05,
+      "loss": 0.4353,
+      "step": 5466
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4530726874095192e-05,
+      "loss": 0.8315,
+      "step": 5467
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4528773461654513e-05,
+      "loss": 0.7968,
+      "step": 5468
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4526819831791983e-05,
+      "loss": 0.7492,
+      "step": 5469
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.452486598460139e-05,
+      "loss": 0.7851,
+      "step": 5470
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4522911920176544e-05,
+      "loss": 0.8028,
+      "step": 5471
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4520957638611252e-05,
+      "loss": 0.4594,
+      "step": 5472
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.451900313999934e-05,
+      "loss": 0.7714,
+      "step": 5473
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4517048424434638e-05,
+      "loss": 0.7885,
+      "step": 5474
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4515093492010997e-05,
+      "loss": 0.7809,
+      "step": 5475
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4513138342822264e-05,
+      "loss": 0.7908,
+      "step": 5476
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4511182976962306e-05,
+      "loss": 0.4531,
+      "step": 5477
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4509227394525001e-05,
+      "loss": 0.7854,
+      "step": 5478
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4507271595604229e-05,
+      "loss": 0.8149,
+      "step": 5479
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.450531558029389e-05,
+      "loss": 0.8189,
+      "step": 5480
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4503359348687891e-05,
+      "loss": 0.452,
+      "step": 5481
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4501402900880143e-05,
+      "loss": 0.4389,
+      "step": 5482
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4499446236964583e-05,
+      "loss": 0.8128,
+      "step": 5483
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4497489357035139e-05,
+      "loss": 0.4518,
+      "step": 5484
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4495532261185763e-05,
+      "loss": 0.466,
+      "step": 5485
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4493574949510412e-05,
+      "loss": 0.79,
+      "step": 5486
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4491617422103055e-05,
+      "loss": 0.7841,
+      "step": 5487
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4489659679057675e-05,
+      "loss": 0.7834,
+      "step": 5488
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4487701720468252e-05,
+      "loss": 0.7775,
+      "step": 5489
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4485743546428796e-05,
+      "loss": 0.4254,
+      "step": 5490
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4483785157033307e-05,
+      "loss": 0.857,
+      "step": 5491
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4481826552375815e-05,
+      "loss": 0.7703,
+      "step": 5492
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.447986773255034e-05,
+      "loss": 0.8293,
+      "step": 5493
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4477908697650936e-05,
+      "loss": 0.4388,
+      "step": 5494
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4475949447771642e-05,
+      "loss": 0.8471,
+      "step": 5495
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4473989983006527e-05,
+      "loss": 0.4791,
+      "step": 5496
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4472030303449662e-05,
+      "loss": 0.7518,
+      "step": 5497
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4470070409195127e-05,
+      "loss": 0.8235,
+      "step": 5498
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4468110300337014e-05,
+      "loss": 0.4428,
+      "step": 5499
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4466149976969429e-05,
+      "loss": 0.765,
+      "step": 5500
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4464189439186483e-05,
+      "loss": 0.7661,
+      "step": 5501
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4462228687082298e-05,
+      "loss": 0.8065,
+      "step": 5502
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4460267720751012e-05,
+      "loss": 0.8235,
+      "step": 5503
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4458306540286767e-05,
+      "loss": 0.4358,
+      "step": 5504
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4456345145783717e-05,
+      "loss": 0.7743,
+      "step": 5505
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4454383537336025e-05,
+      "loss": 0.7938,
+      "step": 5506
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4452421715037868e-05,
+      "loss": 0.8331,
+      "step": 5507
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4450459678983433e-05,
+      "loss": 0.7716,
+      "step": 5508
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4448497429266908e-05,
+      "loss": 0.4291,
+      "step": 5509
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4446534965982508e-05,
+      "loss": 0.8511,
+      "step": 5510
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.444457228922444e-05,
+      "loss": 0.8166,
+      "step": 5511
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.444260939908694e-05,
+      "loss": 0.7431,
+      "step": 5512
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4440646295664236e-05,
+      "loss": 0.819,
+      "step": 5513
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4438682979050578e-05,
+      "loss": 0.8031,
+      "step": 5514
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4436719449340223e-05,
+      "loss": 0.7766,
+      "step": 5515
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4434755706627437e-05,
+      "loss": 0.446,
+      "step": 5516
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4432791751006496e-05,
+      "loss": 0.433,
+      "step": 5517
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4430827582571692e-05,
+      "loss": 0.4556,
+      "step": 5518
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4428863201417319e-05,
+      "loss": 0.8622,
+      "step": 5519
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4426898607637684e-05,
+      "loss": 0.4404,
+      "step": 5520
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4424933801327112e-05,
+      "loss": 0.7707,
+      "step": 5521
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4422968782579923e-05,
+      "loss": 0.7929,
+      "step": 5522
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.442100355149046e-05,
+      "loss": 0.4908,
+      "step": 5523
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4419038108153073e-05,
+      "loss": 0.7838,
+      "step": 5524
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4417072452662115e-05,
+      "loss": 0.7467,
+      "step": 5525
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4415106585111962e-05,
+      "loss": 0.7952,
+      "step": 5526
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.441314050559699e-05,
+      "loss": 0.7977,
+      "step": 5527
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4411174214211587e-05,
+      "loss": 0.4238,
+      "step": 5528
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4409207711050154e-05,
+      "loss": 0.4673,
+      "step": 5529
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4407240996207105e-05,
+      "loss": 0.7737,
+      "step": 5530
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4405274069776856e-05,
+      "loss": 0.831,
+      "step": 5531
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4403306931853833e-05,
+      "loss": 0.8352,
+      "step": 5532
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4401339582532485e-05,
+      "loss": 0.7617,
+      "step": 5533
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4399372021907257e-05,
+      "loss": 0.7908,
+      "step": 5534
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4397404250072611e-05,
+      "loss": 0.7737,
+      "step": 5535
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4395436267123017e-05,
+      "loss": 0.7606,
+      "step": 5536
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4393468073152956e-05,
+      "loss": 0.7318,
+      "step": 5537
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4391499668256921e-05,
+      "loss": 0.7782,
+      "step": 5538
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.438953105252941e-05,
+      "loss": 0.7654,
+      "step": 5539
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4387562226064939e-05,
+      "loss": 0.788,
+      "step": 5540
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4385593188958023e-05,
+      "loss": 0.7893,
+      "step": 5541
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.43836239413032e-05,
+      "loss": 0.7868,
+      "step": 5542
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4381654483195004e-05,
+      "loss": 0.4683,
+      "step": 5543
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4379684814727995e-05,
+      "loss": 0.4623,
+      "step": 5544
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4377714935996727e-05,
+      "loss": 0.8005,
+      "step": 5545
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4375744847095779e-05,
+      "loss": 0.7861,
+      "step": 5546
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.437377454811973e-05,
+      "loss": 0.8124,
+      "step": 5547
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.437180403916317e-05,
+      "loss": 0.4733,
+      "step": 5548
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4369833320320703e-05,
+      "loss": 0.7737,
+      "step": 5549
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4367862391686942e-05,
+      "loss": 0.7934,
+      "step": 5550
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4365891253356507e-05,
+      "loss": 0.8071,
+      "step": 5551
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4363919905424033e-05,
+      "loss": 0.859,
+      "step": 5552
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4361948347984162e-05,
+      "loss": 0.8014,
+      "step": 5553
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4359976581131544e-05,
+      "loss": 0.7969,
+      "step": 5554
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4358004604960844e-05,
+      "loss": 0.8308,
+      "step": 5555
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4356032419566734e-05,
+      "loss": 0.4293,
+      "step": 5556
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4354060025043897e-05,
+      "loss": 0.7826,
+      "step": 5557
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4352087421487027e-05,
+      "loss": 0.6786,
+      "step": 5558
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4350114608990819e-05,
+      "loss": 0.8406,
+      "step": 5559
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4348141587649998e-05,
+      "loss": 0.7681,
+      "step": 5560
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4346168357559276e-05,
+      "loss": 0.4471,
+      "step": 5561
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4344194918813392e-05,
+      "loss": 0.4341,
+      "step": 5562
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4342221271507086e-05,
+      "loss": 0.7906,
+      "step": 5563
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4340247415735111e-05,
+      "loss": 0.8391,
+      "step": 5564
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4338273351592235e-05,
+      "loss": 0.8147,
+      "step": 5565
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4336299079173224e-05,
+      "loss": 0.8132,
+      "step": 5566
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4334324598572864e-05,
+      "loss": 0.7983,
+      "step": 5567
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4332349909885945e-05,
+      "loss": 0.8048,
+      "step": 5568
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4330375013207274e-05,
+      "loss": 0.7824,
+      "step": 5569
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4328399908631659e-05,
+      "loss": 0.7922,
+      "step": 5570
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4326424596253927e-05,
+      "loss": 0.7613,
+      "step": 5571
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.432444907616891e-05,
+      "loss": 0.4598,
+      "step": 5572
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.432247334847145e-05,
+      "loss": 0.4317,
+      "step": 5573
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.43204974132564e-05,
+      "loss": 0.7957,
+      "step": 5574
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4318521270618622e-05,
+      "loss": 0.8402,
+      "step": 5575
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4316544920652988e-05,
+      "loss": 0.438,
+      "step": 5576
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4314568363454381e-05,
+      "loss": 0.8225,
+      "step": 5577
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4312591599117696e-05,
+      "loss": 0.4373,
+      "step": 5578
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4310614627737833e-05,
+      "loss": 0.8104,
+      "step": 5579
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4308637449409705e-05,
+      "loss": 0.7657,
+      "step": 5580
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4306660064228232e-05,
+      "loss": 0.4411,
+      "step": 5581
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.430468247228835e-05,
+      "loss": 0.8165,
+      "step": 5582
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4302704673685004e-05,
+      "loss": 0.7874,
+      "step": 5583
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4300726668513137e-05,
+      "loss": 0.7768,
+      "step": 5584
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4298748456867717e-05,
+      "loss": 0.8325,
+      "step": 5585
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4296770038843714e-05,
+      "loss": 0.7887,
+      "step": 5586
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4294791414536113e-05,
+      "loss": 0.7898,
+      "step": 5587
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4292812584039901e-05,
+      "loss": 0.7896,
+      "step": 5588
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4290833547450085e-05,
+      "loss": 0.8568,
+      "step": 5589
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4288854304861673e-05,
+      "loss": 0.8242,
+      "step": 5590
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4286874856369684e-05,
+      "loss": 0.8185,
+      "step": 5591
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4284895202069153e-05,
+      "loss": 0.8287,
+      "step": 5592
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.428291534205512e-05,
+      "loss": 0.8278,
+      "step": 5593
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4280935276422643e-05,
+      "loss": 0.8366,
+      "step": 5594
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4278955005266773e-05,
+      "loss": 0.4358,
+      "step": 5595
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4276974528682583e-05,
+      "loss": 0.7989,
+      "step": 5596
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4274993846765155e-05,
+      "loss": 0.4368,
+      "step": 5597
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4273012959609582e-05,
+      "loss": 0.7919,
+      "step": 5598
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4271031867310961e-05,
+      "loss": 0.7761,
+      "step": 5599
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4269050569964403e-05,
+      "loss": 0.8068,
+      "step": 5600
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.426706906766503e-05,
+      "loss": 0.8227,
+      "step": 5601
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.426508736050797e-05,
+      "loss": 0.8251,
+      "step": 5602
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4263105448588362e-05,
+      "loss": 0.7982,
+      "step": 5603
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4261123332001354e-05,
+      "loss": 0.817,
+      "step": 5604
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4259141010842114e-05,
+      "loss": 0.8283,
+      "step": 5605
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4257158485205805e-05,
+      "loss": 0.458,
+      "step": 5606
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.42551757551876e-05,
+      "loss": 0.7692,
+      "step": 5607
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.42531928208827e-05,
+      "loss": 0.4555,
+      "step": 5608
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4251209682386298e-05,
+      "loss": 0.7996,
+      "step": 5609
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4249226339793601e-05,
+      "loss": 0.7854,
+      "step": 5610
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4247242793199827e-05,
+      "loss": 0.761,
+      "step": 5611
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4245259042700208e-05,
+      "loss": 0.8074,
+      "step": 5612
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.424327508838998e-05,
+      "loss": 0.4517,
+      "step": 5613
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4241290930364389e-05,
+      "loss": 0.7795,
+      "step": 5614
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4239306568718696e-05,
+      "loss": 0.7784,
+      "step": 5615
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4237322003548164e-05,
+      "loss": 0.8475,
+      "step": 5616
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4235337234948075e-05,
+      "loss": 0.7985,
+      "step": 5617
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4233352263013709e-05,
+      "loss": 0.7816,
+      "step": 5618
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4231367087840368e-05,
+      "loss": 0.7393,
+      "step": 5619
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4229381709523353e-05,
+      "loss": 0.8328,
+      "step": 5620
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4227396128157988e-05,
+      "loss": 0.7901,
+      "step": 5621
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4225410343839594e-05,
+      "loss": 0.7576,
+      "step": 5622
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4223424356663505e-05,
+      "loss": 0.811,
+      "step": 5623
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4221438166725071e-05,
+      "loss": 0.788,
+      "step": 5624
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4219451774119639e-05,
+      "loss": 0.8067,
+      "step": 5625
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4217465178942583e-05,
+      "loss": 0.7671,
+      "step": 5626
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.421547838128927e-05,
+      "loss": 0.7468,
+      "step": 5627
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4213491381255089e-05,
+      "loss": 0.4082,
+      "step": 5628
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.421150417893543e-05,
+      "loss": 0.8224,
+      "step": 5629
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4209516774425702e-05,
+      "loss": 0.4308,
+      "step": 5630
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4207529167821315e-05,
+      "loss": 0.7687,
+      "step": 5631
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.420554135921769e-05,
+      "loss": 0.7976,
+      "step": 5632
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4203553348710261e-05,
+      "loss": 0.4263,
+      "step": 5633
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4201565136394474e-05,
+      "loss": 0.8419,
+      "step": 5634
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4199576722365776e-05,
+      "loss": 0.4784,
+      "step": 5635
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4197588106719629e-05,
+      "loss": 0.4105,
+      "step": 5636
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4195599289551509e-05,
+      "loss": 0.805,
+      "step": 5637
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4193610270956895e-05,
+      "loss": 0.7329,
+      "step": 5638
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4191621051031278e-05,
+      "loss": 0.7903,
+      "step": 5639
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4189631629870156e-05,
+      "loss": 0.8091,
+      "step": 5640
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4187642007569038e-05,
+      "loss": 0.4476,
+      "step": 5641
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4185652184223451e-05,
+      "loss": 0.793,
+      "step": 5642
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4183662159928915e-05,
+      "loss": 0.7997,
+      "step": 5643
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4181671934780979e-05,
+      "loss": 0.7161,
+      "step": 5644
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4179681508875183e-05,
+      "loss": 0.7795,
+      "step": 5645
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4177690882307093e-05,
+      "loss": 0.4166,
+      "step": 5646
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.417570005517227e-05,
+      "loss": 0.4044,
+      "step": 5647
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4173709027566295e-05,
+      "loss": 0.7913,
+      "step": 5648
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4171717799584759e-05,
+      "loss": 0.7983,
+      "step": 5649
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4169726371323247e-05,
+      "loss": 0.7972,
+      "step": 5650
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.416773474287738e-05,
+      "loss": 0.4241,
+      "step": 5651
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4165742914342761e-05,
+      "loss": 0.7645,
+      "step": 5652
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4163750885815026e-05,
+      "loss": 0.4827,
+      "step": 5653
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4161758657389808e-05,
+      "loss": 0.4242,
+      "step": 5654
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4159766229162748e-05,
+      "loss": 0.8428,
+      "step": 5655
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4157773601229502e-05,
+      "loss": 0.8191,
+      "step": 5656
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4155780773685735e-05,
+      "loss": 0.7957,
+      "step": 5657
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.415378774662712e-05,
+      "loss": 0.8039,
+      "step": 5658
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.415179452014934e-05,
+      "loss": 0.749,
+      "step": 5659
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4149801094348089e-05,
+      "loss": 0.4225,
+      "step": 5660
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.414780746931907e-05,
+      "loss": 0.8048,
+      "step": 5661
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.414581364515799e-05,
+      "loss": 0.8357,
+      "step": 5662
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4143819621960578e-05,
+      "loss": 0.8052,
+      "step": 5663
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4141825399822559e-05,
+      "loss": 0.4297,
+      "step": 5664
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4139830978839677e-05,
+      "loss": 0.4542,
+      "step": 5665
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4137836359107677e-05,
+      "loss": 0.4416,
+      "step": 5666
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4135841540722327e-05,
+      "loss": 0.7935,
+      "step": 5667
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4133846523779388e-05,
+      "loss": 0.8133,
+      "step": 5668
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4131851308374649e-05,
+      "loss": 0.7792,
+      "step": 5669
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4129855894603885e-05,
+      "loss": 0.8217,
+      "step": 5670
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4127860282562905e-05,
+      "loss": 0.3858,
+      "step": 5671
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4125864472347508e-05,
+      "loss": 0.7995,
+      "step": 5672
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4123868464053517e-05,
+      "loss": 0.7592,
+      "step": 5673
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4121872257776757e-05,
+      "loss": 0.4211,
+      "step": 5674
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4119875853613058e-05,
+      "loss": 0.813,
+      "step": 5675
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4117879251658278e-05,
+      "loss": 0.7809,
+      "step": 5676
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4115882452008259e-05,
+      "loss": 0.8053,
+      "step": 5677
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4113885454758873e-05,
+      "loss": 0.8002,
+      "step": 5678
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4111888260005987e-05,
+      "loss": 0.4584,
+      "step": 5679
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4109890867845493e-05,
+      "loss": 0.7921,
+      "step": 5680
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4107893278373279e-05,
+      "loss": 0.4245,
+      "step": 5681
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4105895491685244e-05,
+      "loss": 0.7796,
+      "step": 5682
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4103897507877307e-05,
+      "loss": 0.7358,
+      "step": 5683
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4101899327045386e-05,
+      "loss": 0.7856,
+      "step": 5684
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.409990094928541e-05,
+      "loss": 0.81,
+      "step": 5685
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.409790237469332e-05,
+      "loss": 0.477,
+      "step": 5686
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4095903603365067e-05,
+      "loss": 0.4236,
+      "step": 5687
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4093904635396605e-05,
+      "loss": 0.7925,
+      "step": 5688
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4091905470883908e-05,
+      "loss": 0.8378,
+      "step": 5689
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4089906109922957e-05,
+      "loss": 0.7838,
+      "step": 5690
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4087906552609729e-05,
+      "loss": 0.7732,
+      "step": 5691
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.408590679904023e-05,
+      "loss": 0.7465,
+      "step": 5692
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.408390684931046e-05,
+      "loss": 0.7793,
+      "step": 5693
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4081906703516438e-05,
+      "loss": 0.7922,
+      "step": 5694
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4079906361754186e-05,
+      "loss": 0.7902,
+      "step": 5695
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4077905824119742e-05,
+      "loss": 0.7982,
+      "step": 5696
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4075905090709149e-05,
+      "loss": 0.7493,
+      "step": 5697
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4073904161618457e-05,
+      "loss": 0.429,
+      "step": 5698
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4071903036943735e-05,
+      "loss": 0.7911,
+      "step": 5699
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4069901716781047e-05,
+      "loss": 0.7458,
+      "step": 5700
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.406790020122648e-05,
+      "loss": 0.8298,
+      "step": 5701
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.406589849037612e-05,
+      "loss": 0.4084,
+      "step": 5702
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4063896584326073e-05,
+      "loss": 0.7881,
+      "step": 5703
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4061894483172442e-05,
+      "loss": 0.7781,
+      "step": 5704
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4059892187011355e-05,
+      "loss": 0.8254,
+      "step": 5705
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.405788969593893e-05,
+      "loss": 0.4414,
+      "step": 5706
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4055887010051315e-05,
+      "loss": 0.8011,
+      "step": 5707
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4053884129444648e-05,
+      "loss": 0.4708,
+      "step": 5708
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4051881054215086e-05,
+      "loss": 0.4655,
+      "step": 5709
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4049877784458801e-05,
+      "loss": 0.7873,
+      "step": 5710
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4047874320271962e-05,
+      "loss": 0.7939,
+      "step": 5711
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4045870661750756e-05,
+      "loss": 0.4573,
+      "step": 5712
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4043866808991376e-05,
+      "loss": 0.4335,
+      "step": 5713
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4041862762090023e-05,
+      "loss": 0.8603,
+      "step": 5714
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4039858521142915e-05,
+      "loss": 0.823,
+      "step": 5715
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4037854086246266e-05,
+      "loss": 0.4463,
+      "step": 5716
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4035849457496313e-05,
+      "loss": 0.8051,
+      "step": 5717
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.403384463498929e-05,
+      "loss": 0.439,
+      "step": 5718
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4031839618821456e-05,
+      "loss": 0.7958,
+      "step": 5719
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.402983440908906e-05,
+      "loss": 0.4296,
+      "step": 5720
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.402782900588838e-05,
+      "loss": 0.7723,
+      "step": 5721
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4025823409315682e-05,
+      "loss": 0.4222,
+      "step": 5722
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4023817619467262e-05,
+      "loss": 0.798,
+      "step": 5723
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4021811636439411e-05,
+      "loss": 0.7775,
+      "step": 5724
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4019805460328434e-05,
+      "loss": 0.7861,
+      "step": 5725
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4017799091230648e-05,
+      "loss": 0.8561,
+      "step": 5726
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4015792529242376e-05,
+      "loss": 0.7872,
+      "step": 5727
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4013785774459954e-05,
+      "loss": 0.794,
+      "step": 5728
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4011778826979713e-05,
+      "loss": 0.4067,
+      "step": 5729
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.400977168689802e-05,
+      "loss": 0.8074,
+      "step": 5730
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4007764354311224e-05,
+      "loss": 0.7227,
+      "step": 5731
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4005756829315705e-05,
+      "loss": 0.7669,
+      "step": 5732
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4003749112007832e-05,
+      "loss": 0.8065,
+      "step": 5733
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4001741202483997e-05,
+      "loss": 0.8257,
+      "step": 5734
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3999733100840602e-05,
+      "loss": 0.7666,
+      "step": 5735
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.399772480717405e-05,
+      "loss": 0.7704,
+      "step": 5736
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.399571632158076e-05,
+      "loss": 0.8157,
+      "step": 5737
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3993707644157152e-05,
+      "loss": 0.4797,
+      "step": 5738
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3991698774999668e-05,
+      "loss": 0.7852,
+      "step": 5739
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3989689714204745e-05,
+      "loss": 0.8254,
+      "step": 5740
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3987680461868835e-05,
+      "loss": 0.8191,
+      "step": 5741
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3985671018088409e-05,
+      "loss": 0.7712,
+      "step": 5742
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3983661382959933e-05,
+      "loss": 0.7611,
+      "step": 5743
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3981651556579888e-05,
+      "loss": 0.7352,
+      "step": 5744
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3979641539044759e-05,
+      "loss": 0.8195,
+      "step": 5745
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3977631330451054e-05,
+      "loss": 0.7834,
+      "step": 5746
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3975620930895272e-05,
+      "loss": 0.8054,
+      "step": 5747
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.397361034047394e-05,
+      "loss": 0.8231,
+      "step": 5748
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3971599559283577e-05,
+      "loss": 0.781,
+      "step": 5749
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.396958858742072e-05,
+      "loss": 0.7956,
+      "step": 5750
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3967577424981917e-05,
+      "loss": 0.7702,
+      "step": 5751
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3965566072063716e-05,
+      "loss": 0.7871,
+      "step": 5752
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3963554528762686e-05,
+      "loss": 0.7882,
+      "step": 5753
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3961542795175396e-05,
+      "loss": 0.8024,
+      "step": 5754
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3959530871398427e-05,
+      "loss": 0.823,
+      "step": 5755
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3957518757528372e-05,
+      "loss": 0.8542,
+      "step": 5756
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3955506453661828e-05,
+      "loss": 0.4514,
+      "step": 5757
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3953493959895405e-05,
+      "loss": 0.7648,
+      "step": 5758
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.395148127632572e-05,
+      "loss": 0.8104,
+      "step": 5759
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.39494684030494e-05,
+      "loss": 0.4712,
+      "step": 5760
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.394745534016308e-05,
+      "loss": 0.7853,
+      "step": 5761
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3945442087763408e-05,
+      "loss": 0.831,
+      "step": 5762
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3943428645947035e-05,
+      "loss": 0.8651,
+      "step": 5763
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3941415014810628e-05,
+      "loss": 0.4332,
+      "step": 5764
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3939401194450857e-05,
+      "loss": 0.7747,
+      "step": 5765
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3937387184964405e-05,
+      "loss": 0.7661,
+      "step": 5766
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.393537298644796e-05,
+      "loss": 0.8092,
+      "step": 5767
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3933358598998223e-05,
+      "loss": 0.7688,
+      "step": 5768
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3931344022711903e-05,
+      "loss": 0.8187,
+      "step": 5769
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3929329257685717e-05,
+      "loss": 0.8304,
+      "step": 5770
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3927314304016392e-05,
+      "loss": 0.7919,
+      "step": 5771
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3925299161800666e-05,
+      "loss": 0.8062,
+      "step": 5772
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3923283831135282e-05,
+      "loss": 0.7795,
+      "step": 5773
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3921268312116993e-05,
+      "loss": 0.8217,
+      "step": 5774
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3919252604842564e-05,
+      "loss": 0.7712,
+      "step": 5775
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3917236709408765e-05,
+      "loss": 0.8043,
+      "step": 5776
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.391522062591238e-05,
+      "loss": 0.7677,
+      "step": 5777
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3913204354450198e-05,
+      "loss": 0.4536,
+      "step": 5778
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3911187895119015e-05,
+      "loss": 0.4369,
+      "step": 5779
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3909171248015648e-05,
+      "loss": 0.4177,
+      "step": 5780
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3907154413236901e-05,
+      "loss": 0.8187,
+      "step": 5781
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3905137390879611e-05,
+      "loss": 0.4756,
+      "step": 5782
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.390312018104061e-05,
+      "loss": 0.4116,
+      "step": 5783
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3901102783816737e-05,
+      "loss": 0.4593,
+      "step": 5784
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3899085199304856e-05,
+      "loss": 0.7422,
+      "step": 5785
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3897067427601818e-05,
+      "loss": 0.8004,
+      "step": 5786
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3895049468804505e-05,
+      "loss": 0.7736,
+      "step": 5787
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3893031323009787e-05,
+      "loss": 0.7088,
+      "step": 5788
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3891012990314561e-05,
+      "loss": 0.4904,
+      "step": 5789
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.388899447081572e-05,
+      "loss": 0.3953,
+      "step": 5790
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3886975764610171e-05,
+      "loss": 0.7965,
+      "step": 5791
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3884956871794835e-05,
+      "loss": 0.8078,
+      "step": 5792
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.388293779246663e-05,
+      "loss": 0.831,
+      "step": 5793
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3880918526722497e-05,
+      "loss": 0.457,
+      "step": 5794
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3878899074659377e-05,
+      "loss": 0.7881,
+      "step": 5795
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.387687943637422e-05,
+      "loss": 0.7748,
+      "step": 5796
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3874859611963984e-05,
+      "loss": 0.8139,
+      "step": 5797
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3872839601525647e-05,
+      "loss": 0.4067,
+      "step": 5798
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3870819405156179e-05,
+      "loss": 0.7848,
+      "step": 5799
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3868799022952573e-05,
+      "loss": 0.7729,
+      "step": 5800
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3866778455011824e-05,
+      "loss": 0.8019,
+      "step": 5801
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3864757701430935e-05,
+      "loss": 0.7495,
+      "step": 5802
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.386273676230693e-05,
+      "loss": 0.7649,
+      "step": 5803
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3860715637736817e-05,
+      "loss": 0.7799,
+      "step": 5804
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.385869432781764e-05,
+      "loss": 0.7632,
+      "step": 5805
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3856672832646434e-05,
+      "loss": 0.7938,
+      "step": 5806
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3854651152320253e-05,
+      "loss": 0.4538,
+      "step": 5807
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3852629286936154e-05,
+      "loss": 0.8341,
+      "step": 5808
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3850607236591203e-05,
+      "loss": 0.4676,
+      "step": 5809
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3848585001382482e-05,
+      "loss": 0.4407,
+      "step": 5810
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3846562581407067e-05,
+      "loss": 0.7553,
+      "step": 5811
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3844539976762063e-05,
+      "loss": 0.787,
+      "step": 5812
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3842517187544567e-05,
+      "loss": 0.7927,
+      "step": 5813
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3840494213851692e-05,
+      "loss": 0.7267,
+      "step": 5814
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.383847105578056e-05,
+      "loss": 0.7593,
+      "step": 5815
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3836447713428298e-05,
+      "loss": 0.4281,
+      "step": 5816
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3834424186892048e-05,
+      "loss": 0.834,
+      "step": 5817
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3832400476268957e-05,
+      "loss": 0.8108,
+      "step": 5818
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3830376581656181e-05,
+      "loss": 0.793,
+      "step": 5819
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3828352503150884e-05,
+      "loss": 0.4078,
+      "step": 5820
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3826328240850239e-05,
+      "loss": 0.7988,
+      "step": 5821
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.382430379485143e-05,
+      "loss": 0.8082,
+      "step": 5822
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3822279165251651e-05,
+      "loss": 0.7562,
+      "step": 5823
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3820254352148101e-05,
+      "loss": 0.7808,
+      "step": 5824
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.381822935563799e-05,
+      "loss": 0.4199,
+      "step": 5825
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3816204175818533e-05,
+      "loss": 0.457,
+      "step": 5826
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3814178812786955e-05,
+      "loss": 0.4433,
+      "step": 5827
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.38121532666405e-05,
+      "loss": 0.4248,
+      "step": 5828
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3810127537476406e-05,
+      "loss": 0.7865,
+      "step": 5829
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3808101625391928e-05,
+      "loss": 0.8079,
+      "step": 5830
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3806075530484329e-05,
+      "loss": 0.8142,
+      "step": 5831
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3804049252850877e-05,
+      "loss": 0.4519,
+      "step": 5832
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3802022792588854e-05,
+      "loss": 0.8249,
+      "step": 5833
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3799996149795545e-05,
+      "loss": 0.8648,
+      "step": 5834
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3797969324568252e-05,
+      "loss": 0.7927,
+      "step": 5835
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3795942317004279e-05,
+      "loss": 0.7918,
+      "step": 5836
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3793915127200938e-05,
+      "loss": 0.8099,
+      "step": 5837
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3791887755255556e-05,
+      "loss": 0.7761,
+      "step": 5838
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.378986020126546e-05,
+      "loss": 0.8171,
+      "step": 5839
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3787832465328e-05,
+      "loss": 0.8432,
+      "step": 5840
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3785804547540517e-05,
+      "loss": 0.4218,
+      "step": 5841
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3783776448000372e-05,
+      "loss": 0.7885,
+      "step": 5842
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3781748166804931e-05,
+      "loss": 0.8348,
+      "step": 5843
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3779719704051573e-05,
+      "loss": 0.7817,
+      "step": 5844
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.377769105983768e-05,
+      "loss": 0.7819,
+      "step": 5845
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3775662234260643e-05,
+      "loss": 0.443,
+      "step": 5846
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.377363322741787e-05,
+      "loss": 0.7912,
+      "step": 5847
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3771604039406766e-05,
+      "loss": 0.8139,
+      "step": 5848
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3769574670324753e-05,
+      "loss": 0.7661,
+      "step": 5849
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3767545120269257e-05,
+      "loss": 0.7893,
+      "step": 5850
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3765515389337715e-05,
+      "loss": 0.4487,
+      "step": 5851
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3763485477627575e-05,
+      "loss": 0.7821,
+      "step": 5852
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3761455385236289e-05,
+      "loss": 0.4351,
+      "step": 5853
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3759425112261318e-05,
+      "loss": 0.827,
+      "step": 5854
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.375739465880014e-05,
+      "loss": 0.7462,
+      "step": 5855
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3755364024950225e-05,
+      "loss": 0.784,
+      "step": 5856
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3753333210809068e-05,
+      "loss": 0.4563,
+      "step": 5857
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3751302216474168e-05,
+      "loss": 0.8282,
+      "step": 5858
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3749271042043024e-05,
+      "loss": 0.7476,
+      "step": 5859
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3747239687613159e-05,
+      "loss": 0.7628,
+      "step": 5860
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3745208153282088e-05,
+      "loss": 0.8164,
+      "step": 5861
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3743176439147352e-05,
+      "loss": 0.7747,
+      "step": 5862
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3741144545306484e-05,
+      "loss": 0.437,
+      "step": 5863
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3739112471857035e-05,
+      "loss": 0.7496,
+      "step": 5864
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3737080218896566e-05,
+      "loss": 0.7861,
+      "step": 5865
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3735047786522638e-05,
+      "loss": 0.4388,
+      "step": 5866
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3733015174832831e-05,
+      "loss": 0.7631,
+      "step": 5867
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3730982383924724e-05,
+      "loss": 0.7636,
+      "step": 5868
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3728949413895916e-05,
+      "loss": 0.4191,
+      "step": 5869
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3726916264844002e-05,
+      "loss": 0.7891,
+      "step": 5870
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3724882936866596e-05,
+      "loss": 0.7366,
+      "step": 5871
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3722849430061307e-05,
+      "loss": 0.4099,
+      "step": 5872
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3720815744525772e-05,
+      "loss": 0.7839,
+      "step": 5873
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3718781880357623e-05,
+      "loss": 0.8028,
+      "step": 5874
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.37167478376545e-05,
+      "loss": 0.4403,
+      "step": 5875
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3714713616514059e-05,
+      "loss": 0.7208,
+      "step": 5876
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3712679217033962e-05,
+      "loss": 0.7899,
+      "step": 5877
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3710644639311874e-05,
+      "loss": 0.8136,
+      "step": 5878
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3708609883445477e-05,
+      "loss": 0.735,
+      "step": 5879
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3706574949532456e-05,
+      "loss": 0.4539,
+      "step": 5880
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3704539837670505e-05,
+      "loss": 0.455,
+      "step": 5881
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3702504547957332e-05,
+      "loss": 0.7513,
+      "step": 5882
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3700469080490643e-05,
+      "loss": 0.8407,
+      "step": 5883
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3698433435368163e-05,
+      "loss": 0.8253,
+      "step": 5884
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3696397612687621e-05,
+      "loss": 0.7978,
+      "step": 5885
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3694361612546754e-05,
+      "loss": 0.7271,
+      "step": 5886
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3692325435043307e-05,
+      "loss": 0.7567,
+      "step": 5887
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3690289080275034e-05,
+      "loss": 0.7889,
+      "step": 5888
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3688252548339705e-05,
+      "loss": 0.4525,
+      "step": 5889
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3686215839335087e-05,
+      "loss": 0.7837,
+      "step": 5890
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3684178953358957e-05,
+      "loss": 0.4393,
+      "step": 5891
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.368214189050911e-05,
+      "loss": 0.7874,
+      "step": 5892
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3680104650883341e-05,
+      "loss": 0.4124,
+      "step": 5893
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3678067234579456e-05,
+      "loss": 0.4471,
+      "step": 5894
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3676029641695266e-05,
+      "loss": 0.4403,
+      "step": 5895
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3673991872328599e-05,
+      "loss": 0.4003,
+      "step": 5896
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3671953926577281e-05,
+      "loss": 0.4464,
+      "step": 5897
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3669915804539158e-05,
+      "loss": 0.8449,
+      "step": 5898
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3667877506312074e-05,
+      "loss": 0.7617,
+      "step": 5899
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3665839031993888e-05,
+      "loss": 0.4463,
+      "step": 5900
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3663800381682465e-05,
+      "loss": 0.8022,
+      "step": 5901
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3661761555475672e-05,
+      "loss": 0.4173,
+      "step": 5902
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.36597225534714e-05,
+      "loss": 0.8285,
+      "step": 5903
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3657683375767532e-05,
+      "loss": 0.7458,
+      "step": 5904
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3655644022461974e-05,
+      "loss": 0.8117,
+      "step": 5905
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3653604493652628e-05,
+      "loss": 0.7878,
+      "step": 5906
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3651564789437415e-05,
+      "loss": 0.4216,
+      "step": 5907
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3649524909914252e-05,
+      "loss": 0.8289,
+      "step": 5908
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3647484855181074e-05,
+      "loss": 0.8043,
+      "step": 5909
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3645444625335826e-05,
+      "loss": 0.7837,
+      "step": 5910
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3643404220476452e-05,
+      "loss": 0.8024,
+      "step": 5911
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3641363640700915e-05,
+      "loss": 0.7897,
+      "step": 5912
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3639322886107175e-05,
+      "loss": 0.7484,
+      "step": 5913
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3637281956793218e-05,
+      "loss": 0.4099,
+      "step": 5914
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3635240852857011e-05,
+      "loss": 0.827,
+      "step": 5915
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3633199574396556e-05,
+      "loss": 0.8806,
+      "step": 5916
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.363115812150985e-05,
+      "loss": 0.4067,
+      "step": 5917
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3629116494294901e-05,
+      "loss": 0.7431,
+      "step": 5918
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3627074692849726e-05,
+      "loss": 0.4256,
+      "step": 5919
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3625032717272348e-05,
+      "loss": 0.4376,
+      "step": 5920
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3622990567660806e-05,
+      "loss": 0.7276,
+      "step": 5921
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3620948244113135e-05,
+      "loss": 0.7626,
+      "step": 5922
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3618905746727387e-05,
+      "loss": 0.7439,
+      "step": 5923
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3616863075601618e-05,
+      "loss": 0.7839,
+      "step": 5924
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3614820230833898e-05,
+      "loss": 0.7361,
+      "step": 5925
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3612777212522303e-05,
+      "loss": 0.775,
+      "step": 5926
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.361073402076491e-05,
+      "loss": 0.8402,
+      "step": 5927
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.360869065565982e-05,
+      "loss": 0.7427,
+      "step": 5928
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3606647117305125e-05,
+      "loss": 0.4146,
+      "step": 5929
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3604603405798938e-05,
+      "loss": 0.7324,
+      "step": 5930
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.360255952123937e-05,
+      "loss": 0.8014,
+      "step": 5931
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.360051546372455e-05,
+      "loss": 0.442,
+      "step": 5932
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3598471233352611e-05,
+      "loss": 0.7993,
+      "step": 5933
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3596426830221691e-05,
+      "loss": 0.7985,
+      "step": 5934
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3594382254429948e-05,
+      "loss": 0.4677,
+      "step": 5935
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.359233750607553e-05,
+      "loss": 0.3875,
+      "step": 5936
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3590292585256613e-05,
+      "loss": 0.7689,
+      "step": 5937
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.358824749207136e-05,
+      "loss": 0.7934,
+      "step": 5938
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3586202226617964e-05,
+      "loss": 0.8043,
+      "step": 5939
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3584156788994611e-05,
+      "loss": 0.8243,
+      "step": 5940
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3582111179299502e-05,
+      "loss": 0.8016,
+      "step": 5941
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3580065397630847e-05,
+      "loss": 0.7895,
+      "step": 5942
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3578019444086855e-05,
+      "loss": 0.864,
+      "step": 5943
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3575973318765759e-05,
+      "loss": 0.7954,
+      "step": 5944
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3573927021765783e-05,
+      "loss": 0.741,
+      "step": 5945
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3571880553185175e-05,
+      "loss": 0.7664,
+      "step": 5946
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3569833913122177e-05,
+      "loss": 0.7521,
+      "step": 5947
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3567787101675052e-05,
+      "loss": 0.7803,
+      "step": 5948
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3565740118942063e-05,
+      "loss": 0.8133,
+      "step": 5949
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3563692965021482e-05,
+      "loss": 0.8083,
+      "step": 5950
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3561645640011599e-05,
+      "loss": 0.7949,
+      "step": 5951
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3559598144010688e-05,
+      "loss": 0.7691,
+      "step": 5952
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3557550477117063e-05,
+      "loss": 0.7626,
+      "step": 5953
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3555502639429019e-05,
+      "loss": 0.4571,
+      "step": 5954
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.355345463104488e-05,
+      "loss": 0.7595,
+      "step": 5955
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3551406452062961e-05,
+      "loss": 0.8062,
+      "step": 5956
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.35493581025816e-05,
+      "loss": 0.7938,
+      "step": 5957
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3547309582699131e-05,
+      "loss": 0.722,
+      "step": 5958
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3545260892513903e-05,
+      "loss": 0.7351,
+      "step": 5959
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3543212032124274e-05,
+      "loss": 0.4153,
+      "step": 5960
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3541163001628601e-05,
+      "loss": 0.7765,
+      "step": 5961
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3539113801125264e-05,
+      "loss": 0.4963,
+      "step": 5962
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3537064430712636e-05,
+      "loss": 0.8236,
+      "step": 5963
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3535014890489109e-05,
+      "loss": 0.7574,
+      "step": 5964
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3532965180553082e-05,
+      "loss": 0.7913,
+      "step": 5965
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3530915301002953e-05,
+      "loss": 0.782,
+      "step": 5966
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3528865251937142e-05,
+      "loss": 0.7486,
+      "step": 5967
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3526815033454061e-05,
+      "loss": 0.8011,
+      "step": 5968
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3524764645652146e-05,
+      "loss": 0.7952,
+      "step": 5969
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3522714088629831e-05,
+      "loss": 0.7633,
+      "step": 5970
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3520663362485564e-05,
+      "loss": 0.7948,
+      "step": 5971
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3518612467317793e-05,
+      "loss": 0.7511,
+      "step": 5972
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3516561403224987e-05,
+      "loss": 0.764,
+      "step": 5973
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3514510170305608e-05,
+      "loss": 0.4097,
+      "step": 5974
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3512458768658139e-05,
+      "loss": 0.7689,
+      "step": 5975
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3510407198381062e-05,
+      "loss": 0.4386,
+      "step": 5976
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.350835545957287e-05,
+      "loss": 0.4766,
+      "step": 5977
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3506303552332072e-05,
+      "loss": 0.8164,
+      "step": 5978
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3504251476757168e-05,
+      "loss": 0.7204,
+      "step": 5979
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3502199232946685e-05,
+      "loss": 0.7903,
+      "step": 5980
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3500146820999148e-05,
+      "loss": 0.7903,
+      "step": 5981
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3498094241013085e-05,
+      "loss": 0.4255,
+      "step": 5982
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3496041493087042e-05,
+      "loss": 0.8287,
+      "step": 5983
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.349398857731957e-05,
+      "loss": 0.7726,
+      "step": 5984
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3491935493809227e-05,
+      "loss": 0.7645,
+      "step": 5985
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3489882242654577e-05,
+      "loss": 0.4129,
+      "step": 5986
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.34878288239542e-05,
+      "loss": 0.7631,
+      "step": 5987
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3485775237806674e-05,
+      "loss": 0.4202,
+      "step": 5988
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3483721484310593e-05,
+      "loss": 0.4641,
+      "step": 5989
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3481667563564552e-05,
+      "loss": 0.437,
+      "step": 5990
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3479613475667157e-05,
+      "loss": 0.7805,
+      "step": 5991
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3477559220717029e-05,
+      "loss": 0.7744,
+      "step": 5992
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3475504798812784e-05,
+      "loss": 0.7644,
+      "step": 5993
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3473450210053057e-05,
+      "loss": 0.7878,
+      "step": 5994
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3471395454536483e-05,
+      "loss": 0.7573,
+      "step": 5995
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3469340532361714e-05,
+      "loss": 0.818,
+      "step": 5996
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3467285443627399e-05,
+      "loss": 0.8015,
+      "step": 5997
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3465230188432203e-05,
+      "loss": 0.8001,
+      "step": 5998
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.34631747668748e-05,
+      "loss": 0.8116,
+      "step": 5999
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3461119179053864e-05,
+      "loss": 0.8077,
+      "step": 6000
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3459063425068082e-05,
+      "loss": 0.7778,
+      "step": 6001
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.345700750501615e-05,
+      "loss": 0.4424,
+      "step": 6002
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3454951418996775e-05,
+      "loss": 0.7824,
+      "step": 6003
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3452895167108661e-05,
+      "loss": 0.7676,
+      "step": 6004
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3450838749450529e-05,
+      "loss": 0.7777,
+      "step": 6005
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3448782166121106e-05,
+      "loss": 0.7352,
+      "step": 6006
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3446725417219124e-05,
+      "loss": 0.8059,
+      "step": 6007
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.344466850284333e-05,
+      "loss": 0.8076,
+      "step": 6008
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3442611423092472e-05,
+      "loss": 0.434,
+      "step": 6009
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3440554178065306e-05,
+      "loss": 0.8098,
+      "step": 6010
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3438496767860602e-05,
+      "loss": 0.7836,
+      "step": 6011
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3436439192577132e-05,
+      "loss": 0.8237,
+      "step": 6012
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.343438145231368e-05,
+      "loss": 0.4349,
+      "step": 6013
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3432323547169035e-05,
+      "loss": 0.7702,
+      "step": 6014
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3430265477241995e-05,
+      "loss": 0.7531,
+      "step": 6015
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3428207242631363e-05,
+      "loss": 0.7405,
+      "step": 6016
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3426148843435957e-05,
+      "loss": 0.7565,
+      "step": 6017
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3424090279754598e-05,
+      "loss": 0.7438,
+      "step": 6018
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3422031551686114e-05,
+      "loss": 0.8389,
+      "step": 6019
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3419972659329342e-05,
+      "loss": 0.7933,
+      "step": 6020
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3417913602783128e-05,
+      "loss": 0.7772,
+      "step": 6021
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3415854382146328e-05,
+      "loss": 0.8129,
+      "step": 6022
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3413794997517799e-05,
+      "loss": 0.8617,
+      "step": 6023
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3411735448996412e-05,
+      "loss": 0.8109,
+      "step": 6024
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3409675736681042e-05,
+      "loss": 0.7676,
+      "step": 6025
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3407615860670582e-05,
+      "loss": 0.8315,
+      "step": 6026
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3405555821063911e-05,
+      "loss": 0.7605,
+      "step": 6027
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.340349561795994e-05,
+      "loss": 0.7971,
+      "step": 6028
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.340143525145757e-05,
+      "loss": 0.8134,
+      "step": 6029
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3399374721655725e-05,
+      "loss": 0.8089,
+      "step": 6030
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3397314028653323e-05,
+      "loss": 0.7574,
+      "step": 6031
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3395253172549297e-05,
+      "loss": 0.7961,
+      "step": 6032
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3393192153442587e-05,
+      "loss": 0.4075,
+      "step": 6033
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3391130971432143e-05,
+      "loss": 0.8145,
+      "step": 6034
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3389069626616914e-05,
+      "loss": 0.8179,
+      "step": 6035
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3387008119095867e-05,
+      "loss": 0.7974,
+      "step": 6036
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3384946448967978e-05,
+      "loss": 0.8235,
+      "step": 6037
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3382884616332214e-05,
+      "loss": 0.7628,
+      "step": 6038
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3380822621287573e-05,
+      "loss": 0.7205,
+      "step": 6039
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3378760463933042e-05,
+      "loss": 0.7836,
+      "step": 6040
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3376698144367626e-05,
+      "loss": 0.7255,
+      "step": 6041
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3374635662690336e-05,
+      "loss": 0.819,
+      "step": 6042
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3372573019000183e-05,
+      "loss": 0.7591,
+      "step": 6043
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3370510213396201e-05,
+      "loss": 0.8307,
+      "step": 6044
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.336844724597742e-05,
+      "loss": 0.8123,
+      "step": 6045
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3366384116842878e-05,
+      "loss": 0.7676,
+      "step": 6046
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3364320826091628e-05,
+      "loss": 0.4628,
+      "step": 6047
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.336225737382273e-05,
+      "loss": 0.7776,
+      "step": 6048
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3360193760135236e-05,
+      "loss": 0.7888,
+      "step": 6049
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3358129985128228e-05,
+      "loss": 0.732,
+      "step": 6050
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3356066048900782e-05,
+      "loss": 0.7894,
+      "step": 6051
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3354001951551987e-05,
+      "loss": 0.427,
+      "step": 6052
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3351937693180937e-05,
+      "loss": 0.775,
+      "step": 6053
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3349873273886735e-05,
+      "loss": 0.7571,
+      "step": 6054
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3347808693768496e-05,
+      "loss": 0.8077,
+      "step": 6055
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3345743952925331e-05,
+      "loss": 0.4637,
+      "step": 6056
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3343679051456372e-05,
+      "loss": 0.7724,
+      "step": 6057
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3341613989460751e-05,
+      "loss": 0.8044,
+      "step": 6058
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3339548767037607e-05,
+      "loss": 0.8073,
+      "step": 6059
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3337483384286091e-05,
+      "loss": 0.7717,
+      "step": 6060
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3335417841305362e-05,
+      "loss": 0.7858,
+      "step": 6061
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3333352138194585e-05,
+      "loss": 0.4223,
+      "step": 6062
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3331286275052929e-05,
+      "loss": 0.4641,
+      "step": 6063
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3329220251979575e-05,
+      "loss": 0.7841,
+      "step": 6064
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3327154069073708e-05,
+      "loss": 0.7606,
+      "step": 6065
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.332508772643453e-05,
+      "loss": 0.7472,
+      "step": 6066
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3323021224161244e-05,
+      "loss": 0.8063,
+      "step": 6067
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.332095456235305e-05,
+      "loss": 0.8112,
+      "step": 6068
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3318887741109177e-05,
+      "loss": 0.8007,
+      "step": 6069
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3316820760528848e-05,
+      "loss": 0.7667,
+      "step": 6070
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3314753620711297e-05,
+      "loss": 0.4688,
+      "step": 6071
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.331268632175576e-05,
+      "loss": 0.4539,
+      "step": 6072
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3310618863761496e-05,
+      "loss": 0.8056,
+      "step": 6073
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3308551246827751e-05,
+      "loss": 0.8324,
+      "step": 6074
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3306483471053798e-05,
+      "loss": 0.7314,
+      "step": 6075
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3304415536538904e-05,
+      "loss": 0.7754,
+      "step": 6076
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3302347443382346e-05,
+      "loss": 0.4261,
+      "step": 6077
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.330027919168342e-05,
+      "loss": 0.4377,
+      "step": 6078
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3298210781541412e-05,
+      "loss": 0.4195,
+      "step": 6079
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.329614221305563e-05,
+      "loss": 0.4303,
+      "step": 6080
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3294073486325378e-05,
+      "loss": 0.7629,
+      "step": 6081
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.329200460144998e-05,
+      "loss": 0.8186,
+      "step": 6082
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3289935558528757e-05,
+      "loss": 0.8255,
+      "step": 6083
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3287866357661042e-05,
+      "loss": 0.4532,
+      "step": 6084
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3285796998946178e-05,
+      "loss": 0.7396,
+      "step": 6085
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3283727482483508e-05,
+      "loss": 0.8322,
+      "step": 6086
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3281657808372391e-05,
+      "loss": 0.4256,
+      "step": 6087
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3279587976712187e-05,
+      "loss": 0.7765,
+      "step": 6088
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.327751798760227e-05,
+      "loss": 0.7625,
+      "step": 6089
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3275447841142021e-05,
+      "loss": 0.8009,
+      "step": 6090
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3273377537430815e-05,
+      "loss": 0.8633,
+      "step": 6091
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3271307076568057e-05,
+      "loss": 0.4789,
+      "step": 6092
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.326923645865314e-05,
+      "loss": 0.4288,
+      "step": 6093
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3267165683785475e-05,
+      "loss": 0.7748,
+      "step": 6094
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3265094752064473e-05,
+      "loss": 0.8038,
+      "step": 6095
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3263023663589568e-05,
+      "loss": 0.7416,
+      "step": 6096
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3260952418460181e-05,
+      "loss": 0.8406,
+      "step": 6097
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3258881016775756e-05,
+      "loss": 0.7301,
+      "step": 6098
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3256809458635735e-05,
+      "loss": 0.8333,
+      "step": 6099
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3254737744139576e-05,
+      "loss": 0.8304,
+      "step": 6100
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3252665873386736e-05,
+      "loss": 0.8525,
+      "step": 6101
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3250593846476683e-05,
+      "loss": 0.4881,
+      "step": 6102
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3248521663508895e-05,
+      "loss": 0.7979,
+      "step": 6103
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3246449324582856e-05,
+      "loss": 0.7839,
+      "step": 6104
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3244376829798059e-05,
+      "loss": 0.8147,
+      "step": 6105
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3242304179253996e-05,
+      "loss": 0.7709,
+      "step": 6106
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3240231373050172e-05,
+      "loss": 0.8492,
+      "step": 6107
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.323815841128611e-05,
+      "loss": 0.8335,
+      "step": 6108
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3236085294061326e-05,
+      "loss": 0.8051,
+      "step": 6109
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3234012021475347e-05,
+      "loss": 0.8006,
+      "step": 6110
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3231938593627706e-05,
+      "loss": 0.4159,
+      "step": 6111
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.322986501061795e-05,
+      "loss": 0.855,
+      "step": 6112
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3227791272545632e-05,
+      "loss": 0.4157,
+      "step": 6113
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3225717379510306e-05,
+      "loss": 0.7741,
+      "step": 6114
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3223643331611538e-05,
+      "loss": 0.5044,
+      "step": 6115
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3221569128948903e-05,
+      "loss": 0.4155,
+      "step": 6116
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3219494771621981e-05,
+      "loss": 0.7674,
+      "step": 6117
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3217420259730356e-05,
+      "loss": 0.796,
+      "step": 6118
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3215345593373627e-05,
+      "loss": 0.7703,
+      "step": 6119
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3213270772651395e-05,
+      "loss": 0.8034,
+      "step": 6120
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3211195797663272e-05,
+      "loss": 0.4327,
+      "step": 6121
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3209120668508875e-05,
+      "loss": 0.7618,
+      "step": 6122
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3207045385287827e-05,
+      "loss": 0.7129,
+      "step": 6123
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3204969948099761e-05,
+      "loss": 0.8162,
+      "step": 6124
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3202894357044318e-05,
+      "loss": 0.8015,
+      "step": 6125
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3200818612221145e-05,
+      "loss": 0.8,
+      "step": 6126
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3198742713729891e-05,
+      "loss": 0.4833,
+      "step": 6127
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3196666661670228e-05,
+      "loss": 0.7945,
+      "step": 6128
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3194590456141814e-05,
+      "loss": 0.4175,
+      "step": 6129
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3192514097244335e-05,
+      "loss": 0.8003,
+      "step": 6130
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3190437585077469e-05,
+      "loss": 0.806,
+      "step": 6131
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.318836091974091e-05,
+      "loss": 0.4183,
+      "step": 6132
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3186284101334357e-05,
+      "loss": 0.4589,
+      "step": 6133
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3184207129957511e-05,
+      "loss": 0.7995,
+      "step": 6134
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3182130005710092e-05,
+      "loss": 0.8127,
+      "step": 6135
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3180052728691814e-05,
+      "loss": 0.8141,
+      "step": 6136
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3177975299002412e-05,
+      "loss": 0.4126,
+      "step": 6137
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3175897716741616e-05,
+      "loss": 0.8014,
+      "step": 6138
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3173819982009173e-05,
+      "loss": 0.7651,
+      "step": 6139
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3171742094904827e-05,
+      "loss": 0.7516,
+      "step": 6140
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3169664055528341e-05,
+      "loss": 0.797,
+      "step": 6141
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.316758586397948e-05,
+      "loss": 0.7846,
+      "step": 6142
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3165507520358007e-05,
+      "loss": 0.417,
+      "step": 6143
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3163429024763712e-05,
+      "loss": 0.8155,
+      "step": 6144
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3161350377296377e-05,
+      "loss": 0.8002,
+      "step": 6145
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3159271578055795e-05,
+      "loss": 0.7697,
+      "step": 6146
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3157192627141764e-05,
+      "loss": 0.442,
+      "step": 6147
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.31551135246541e-05,
+      "loss": 0.7979,
+      "step": 6148
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3153034270692615e-05,
+      "loss": 0.8096,
+      "step": 6149
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3150954865357129e-05,
+      "loss": 0.817,
+      "step": 6150
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3148875308747475e-05,
+      "loss": 0.8078,
+      "step": 6151
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3146795600963492e-05,
+      "loss": 0.7936,
+      "step": 6152
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3144715742105024e-05,
+      "loss": 0.8045,
+      "step": 6153
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3142635732271915e-05,
+      "loss": 0.7866,
+      "step": 6154
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3140555571564037e-05,
+      "loss": 0.4837,
+      "step": 6155
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3138475260081248e-05,
+      "loss": 0.4344,
+      "step": 6156
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3136394797923424e-05,
+      "loss": 0.4899,
+      "step": 6157
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.313431418519045e-05,
+      "loss": 0.7463,
+      "step": 6158
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3132233421982204e-05,
+      "loss": 0.397,
+      "step": 6159
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3130152508398593e-05,
+      "loss": 0.8185,
+      "step": 6160
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3128071444539507e-05,
+      "loss": 0.7479,
+      "step": 6161
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3125990230504869e-05,
+      "loss": 0.7758,
+      "step": 6162
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3123908866394584e-05,
+      "loss": 0.4006,
+      "step": 6163
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3121827352308585e-05,
+      "loss": 0.7931,
+      "step": 6164
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.31197456883468e-05,
+      "loss": 0.4701,
+      "step": 6165
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3117663874609167e-05,
+      "loss": 0.7616,
+      "step": 6166
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3115581911195635e-05,
+      "loss": 0.4492,
+      "step": 6167
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3113499798206153e-05,
+      "loss": 0.754,
+      "step": 6168
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3111417535740685e-05,
+      "loss": 0.8011,
+      "step": 6169
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3109335123899191e-05,
+      "loss": 0.4516,
+      "step": 6170
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3107252562781656e-05,
+      "loss": 0.8217,
+      "step": 6171
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3105169852488051e-05,
+      "loss": 0.7911,
+      "step": 6172
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3103086993118375e-05,
+      "loss": 0.8044,
+      "step": 6173
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3101003984772617e-05,
+      "loss": 0.7735,
+      "step": 6174
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3098920827550784e-05,
+      "loss": 0.8628,
+      "step": 6175
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3096837521552881e-05,
+      "loss": 0.8625,
+      "step": 6176
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3094754066878928e-05,
+      "loss": 0.8051,
+      "step": 6177
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3092670463628954e-05,
+      "loss": 0.7814,
+      "step": 6178
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3090586711902984e-05,
+      "loss": 0.4646,
+      "step": 6179
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3088502811801063e-05,
+      "loss": 0.8198,
+      "step": 6180
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3086418763423231e-05,
+      "loss": 0.734,
+      "step": 6181
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3084334566869548e-05,
+      "loss": 0.4599,
+      "step": 6182
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3082250222240065e-05,
+      "loss": 0.4052,
+      "step": 6183
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.308016572963486e-05,
+      "loss": 0.7789,
+      "step": 6184
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3078081089153998e-05,
+      "loss": 0.7924,
+      "step": 6185
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3075996300897562e-05,
+      "loss": 0.8175,
+      "step": 6186
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3073911364965648e-05,
+      "loss": 0.7198,
+      "step": 6187
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3071826281458342e-05,
+      "loss": 0.809,
+      "step": 6188
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3069741050475757e-05,
+      "loss": 0.4068,
+      "step": 6189
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3067655672117991e-05,
+      "loss": 0.7567,
+      "step": 6190
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3065570146485172e-05,
+      "loss": 0.7618,
+      "step": 6191
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3063484473677417e-05,
+      "loss": 0.7235,
+      "step": 6192
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3061398653794857e-05,
+      "loss": 0.8297,
+      "step": 6193
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3059312686937635e-05,
+      "loss": 0.8005,
+      "step": 6194
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.305722657320589e-05,
+      "loss": 0.7417,
+      "step": 6195
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.305514031269978e-05,
+      "loss": 0.7756,
+      "step": 6196
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3053053905519465e-05,
+      "loss": 0.4345,
+      "step": 6197
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3050967351765104e-05,
+      "loss": 0.7897,
+      "step": 6198
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3048880651536876e-05,
+      "loss": 0.8024,
+      "step": 6199
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3046793804934959e-05,
+      "loss": 0.7808,
+      "step": 6200
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3044706812059542e-05,
+      "loss": 0.774,
+      "step": 6201
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3042619673010817e-05,
+      "loss": 0.7774,
+      "step": 6202
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.304053238788899e-05,
+      "loss": 0.8116,
+      "step": 6203
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3038444956794262e-05,
+      "loss": 0.8523,
+      "step": 6204
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3036357379826858e-05,
+      "loss": 0.7787,
+      "step": 6205
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3034269657086993e-05,
+      "loss": 0.7935,
+      "step": 6206
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.30321817886749e-05,
+      "loss": 0.8207,
+      "step": 6207
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3030093774690813e-05,
+      "loss": 0.4182,
+      "step": 6208
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3028005615234978e-05,
+      "loss": 0.4755,
+      "step": 6209
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3025917310407646e-05,
+      "loss": 0.7502,
+      "step": 6210
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3023828860309066e-05,
+      "loss": 0.4435,
+      "step": 6211
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3021740265039517e-05,
+      "loss": 0.8115,
+      "step": 6212
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.301965152469926e-05,
+      "loss": 0.7931,
+      "step": 6213
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3017562639388575e-05,
+      "loss": 0.8105,
+      "step": 6214
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3015473609207748e-05,
+      "loss": 0.7752,
+      "step": 6215
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3013384434257071e-05,
+      "loss": 0.7662,
+      "step": 6216
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3011295114636844e-05,
+      "loss": 0.7792,
+      "step": 6217
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3009205650447372e-05,
+      "loss": 0.4232,
+      "step": 6218
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3007116041788972e-05,
+      "loss": 0.8011,
+      "step": 6219
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3005026288761956e-05,
+      "loss": 0.7961,
+      "step": 6220
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3002936391466658e-05,
+      "loss": 0.4618,
+      "step": 6221
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.300084635000341e-05,
+      "loss": 0.7916,
+      "step": 6222
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2998756164472551e-05,
+      "loss": 0.8031,
+      "step": 6223
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2996665834974432e-05,
+      "loss": 0.7558,
+      "step": 6224
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2994575361609401e-05,
+      "loss": 0.4355,
+      "step": 6225
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2992484744477828e-05,
+      "loss": 0.8172,
+      "step": 6226
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2990393983680076e-05,
+      "loss": 0.7711,
+      "step": 6227
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2988303079316525e-05,
+      "loss": 0.7702,
+      "step": 6228
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.298621203148755e-05,
+      "loss": 0.7896,
+      "step": 6229
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2984120840293546e-05,
+      "loss": 0.8142,
+      "step": 6230
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2982029505834906e-05,
+      "loss": 0.784,
+      "step": 6231
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2979938028212033e-05,
+      "loss": 0.4572,
+      "step": 6232
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.297784640752534e-05,
+      "loss": 0.7543,
+      "step": 6233
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2975754643875242e-05,
+      "loss": 0.7489,
+      "step": 6234
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2973662737362162e-05,
+      "loss": 0.4253,
+      "step": 6235
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2971570688086528e-05,
+      "loss": 0.784,
+      "step": 6236
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.296947849614878e-05,
+      "loss": 0.7707,
+      "step": 6237
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.296738616164936e-05,
+      "loss": 0.726,
+      "step": 6238
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2965293684688722e-05,
+      "loss": 0.731,
+      "step": 6239
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.296320106536732e-05,
+      "loss": 0.771,
+      "step": 6240
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2961108303785626e-05,
+      "loss": 0.4766,
+      "step": 6241
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.29590154000441e-05,
+      "loss": 0.8184,
+      "step": 6242
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2956922354243229e-05,
+      "loss": 0.7562,
+      "step": 6243
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2954829166483493e-05,
+      "loss": 0.7489,
+      "step": 6244
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2952735836865387e-05,
+      "loss": 0.7526,
+      "step": 6245
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2950642365489407e-05,
+      "loss": 0.7485,
+      "step": 6246
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2948548752456063e-05,
+      "loss": 0.4225,
+      "step": 6247
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.294645499786586e-05,
+      "loss": 0.766,
+      "step": 6248
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2944361101819328e-05,
+      "loss": 0.7974,
+      "step": 6249
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.294226706441698e-05,
+      "loss": 0.8142,
+      "step": 6250
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2940172885759358e-05,
+      "loss": 0.7956,
+      "step": 6251
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2938078565946997e-05,
+      "loss": 0.7707,
+      "step": 6252
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2935984105080443e-05,
+      "loss": 0.7936,
+      "step": 6253
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2933889503260251e-05,
+      "loss": 0.7987,
+      "step": 6254
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2931794760586981e-05,
+      "loss": 0.4013,
+      "step": 6255
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.29296998771612e-05,
+      "loss": 0.3845,
+      "step": 6256
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2927604853083477e-05,
+      "loss": 0.8122,
+      "step": 6257
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2925509688454395e-05,
+      "loss": 0.8109,
+      "step": 6258
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2923414383374541e-05,
+      "loss": 0.4064,
+      "step": 6259
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.292131893794451e-05,
+      "loss": 0.7864,
+      "step": 6260
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2919223352264895e-05,
+      "loss": 0.7899,
+      "step": 6261
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2917127626436314e-05,
+      "loss": 0.7903,
+      "step": 6262
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2915031760559371e-05,
+      "loss": 0.7617,
+      "step": 6263
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2912935754734695e-05,
+      "loss": 0.7677,
+      "step": 6264
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2910839609062902e-05,
+      "loss": 0.7655,
+      "step": 6265
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2908743323644636e-05,
+      "loss": 0.7876,
+      "step": 6266
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2906646898580533e-05,
+      "loss": 0.4152,
+      "step": 6267
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.290455033397124e-05,
+      "loss": 0.788,
+      "step": 6268
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2902453629917414e-05,
+      "loss": 0.817,
+      "step": 6269
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.290035678651971e-05,
+      "loss": 0.8059,
+      "step": 6270
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2898259803878805e-05,
+      "loss": 0.819,
+      "step": 6271
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2896162682095361e-05,
+      "loss": 0.8262,
+      "step": 6272
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.289406542127007e-05,
+      "loss": 0.7534,
+      "step": 6273
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2891968021503609e-05,
+      "loss": 0.7632,
+      "step": 6274
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.288987048289668e-05,
+      "loss": 0.82,
+      "step": 6275
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.288777280554998e-05,
+      "loss": 0.4377,
+      "step": 6276
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2885674989564216e-05,
+      "loss": 0.7743,
+      "step": 6277
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2883577035040107e-05,
+      "loss": 0.8212,
+      "step": 6278
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2881478942078367e-05,
+      "loss": 0.7301,
+      "step": 6279
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2879380710779728e-05,
+      "loss": 0.4507,
+      "step": 6280
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.287728234124492e-05,
+      "loss": 0.374,
+      "step": 6281
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2875183833574687e-05,
+      "loss": 0.7654,
+      "step": 6282
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2873085187869776e-05,
+      "loss": 0.7645,
+      "step": 6283
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2870986404230937e-05,
+      "loss": 0.7711,
+      "step": 6284
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2868887482758936e-05,
+      "loss": 0.7577,
+      "step": 6285
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2866788423554537e-05,
+      "loss": 0.7397,
+      "step": 6286
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2864689226718516e-05,
+      "loss": 0.4597,
+      "step": 6287
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2862589892351649e-05,
+      "loss": 0.7687,
+      "step": 6288
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2860490420554727e-05,
+      "loss": 0.7934,
+      "step": 6289
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.285839081142854e-05,
+      "loss": 0.811,
+      "step": 6290
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2856291065073892e-05,
+      "loss": 0.79,
+      "step": 6291
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2854191181591587e-05,
+      "loss": 0.76,
+      "step": 6292
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.285209116108244e-05,
+      "loss": 0.4185,
+      "step": 6293
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.284999100364727e-05,
+      "loss": 0.7891,
+      "step": 6294
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.28478907093869e-05,
+      "loss": 0.4273,
+      "step": 6295
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2845790278402173e-05,
+      "loss": 0.7948,
+      "step": 6296
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2843689710793917e-05,
+      "loss": 0.8332,
+      "step": 6297
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2841589006662984e-05,
+      "loss": 0.7157,
+      "step": 6298
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2839488166110227e-05,
+      "loss": 0.4588,
+      "step": 6299
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2837387189236503e-05,
+      "loss": 0.8002,
+      "step": 6300
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2835286076142679e-05,
+      "loss": 0.8202,
+      "step": 6301
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.283318482692963e-05,
+      "loss": 0.428,
+      "step": 6302
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2831083441698229e-05,
+      "loss": 0.4209,
+      "step": 6303
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2828981920549364e-05,
+      "loss": 0.7878,
+      "step": 6304
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.282688026358393e-05,
+      "loss": 0.8287,
+      "step": 6305
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2824778470902822e-05,
+      "loss": 0.7214,
+      "step": 6306
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2822676542606944e-05,
+      "loss": 0.7749,
+      "step": 6307
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.282057447879721e-05,
+      "loss": 0.4245,
+      "step": 6308
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2818472279574538e-05,
+      "loss": 0.7495,
+      "step": 6309
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2816369945039853e-05,
+      "loss": 0.7669,
+      "step": 6310
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.281426747529408e-05,
+      "loss": 0.8178,
+      "step": 6311
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2812164870438166e-05,
+      "loss": 0.8007,
+      "step": 6312
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2810062130573047e-05,
+      "loss": 0.4244,
+      "step": 6313
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2807959255799677e-05,
+      "loss": 0.7681,
+      "step": 6314
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2805856246219013e-05,
+      "loss": 0.4139,
+      "step": 6315
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2803753101932017e-05,
+      "loss": 0.7785,
+      "step": 6316
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2801649823039658e-05,
+      "loss": 0.8255,
+      "step": 6317
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2799546409642918e-05,
+      "loss": 0.4376,
+      "step": 6318
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2797442861842773e-05,
+      "loss": 0.7922,
+      "step": 6319
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2795339179740215e-05,
+      "loss": 0.8066,
+      "step": 6320
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.279323536343624e-05,
+      "loss": 0.8209,
+      "step": 6321
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2791131413031847e-05,
+      "loss": 0.7395,
+      "step": 6322
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2789027328628053e-05,
+      "loss": 0.4541,
+      "step": 6323
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2786923110325863e-05,
+      "loss": 0.8331,
+      "step": 6324
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.27848187582263e-05,
+      "loss": 0.8416,
+      "step": 6325
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.27827142724304e-05,
+      "loss": 0.4664,
+      "step": 6326
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2780609653039187e-05,
+      "loss": 0.7478,
+      "step": 6327
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2778504900153708e-05,
+      "loss": 0.8007,
+      "step": 6328
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2776400013875006e-05,
+      "loss": 0.7538,
+      "step": 6329
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2774294994304139e-05,
+      "loss": 0.8563,
+      "step": 6330
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2772189841542164e-05,
+      "loss": 0.768,
+      "step": 6331
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.277008455569015e-05,
+      "loss": 0.8479,
+      "step": 6332
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2767979136849165e-05,
+      "loss": 0.7918,
+      "step": 6333
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2765873585120292e-05,
+      "loss": 0.7651,
+      "step": 6334
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2763767900604616e-05,
+      "loss": 0.8015,
+      "step": 6335
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2761662083403224e-05,
+      "loss": 0.7943,
+      "step": 6336
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2759556133617218e-05,
+      "loss": 0.7716,
+      "step": 6337
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2757450051347706e-05,
+      "loss": 0.797,
+      "step": 6338
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2755343836695795e-05,
+      "loss": 0.41,
+      "step": 6339
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.27532374897626e-05,
+      "loss": 0.7758,
+      "step": 6340
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.275113101064925e-05,
+      "loss": 0.7779,
+      "step": 6341
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.274902439945687e-05,
+      "loss": 0.4056,
+      "step": 6342
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2746917656286597e-05,
+      "loss": 0.8199,
+      "step": 6343
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2744810781239578e-05,
+      "loss": 0.7866,
+      "step": 6344
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2742703774416956e-05,
+      "loss": 0.8022,
+      "step": 6345
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2740596635919896e-05,
+      "loss": 0.7783,
+      "step": 6346
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2738489365849544e-05,
+      "loss": 0.8227,
+      "step": 6347
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2736381964307083e-05,
+      "loss": 0.4657,
+      "step": 6348
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2734274431393677e-05,
+      "loss": 0.4394,
+      "step": 6349
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2732166767210512e-05,
+      "loss": 0.8034,
+      "step": 6350
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2730058971858773e-05,
+      "loss": 0.7862,
+      "step": 6351
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2727951045439655e-05,
+      "loss": 0.811,
+      "step": 6352
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2725842988054357e-05,
+      "loss": 0.8263,
+      "step": 6353
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2723734799804079e-05,
+      "loss": 0.4329,
+      "step": 6354
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2721626480790041e-05,
+      "loss": 0.8307,
+      "step": 6355
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2719518031113456e-05,
+      "loss": 0.808,
+      "step": 6356
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2717409450875552e-05,
+      "loss": 0.8074,
+      "step": 6357
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2715300740177559e-05,
+      "loss": 0.7692,
+      "step": 6358
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2713191899120709e-05,
+      "loss": 0.7975,
+      "step": 6359
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2711082927806255e-05,
+      "loss": 0.8002,
+      "step": 6360
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2708973826335437e-05,
+      "loss": 0.4393,
+      "step": 6361
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2706864594809517e-05,
+      "loss": 0.7735,
+      "step": 6362
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2704755233329755e-05,
+      "loss": 0.7747,
+      "step": 6363
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2702645741997422e-05,
+      "loss": 0.4173,
+      "step": 6364
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2700536120913784e-05,
+      "loss": 0.7944,
+      "step": 6365
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2698426370180132e-05,
+      "loss": 0.825,
+      "step": 6366
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2696316489897749e-05,
+      "loss": 0.8071,
+      "step": 6367
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2694206480167927e-05,
+      "loss": 0.8279,
+      "step": 6368
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2692096341091968e-05,
+      "loss": 0.7771,
+      "step": 6369
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2689986072771173e-05,
+      "loss": 0.7841,
+      "step": 6370
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2687875675306857e-05,
+      "loss": 0.7371,
+      "step": 6371
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.268576514880034e-05,
+      "loss": 0.8073,
+      "step": 6372
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2683654493352944e-05,
+      "loss": 0.7987,
+      "step": 6373
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2681543709066e-05,
+      "loss": 0.7255,
+      "step": 6374
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2679432796040844e-05,
+      "loss": 0.8245,
+      "step": 6375
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2677321754378819e-05,
+      "loss": 0.7792,
+      "step": 6376
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2675210584181274e-05,
+      "loss": 0.7939,
+      "step": 6377
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2673099285549567e-05,
+      "loss": 0.3994,
+      "step": 6378
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.267098785858505e-05,
+      "loss": 0.7564,
+      "step": 6379
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2668876303389103e-05,
+      "loss": 0.4139,
+      "step": 6380
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2666764620063092e-05,
+      "loss": 0.417,
+      "step": 6381
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2664652808708397e-05,
+      "loss": 0.7829,
+      "step": 6382
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.266254086942641e-05,
+      "loss": 0.773,
+      "step": 6383
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2660428802318517e-05,
+      "loss": 0.7902,
+      "step": 6384
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2658316607486115e-05,
+      "loss": 0.7557,
+      "step": 6385
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.265620428503061e-05,
+      "loss": 0.8282,
+      "step": 6386
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2654091835053416e-05,
+      "loss": 0.8031,
+      "step": 6387
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2651979257655946e-05,
+      "loss": 0.7928,
+      "step": 6388
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2649866552939628e-05,
+      "loss": 0.7977,
+      "step": 6389
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2647753721005884e-05,
+      "loss": 0.4383,
+      "step": 6390
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.264564076195615e-05,
+      "loss": 0.7563,
+      "step": 6391
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2643527675891869e-05,
+      "loss": 0.7643,
+      "step": 6392
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2641414462914489e-05,
+      "loss": 0.7997,
+      "step": 6393
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2639301123125462e-05,
+      "loss": 0.7715,
+      "step": 6394
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2637187656626247e-05,
+      "loss": 0.4341,
+      "step": 6395
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.263507406351831e-05,
+      "loss": 0.8028,
+      "step": 6396
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.263296034390312e-05,
+      "loss": 0.784,
+      "step": 6397
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2630846497882163e-05,
+      "loss": 0.7795,
+      "step": 6398
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.262873252555691e-05,
+      "loss": 0.4741,
+      "step": 6399
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2626618427028861e-05,
+      "loss": 0.7257,
+      "step": 6400
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2624504202399505e-05,
+      "loss": 0.7374,
+      "step": 6401
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2622389851770346e-05,
+      "loss": 0.4159,
+      "step": 6402
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2620275375242897e-05,
+      "loss": 0.4322,
+      "step": 6403
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2618160772918662e-05,
+      "loss": 0.3986,
+      "step": 6404
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.261604604489917e-05,
+      "loss": 0.7539,
+      "step": 6405
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2613931191285941e-05,
+      "loss": 0.7405,
+      "step": 6406
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.261181621218051e-05,
+      "loss": 0.4157,
+      "step": 6407
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2609701107684413e-05,
+      "loss": 0.4079,
+      "step": 6408
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2607585877899197e-05,
+      "loss": 0.4464,
+      "step": 6409
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2605470522926413e-05,
+      "loss": 0.787,
+      "step": 6410
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.260335504286761e-05,
+      "loss": 0.8192,
+      "step": 6411
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2601239437824358e-05,
+      "loss": 0.7951,
+      "step": 6412
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2599123707898215e-05,
+      "loss": 0.7109,
+      "step": 6413
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2597007853190767e-05,
+      "loss": 0.4437,
+      "step": 6414
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2594891873803587e-05,
+      "loss": 0.7856,
+      "step": 6415
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2592775769838263e-05,
+      "loss": 0.8001,
+      "step": 6416
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2590659541396388e-05,
+      "loss": 0.4281,
+      "step": 6417
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2588543188579555e-05,
+      "loss": 0.4616,
+      "step": 6418
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2586426711489374e-05,
+      "loss": 0.7998,
+      "step": 6419
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2584310110227453e-05,
+      "loss": 0.7547,
+      "step": 6420
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2582193384895409e-05,
+      "loss": 0.7157,
+      "step": 6421
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2580076535594858e-05,
+      "loss": 0.7994,
+      "step": 6422
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2577959562427436e-05,
+      "loss": 0.7604,
+      "step": 6423
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.257584246549477e-05,
+      "loss": 0.791,
+      "step": 6424
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2573725244898506e-05,
+      "loss": 0.8266,
+      "step": 6425
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2571607900740283e-05,
+      "loss": 0.738,
+      "step": 6426
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2569490433121758e-05,
+      "loss": 0.4583,
+      "step": 6427
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2567372842144586e-05,
+      "loss": 0.7418,
+      "step": 6428
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2565255127910427e-05,
+      "loss": 0.4139,
+      "step": 6429
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2563137290520957e-05,
+      "loss": 0.7833,
+      "step": 6430
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2561019330077846e-05,
+      "loss": 0.7605,
+      "step": 6431
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.255890124668278e-05,
+      "loss": 0.7799,
+      "step": 6432
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2556783040437444e-05,
+      "loss": 0.4456,
+      "step": 6433
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2554664711443527e-05,
+      "loss": 0.7514,
+      "step": 6434
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2552546259802736e-05,
+      "loss": 0.8459,
+      "step": 6435
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2550427685616767e-05,
+      "loss": 0.4391,
+      "step": 6436
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2548308988987334e-05,
+      "loss": 0.7711,
+      "step": 6437
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2546190170016155e-05,
+      "loss": 0.7879,
+      "step": 6438
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2544071228804952e-05,
+      "loss": 0.4278,
+      "step": 6439
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2541952165455452e-05,
+      "loss": 0.7707,
+      "step": 6440
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2539832980069391e-05,
+      "loss": 0.761,
+      "step": 6441
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2537713672748506e-05,
+      "loss": 0.7202,
+      "step": 6442
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2535594243594547e-05,
+      "loss": 0.7874,
+      "step": 6443
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2533474692709262e-05,
+      "loss": 0.7803,
+      "step": 6444
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2531355020194407e-05,
+      "loss": 0.8205,
+      "step": 6445
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2529235226151752e-05,
+      "loss": 0.7662,
+      "step": 6446
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2527115310683057e-05,
+      "loss": 0.8741,
+      "step": 6447
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2524995273890108e-05,
+      "loss": 0.7709,
+      "step": 6448
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2522875115874678e-05,
+      "loss": 0.4441,
+      "step": 6449
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2520754836738556e-05,
+      "loss": 0.8008,
+      "step": 6450
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.251863443658353e-05,
+      "loss": 0.8031,
+      "step": 6451
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2516513915511407e-05,
+      "loss": 0.8066,
+      "step": 6452
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2514393273623986e-05,
+      "loss": 0.8004,
+      "step": 6453
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2512272511023075e-05,
+      "loss": 0.403,
+      "step": 6454
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2510151627810494e-05,
+      "loss": 0.7843,
+      "step": 6455
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2508030624088061e-05,
+      "loss": 0.7552,
+      "step": 6456
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2505909499957607e-05,
+      "loss": 0.8131,
+      "step": 6457
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.250378825552096e-05,
+      "loss": 0.4751,
+      "step": 6458
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2501666890879965e-05,
+      "loss": 0.4444,
+      "step": 6459
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2499545406136461e-05,
+      "loss": 0.7854,
+      "step": 6460
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.24974238013923e-05,
+      "loss": 0.7697,
+      "step": 6461
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2495302076749338e-05,
+      "loss": 0.7879,
+      "step": 6462
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.249318023230944e-05,
+      "loss": 0.791,
+      "step": 6463
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2491058268174468e-05,
+      "loss": 0.7752,
+      "step": 6464
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2488936184446302e-05,
+      "loss": 0.7761,
+      "step": 6465
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2486813981226814e-05,
+      "loss": 0.8038,
+      "step": 6466
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2484691658617893e-05,
+      "loss": 0.4377,
+      "step": 6467
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2482569216721431e-05,
+      "loss": 0.7731,
+      "step": 6468
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2480446655639323e-05,
+      "loss": 0.717,
+      "step": 6469
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2478323975473468e-05,
+      "loss": 0.463,
+      "step": 6470
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2476201176325775e-05,
+      "loss": 0.7422,
+      "step": 6471
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2474078258298162e-05,
+      "loss": 0.7898,
+      "step": 6472
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2471955221492541e-05,
+      "loss": 0.7745,
+      "step": 6473
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2469832066010843e-05,
+      "loss": 0.4489,
+      "step": 6474
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2467708791954995e-05,
+      "loss": 0.4645,
+      "step": 6475
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2465585399426937e-05,
+      "loss": 0.4489,
+      "step": 6476
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2463461888528604e-05,
+      "loss": 0.7262,
+      "step": 6477
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2461338259361949e-05,
+      "loss": 0.7761,
+      "step": 6478
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2459214512028926e-05,
+      "loss": 0.7806,
+      "step": 6479
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2457090646631488e-05,
+      "loss": 0.4339,
+      "step": 6480
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2454966663271606e-05,
+      "loss": 0.7396,
+      "step": 6481
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.245284256205125e-05,
+      "loss": 0.444,
+      "step": 6482
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.245071834307239e-05,
+      "loss": 0.7822,
+      "step": 6483
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2448594006437013e-05,
+      "loss": 0.4133,
+      "step": 6484
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.244646955224711e-05,
+      "loss": 0.4687,
+      "step": 6485
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2444344980604662e-05,
+      "loss": 0.405,
+      "step": 6486
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2442220291611679e-05,
+      "loss": 0.4582,
+      "step": 6487
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2440095485370156e-05,
+      "loss": 0.7725,
+      "step": 6488
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2437970561982111e-05,
+      "loss": 0.8304,
+      "step": 6489
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2435845521549555e-05,
+      "loss": 0.8173,
+      "step": 6490
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2433720364174508e-05,
+      "loss": 0.789,
+      "step": 6491
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2431595089959001e-05,
+      "loss": 0.4181,
+      "step": 6492
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2429469699005061e-05,
+      "loss": 0.4832,
+      "step": 6493
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2427344191414732e-05,
+      "loss": 0.7939,
+      "step": 6494
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2425218567290049e-05,
+      "loss": 0.7561,
+      "step": 6495
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2423092826733072e-05,
+      "loss": 0.7623,
+      "step": 6496
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2420966969845844e-05,
+      "loss": 0.772,
+      "step": 6497
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2418840996730435e-05,
+      "loss": 0.421,
+      "step": 6498
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2416714907488905e-05,
+      "loss": 0.7792,
+      "step": 6499
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2414588702223329e-05,
+      "loss": 0.4333,
+      "step": 6500
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2412462381035782e-05,
+      "loss": 0.834,
+      "step": 6501
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2410335944028346e-05,
+      "loss": 0.4513,
+      "step": 6502
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2408209391303112e-05,
+      "loss": 0.8408,
+      "step": 6503
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2406082722962167e-05,
+      "loss": 0.4444,
+      "step": 6504
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.240395593910762e-05,
+      "loss": 0.7429,
+      "step": 6505
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2401829039841565e-05,
+      "loss": 0.4722,
+      "step": 6506
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.239970202526612e-05,
+      "loss": 0.4092,
+      "step": 6507
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.23975748954834e-05,
+      "loss": 0.8092,
+      "step": 6508
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2395447650595524e-05,
+      "loss": 0.7437,
+      "step": 6509
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2393320290704623e-05,
+      "loss": 0.7617,
+      "step": 6510
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.239119281591282e-05,
+      "loss": 0.8496,
+      "step": 6511
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2389065226322266e-05,
+      "loss": 0.4512,
+      "step": 6512
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2386937522035092e-05,
+      "loss": 0.7999,
+      "step": 6513
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2384809703153456e-05,
+      "loss": 0.7741,
+      "step": 6514
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2382681769779506e-05,
+      "loss": 0.4347,
+      "step": 6515
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2380553722015405e-05,
+      "loss": 0.7447,
+      "step": 6516
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.237842555996332e-05,
+      "loss": 0.7656,
+      "step": 6517
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2376297283725423e-05,
+      "loss": 0.4157,
+      "step": 6518
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2374168893403886e-05,
+      "loss": 0.7931,
+      "step": 6519
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.237204038910089e-05,
+      "loss": 0.4622,
+      "step": 6520
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2369911770918626e-05,
+      "loss": 0.4429,
+      "step": 6521
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2367783038959284e-05,
+      "loss": 0.4426,
+      "step": 6522
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2365654193325067e-05,
+      "loss": 0.7958,
+      "step": 6523
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2363525234118176e-05,
+      "loss": 0.8009,
+      "step": 6524
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.236139616144082e-05,
+      "loss": 0.7627,
+      "step": 6525
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.235926697539521e-05,
+      "loss": 0.4655,
+      "step": 6526
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2357137676083572e-05,
+      "loss": 0.7625,
+      "step": 6527
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.235500826360813e-05,
+      "loss": 0.8045,
+      "step": 6528
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2352878738071113e-05,
+      "loss": 0.7806,
+      "step": 6529
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.235074909957476e-05,
+      "loss": 0.8005,
+      "step": 6530
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2348619348221309e-05,
+      "loss": 0.7745,
+      "step": 6531
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2346489484113012e-05,
+      "loss": 0.8054,
+      "step": 6532
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2344359507352115e-05,
+      "loss": 0.7402,
+      "step": 6533
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2342229418040883e-05,
+      "loss": 0.8109,
+      "step": 6534
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2340099216281575e-05,
+      "loss": 0.8293,
+      "step": 6535
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2337968902176463e-05,
+      "loss": 0.7391,
+      "step": 6536
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.233583847582782e-05,
+      "loss": 0.7774,
+      "step": 6537
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2333707937337923e-05,
+      "loss": 0.4356,
+      "step": 6538
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2331577286809062e-05,
+      "loss": 0.7627,
+      "step": 6539
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2329446524343522e-05,
+      "loss": 0.7475,
+      "step": 6540
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2327315650043605e-05,
+      "loss": 0.796,
+      "step": 6541
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2325184664011604e-05,
+      "loss": 0.4525,
+      "step": 6542
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2323053566349834e-05,
+      "loss": 0.7804,
+      "step": 6543
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2320922357160601e-05,
+      "loss": 0.4498,
+      "step": 6544
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2318791036546223e-05,
+      "loss": 0.7881,
+      "step": 6545
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2316659604609029e-05,
+      "loss": 0.806,
+      "step": 6546
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2314528061451336e-05,
+      "loss": 0.4497,
+      "step": 6547
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2312396407175483e-05,
+      "loss": 0.7662,
+      "step": 6548
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2310264641883812e-05,
+      "loss": 0.8301,
+      "step": 6549
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2308132765678659e-05,
+      "loss": 0.4317,
+      "step": 6550
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2306000778662381e-05,
+      "loss": 0.4255,
+      "step": 6551
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2303868680937328e-05,
+      "loss": 0.7988,
+      "step": 6552
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2301736472605858e-05,
+      "loss": 0.7732,
+      "step": 6553
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2299604153770344e-05,
+      "loss": 0.7542,
+      "step": 6554
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.229747172453315e-05,
+      "loss": 0.7869,
+      "step": 6555
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.229533918499665e-05,
+      "loss": 0.7666,
+      "step": 6556
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2293206535263234e-05,
+      "loss": 0.7623,
+      "step": 6557
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2291073775435276e-05,
+      "loss": 0.4513,
+      "step": 6558
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.228894090561518e-05,
+      "loss": 0.4673,
+      "step": 6559
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2286807925905335e-05,
+      "loss": 0.442,
+      "step": 6560
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2284674836408149e-05,
+      "loss": 0.4382,
+      "step": 6561
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2282541637226024e-05,
+      "loss": 0.823,
+      "step": 6562
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2280408328461371e-05,
+      "loss": 0.7774,
+      "step": 6563
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2278274910216617e-05,
+      "loss": 0.8356,
+      "step": 6564
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2276141382594175e-05,
+      "loss": 0.7658,
+      "step": 6565
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2274007745696484e-05,
+      "loss": 0.8583,
+      "step": 6566
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.227187399962597e-05,
+      "loss": 0.7495,
+      "step": 6567
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2269740144485072e-05,
+      "loss": 0.8065,
+      "step": 6568
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2267606180376241e-05,
+      "loss": 0.7558,
+      "step": 6569
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2265472107401919e-05,
+      "loss": 0.8142,
+      "step": 6570
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2263337925664565e-05,
+      "loss": 0.7667,
+      "step": 6571
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2261203635266637e-05,
+      "loss": 0.7927,
+      "step": 6572
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2259069236310601e-05,
+      "loss": 0.8082,
+      "step": 6573
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2256934728898927e-05,
+      "loss": 0.8078,
+      "step": 6574
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.225480011313409e-05,
+      "loss": 0.7955,
+      "step": 6575
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2252665389118572e-05,
+      "loss": 0.8011,
+      "step": 6576
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.225053055695486e-05,
+      "loss": 0.7648,
+      "step": 6577
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2248395616745444e-05,
+      "loss": 0.8154,
+      "step": 6578
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2246260568592813e-05,
+      "loss": 0.7635,
+      "step": 6579
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2244125412599481e-05,
+      "loss": 0.7568,
+      "step": 6580
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2241990148867948e-05,
+      "loss": 0.7934,
+      "step": 6581
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2239854777500728e-05,
+      "loss": 0.8145,
+      "step": 6582
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2237719298600337e-05,
+      "loss": 0.7763,
+      "step": 6583
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2235583712269296e-05,
+      "loss": 0.4123,
+      "step": 6584
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2233448018610134e-05,
+      "loss": 0.4776,
+      "step": 6585
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2231312217725381e-05,
+      "loss": 0.747,
+      "step": 6586
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2229176309717581e-05,
+      "loss": 0.4393,
+      "step": 6587
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2227040294689267e-05,
+      "loss": 0.7851,
+      "step": 6588
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2224904172742994e-05,
+      "loss": 0.4739,
+      "step": 6589
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2222767943981315e-05,
+      "loss": 0.404,
+      "step": 6590
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2220631608506788e-05,
+      "loss": 0.4122,
+      "step": 6591
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.221849516642197e-05,
+      "loss": 0.7951,
+      "step": 6592
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.221635861782944e-05,
+      "loss": 0.7158,
+      "step": 6593
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2214221962831763e-05,
+      "loss": 0.4645,
+      "step": 6594
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2212085201531519e-05,
+      "loss": 0.397,
+      "step": 6595
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2209948334031296e-05,
+      "loss": 0.4034,
+      "step": 6596
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2207811360433677e-05,
+      "loss": 0.401,
+      "step": 6597
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2205674280841264e-05,
+      "loss": 0.4263,
+      "step": 6598
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2203537095356647e-05,
+      "loss": 0.746,
+      "step": 6599
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2201399804082436e-05,
+      "loss": 0.7474,
+      "step": 6600
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2199262407121237e-05,
+      "loss": 0.4132,
+      "step": 6601
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2197124904575668e-05,
+      "loss": 0.4205,
+      "step": 6602
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2194987296548344e-05,
+      "loss": 0.802,
+      "step": 6603
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2192849583141891e-05,
+      "loss": 0.8018,
+      "step": 6604
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.219071176445894e-05,
+      "loss": 0.4555,
+      "step": 6605
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2188573840602127e-05,
+      "loss": 0.7818,
+      "step": 6606
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2186435811674087e-05,
+      "loss": 0.7946,
+      "step": 6607
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2184297677777463e-05,
+      "loss": 0.7458,
+      "step": 6608
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2182159439014913e-05,
+      "loss": 0.7589,
+      "step": 6609
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2180021095489088e-05,
+      "loss": 0.4407,
+      "step": 6610
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.217788264730264e-05,
+      "loss": 0.7456,
+      "step": 6611
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2175744094558248e-05,
+      "loss": 0.4319,
+      "step": 6612
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2173605437358571e-05,
+      "loss": 0.8354,
+      "step": 6613
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2171466675806288e-05,
+      "loss": 0.8001,
+      "step": 6614
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2169327810004075e-05,
+      "loss": 0.8304,
+      "step": 6615
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2167188840054622e-05,
+      "loss": 0.7937,
+      "step": 6616
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2165049766060614e-05,
+      "loss": 0.7689,
+      "step": 6617
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.216291058812475e-05,
+      "loss": 0.8976,
+      "step": 6618
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.216077130634973e-05,
+      "loss": 0.4121,
+      "step": 6619
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.215863192083825e-05,
+      "loss": 0.8006,
+      "step": 6620
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2156492431693036e-05,
+      "loss": 0.8216,
+      "step": 6621
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2154352839016785e-05,
+      "loss": 0.7796,
+      "step": 6622
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.215221314291223e-05,
+      "loss": 0.8196,
+      "step": 6623
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2150073343482087e-05,
+      "loss": 0.7941,
+      "step": 6624
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2147933440829091e-05,
+      "loss": 0.7741,
+      "step": 6625
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2145793435055975e-05,
+      "loss": 0.8159,
+      "step": 6626
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2143653326265477e-05,
+      "loss": 0.8022,
+      "step": 6627
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2141513114560347e-05,
+      "loss": 0.8064,
+      "step": 6628
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2139372800043323e-05,
+      "loss": 0.8254,
+      "step": 6629
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.213723238281717e-05,
+      "loss": 0.7342,
+      "step": 6630
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.213509186298464e-05,
+      "loss": 0.7798,
+      "step": 6631
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2132951240648508e-05,
+      "loss": 0.8184,
+      "step": 6632
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2130810515911529e-05,
+      "loss": 0.7568,
+      "step": 6633
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2128669688876488e-05,
+      "loss": 0.7809,
+      "step": 6634
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2126528759646159e-05,
+      "loss": 0.8106,
+      "step": 6635
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2124387728323328e-05,
+      "loss": 0.8003,
+      "step": 6636
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2122246595010782e-05,
+      "loss": 0.4849,
+      "step": 6637
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2120105359811313e-05,
+      "loss": 0.4096,
+      "step": 6638
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2117964022827726e-05,
+      "loss": 0.7774,
+      "step": 6639
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2115822584162818e-05,
+      "loss": 0.8247,
+      "step": 6640
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2113681043919399e-05,
+      "loss": 0.463,
+      "step": 6641
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2111539402200282e-05,
+      "loss": 0.8034,
+      "step": 6642
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2109397659108291e-05,
+      "loss": 0.4735,
+      "step": 6643
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2107255814746239e-05,
+      "loss": 0.8186,
+      "step": 6644
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2105113869216958e-05,
+      "loss": 0.8051,
+      "step": 6645
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2102971822623284e-05,
+      "loss": 0.7995,
+      "step": 6646
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.210082967506805e-05,
+      "loss": 0.8057,
+      "step": 6647
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2098687426654098e-05,
+      "loss": 0.7597,
+      "step": 6648
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2096545077484278e-05,
+      "loss": 0.4347,
+      "step": 6649
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2094402627661447e-05,
+      "loss": 0.7615,
+      "step": 6650
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.209226007728845e-05,
+      "loss": 0.8064,
+      "step": 6651
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2090117426468158e-05,
+      "loss": 0.7727,
+      "step": 6652
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2087974675303433e-05,
+      "loss": 0.7835,
+      "step": 6653
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2085831823897145e-05,
+      "loss": 0.707,
+      "step": 6654
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2083688872352178e-05,
+      "loss": 0.4411,
+      "step": 6655
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2081545820771403e-05,
+      "loss": 0.7857,
+      "step": 6656
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2079402669257712e-05,
+      "loss": 0.3996,
+      "step": 6657
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2077259417913996e-05,
+      "loss": 0.8221,
+      "step": 6658
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2075116066843147e-05,
+      "loss": 0.446,
+      "step": 6659
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2072972616148066e-05,
+      "loss": 0.7857,
+      "step": 6660
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2070829065931658e-05,
+      "loss": 0.7618,
+      "step": 6661
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2068685416296833e-05,
+      "loss": 0.3735,
+      "step": 6662
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2066541667346506e-05,
+      "loss": 0.7861,
+      "step": 6663
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2064397819183596e-05,
+      "loss": 0.8063,
+      "step": 6664
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2062253871911024e-05,
+      "loss": 0.8075,
+      "step": 6665
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2060109825631728e-05,
+      "loss": 0.8387,
+      "step": 6666
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2057965680448627e-05,
+      "loss": 0.7661,
+      "step": 6667
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2055821436464671e-05,
+      "loss": 0.8292,
+      "step": 6668
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.20536770937828e-05,
+      "loss": 0.4491,
+      "step": 6669
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.205153265250596e-05,
+      "loss": 0.7741,
+      "step": 6670
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2049388112737104e-05,
+      "loss": 0.7657,
+      "step": 6671
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2047243474579189e-05,
+      "loss": 0.3782,
+      "step": 6672
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2045098738135182e-05,
+      "loss": 0.7889,
+      "step": 6673
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2042953903508038e-05,
+      "loss": 0.7729,
+      "step": 6674
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2040808970800741e-05,
+      "loss": 0.7838,
+      "step": 6675
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.203866394011626e-05,
+      "loss": 0.4382,
+      "step": 6676
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2036518811557576e-05,
+      "loss": 0.7226,
+      "step": 6677
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.203437358522768e-05,
+      "loss": 0.8545,
+      "step": 6678
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2032228261229553e-05,
+      "loss": 0.7872,
+      "step": 6679
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2030082839666198e-05,
+      "loss": 0.4244,
+      "step": 6680
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2027937320640609e-05,
+      "loss": 0.7315,
+      "step": 6681
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2025791704255794e-05,
+      "loss": 0.7828,
+      "step": 6682
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2023645990614758e-05,
+      "loss": 0.795,
+      "step": 6683
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.202150017982052e-05,
+      "loss": 0.4234,
+      "step": 6684
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2019354271976094e-05,
+      "loss": 0.7427,
+      "step": 6685
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2017208267184501e-05,
+      "loss": 0.7553,
+      "step": 6686
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2015062165548777e-05,
+      "loss": 0.462,
+      "step": 6687
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2012915967171947e-05,
+      "loss": 0.7635,
+      "step": 6688
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.201076967215705e-05,
+      "loss": 0.4312,
+      "step": 6689
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2008623280607123e-05,
+      "loss": 0.7437,
+      "step": 6690
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.200647679262522e-05,
+      "loss": 0.7939,
+      "step": 6691
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2004330208314385e-05,
+      "loss": 0.7786,
+      "step": 6692
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2002183527777678e-05,
+      "loss": 0.744,
+      "step": 6693
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2000036751118159e-05,
+      "loss": 0.4389,
+      "step": 6694
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1997889878438888e-05,
+      "loss": 0.7792,
+      "step": 6695
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.199574290984294e-05,
+      "loss": 0.7834,
+      "step": 6696
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1993595845433381e-05,
+      "loss": 0.7082,
+      "step": 6697
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1991448685313298e-05,
+      "loss": 0.43,
+      "step": 6698
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1989301429585766e-05,
+      "loss": 0.7705,
+      "step": 6699
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1987154078353882e-05,
+      "loss": 0.7935,
+      "step": 6700
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.198500663172073e-05,
+      "loss": 0.7801,
+      "step": 6701
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1982859089789409e-05,
+      "loss": 0.8178,
+      "step": 6702
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1980711452663023e-05,
+      "loss": 0.7439,
+      "step": 6703
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1978563720444674e-05,
+      "loss": 0.7514,
+      "step": 6704
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1976415893237477e-05,
+      "loss": 0.4578,
+      "step": 6705
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.197426797114454e-05,
+      "loss": 0.4554,
+      "step": 6706
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.197211995426899e-05,
+      "loss": 0.7727,
+      "step": 6707
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1969971842713946e-05,
+      "loss": 0.7725,
+      "step": 6708
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.196782363658254e-05,
+      "loss": 0.7988,
+      "step": 6709
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1965675335977905e-05,
+      "loss": 0.7319,
+      "step": 6710
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1963526941003176e-05,
+      "loss": 0.7536,
+      "step": 6711
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1961378451761499e-05,
+      "loss": 0.7501,
+      "step": 6712
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1959229868356017e-05,
+      "loss": 0.7606,
+      "step": 6713
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1957081190889883e-05,
+      "loss": 0.8018,
+      "step": 6714
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1954932419466252e-05,
+      "loss": 0.7594,
+      "step": 6715
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1952783554188287e-05,
+      "loss": 0.7475,
+      "step": 6716
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1950634595159152e-05,
+      "loss": 0.8196,
+      "step": 6717
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1948485542482014e-05,
+      "loss": 0.8231,
+      "step": 6718
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1946336396260049e-05,
+      "loss": 0.7132,
+      "step": 6719
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1944187156596432e-05,
+      "loss": 0.8118,
+      "step": 6720
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1942037823594353e-05,
+      "loss": 0.6988,
+      "step": 6721
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1939888397356989e-05,
+      "loss": 0.7777,
+      "step": 6722
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1937738877987544e-05,
+      "loss": 0.8047,
+      "step": 6723
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1935589265589203e-05,
+      "loss": 0.4734,
+      "step": 6724
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1933439560265174e-05,
+      "loss": 0.7485,
+      "step": 6725
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1931289762118659e-05,
+      "loss": 0.7464,
+      "step": 6726
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1929139871252869e-05,
+      "loss": 0.4394,
+      "step": 6727
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1926989887771018e-05,
+      "loss": 0.7759,
+      "step": 6728
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.192483981177632e-05,
+      "loss": 0.8066,
+      "step": 6729
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1922689643372007e-05,
+      "loss": 0.748,
+      "step": 6730
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1920539382661298e-05,
+      "loss": 0.8147,
+      "step": 6731
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1918389029747432e-05,
+      "loss": 0.7604,
+      "step": 6732
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1916238584733639e-05,
+      "loss": 0.435,
+      "step": 6733
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1914088047723165e-05,
+      "loss": 0.4437,
+      "step": 6734
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.191193741881925e-05,
+      "loss": 0.466,
+      "step": 6735
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1909786698125145e-05,
+      "loss": 0.8156,
+      "step": 6736
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1907635885744106e-05,
+      "loss": 0.7933,
+      "step": 6737
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.190548498177939e-05,
+      "loss": 0.7695,
+      "step": 6738
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.190333398633426e-05,
+      "loss": 0.7366,
+      "step": 6739
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1901182899511983e-05,
+      "loss": 0.431,
+      "step": 6740
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1899031721415832e-05,
+      "loss": 0.8415,
+      "step": 6741
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1896880452149077e-05,
+      "loss": 0.7553,
+      "step": 6742
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1894729091815008e-05,
+      "loss": 0.8216,
+      "step": 6743
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1892577640516903e-05,
+      "loss": 0.7319,
+      "step": 6744
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.189042609835805e-05,
+      "loss": 0.7255,
+      "step": 6745
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1888274465441747e-05,
+      "loss": 0.7314,
+      "step": 6746
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.188612274187129e-05,
+      "loss": 0.4178,
+      "step": 6747
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.188397092774998e-05,
+      "loss": 0.4344,
+      "step": 6748
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.188181902318112e-05,
+      "loss": 0.8034,
+      "step": 6749
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.187966702826803e-05,
+      "loss": 0.4225,
+      "step": 6750
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1877514943114016e-05,
+      "loss": 0.762,
+      "step": 6751
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1875362767822405e-05,
+      "loss": 0.7767,
+      "step": 6752
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1873210502496515e-05,
+      "loss": 0.452,
+      "step": 6753
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1871058147239675e-05,
+      "loss": 0.7795,
+      "step": 6754
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1868905702155222e-05,
+      "loss": 0.7263,
+      "step": 6755
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1866753167346488e-05,
+      "loss": 0.8172,
+      "step": 6756
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1864600542916813e-05,
+      "loss": 0.4432,
+      "step": 6757
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1862447828969545e-05,
+      "loss": 0.4219,
+      "step": 6758
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1860295025608036e-05,
+      "loss": 0.4558,
+      "step": 6759
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1858142132935635e-05,
+      "loss": 0.7857,
+      "step": 6760
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1855989151055702e-05,
+      "loss": 0.8281,
+      "step": 6761
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1853836080071603e-05,
+      "loss": 0.8199,
+      "step": 6762
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1851682920086701e-05,
+      "loss": 0.7486,
+      "step": 6763
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1849529671204366e-05,
+      "loss": 0.7832,
+      "step": 6764
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1847376333527973e-05,
+      "loss": 0.7233,
+      "step": 6765
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1845222907160907e-05,
+      "loss": 0.5057,
+      "step": 6766
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1843069392206545e-05,
+      "loss": 0.7608,
+      "step": 6767
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1840915788768281e-05,
+      "loss": 0.7622,
+      "step": 6768
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1838762096949506e-05,
+      "loss": 0.7626,
+      "step": 6769
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1836608316853616e-05,
+      "loss": 0.7625,
+      "step": 6770
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1834454448584008e-05,
+      "loss": 0.8188,
+      "step": 6771
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.183230049224409e-05,
+      "loss": 0.7598,
+      "step": 6772
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1830146447937273e-05,
+      "loss": 0.7624,
+      "step": 6773
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1827992315766967e-05,
+      "loss": 0.6961,
+      "step": 6774
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1825838095836594e-05,
+      "loss": 0.7388,
+      "step": 6775
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1823683788249576e-05,
+      "loss": 0.7668,
+      "step": 6776
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1821529393109334e-05,
+      "loss": 0.7669,
+      "step": 6777
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1819374910519303e-05,
+      "loss": 0.7996,
+      "step": 6778
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1817220340582911e-05,
+      "loss": 0.7977,
+      "step": 6779
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1815065683403606e-05,
+      "loss": 0.7863,
+      "step": 6780
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1812910939084823e-05,
+      "loss": 0.8393,
+      "step": 6781
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1810756107730014e-05,
+      "loss": 0.4552,
+      "step": 6782
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1808601189442627e-05,
+      "loss": 0.7732,
+      "step": 6783
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1806446184326126e-05,
+      "loss": 0.7589,
+      "step": 6784
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1804291092483957e-05,
+      "loss": 0.4313,
+      "step": 6785
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1802135914019592e-05,
+      "loss": 0.7849,
+      "step": 6786
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.17999806490365e-05,
+      "loss": 0.3993,
+      "step": 6787
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1797825297638146e-05,
+      "loss": 0.7651,
+      "step": 6788
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1795669859928014e-05,
+      "loss": 0.4496,
+      "step": 6789
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.179351433600958e-05,
+      "loss": 0.7674,
+      "step": 6790
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.179135872598633e-05,
+      "loss": 0.7677,
+      "step": 6791
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1789203029961753e-05,
+      "loss": 0.7765,
+      "step": 6792
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1787047248039342e-05,
+      "loss": 0.4124,
+      "step": 6793
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1784891380322594e-05,
+      "loss": 0.7867,
+      "step": 6794
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1782735426915006e-05,
+      "loss": 0.7693,
+      "step": 6795
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.178057938792009e-05,
+      "loss": 0.7556,
+      "step": 6796
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1778423263441348e-05,
+      "loss": 0.7015,
+      "step": 6797
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1776267053582297e-05,
+      "loss": 0.7384,
+      "step": 6798
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1774110758446457e-05,
+      "loss": 0.7992,
+      "step": 6799
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1771954378137349e-05,
+      "loss": 0.7856,
+      "step": 6800
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1769797912758491e-05,
+      "loss": 0.4553,
+      "step": 6801
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1767641362413422e-05,
+      "loss": 0.4483,
+      "step": 6802
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1765484727205672e-05,
+      "loss": 0.7821,
+      "step": 6803
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1763328007238777e-05,
+      "loss": 0.7949,
+      "step": 6804
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1761171202616284e-05,
+      "loss": 0.7553,
+      "step": 6805
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1759014313441732e-05,
+      "loss": 0.7505,
+      "step": 6806
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1756857339818681e-05,
+      "loss": 0.7856,
+      "step": 6807
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1754700281850675e-05,
+      "loss": 0.4378,
+      "step": 6808
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1752543139641276e-05,
+      "loss": 0.7921,
+      "step": 6809
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1750385913294049e-05,
+      "loss": 0.7691,
+      "step": 6810
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1748228602912557e-05,
+      "loss": 0.8054,
+      "step": 6811
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.174607120860037e-05,
+      "loss": 0.7624,
+      "step": 6812
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1743913730461065e-05,
+      "loss": 0.7636,
+      "step": 6813
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1741756168598221e-05,
+      "loss": 0.8026,
+      "step": 6814
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1739598523115413e-05,
+      "loss": 0.784,
+      "step": 6815
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1737440794116238e-05,
+      "loss": 0.7676,
+      "step": 6816
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1735282981704278e-05,
+      "loss": 0.7663,
+      "step": 6817
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.173312508598313e-05,
+      "loss": 0.742,
+      "step": 6818
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1730967107056394e-05,
+      "loss": 0.7904,
+      "step": 6819
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1728809045027672e-05,
+      "loss": 0.4316,
+      "step": 6820
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1726650900000571e-05,
+      "loss": 0.8003,
+      "step": 6821
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1724492672078696e-05,
+      "loss": 0.7391,
+      "step": 6822
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1722334361365668e-05,
+      "loss": 0.4697,
+      "step": 6823
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.17201759679651e-05,
+      "loss": 0.4354,
+      "step": 6824
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.171801749198062e-05,
+      "loss": 0.4666,
+      "step": 6825
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1715858933515847e-05,
+      "loss": 0.424,
+      "step": 6826
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.171370029267442e-05,
+      "loss": 0.7519,
+      "step": 6827
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1711541569559968e-05,
+      "loss": 0.7757,
+      "step": 6828
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.170938276427613e-05,
+      "loss": 0.7778,
+      "step": 6829
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1707223876926546e-05,
+      "loss": 0.4343,
+      "step": 6830
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1705064907614862e-05,
+      "loss": 0.7832,
+      "step": 6831
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1702905856444735e-05,
+      "loss": 0.8365,
+      "step": 6832
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.170074672351981e-05,
+      "loss": 0.7851,
+      "step": 6833
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1698587508943749e-05,
+      "loss": 0.8012,
+      "step": 6834
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1696428212820211e-05,
+      "loss": 0.7435,
+      "step": 6835
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1694268835252869e-05,
+      "loss": 0.4287,
+      "step": 6836
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1692109376345384e-05,
+      "loss": 0.8072,
+      "step": 6837
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1689949836201433e-05,
+      "loss": 0.7744,
+      "step": 6838
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1687790214924693e-05,
+      "loss": 0.7604,
+      "step": 6839
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1685630512618843e-05,
+      "loss": 0.7631,
+      "step": 6840
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1683470729387574e-05,
+      "loss": 0.7634,
+      "step": 6841
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.168131086533457e-05,
+      "loss": 0.4225,
+      "step": 6842
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1679150920563527e-05,
+      "loss": 0.803,
+      "step": 6843
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.167699089517814e-05,
+      "loss": 0.7655,
+      "step": 6844
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1674830789282108e-05,
+      "loss": 0.7534,
+      "step": 6845
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1672670602979139e-05,
+      "loss": 0.7219,
+      "step": 6846
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1670510336372938e-05,
+      "loss": 0.7292,
+      "step": 6847
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.166834998956722e-05,
+      "loss": 0.8301,
+      "step": 6848
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1666189562665698e-05,
+      "loss": 0.8147,
+      "step": 6849
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1664029055772095e-05,
+      "loss": 0.7777,
+      "step": 6850
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1661868468990135e-05,
+      "loss": 0.7496,
+      "step": 6851
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1659707802423546e-05,
+      "loss": 0.8143,
+      "step": 6852
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1657547056176055e-05,
+      "loss": 0.4376,
+      "step": 6853
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1655386230351398e-05,
+      "loss": 0.7195,
+      "step": 6854
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1653225325053319e-05,
+      "loss": 0.7726,
+      "step": 6855
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1651064340385554e-05,
+      "loss": 0.8194,
+      "step": 6856
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1648903276451856e-05,
+      "loss": 0.757,
+      "step": 6857
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1646742133355974e-05,
+      "loss": 0.734,
+      "step": 6858
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1644580911201663e-05,
+      "loss": 0.7948,
+      "step": 6859
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1642419610092678e-05,
+      "loss": 0.7784,
+      "step": 6860
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1640258230132781e-05,
+      "loss": 0.8385,
+      "step": 6861
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.163809677142574e-05,
+      "loss": 0.7543,
+      "step": 6862
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1635935234075321e-05,
+      "loss": 0.7742,
+      "step": 6863
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1633773618185302e-05,
+      "loss": 0.7701,
+      "step": 6864
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1631611923859458e-05,
+      "loss": 0.8171,
+      "step": 6865
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.162945015120157e-05,
+      "loss": 0.7774,
+      "step": 6866
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.162728830031542e-05,
+      "loss": 0.7553,
+      "step": 6867
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.16251263713048e-05,
+      "loss": 0.4192,
+      "step": 6868
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1622964364273502e-05,
+      "loss": 0.4088,
+      "step": 6869
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1620802279325316e-05,
+      "loss": 0.4621,
+      "step": 6870
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1618640116564048e-05,
+      "loss": 0.4615,
+      "step": 6871
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1616477876093497e-05,
+      "loss": 0.8307,
+      "step": 6872
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1614315558017473e-05,
+      "loss": 0.7831,
+      "step": 6873
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1612153162439788e-05,
+      "loss": 0.7978,
+      "step": 6874
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1609990689464253e-05,
+      "loss": 0.8161,
+      "step": 6875
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1607828139194683e-05,
+      "loss": 0.4484,
+      "step": 6876
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1605665511734907e-05,
+      "loss": 0.4288,
+      "step": 6877
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1603502807188748e-05,
+      "loss": 0.7903,
+      "step": 6878
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1601340025660033e-05,
+      "loss": 0.7466,
+      "step": 6879
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.15991771672526e-05,
+      "loss": 0.4079,
+      "step": 6880
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1597014232070281e-05,
+      "loss": 0.4247,
+      "step": 6881
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1594851220216919e-05,
+      "loss": 0.7602,
+      "step": 6882
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1592688131796352e-05,
+      "loss": 0.7598,
+      "step": 6883
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1590524966912437e-05,
+      "loss": 0.8066,
+      "step": 6884
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1588361725669022e-05,
+      "loss": 0.8415,
+      "step": 6885
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1586198408169957e-05,
+      "loss": 0.8412,
+      "step": 6886
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1584035014519106e-05,
+      "loss": 0.7714,
+      "step": 6887
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.158187154482033e-05,
+      "loss": 0.7898,
+      "step": 6888
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.15797079991775e-05,
+      "loss": 0.4024,
+      "step": 6889
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1577544377694472e-05,
+      "loss": 0.7895,
+      "step": 6890
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1575380680475133e-05,
+      "loss": 0.7973,
+      "step": 6891
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1573216907623352e-05,
+      "loss": 0.393,
+      "step": 6892
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1571053059243014e-05,
+      "loss": 0.7676,
+      "step": 6893
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1568889135438002e-05,
+      "loss": 0.791,
+      "step": 6894
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1566725136312203e-05,
+      "loss": 0.7967,
+      "step": 6895
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.156456106196951e-05,
+      "loss": 0.7546,
+      "step": 6896
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1562396912513814e-05,
+      "loss": 0.799,
+      "step": 6897
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.156023268804902e-05,
+      "loss": 0.7627,
+      "step": 6898
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1558068388679023e-05,
+      "loss": 0.7591,
+      "step": 6899
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1555904014507735e-05,
+      "loss": 0.772,
+      "step": 6900
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1553739565639059e-05,
+      "loss": 0.7579,
+      "step": 6901
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1551575042176918e-05,
+      "loss": 0.7882,
+      "step": 6902
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1549410444225217e-05,
+      "loss": 0.7953,
+      "step": 6903
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1547245771887887e-05,
+      "loss": 0.7995,
+      "step": 6904
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1545081025268843e-05,
+      "loss": 0.8146,
+      "step": 6905
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1542916204472017e-05,
+      "loss": 0.7657,
+      "step": 6906
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1540751309601337e-05,
+      "loss": 0.7718,
+      "step": 6907
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.153858634076074e-05,
+      "loss": 0.4451,
+      "step": 6908
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1536421298054163e-05,
+      "loss": 0.7758,
+      "step": 6909
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1534256181585548e-05,
+      "loss": 0.7551,
+      "step": 6910
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.153209099145884e-05,
+      "loss": 0.7641,
+      "step": 6911
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1529925727777986e-05,
+      "loss": 0.7685,
+      "step": 6912
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1527760390646939e-05,
+      "loss": 0.7954,
+      "step": 6913
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1525594980169656e-05,
+      "loss": 0.4415,
+      "step": 6914
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1523429496450093e-05,
+      "loss": 0.4252,
+      "step": 6915
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1521263939592218e-05,
+      "loss": 0.4221,
+      "step": 6916
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.151909830969999e-05,
+      "loss": 0.8096,
+      "step": 6917
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.151693260687739e-05,
+      "loss": 0.7274,
+      "step": 6918
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1514766831228377e-05,
+      "loss": 0.7614,
+      "step": 6919
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1512600982856936e-05,
+      "loss": 0.4172,
+      "step": 6920
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1510435061867049e-05,
+      "loss": 0.4572,
+      "step": 6921
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1508269068362692e-05,
+      "loss": 0.43,
+      "step": 6922
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1506103002447859e-05,
+      "loss": 0.7732,
+      "step": 6923
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1503936864226537e-05,
+      "loss": 0.8047,
+      "step": 6924
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1501770653802725e-05,
+      "loss": 0.7759,
+      "step": 6925
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1499604371280414e-05,
+      "loss": 0.78,
+      "step": 6926
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.149743801676361e-05,
+      "loss": 0.7742,
+      "step": 6927
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1495271590356317e-05,
+      "loss": 0.7336,
+      "step": 6928
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1493105092162535e-05,
+      "loss": 0.754,
+      "step": 6929
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1490938522286288e-05,
+      "loss": 0.7775,
+      "step": 6930
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1488771880831581e-05,
+      "loss": 0.8218,
+      "step": 6931
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1486605167902438e-05,
+      "loss": 0.8139,
+      "step": 6932
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.148443838360288e-05,
+      "loss": 0.7302,
+      "step": 6933
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.148227152803693e-05,
+      "loss": 0.7853,
+      "step": 6934
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1480104601308614e-05,
+      "loss": 0.4473,
+      "step": 6935
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1477937603521971e-05,
+      "loss": 0.3993,
+      "step": 6936
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1475770534781035e-05,
+      "loss": 0.7873,
+      "step": 6937
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1473603395189836e-05,
+      "loss": 0.4378,
+      "step": 6938
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1471436184852427e-05,
+      "loss": 0.7575,
+      "step": 6939
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.146926890387285e-05,
+      "loss": 0.7284,
+      "step": 6940
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1467101552355154e-05,
+      "loss": 0.7526,
+      "step": 6941
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1464934130403384e-05,
+      "loss": 0.7615,
+      "step": 6942
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.146276663812161e-05,
+      "loss": 0.8014,
+      "step": 6943
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1460599075613881e-05,
+      "loss": 0.7927,
+      "step": 6944
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1458431442984259e-05,
+      "loss": 0.4755,
+      "step": 6945
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1456263740336814e-05,
+      "loss": 0.7381,
+      "step": 6946
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1454095967775616e-05,
+      "loss": 0.7963,
+      "step": 6947
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1451928125404738e-05,
+      "loss": 0.7428,
+      "step": 6948
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1449760213328249e-05,
+      "loss": 0.7437,
+      "step": 6949
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1447592231650235e-05,
+      "loss": 0.7797,
+      "step": 6950
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1445424180474775e-05,
+      "loss": 0.8428,
+      "step": 6951
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1443256059905959e-05,
+      "loss": 0.7816,
+      "step": 6952
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1441087870047873e-05,
+      "loss": 0.7777,
+      "step": 6953
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1438919611004608e-05,
+      "loss": 0.7357,
+      "step": 6954
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1436751282880268e-05,
+      "loss": 0.7678,
+      "step": 6955
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.143458288577894e-05,
+      "loss": 0.7796,
+      "step": 6956
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1432414419804735e-05,
+      "loss": 0.801,
+      "step": 6957
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1430245885061758e-05,
+      "loss": 0.4415,
+      "step": 6958
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1428077281654115e-05,
+      "loss": 0.7844,
+      "step": 6959
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1425908609685923e-05,
+      "loss": 0.4194,
+      "step": 6960
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1423739869261295e-05,
+      "loss": 0.7203,
+      "step": 6961
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.142157106048435e-05,
+      "loss": 0.7148,
+      "step": 6962
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1419402183459212e-05,
+      "loss": 0.8061,
+      "step": 6963
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1417233238290004e-05,
+      "loss": 0.4206,
+      "step": 6964
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1415064225080854e-05,
+      "loss": 0.7228,
+      "step": 6965
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.14128951439359e-05,
+      "loss": 0.7283,
+      "step": 6966
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1410725994959272e-05,
+      "loss": 0.7419,
+      "step": 6967
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.140855677825511e-05,
+      "loss": 0.4279,
+      "step": 6968
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1406387493927555e-05,
+      "loss": 0.7668,
+      "step": 6969
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1404218142080759e-05,
+      "loss": 0.7804,
+      "step": 6970
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1402048722818862e-05,
+      "loss": 0.7919,
+      "step": 6971
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1399879236246013e-05,
+      "loss": 0.4332,
+      "step": 6972
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1397709682466381e-05,
+      "loss": 0.7686,
+      "step": 6973
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1395540061584107e-05,
+      "loss": 0.8133,
+      "step": 6974
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1393370373703367e-05,
+      "loss": 0.7951,
+      "step": 6975
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1391200618928315e-05,
+      "loss": 0.4635,
+      "step": 6976
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1389030797363128e-05,
+      "loss": 0.8106,
+      "step": 6977
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1386860909111968e-05,
+      "loss": 0.7752,
+      "step": 6978
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1384690954279017e-05,
+      "loss": 0.7342,
+      "step": 6979
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1382520932968446e-05,
+      "loss": 0.7602,
+      "step": 6980
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1380350845284436e-05,
+      "loss": 0.4245,
+      "step": 6981
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1378180691331176e-05,
+      "loss": 0.7708,
+      "step": 6982
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1376010471212846e-05,
+      "loss": 0.7672,
+      "step": 6983
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1373840185033641e-05,
+      "loss": 0.8046,
+      "step": 6984
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1371669832897756e-05,
+      "loss": 0.7328,
+      "step": 6985
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1369499414909381e-05,
+      "loss": 0.424,
+      "step": 6986
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1367328931172721e-05,
+      "loss": 0.7524,
+      "step": 6987
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1365158381791974e-05,
+      "loss": 0.7369,
+      "step": 6988
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1362987766871352e-05,
+      "loss": 0.7687,
+      "step": 6989
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1360817086515057e-05,
+      "loss": 0.765,
+      "step": 6990
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1358646340827309e-05,
+      "loss": 0.8012,
+      "step": 6991
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1356475529912315e-05,
+      "loss": 0.8328,
+      "step": 6992
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1354304653874302e-05,
+      "loss": 0.7421,
+      "step": 6993
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1352133712817483e-05,
+      "loss": 0.7838,
+      "step": 6994
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1349962706846087e-05,
+      "loss": 0.403,
+      "step": 6995
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1347791636064345e-05,
+      "loss": 0.7167,
+      "step": 6996
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1345620500576482e-05,
+      "loss": 0.7445,
+      "step": 6997
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1343449300486735e-05,
+      "loss": 0.7225,
+      "step": 6998
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1341278035899339e-05,
+      "loss": 0.7735,
+      "step": 6999
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1339106706918542e-05,
+      "loss": 0.8199,
+      "step": 7000
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1336935313648574e-05,
+      "loss": 0.8461,
+      "step": 7001
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.133476385619369e-05,
+      "loss": 0.792,
+      "step": 7002
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.133259233465814e-05,
+      "loss": 0.7672,
+      "step": 7003
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1330420749146173e-05,
+      "loss": 0.4255,
+      "step": 7004
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.132824909976205e-05,
+      "loss": 0.7811,
+      "step": 7005
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1326077386610022e-05,
+      "loss": 0.7374,
+      "step": 7006
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1323905609794359e-05,
+      "loss": 0.7769,
+      "step": 7007
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1321733769419319e-05,
+      "loss": 0.7596,
+      "step": 7008
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1319561865589173e-05,
+      "loss": 0.7915,
+      "step": 7009
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1317389898408188e-05,
+      "loss": 0.7672,
+      "step": 7010
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1315217867980646e-05,
+      "loss": 0.3987,
+      "step": 7011
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1313045774410819e-05,
+      "loss": 0.7339,
+      "step": 7012
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1310873617802982e-05,
+      "loss": 0.3908,
+      "step": 7013
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1308701398261427e-05,
+      "loss": 0.7983,
+      "step": 7014
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1306529115890438e-05,
+      "loss": 0.8145,
+      "step": 7015
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1304356770794303e-05,
+      "loss": 0.7317,
+      "step": 7016
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1302184363077311e-05,
+      "loss": 0.7422,
+      "step": 7017
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.130001189284376e-05,
+      "loss": 0.8081,
+      "step": 7018
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1297839360197951e-05,
+      "loss": 0.8495,
+      "step": 7019
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1295666765244177e-05,
+      "loss": 0.7903,
+      "step": 7020
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.129349410808675e-05,
+      "loss": 0.7745,
+      "step": 7021
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1291321388829974e-05,
+      "loss": 0.4654,
+      "step": 7022
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1289148607578159e-05,
+      "loss": 0.4595,
+      "step": 7023
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1286975764435617e-05,
+      "loss": 0.7779,
+      "step": 7024
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1284802859506667e-05,
+      "loss": 0.7315,
+      "step": 7025
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1282629892895624e-05,
+      "loss": 0.8132,
+      "step": 7026
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1280456864706814e-05,
+      "loss": 0.7795,
+      "step": 7027
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1278283775044563e-05,
+      "loss": 0.7314,
+      "step": 7028
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.127611062401319e-05,
+      "loss": 0.7516,
+      "step": 7029
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1273937411717037e-05,
+      "loss": 0.7732,
+      "step": 7030
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1271764138260428e-05,
+      "loss": 0.4398,
+      "step": 7031
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1269590803747708e-05,
+      "loss": 0.7446,
+      "step": 7032
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.126741740828321e-05,
+      "loss": 0.7301,
+      "step": 7033
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.126524395197128e-05,
+      "loss": 0.8064,
+      "step": 7034
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1263070434916263e-05,
+      "loss": 0.7462,
+      "step": 7035
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1260896857222505e-05,
+      "loss": 0.7737,
+      "step": 7036
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1258723218994364e-05,
+      "loss": 0.7178,
+      "step": 7037
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1256549520336187e-05,
+      "loss": 0.8001,
+      "step": 7038
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1254375761352332e-05,
+      "loss": 0.7681,
+      "step": 7039
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1252201942147161e-05,
+      "loss": 0.7671,
+      "step": 7040
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1250028062825036e-05,
+      "loss": 0.7574,
+      "step": 7041
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1247854123490323e-05,
+      "loss": 0.4734,
+      "step": 7042
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1245680124247394e-05,
+      "loss": 0.4504,
+      "step": 7043
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1243506065200615e-05,
+      "loss": 0.7678,
+      "step": 7044
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1241331946454363e-05,
+      "loss": 0.4445,
+      "step": 7045
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1239157768113015e-05,
+      "loss": 0.8009,
+      "step": 7046
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1236983530280951e-05,
+      "loss": 0.7448,
+      "step": 7047
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1234809233062556e-05,
+      "loss": 0.4333,
+      "step": 7048
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.123263487656221e-05,
+      "loss": 0.7891,
+      "step": 7049
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1230460460884309e-05,
+      "loss": 0.4432,
+      "step": 7050
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1228285986133241e-05,
+      "loss": 0.414,
+      "step": 7051
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.12261114524134e-05,
+      "loss": 0.786,
+      "step": 7052
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1223936859829179e-05,
+      "loss": 0.7839,
+      "step": 7053
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1221762208484987e-05,
+      "loss": 0.7644,
+      "step": 7054
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1219587498485223e-05,
+      "loss": 0.7771,
+      "step": 7055
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.121741272993429e-05,
+      "loss": 0.7927,
+      "step": 7056
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1215237902936598e-05,
+      "loss": 0.7944,
+      "step": 7057
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1213063017596559e-05,
+      "loss": 0.7875,
+      "step": 7058
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.121088807401859e-05,
+      "loss": 0.4005,
+      "step": 7059
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.12087130723071e-05,
+      "loss": 0.8293,
+      "step": 7060
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1206538012566517e-05,
+      "loss": 0.4748,
+      "step": 7061
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1204362894901258e-05,
+      "loss": 0.7132,
+      "step": 7062
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1202187719415748e-05,
+      "loss": 0.7769,
+      "step": 7063
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.120001248621442e-05,
+      "loss": 0.7923,
+      "step": 7064
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.11978371954017e-05,
+      "loss": 0.7635,
+      "step": 7065
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1195661847082023e-05,
+      "loss": 0.7292,
+      "step": 7066
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1193486441359826e-05,
+      "loss": 0.7586,
+      "step": 7067
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1191310978339549e-05,
+      "loss": 0.7329,
+      "step": 7068
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1189135458125628e-05,
+      "loss": 0.4079,
+      "step": 7069
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1186959880822517e-05,
+      "loss": 0.4138,
+      "step": 7070
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1184784246534655e-05,
+      "loss": 0.7904,
+      "step": 7071
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1182608555366493e-05,
+      "loss": 0.4687,
+      "step": 7072
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.118043280742249e-05,
+      "loss": 0.7511,
+      "step": 7073
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1178257002807096e-05,
+      "loss": 0.418,
+      "step": 7074
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.117608114162477e-05,
+      "loss": 0.8247,
+      "step": 7075
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1173905223979972e-05,
+      "loss": 0.8073,
+      "step": 7076
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1171729249977171e-05,
+      "loss": 0.7557,
+      "step": 7077
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1169553219720828e-05,
+      "loss": 0.7716,
+      "step": 7078
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.116737713331541e-05,
+      "loss": 0.7677,
+      "step": 7079
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1165200990865394e-05,
+      "loss": 0.4431,
+      "step": 7080
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1163024792475252e-05,
+      "loss": 0.834,
+      "step": 7081
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1160848538249464e-05,
+      "loss": 0.4487,
+      "step": 7082
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1158672228292503e-05,
+      "loss": 0.7211,
+      "step": 7083
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1156495862708861e-05,
+      "loss": 0.797,
+      "step": 7084
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1154319441603012e-05,
+      "loss": 0.7502,
+      "step": 7085
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1152142965079453e-05,
+      "loss": 0.4474,
+      "step": 7086
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1149966433242672e-05,
+      "loss": 0.4772,
+      "step": 7087
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.114778984619716e-05,
+      "loss": 0.7618,
+      "step": 7088
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1145613204047416e-05,
+      "loss": 0.424,
+      "step": 7089
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1143436506897934e-05,
+      "loss": 0.7599,
+      "step": 7090
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1141259754853217e-05,
+      "loss": 0.458,
+      "step": 7091
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1139082948017768e-05,
+      "loss": 0.778,
+      "step": 7092
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1136906086496097e-05,
+      "loss": 0.8015,
+      "step": 7093
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.113472917039271e-05,
+      "loss": 0.4797,
+      "step": 7094
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1132552199812117e-05,
+      "loss": 0.7803,
+      "step": 7095
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1130375174858836e-05,
+      "loss": 0.7074,
+      "step": 7096
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1128198095637381e-05,
+      "loss": 0.7505,
+      "step": 7097
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1126020962252274e-05,
+      "loss": 0.7556,
+      "step": 7098
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.112384377480803e-05,
+      "loss": 0.7548,
+      "step": 7099
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1121666533409184e-05,
+      "loss": 0.7407,
+      "step": 7100
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1119489238160255e-05,
+      "loss": 0.7891,
+      "step": 7101
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1117311889165777e-05,
+      "loss": 0.7261,
+      "step": 7102
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.111513448653028e-05,
+      "loss": 0.7931,
+      "step": 7103
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.11129570303583e-05,
+      "loss": 0.7917,
+      "step": 7104
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1110779520754375e-05,
+      "loss": 0.8009,
+      "step": 7105
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1108601957823044e-05,
+      "loss": 0.3965,
+      "step": 7106
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.110642434166885e-05,
+      "loss": 0.7299,
+      "step": 7107
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1104246672396338e-05,
+      "loss": 0.777,
+      "step": 7108
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1102068950110056e-05,
+      "loss": 0.8306,
+      "step": 7109
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1099891174914554e-05,
+      "loss": 0.7557,
+      "step": 7110
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1097713346914384e-05,
+      "loss": 0.7975,
+      "step": 7111
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1095535466214105e-05,
+      "loss": 0.748,
+      "step": 7112
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.109335753291827e-05,
+      "loss": 0.7885,
+      "step": 7113
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1091179547131443e-05,
+      "loss": 0.7592,
+      "step": 7114
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1089001508958184e-05,
+      "loss": 0.7692,
+      "step": 7115
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1086823418503058e-05,
+      "loss": 0.4407,
+      "step": 7116
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1084645275870639e-05,
+      "loss": 0.4551,
+      "step": 7117
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.108246708116549e-05,
+      "loss": 0.4556,
+      "step": 7118
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1080288834492188e-05,
+      "loss": 0.7973,
+      "step": 7119
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1078110535955309e-05,
+      "loss": 0.7974,
+      "step": 7120
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1075932185659428e-05,
+      "loss": 0.7239,
+      "step": 7121
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1073753783709125e-05,
+      "loss": 0.7309,
+      "step": 7122
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1071575330208989e-05,
+      "loss": 0.7794,
+      "step": 7123
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1069396825263596e-05,
+      "loss": 0.7726,
+      "step": 7124
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1067218268977542e-05,
+      "loss": 0.4506,
+      "step": 7125
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1065039661455415e-05,
+      "loss": 0.4404,
+      "step": 7126
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1062861002801806e-05,
+      "loss": 0.7627,
+      "step": 7127
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1060682293121308e-05,
+      "loss": 0.4262,
+      "step": 7128
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1058503532518525e-05,
+      "loss": 0.3907,
+      "step": 7129
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1056324721098056e-05,
+      "loss": 0.7738,
+      "step": 7130
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1054145858964495e-05,
+      "loss": 0.7589,
+      "step": 7131
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.105196694622246e-05,
+      "loss": 0.448,
+      "step": 7132
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1049787982976548e-05,
+      "loss": 0.4417,
+      "step": 7133
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1047608969331375e-05,
+      "loss": 0.7968,
+      "step": 7134
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1045429905391549e-05,
+      "loss": 0.7711,
+      "step": 7135
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1043250791261689e-05,
+      "loss": 0.7486,
+      "step": 7136
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.104107162704641e-05,
+      "loss": 0.7522,
+      "step": 7137
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1038892412850327e-05,
+      "loss": 0.3993,
+      "step": 7138
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1036713148778069e-05,
+      "loss": 0.7541,
+      "step": 7139
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1034533834934253e-05,
+      "loss": 0.8202,
+      "step": 7140
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1032354471423516e-05,
+      "loss": 0.7359,
+      "step": 7141
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1030175058350478e-05,
+      "loss": 0.8101,
+      "step": 7142
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1027995595819773e-05,
+      "loss": 0.7435,
+      "step": 7143
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1025816083936036e-05,
+      "loss": 0.4107,
+      "step": 7144
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1023636522803903e-05,
+      "loss": 0.8075,
+      "step": 7145
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1021456912528012e-05,
+      "loss": 0.7718,
+      "step": 7146
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1019277253213002e-05,
+      "loss": 0.8084,
+      "step": 7147
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1017097544963519e-05,
+      "loss": 0.7537,
+      "step": 7148
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.101491778788421e-05,
+      "loss": 0.7478,
+      "step": 7149
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1012737982079718e-05,
+      "loss": 0.4806,
+      "step": 7150
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1010558127654693e-05,
+      "loss": 0.7806,
+      "step": 7151
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1008378224713796e-05,
+      "loss": 0.4025,
+      "step": 7152
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1006198273361675e-05,
+      "loss": 0.7943,
+      "step": 7153
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1004018273702985e-05,
+      "loss": 0.7658,
+      "step": 7154
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1001838225842393e-05,
+      "loss": 0.745,
+      "step": 7155
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0999658129884558e-05,
+      "loss": 0.7057,
+      "step": 7156
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0997477985934141e-05,
+      "loss": 0.7259,
+      "step": 7157
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0995297794095812e-05,
+      "loss": 0.4266,
+      "step": 7158
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.099311755447424e-05,
+      "loss": 0.4641,
+      "step": 7159
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0990937267174091e-05,
+      "loss": 0.4236,
+      "step": 7160
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0988756932300044e-05,
+      "loss": 0.7839,
+      "step": 7161
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0986576549956775e-05,
+      "loss": 0.7801,
+      "step": 7162
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0984396120248956e-05,
+      "loss": 0.7613,
+      "step": 7163
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0982215643281278e-05,
+      "loss": 0.4324,
+      "step": 7164
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.098003511915841e-05,
+      "loss": 0.7984,
+      "step": 7165
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0977854547985046e-05,
+      "loss": 0.7604,
+      "step": 7166
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0975673929865868e-05,
+      "loss": 0.831,
+      "step": 7167
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.097349326490557e-05,
+      "loss": 0.7882,
+      "step": 7168
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0971312553208841e-05,
+      "loss": 0.7647,
+      "step": 7169
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0969131794880374e-05,
+      "loss": 0.4163,
+      "step": 7170
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0966950990024867e-05,
+      "loss": 0.794,
+      "step": 7171
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0964770138747019e-05,
+      "loss": 0.7539,
+      "step": 7172
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0962589241151526e-05,
+      "loss": 0.7652,
+      "step": 7173
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0960408297343092e-05,
+      "loss": 0.7438,
+      "step": 7174
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0958227307426426e-05,
+      "loss": 0.7658,
+      "step": 7175
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.095604627150623e-05,
+      "loss": 0.7813,
+      "step": 7176
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0953865189687219e-05,
+      "loss": 0.7157,
+      "step": 7177
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.09516840620741e-05,
+      "loss": 0.4034,
+      "step": 7178
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0949502888771588e-05,
+      "loss": 0.7852,
+      "step": 7179
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.09473216698844e-05,
+      "loss": 0.7422,
+      "step": 7180
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0945140405517251e-05,
+      "loss": 0.7273,
+      "step": 7181
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0942959095774865e-05,
+      "loss": 0.7475,
+      "step": 7182
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0940777740761962e-05,
+      "loss": 0.7695,
+      "step": 7183
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0938596340583267e-05,
+      "loss": 0.7741,
+      "step": 7184
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.093641489534351e-05,
+      "loss": 0.381,
+      "step": 7185
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0934233405147417e-05,
+      "loss": 0.78,
+      "step": 7186
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0932051870099715e-05,
+      "loss": 0.7839,
+      "step": 7187
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0929870290305147e-05,
+      "loss": 0.4041,
+      "step": 7188
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.092768866586844e-05,
+      "loss": 0.8206,
+      "step": 7189
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0925506996894335e-05,
+      "loss": 0.7724,
+      "step": 7190
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0923325283487573e-05,
+      "loss": 0.469,
+      "step": 7191
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0921143525752893e-05,
+      "loss": 0.7913,
+      "step": 7192
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0918961723795046e-05,
+      "loss": 0.7411,
+      "step": 7193
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0916779877718767e-05,
+      "loss": 0.7339,
+      "step": 7194
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.091459798762881e-05,
+      "loss": 0.8077,
+      "step": 7195
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0912416053629926e-05,
+      "loss": 0.7489,
+      "step": 7196
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0910234075826866e-05,
+      "loss": 0.7945,
+      "step": 7197
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0908052054324387e-05,
+      "loss": 0.4046,
+      "step": 7198
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.090586998922724e-05,
+      "loss": 0.4438,
+      "step": 7199
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0903687880640192e-05,
+      "loss": 0.4596,
+      "step": 7200
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0901505728667998e-05,
+      "loss": 0.8165,
+      "step": 7201
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0899323533415423e-05,
+      "loss": 0.4278,
+      "step": 7202
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0897141294987228e-05,
+      "loss": 0.7617,
+      "step": 7203
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0894959013488184e-05,
+      "loss": 0.7719,
+      "step": 7204
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0892776689023064e-05,
+      "loss": 0.8073,
+      "step": 7205
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0890594321696627e-05,
+      "loss": 0.743,
+      "step": 7206
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.088841191161366e-05,
+      "loss": 0.7664,
+      "step": 7207
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0886229458878932e-05,
+      "loss": 0.7568,
+      "step": 7208
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.088404696359722e-05,
+      "loss": 0.7414,
+      "step": 7209
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0881864425873304e-05,
+      "loss": 0.4305,
+      "step": 7210
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0879681845811964e-05,
+      "loss": 0.4286,
+      "step": 7211
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0877499223517986e-05,
+      "loss": 0.715,
+      "step": 7212
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0875316559096155e-05,
+      "loss": 0.751,
+      "step": 7213
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0873133852651257e-05,
+      "loss": 0.8034,
+      "step": 7214
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0870951104288081e-05,
+      "loss": 0.7508,
+      "step": 7215
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0868768314111424e-05,
+      "loss": 0.3862,
+      "step": 7216
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0866585482226071e-05,
+      "loss": 0.4647,
+      "step": 7217
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0864402608736825e-05,
+      "loss": 0.8176,
+      "step": 7218
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0862219693748479e-05,
+      "loss": 0.7748,
+      "step": 7219
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0860036737365836e-05,
+      "loss": 0.808,
+      "step": 7220
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0857853739693696e-05,
+      "loss": 0.4057,
+      "step": 7221
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0855670700836858e-05,
+      "loss": 0.7811,
+      "step": 7222
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0853487620900138e-05,
+      "loss": 0.77,
+      "step": 7223
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0851304499988333e-05,
+      "loss": 0.7828,
+      "step": 7224
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.084912133820626e-05,
+      "loss": 0.7972,
+      "step": 7225
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0846938135658722e-05,
+      "loss": 0.7611,
+      "step": 7226
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0844754892450541e-05,
+      "loss": 0.8188,
+      "step": 7227
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.084257160868653e-05,
+      "loss": 0.7787,
+      "step": 7228
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0840388284471502e-05,
+      "loss": 0.7721,
+      "step": 7229
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0838204919910283e-05,
+      "loss": 0.7733,
+      "step": 7230
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0836021515107686e-05,
+      "loss": 0.766,
+      "step": 7231
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0833838070168543e-05,
+      "loss": 0.4347,
+      "step": 7232
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0831654585197668e-05,
+      "loss": 0.7721,
+      "step": 7233
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.08294710602999e-05,
+      "loss": 0.7754,
+      "step": 7234
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0827287495580058e-05,
+      "loss": 0.7326,
+      "step": 7235
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0825103891142981e-05,
+      "loss": 0.7299,
+      "step": 7236
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0822920247093496e-05,
+      "loss": 0.4104,
+      "step": 7237
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0820736563536439e-05,
+      "loss": 0.4139,
+      "step": 7238
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0818552840576646e-05,
+      "loss": 0.763,
+      "step": 7239
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0816369078318956e-05,
+      "loss": 0.732,
+      "step": 7240
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.081418527686821e-05,
+      "loss": 0.4481,
+      "step": 7241
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0812001436329246e-05,
+      "loss": 0.7542,
+      "step": 7242
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0809817556806914e-05,
+      "loss": 0.7461,
+      "step": 7243
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0807633638406059e-05,
+      "loss": 0.7301,
+      "step": 7244
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0805449681231524e-05,
+      "loss": 0.4464,
+      "step": 7245
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.080326568538816e-05,
+      "loss": 0.8392,
+      "step": 7246
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0801081650980823e-05,
+      "loss": 0.45,
+      "step": 7247
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.079889757811436e-05,
+      "loss": 0.8414,
+      "step": 7248
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.079671346689363e-05,
+      "loss": 0.7648,
+      "step": 7249
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.079452931742349e-05,
+      "loss": 0.784,
+      "step": 7250
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0792345129808797e-05,
+      "loss": 0.7759,
+      "step": 7251
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0790160904154414e-05,
+      "loss": 0.4644,
+      "step": 7252
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0787976640565204e-05,
+      "loss": 0.8003,
+      "step": 7253
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0785792339146026e-05,
+      "loss": 0.3902,
+      "step": 7254
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0783608000001753e-05,
+      "loss": 0.8041,
+      "step": 7255
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0781423623237247e-05,
+      "loss": 0.7542,
+      "step": 7256
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.077923920895738e-05,
+      "loss": 0.7418,
+      "step": 7257
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0777054757267021e-05,
+      "loss": 0.7678,
+      "step": 7258
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0774870268271049e-05,
+      "loss": 0.7645,
+      "step": 7259
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0772685742074339e-05,
+      "loss": 0.7821,
+      "step": 7260
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0770501178781763e-05,
+      "loss": 0.7591,
+      "step": 7261
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0768316578498199e-05,
+      "loss": 0.7255,
+      "step": 7262
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0766131941328533e-05,
+      "loss": 0.4382,
+      "step": 7263
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0763947267377643e-05,
+      "loss": 0.7799,
+      "step": 7264
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0761762556750415e-05,
+      "loss": 0.7528,
+      "step": 7265
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0759577809551734e-05,
+      "loss": 0.8626,
+      "step": 7266
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0757393025886486e-05,
+      "loss": 0.7749,
+      "step": 7267
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0755208205859566e-05,
+      "loss": 0.7641,
+      "step": 7268
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0753023349575857e-05,
+      "loss": 0.4787,
+      "step": 7269
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.075083845714026e-05,
+      "loss": 0.7793,
+      "step": 7270
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0748653528657663e-05,
+      "loss": 0.7991,
+      "step": 7271
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0746468564232963e-05,
+      "loss": 0.7847,
+      "step": 7272
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.074428356397106e-05,
+      "loss": 0.7985,
+      "step": 7273
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0742098527976852e-05,
+      "loss": 0.7929,
+      "step": 7274
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0739913456355247e-05,
+      "loss": 0.7743,
+      "step": 7275
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0737728349211136e-05,
+      "loss": 0.4287,
+      "step": 7276
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0735543206649432e-05,
+      "loss": 0.7521,
+      "step": 7277
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.073335802877504e-05,
+      "loss": 0.7897,
+      "step": 7278
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.073117281569287e-05,
+      "loss": 0.7401,
+      "step": 7279
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0728987567507827e-05,
+      "loss": 0.437,
+      "step": 7280
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0726802284324826e-05,
+      "loss": 0.7631,
+      "step": 7281
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.072461696624878e-05,
+      "loss": 0.7891,
+      "step": 7282
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0722431613384605e-05,
+      "loss": 0.7627,
+      "step": 7283
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0720246225837214e-05,
+      "loss": 0.765,
+      "step": 7284
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0718060803711527e-05,
+      "loss": 0.7158,
+      "step": 7285
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0715875347112467e-05,
+      "loss": 0.7835,
+      "step": 7286
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.071368985614495e-05,
+      "loss": 0.446,
+      "step": 7287
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0711504330913904e-05,
+      "loss": 0.7484,
+      "step": 7288
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0709318771524253e-05,
+      "loss": 0.7427,
+      "step": 7289
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.070713317808092e-05,
+      "loss": 0.8162,
+      "step": 7290
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0704947550688838e-05,
+      "loss": 0.4285,
+      "step": 7291
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.070276188945293e-05,
+      "loss": 0.7166,
+      "step": 7292
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0700576194478136e-05,
+      "loss": 0.824,
+      "step": 7293
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0698390465869381e-05,
+      "loss": 0.8034,
+      "step": 7294
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0696204703731607e-05,
+      "loss": 0.8009,
+      "step": 7295
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0694018908169746e-05,
+      "loss": 0.7974,
+      "step": 7296
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0691833079288738e-05,
+      "loss": 0.7598,
+      "step": 7297
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0689647217193519e-05,
+      "loss": 0.4032,
+      "step": 7298
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.068746132198903e-05,
+      "loss": 0.752,
+      "step": 7299
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0685275393780219e-05,
+      "loss": 0.7397,
+      "step": 7300
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0683089432672023e-05,
+      "loss": 0.788,
+      "step": 7301
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0680903438769394e-05,
+      "loss": 0.4603,
+      "step": 7302
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0678717412177277e-05,
+      "loss": 0.441,
+      "step": 7303
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0676531353000618e-05,
+      "loss": 0.4577,
+      "step": 7304
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0674345261344376e-05,
+      "loss": 0.4499,
+      "step": 7305
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0672159137313494e-05,
+      "loss": 0.7357,
+      "step": 7306
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.066997298101293e-05,
+      "loss": 0.7557,
+      "step": 7307
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0667786792547636e-05,
+      "loss": 0.7658,
+      "step": 7308
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0665600572022574e-05,
+      "loss": 0.769,
+      "step": 7309
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0663414319542695e-05,
+      "loss": 0.7546,
+      "step": 7310
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0661228035212968e-05,
+      "loss": 0.8153,
+      "step": 7311
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0659041719138348e-05,
+      "loss": 0.7884,
+      "step": 7312
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0656855371423798e-05,
+      "loss": 0.7471,
+      "step": 7313
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0654668992174285e-05,
+      "loss": 0.7645,
+      "step": 7314
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.065248258149477e-05,
+      "loss": 0.7897,
+      "step": 7315
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0650296139490225e-05,
+      "loss": 0.7666,
+      "step": 7316
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0648109666265618e-05,
+      "loss": 0.7698,
+      "step": 7317
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.064592316192592e-05,
+      "loss": 0.7402,
+      "step": 7318
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.06437366265761e-05,
+      "loss": 0.7926,
+      "step": 7319
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0641550060321136e-05,
+      "loss": 0.8209,
+      "step": 7320
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0639363463265997e-05,
+      "loss": 0.7152,
+      "step": 7321
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0637176835515664e-05,
+      "loss": 0.7541,
+      "step": 7322
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0634990177175113e-05,
+      "loss": 0.7303,
+      "step": 7323
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0632803488349321e-05,
+      "loss": 0.7539,
+      "step": 7324
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0630616769143274e-05,
+      "loss": 0.7853,
+      "step": 7325
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0628430019661949e-05,
+      "loss": 0.7407,
+      "step": 7326
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0626243240010334e-05,
+      "loss": 0.7949,
+      "step": 7327
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.062405643029341e-05,
+      "loss": 0.7948,
+      "step": 7328
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.062186959061617e-05,
+      "loss": 0.7625,
+      "step": 7329
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0619682721083593e-05,
+      "loss": 0.4379,
+      "step": 7330
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0617495821800674e-05,
+      "loss": 0.7574,
+      "step": 7331
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0615308892872402e-05,
+      "loss": 0.4371,
+      "step": 7332
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.061312193440377e-05,
+      "loss": 0.7485,
+      "step": 7333
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0610934946499774e-05,
+      "loss": 0.794,
+      "step": 7334
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0608747929265405e-05,
+      "loss": 0.7579,
+      "step": 7335
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0606560882805663e-05,
+      "loss": 0.7722,
+      "step": 7336
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.060437380722554e-05,
+      "loss": 0.4393,
+      "step": 7337
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0602186702630044e-05,
+      "loss": 0.7766,
+      "step": 7338
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.059999956912417e-05,
+      "loss": 0.7692,
+      "step": 7339
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.059781240681292e-05,
+      "loss": 0.7307,
+      "step": 7340
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.05956252158013e-05,
+      "loss": 0.774,
+      "step": 7341
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0593437996194313e-05,
+      "loss": 0.8222,
+      "step": 7342
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0591250748096966e-05,
+      "loss": 0.3772,
+      "step": 7343
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0589063471614264e-05,
+      "loss": 0.78,
+      "step": 7344
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0586876166851223e-05,
+      "loss": 0.7657,
+      "step": 7345
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0584688833912847e-05,
+      "loss": 0.7367,
+      "step": 7346
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0582501472904146e-05,
+      "loss": 0.7732,
+      "step": 7347
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.058031408393014e-05,
+      "loss": 0.781,
+      "step": 7348
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0578126667095837e-05,
+      "loss": 0.7927,
+      "step": 7349
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0575939222506258e-05,
+      "loss": 0.7878,
+      "step": 7350
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0573751750266412e-05,
+      "loss": 0.7719,
+      "step": 7351
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0571564250481326e-05,
+      "loss": 0.4172,
+      "step": 7352
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0569376723256014e-05,
+      "loss": 0.782,
+      "step": 7353
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0567189168695501e-05,
+      "loss": 0.7818,
+      "step": 7354
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.056500158690481e-05,
+      "loss": 0.7557,
+      "step": 7355
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0562813977988958e-05,
+      "loss": 0.7919,
+      "step": 7356
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0560626342052978e-05,
+      "loss": 0.8213,
+      "step": 7357
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0558438679201888e-05,
+      "loss": 0.7533,
+      "step": 7358
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0556250989540724e-05,
+      "loss": 0.761,
+      "step": 7359
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0554063273174507e-05,
+      "loss": 0.4385,
+      "step": 7360
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.055187553020827e-05,
+      "loss": 0.7758,
+      "step": 7361
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0549687760747047e-05,
+      "loss": 0.4388,
+      "step": 7362
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0547499964895867e-05,
+      "loss": 0.7648,
+      "step": 7363
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.054531214275977e-05,
+      "loss": 0.7685,
+      "step": 7364
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.054312429444378e-05,
+      "loss": 0.4367,
+      "step": 7365
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0540936420052944e-05,
+      "loss": 0.7922,
+      "step": 7366
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0538748519692292e-05,
+      "loss": 0.7885,
+      "step": 7367
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0536560593466869e-05,
+      "loss": 0.7432,
+      "step": 7368
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0534372641481709e-05,
+      "loss": 0.7585,
+      "step": 7369
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.053218466384186e-05,
+      "loss": 0.7769,
+      "step": 7370
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0529996660652361e-05,
+      "loss": 0.8112,
+      "step": 7371
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.052780863201826e-05,
+      "loss": 0.7889,
+      "step": 7372
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0525620578044595e-05,
+      "loss": 0.7758,
+      "step": 7373
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0523432498836414e-05,
+      "loss": 0.7369,
+      "step": 7374
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0521244394498769e-05,
+      "loss": 0.7782,
+      "step": 7375
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0519056265136702e-05,
+      "loss": 0.4362,
+      "step": 7376
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.051686811085527e-05,
+      "loss": 0.4355,
+      "step": 7377
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0514679931759522e-05,
+      "loss": 0.7592,
+      "step": 7378
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0512491727954506e-05,
+      "loss": 0.7264,
+      "step": 7379
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.051030349954528e-05,
+      "loss": 0.7909,
+      "step": 7380
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0508115246636897e-05,
+      "loss": 0.7415,
+      "step": 7381
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0505926969334412e-05,
+      "loss": 0.7697,
+      "step": 7382
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0503738667742882e-05,
+      "loss": 0.4321,
+      "step": 7383
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0501550341967368e-05,
+      "loss": 0.4233,
+      "step": 7384
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0499361992112926e-05,
+      "loss": 0.7831,
+      "step": 7385
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.049717361828462e-05,
+      "loss": 0.7701,
+      "step": 7386
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0494985220587508e-05,
+      "loss": 0.799,
+      "step": 7387
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0492796799126657e-05,
+      "loss": 0.7827,
+      "step": 7388
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0490608354007127e-05,
+      "loss": 0.8017,
+      "step": 7389
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.048841988533398e-05,
+      "loss": 0.7634,
+      "step": 7390
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0486231393212292e-05,
+      "loss": 0.4351,
+      "step": 7391
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.048404287774712e-05,
+      "loss": 0.8068,
+      "step": 7392
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0481854339043541e-05,
+      "loss": 0.7971,
+      "step": 7393
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.047966577720662e-05,
+      "loss": 0.737,
+      "step": 7394
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0477477192341427e-05,
+      "loss": 0.8002,
+      "step": 7395
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0475288584553034e-05,
+      "loss": 0.7818,
+      "step": 7396
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0473099953946517e-05,
+      "loss": 0.7682,
+      "step": 7397
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.047091130062695e-05,
+      "loss": 0.7961,
+      "step": 7398
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0468722624699401e-05,
+      "loss": 0.7316,
+      "step": 7399
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0466533926268953e-05,
+      "loss": 0.7717,
+      "step": 7400
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0464345205440683e-05,
+      "loss": 0.7403,
+      "step": 7401
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0462156462319669e-05,
+      "loss": 0.7591,
+      "step": 7402
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0459967697010984e-05,
+      "loss": 0.7974,
+      "step": 7403
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0457778909619715e-05,
+      "loss": 0.7629,
+      "step": 7404
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0455590100250943e-05,
+      "loss": 0.7428,
+      "step": 7405
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0453401269009746e-05,
+      "loss": 0.7473,
+      "step": 7406
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0451212416001215e-05,
+      "loss": 0.7568,
+      "step": 7407
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0449023541330428e-05,
+      "loss": 0.777,
+      "step": 7408
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0446834645102477e-05,
+      "loss": 0.7733,
+      "step": 7409
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.044464572742244e-05,
+      "loss": 0.7765,
+      "step": 7410
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0442456788395414e-05,
+      "loss": 0.8627,
+      "step": 7411
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0440267828126478e-05,
+      "loss": 0.7774,
+      "step": 7412
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0438078846720732e-05,
+      "loss": 0.7637,
+      "step": 7413
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0435889844283261e-05,
+      "loss": 0.78,
+      "step": 7414
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0433700820919155e-05,
+      "loss": 0.7709,
+      "step": 7415
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0431511776733514e-05,
+      "loss": 0.7693,
+      "step": 7416
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0429322711831424e-05,
+      "loss": 0.4135,
+      "step": 7417
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0427133626317985e-05,
+      "loss": 0.7636,
+      "step": 7418
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0424944520298288e-05,
+      "loss": 0.423,
+      "step": 7419
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0422755393877437e-05,
+      "loss": 0.7422,
+      "step": 7420
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0420566247160524e-05,
+      "loss": 0.4128,
+      "step": 7421
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0418377080252647e-05,
+      "loss": 0.731,
+      "step": 7422
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.041618789325891e-05,
+      "loss": 0.7613,
+      "step": 7423
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0413998686284412e-05,
+      "loss": 0.4293,
+      "step": 7424
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0411809459434252e-05,
+      "loss": 0.7276,
+      "step": 7425
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0409620212813536e-05,
+      "loss": 0.8129,
+      "step": 7426
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0407430946527365e-05,
+      "loss": 0.799,
+      "step": 7427
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0405241660680847e-05,
+      "loss": 0.4033,
+      "step": 7428
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0403052355379084e-05,
+      "loss": 0.8073,
+      "step": 7429
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0400863030727183e-05,
+      "loss": 0.4368,
+      "step": 7430
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0398673686830255e-05,
+      "loss": 0.7427,
+      "step": 7431
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0396484323793403e-05,
+      "loss": 0.4157,
+      "step": 7432
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0394294941721737e-05,
+      "loss": 0.7722,
+      "step": 7433
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.039210554072037e-05,
+      "loss": 0.7675,
+      "step": 7434
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0389916120894408e-05,
+      "loss": 0.7349,
+      "step": 7435
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0387726682348968e-05,
+      "loss": 0.7802,
+      "step": 7436
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0385537225189163e-05,
+      "loss": 0.4509,
+      "step": 7437
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0383347749520098e-05,
+      "loss": 0.4324,
+      "step": 7438
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0381158255446902e-05,
+      "loss": 0.7675,
+      "step": 7439
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0378968743074675e-05,
+      "loss": 0.76,
+      "step": 7440
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0376779212508543e-05,
+      "loss": 0.4304,
+      "step": 7441
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.037458966385362e-05,
+      "loss": 0.4269,
+      "step": 7442
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0372400097215026e-05,
+      "loss": 0.8257,
+      "step": 7443
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0370210512697876e-05,
+      "loss": 0.7387,
+      "step": 7444
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0368020910407294e-05,
+      "loss": 0.8242,
+      "step": 7445
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.03658312904484e-05,
+      "loss": 0.6791,
+      "step": 7446
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0363641652926315e-05,
+      "loss": 0.7669,
+      "step": 7447
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.036145199794616e-05,
+      "loss": 0.4286,
+      "step": 7448
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0359262325613058e-05,
+      "loss": 0.7312,
+      "step": 7449
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0357072636032132e-05,
+      "loss": 0.4561,
+      "step": 7450
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0354882929308509e-05,
+      "loss": 0.7734,
+      "step": 7451
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0352693205547318e-05,
+      "loss": 0.812,
+      "step": 7452
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0350503464853679e-05,
+      "loss": 0.431,
+      "step": 7453
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0348313707332723e-05,
+      "loss": 0.7947,
+      "step": 7454
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0346123933089576e-05,
+      "loss": 0.4593,
+      "step": 7455
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0343934142229364e-05,
+      "loss": 0.7824,
+      "step": 7456
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0341744334857225e-05,
+      "loss": 0.8041,
+      "step": 7457
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0339554511078282e-05,
+      "loss": 0.429,
+      "step": 7458
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0337364670997671e-05,
+      "loss": 0.7523,
+      "step": 7459
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0335174814720518e-05,
+      "loss": 0.7888,
+      "step": 7460
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0332984942351967e-05,
+      "loss": 0.756,
+      "step": 7461
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.033079505399714e-05,
+      "loss": 0.4483,
+      "step": 7462
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0328605149761176e-05,
+      "loss": 0.6961,
+      "step": 7463
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.032641522974921e-05,
+      "loss": 0.7398,
+      "step": 7464
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0324225294066378e-05,
+      "loss": 0.7694,
+      "step": 7465
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0322035342817818e-05,
+      "loss": 0.436,
+      "step": 7466
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0319845376108663e-05,
+      "loss": 0.4053,
+      "step": 7467
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0317655394044057e-05,
+      "loss": 0.823,
+      "step": 7468
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0315465396729132e-05,
+      "loss": 0.7616,
+      "step": 7469
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0313275384269034e-05,
+      "loss": 0.7467,
+      "step": 7470
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0311085356768898e-05,
+      "loss": 0.7763,
+      "step": 7471
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.030889531433387e-05,
+      "loss": 0.4351,
+      "step": 7472
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0306705257069088e-05,
+      "loss": 0.4448,
+      "step": 7473
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0304515185079693e-05,
+      "loss": 0.7289,
+      "step": 7474
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0302325098470834e-05,
+      "loss": 0.8104,
+      "step": 7475
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.030013499734765e-05,
+      "loss": 0.774,
+      "step": 7476
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.029794488181529e-05,
+      "loss": 0.8152,
+      "step": 7477
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0295754751978893e-05,
+      "loss": 0.7313,
+      "step": 7478
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0293564607943609e-05,
+      "loss": 0.3839,
+      "step": 7479
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0291374449814587e-05,
+      "loss": 0.766,
+      "step": 7480
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0289184277696966e-05,
+      "loss": 0.8077,
+      "step": 7481
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0286994091695903e-05,
+      "loss": 0.7816,
+      "step": 7482
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0284803891916542e-05,
+      "loss": 0.7997,
+      "step": 7483
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0282613678464035e-05,
+      "loss": 0.3851,
+      "step": 7484
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0280423451443526e-05,
+      "loss": 0.7887,
+      "step": 7485
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0278233210960172e-05,
+      "loss": 0.8148,
+      "step": 7486
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0276042957119122e-05,
+      "loss": 0.7809,
+      "step": 7487
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0273852690025525e-05,
+      "loss": 0.3932,
+      "step": 7488
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.027166240978454e-05,
+      "loss": 0.4266,
+      "step": 7489
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0269472116501314e-05,
+      "loss": 0.8024,
+      "step": 7490
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0267281810281006e-05,
+      "loss": 0.4111,
+      "step": 7491
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0265091491228765e-05,
+      "loss": 0.7754,
+      "step": 7492
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.026290115944975e-05,
+      "loss": 0.7522,
+      "step": 7493
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0260710815049112e-05,
+      "loss": 0.7879,
+      "step": 7494
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0258520458132015e-05,
+      "loss": 0.7892,
+      "step": 7495
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.025633008880361e-05,
+      "loss": 0.4103,
+      "step": 7496
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0254139707169053e-05,
+      "loss": 0.7737,
+      "step": 7497
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0251949313333511e-05,
+      "loss": 0.7565,
+      "step": 7498
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.024975890740213e-05,
+      "loss": 0.7597,
+      "step": 7499
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0247568489480078e-05,
+      "loss": 0.7788,
+      "step": 7500
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.024537805967251e-05,
+      "loss": 0.7437,
+      "step": 7501
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0243187618084592e-05,
+      "loss": 0.7842,
+      "step": 7502
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0240997164821477e-05,
+      "loss": 0.4256,
+      "step": 7503
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0238806699988335e-05,
+      "loss": 0.7943,
+      "step": 7504
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0236616223690326e-05,
+      "loss": 0.8092,
+      "step": 7505
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0234425736032607e-05,
+      "loss": 0.7862,
+      "step": 7506
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0232235237120345e-05,
+      "loss": 0.7779,
+      "step": 7507
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0230044727058702e-05,
+      "loss": 0.7516,
+      "step": 7508
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0227854205952846e-05,
+      "loss": 0.4299,
+      "step": 7509
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0225663673907935e-05,
+      "loss": 0.8024,
+      "step": 7510
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0223473131029142e-05,
+      "loss": 0.4761,
+      "step": 7511
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0221282577421629e-05,
+      "loss": 0.8037,
+      "step": 7512
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0219092013190563e-05,
+      "loss": 0.7638,
+      "step": 7513
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.021690143844111e-05,
+      "loss": 0.7686,
+      "step": 7514
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0214710853278433e-05,
+      "loss": 0.4019,
+      "step": 7515
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.021252025780771e-05,
+      "loss": 0.7288,
+      "step": 7516
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0210329652134099e-05,
+      "loss": 0.7594,
+      "step": 7517
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0208139036362777e-05,
+      "loss": 0.4016,
+      "step": 7518
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0205948410598907e-05,
+      "loss": 0.7831,
+      "step": 7519
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0203757774947663e-05,
+      "loss": 0.7557,
+      "step": 7520
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0201567129514215e-05,
+      "loss": 0.7816,
+      "step": 7521
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0199376474403728e-05,
+      "loss": 0.4162,
+      "step": 7522
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0197185809721382e-05,
+      "loss": 0.7706,
+      "step": 7523
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0194995135572339e-05,
+      "loss": 0.3886,
+      "step": 7524
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.019280445206178e-05,
+      "loss": 0.7114,
+      "step": 7525
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.019061375929487e-05,
+      "loss": 0.7758,
+      "step": 7526
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0188423057376789e-05,
+      "loss": 0.7917,
+      "step": 7527
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0186232346412706e-05,
+      "loss": 0.7626,
+      "step": 7528
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0184041626507794e-05,
+      "loss": 0.428,
+      "step": 7529
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0181850897767232e-05,
+      "loss": 0.7719,
+      "step": 7530
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0179660160296187e-05,
+      "loss": 0.4092,
+      "step": 7531
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0177469414199844e-05,
+      "loss": 0.748,
+      "step": 7532
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0175278659583367e-05,
+      "loss": 0.7764,
+      "step": 7533
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0173087896551944e-05,
+      "loss": 0.7142,
+      "step": 7534
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0170897125210744e-05,
+      "loss": 0.4482,
+      "step": 7535
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0168706345664943e-05,
+      "loss": 0.7461,
+      "step": 7536
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0166515558019721e-05,
+      "loss": 0.7444,
+      "step": 7537
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0164324762380255e-05,
+      "loss": 0.3893,
+      "step": 7538
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0162133958851726e-05,
+      "loss": 0.7824,
+      "step": 7539
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0159943147539304e-05,
+      "loss": 0.8035,
+      "step": 7540
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0157752328548176e-05,
+      "loss": 0.4312,
+      "step": 7541
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0155561501983515e-05,
+      "loss": 0.8146,
+      "step": 7542
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0153370667950507e-05,
+      "loss": 0.4387,
+      "step": 7543
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0151179826554324e-05,
+      "loss": 0.7921,
+      "step": 7544
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0148988977900152e-05,
+      "loss": 0.7418,
+      "step": 7545
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0146798122093167e-05,
+      "loss": 0.7869,
+      "step": 7546
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0144607259238555e-05,
+      "loss": 0.7108,
+      "step": 7547
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0142416389441492e-05,
+      "loss": 0.443,
+      "step": 7548
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.014022551280716e-05,
+      "loss": 0.777,
+      "step": 7549
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0138034629440749e-05,
+      "loss": 0.7155,
+      "step": 7550
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0135843739447428e-05,
+      "loss": 0.7344,
+      "step": 7551
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.013365284293239e-05,
+      "loss": 0.7629,
+      "step": 7552
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0131461940000811e-05,
+      "loss": 0.7966,
+      "step": 7553
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0129271030757878e-05,
+      "loss": 0.7499,
+      "step": 7554
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0127080115308774e-05,
+      "loss": 0.7141,
+      "step": 7555
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0124889193758678e-05,
+      "loss": 0.7686,
+      "step": 7556
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.012269826621278e-05,
+      "loss": 0.7815,
+      "step": 7557
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0120507332776263e-05,
+      "loss": 0.8033,
+      "step": 7558
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0118316393554311e-05,
+      "loss": 0.4439,
+      "step": 7559
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0116125448652103e-05,
+      "loss": 0.422,
+      "step": 7560
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0113934498174836e-05,
+      "loss": 0.4449,
+      "step": 7561
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0111743542227683e-05,
+      "loss": 0.7677,
+      "step": 7562
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0109552580915838e-05,
+      "loss": 0.6937,
+      "step": 7563
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0107361614344485e-05,
+      "loss": 0.7382,
+      "step": 7564
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0105170642618807e-05,
+      "loss": 0.7758,
+      "step": 7565
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0102979665843993e-05,
+      "loss": 0.8031,
+      "step": 7566
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0100788684125227e-05,
+      "loss": 0.7267,
+      "step": 7567
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.00985976975677e-05,
+      "loss": 0.767,
+      "step": 7568
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0096406706276595e-05,
+      "loss": 0.7791,
+      "step": 7569
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0094215710357105e-05,
+      "loss": 0.7736,
+      "step": 7570
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.009202470991441e-05,
+      "loss": 0.7494,
+      "step": 7571
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0089833705053703e-05,
+      "loss": 0.7786,
+      "step": 7572
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0087642695880172e-05,
+      "loss": 0.4433,
+      "step": 7573
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0085451682498999e-05,
+      "loss": 0.4257,
+      "step": 7574
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0083260665015382e-05,
+      "loss": 0.7784,
+      "step": 7575
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.00810696435345e-05,
+      "loss": 0.7786,
+      "step": 7576
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0078878618161547e-05,
+      "loss": 0.7631,
+      "step": 7577
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.007668758900171e-05,
+      "loss": 0.7695,
+      "step": 7578
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0074496556160182e-05,
+      "loss": 0.7376,
+      "step": 7579
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.007230551974215e-05,
+      "loss": 0.8128,
+      "step": 7580
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0070114479852801e-05,
+      "loss": 0.7439,
+      "step": 7581
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0067923436597329e-05,
+      "loss": 0.4206,
+      "step": 7582
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0065732390080915e-05,
+      "loss": 0.7681,
+      "step": 7583
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0063541340408762e-05,
+      "loss": 0.7889,
+      "step": 7584
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.006135028768605e-05,
+      "loss": 0.7512,
+      "step": 7585
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0059159232017972e-05,
+      "loss": 0.7963,
+      "step": 7586
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.005696817350972e-05,
+      "loss": 0.7889,
+      "step": 7587
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0054777112266485e-05,
+      "loss": 0.3908,
+      "step": 7588
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0052586048393453e-05,
+      "loss": 0.7953,
+      "step": 7589
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.005039498199582e-05,
+      "loss": 0.777,
+      "step": 7590
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0048203913178773e-05,
+      "loss": 0.7641,
+      "step": 7591
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0046012842047504e-05,
+      "loss": 0.8256,
+      "step": 7592
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0043821768707207e-05,
+      "loss": 0.7712,
+      "step": 7593
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.004163069326307e-05,
+      "loss": 0.7714,
+      "step": 7594
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0039439615820286e-05,
+      "loss": 0.6884,
+      "step": 7595
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0037248536484046e-05,
+      "loss": 0.7071,
+      "step": 7596
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0035057455359541e-05,
+      "loss": 0.7164,
+      "step": 7597
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0032866372551962e-05,
+      "loss": 0.7595,
+      "step": 7598
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0030675288166501e-05,
+      "loss": 0.7551,
+      "step": 7599
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0028484202308353e-05,
+      "loss": 0.4167,
+      "step": 7600
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0026293115082704e-05,
+      "loss": 0.8281,
+      "step": 7601
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0024102026594752e-05,
+      "loss": 0.8272,
+      "step": 7602
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0021910936949684e-05,
+      "loss": 0.7249,
+      "step": 7603
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0019719846252696e-05,
+      "loss": 0.7366,
+      "step": 7604
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0017528754608976e-05,
+      "loss": 0.7458,
+      "step": 7605
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0015337662123717e-05,
+      "loss": 0.7613,
+      "step": 7606
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0013146568902115e-05,
+      "loss": 0.7631,
+      "step": 7607
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0010955475049358e-05,
+      "loss": 0.7606,
+      "step": 7608
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.000876438067064e-05,
+      "loss": 0.8245,
+      "step": 7609
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0006573285871154e-05,
+      "loss": 0.4309,
+      "step": 7610
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.000438219075609e-05,
+      "loss": 0.4133,
+      "step": 7611
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0002191095430643e-05,
+      "loss": 0.7406,
+      "step": 7612
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1e-05,
+      "loss": 0.7592,
+      "step": 7613
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.997808904569362e-06,
+      "loss": 0.7785,
+      "step": 7614
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.995617809243912e-06,
+      "loss": 0.4291,
+      "step": 7615
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.993426714128849e-06,
+      "loss": 0.4141,
+      "step": 7616
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.99123561932936e-06,
+      "loss": 0.4149,
+      "step": 7617
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.989044524950643e-06,
+      "loss": 0.7272,
+      "step": 7618
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.986853431097888e-06,
+      "loss": 0.751,
+      "step": 7619
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.984662337876283e-06,
+      "loss": 0.7772,
+      "step": 7620
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.982471245391029e-06,
+      "loss": 0.7868,
+      "step": 7621
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.98028015374731e-06,
+      "loss": 0.7566,
+      "step": 7622
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.978089063050319e-06,
+      "loss": 0.4306,
+      "step": 7623
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.975897973405253e-06,
+      "loss": 0.7581,
+      "step": 7624
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.973706884917301e-06,
+      "loss": 0.7661,
+      "step": 7625
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.97151579769165e-06,
+      "loss": 0.7457,
+      "step": 7626
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.969324711833502e-06,
+      "loss": 0.7636,
+      "step": 7627
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.967133627448041e-06,
+      "loss": 0.7593,
+      "step": 7628
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.964942544640464e-06,
+      "loss": 0.7664,
+      "step": 7629
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.962751463515958e-06,
+      "loss": 0.761,
+      "step": 7630
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.960560384179716e-06,
+      "loss": 0.7728,
+      "step": 7631
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.958369306736932e-06,
+      "loss": 0.7032,
+      "step": 7632
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.956178231292795e-06,
+      "loss": 0.7482,
+      "step": 7633
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.953987157952498e-06,
+      "loss": 0.7907,
+      "step": 7634
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.95179608682123e-06,
+      "loss": 0.7661,
+      "step": 7635
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.949605018004188e-06,
+      "loss": 0.4226,
+      "step": 7636
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.947413951606552e-06,
+      "loss": 0.7849,
+      "step": 7637
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.945222887733522e-06,
+      "loss": 0.7422,
+      "step": 7638
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.943031826490282e-06,
+      "loss": 0.7432,
+      "step": 7639
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.940840767982031e-06,
+      "loss": 0.7919,
+      "step": 7640
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.938649712313955e-06,
+      "loss": 0.4217,
+      "step": 7641
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.936458659591241e-06,
+      "loss": 0.7701,
+      "step": 7642
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.934267609919086e-06,
+      "loss": 0.7513,
+      "step": 7643
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.932076563402674e-06,
+      "loss": 0.7999,
+      "step": 7644
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.9298855201472e-06,
+      "loss": 0.772,
+      "step": 7645
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.927694480257854e-06,
+      "loss": 0.3729,
+      "step": 7646
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.925503443839818e-06,
+      "loss": 0.7559,
+      "step": 7647
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.923312410998289e-06,
+      "loss": 0.4259,
+      "step": 7648
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.921121381838453e-06,
+      "loss": 0.7523,
+      "step": 7649
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.918930356465501e-06,
+      "loss": 0.7221,
+      "step": 7650
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.916739334984623e-06,
+      "loss": 0.7634,
+      "step": 7651
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.914548317501004e-06,
+      "loss": 0.744,
+      "step": 7652
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.912357304119833e-06,
+      "loss": 0.7617,
+      "step": 7653
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.9101662949463e-06,
+      "loss": 0.7493,
+      "step": 7654
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.907975290085592e-06,
+      "loss": 0.7232,
+      "step": 7655
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.9057842896429e-06,
+      "loss": 0.7438,
+      "step": 7656
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.903593293723406e-06,
+      "loss": 0.7629,
+      "step": 7657
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.901402302432303e-06,
+      "loss": 0.8275,
+      "step": 7658
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.899211315874776e-06,
+      "loss": 0.4321,
+      "step": 7659
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.89702033415601e-06,
+      "loss": 0.7916,
+      "step": 7660
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.894829357381196e-06,
+      "loss": 0.7688,
+      "step": 7661
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.89263838565552e-06,
+      "loss": 0.7355,
+      "step": 7662
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.890447419084163e-06,
+      "loss": 0.7483,
+      "step": 7663
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.888256457772318e-06,
+      "loss": 0.4176,
+      "step": 7664
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.886065501825165e-06,
+      "loss": 0.7858,
+      "step": 7665
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.883874551347899e-06,
+      "loss": 0.8154,
+      "step": 7666
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.881683606445694e-06,
+      "loss": 0.4658,
+      "step": 7667
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.879492667223742e-06,
+      "loss": 0.3923,
+      "step": 7668
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.877301733787223e-06,
+      "loss": 0.7553,
+      "step": 7669
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.875110806241325e-06,
+      "loss": 0.7785,
+      "step": 7670
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.87291988469123e-06,
+      "loss": 0.4277,
+      "step": 7671
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.870728969242124e-06,
+      "loss": 0.7573,
+      "step": 7672
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.868538059999194e-06,
+      "loss": 0.7319,
+      "step": 7673
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.866347157067612e-06,
+      "loss": 0.7481,
+      "step": 7674
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.864156260552573e-06,
+      "loss": 0.7846,
+      "step": 7675
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.861965370559253e-06,
+      "loss": 0.3978,
+      "step": 7676
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.859774487192838e-06,
+      "loss": 0.7894,
+      "step": 7677
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.85758361055851e-06,
+      "loss": 0.8008,
+      "step": 7678
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.855392740761446e-06,
+      "loss": 0.7485,
+      "step": 7679
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.853201877906836e-06,
+      "loss": 0.7935,
+      "step": 7680
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.851011022099853e-06,
+      "loss": 0.7401,
+      "step": 7681
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.848820173445679e-06,
+      "loss": 0.7411,
+      "step": 7682
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.846629332049498e-06,
+      "loss": 0.8058,
+      "step": 7683
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.844438498016488e-06,
+      "loss": 0.396,
+      "step": 7684
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.842247671451827e-06,
+      "loss": 0.7438,
+      "step": 7685
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.8400568524607e-06,
+      "loss": 0.7159,
+      "step": 7686
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.837866041148278e-06,
+      "loss": 0.7777,
+      "step": 7687
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.835675237619746e-06,
+      "loss": 0.8044,
+      "step": 7688
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.833484441980282e-06,
+      "loss": 0.6944,
+      "step": 7689
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.831293654335059e-06,
+      "loss": 0.8116,
+      "step": 7690
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.82910287478926e-06,
+      "loss": 0.754,
+      "step": 7691
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.826912103448058e-06,
+      "loss": 0.7395,
+      "step": 7692
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.824721340416633e-06,
+      "loss": 0.7877,
+      "step": 7693
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.822530585800161e-06,
+      "loss": 0.7432,
+      "step": 7694
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.820339839703817e-06,
+      "loss": 0.4284,
+      "step": 7695
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.818149102232773e-06,
+      "loss": 0.7692,
+      "step": 7696
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.815958373492211e-06,
+      "loss": 0.7499,
+      "step": 7697
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.813767653587299e-06,
+      "loss": 0.7661,
+      "step": 7698
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.811576942623216e-06,
+      "loss": 0.7651,
+      "step": 7699
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.809386240705133e-06,
+      "loss": 0.7288,
+      "step": 7700
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.807195547938224e-06,
+      "loss": 0.7995,
+      "step": 7701
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.805004864427664e-06,
+      "loss": 0.7959,
+      "step": 7702
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.802814190278622e-06,
+      "loss": 0.4296,
+      "step": 7703
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.800623525596275e-06,
+      "loss": 0.8052,
+      "step": 7704
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.79843287048579e-06,
+      "loss": 0.7499,
+      "step": 7705
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.796242225052339e-06,
+      "loss": 0.7292,
+      "step": 7706
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.794051589401094e-06,
+      "loss": 0.3962,
+      "step": 7707
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.791860963637225e-06,
+      "loss": 0.7992,
+      "step": 7708
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.789670347865901e-06,
+      "loss": 0.7877,
+      "step": 7709
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.787479742192295e-06,
+      "loss": 0.423,
+      "step": 7710
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.78528914672157e-06,
+      "loss": 0.7941,
+      "step": 7711
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.783098561558896e-06,
+      "loss": 0.4272,
+      "step": 7712
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.780907986809442e-06,
+      "loss": 0.796,
+      "step": 7713
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.778717422578373e-06,
+      "loss": 0.7809,
+      "step": 7714
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.77652686897086e-06,
+      "loss": 0.7408,
+      "step": 7715
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.774336326092068e-06,
+      "loss": 0.7228,
+      "step": 7716
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.772145794047158e-06,
+      "loss": 0.7902,
+      "step": 7717
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.7699552729413e-06,
+      "loss": 0.7865,
+      "step": 7718
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.767764762879658e-06,
+      "loss": 0.7684,
+      "step": 7719
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.765574263967397e-06,
+      "loss": 0.74,
+      "step": 7720
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.763383776309678e-06,
+      "loss": 0.7748,
+      "step": 7721
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.761193300011665e-06,
+      "loss": 0.749,
+      "step": 7722
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.759002835178523e-06,
+      "loss": 0.7521,
+      "step": 7723
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.756812381915408e-06,
+      "loss": 0.4316,
+      "step": 7724
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.754621940327493e-06,
+      "loss": 0.7788,
+      "step": 7725
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.752431510519927e-06,
+      "loss": 0.7587,
+      "step": 7726
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.750241092597875e-06,
+      "loss": 0.7821,
+      "step": 7727
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.748050686666495e-06,
+      "loss": 0.8115,
+      "step": 7728
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.745860292830948e-06,
+      "loss": 0.7793,
+      "step": 7729
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.743669911196392e-06,
+      "loss": 0.7634,
+      "step": 7730
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.741479541867987e-06,
+      "loss": 0.7452,
+      "step": 7731
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.73928918495089e-06,
+      "loss": 0.3867,
+      "step": 7732
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.737098840550252e-06,
+      "loss": 0.7564,
+      "step": 7733
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.734908508771238e-06,
+      "loss": 0.443,
+      "step": 7734
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.732718189718995e-06,
+      "loss": 0.4411,
+      "step": 7735
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.730527883498688e-06,
+      "loss": 0.7502,
+      "step": 7736
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.728337590215462e-06,
+      "loss": 0.7588,
+      "step": 7737
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.726147309974474e-06,
+      "loss": 0.7691,
+      "step": 7738
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.72395704288088e-06,
+      "loss": 0.7949,
+      "step": 7739
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.721766789039833e-06,
+      "loss": 0.7219,
+      "step": 7740
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.719576548556477e-06,
+      "loss": 0.7345,
+      "step": 7741
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.717386321535971e-06,
+      "loss": 0.727,
+      "step": 7742
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.715196108083463e-06,
+      "loss": 0.4122,
+      "step": 7743
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.713005908304099e-06,
+      "loss": 0.7427,
+      "step": 7744
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.710815722303036e-06,
+      "loss": 0.4482,
+      "step": 7745
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.708625550185418e-06,
+      "loss": 0.7602,
+      "step": 7746
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.706435392056394e-06,
+      "loss": 0.7632,
+      "step": 7747
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.70424524802111e-06,
+      "loss": 0.7842,
+      "step": 7748
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.702055118184713e-06,
+      "loss": 0.4183,
+      "step": 7749
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.699865002652353e-06,
+      "loss": 0.4655,
+      "step": 7750
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.697674901529168e-06,
+      "loss": 0.8181,
+      "step": 7751
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.695484814920309e-06,
+      "loss": 0.7641,
+      "step": 7752
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.693294742930917e-06,
+      "loss": 0.7333,
+      "step": 7753
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.691104685666132e-06,
+      "loss": 0.7852,
+      "step": 7754
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.688914643231106e-06,
+      "loss": 0.7794,
+      "step": 7755
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.68672461573097e-06,
+      "loss": 0.3989,
+      "step": 7756
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.684534603270871e-06,
+      "loss": 0.4139,
+      "step": 7757
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.682344605955948e-06,
+      "loss": 0.723,
+      "step": 7758
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.68015462389134e-06,
+      "loss": 0.7487,
+      "step": 7759
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.677964657182185e-06,
+      "loss": 0.8313,
+      "step": 7760
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.675774705933623e-06,
+      "loss": 0.4136,
+      "step": 7761
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.67358477025079e-06,
+      "loss": 0.7901,
+      "step": 7762
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.671394850238825e-06,
+      "loss": 0.7241,
+      "step": 7763
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.669204946002862e-06,
+      "loss": 0.7587,
+      "step": 7764
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.667015057648034e-06,
+      "loss": 0.8144,
+      "step": 7765
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.66482518527948e-06,
+      "loss": 0.7706,
+      "step": 7766
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.66263532900233e-06,
+      "loss": 0.7434,
+      "step": 7767
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.66044548892172e-06,
+      "loss": 0.395,
+      "step": 7768
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.658255665142778e-06,
+      "loss": 0.4386,
+      "step": 7769
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.656065857770641e-06,
+      "loss": 0.4278,
+      "step": 7770
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.653876066910431e-06,
+      "loss": 0.7593,
+      "step": 7771
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.651686292667284e-06,
+      "loss": 0.4377,
+      "step": 7772
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.649496535146324e-06,
+      "loss": 0.7989,
+      "step": 7773
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.647306794452687e-06,
+      "loss": 0.7633,
+      "step": 7774
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.645117070691493e-06,
+      "loss": 0.8332,
+      "step": 7775
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.64292736396787e-06,
+      "loss": 0.4103,
+      "step": 7776
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.640737674386947e-06,
+      "loss": 0.7787,
+      "step": 7777
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.638548002053842e-06,
+      "loss": 0.3832,
+      "step": 7778
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.636358347073688e-06,
+      "loss": 0.7641,
+      "step": 7779
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.634168709551601e-06,
+      "loss": 0.4055,
+      "step": 7780
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.631979089592706e-06,
+      "loss": 0.4375,
+      "step": 7781
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.629789487302124e-06,
+      "loss": 0.7911,
+      "step": 7782
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.627599902784974e-06,
+      "loss": 0.7293,
+      "step": 7783
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.625410336146382e-06,
+      "loss": 0.4153,
+      "step": 7784
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.623220787491462e-06,
+      "loss": 0.7938,
+      "step": 7785
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.62103125692533e-06,
+      "loss": 0.4627,
+      "step": 7786
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.618841744553105e-06,
+      "loss": 0.4034,
+      "step": 7787
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.616652250479905e-06,
+      "loss": 0.7192,
+      "step": 7788
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.614462774810842e-06,
+      "loss": 0.789,
+      "step": 7789
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.612273317651035e-06,
+      "loss": 0.7916,
+      "step": 7790
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.610083879105596e-06,
+      "loss": 0.764,
+      "step": 7791
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.607894459279635e-06,
+      "loss": 0.7466,
+      "step": 7792
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.605705058278266e-06,
+      "loss": 0.6958,
+      "step": 7793
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.6035156762066e-06,
+      "loss": 0.7598,
+      "step": 7794
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.601326313169748e-06,
+      "loss": 0.7768,
+      "step": 7795
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.599136969272818e-06,
+      "loss": 0.7653,
+      "step": 7796
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.596947644620916e-06,
+      "loss": 0.7532,
+      "step": 7797
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.594758339319154e-06,
+      "loss": 0.4171,
+      "step": 7798
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.592569053472638e-06,
+      "loss": 0.4596,
+      "step": 7799
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.590379787186466e-06,
+      "loss": 0.7719,
+      "step": 7800
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.588190540565751e-06,
+      "loss": 0.7825,
+      "step": 7801
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.586001313715593e-06,
+      "loss": 0.4241,
+      "step": 7802
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.583812106741092e-06,
+      "loss": 0.4264,
+      "step": 7803
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.581622919747356e-06,
+      "loss": 0.7778,
+      "step": 7804
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.57943375283948e-06,
+      "loss": 0.7138,
+      "step": 7805
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.577244606122566e-06,
+      "loss": 0.7423,
+      "step": 7806
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.575055479701713e-06,
+      "loss": 0.4542,
+      "step": 7807
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.572866373682018e-06,
+      "loss": 0.3909,
+      "step": 7808
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.570677288168578e-06,
+      "loss": 0.4433,
+      "step": 7809
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.568488223266488e-06,
+      "loss": 0.7645,
+      "step": 7810
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.566299179080845e-06,
+      "loss": 0.745,
+      "step": 7811
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.56411015571674e-06,
+      "loss": 0.4042,
+      "step": 7812
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.561921153279268e-06,
+      "loss": 0.7754,
+      "step": 7813
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.559732171873524e-06,
+      "loss": 0.737,
+      "step": 7814
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.557543211604592e-06,
+      "loss": 0.7515,
+      "step": 7815
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.555354272577561e-06,
+      "loss": 0.7416,
+      "step": 7816
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.553165354897528e-06,
+      "loss": 0.8103,
+      "step": 7817
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.550976458669576e-06,
+      "loss": 0.4053,
+      "step": 7818
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.548787583998787e-06,
+      "loss": 0.7302,
+      "step": 7819
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.546598730990256e-06,
+      "loss": 0.7447,
+      "step": 7820
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.54440989974906e-06,
+      "loss": 0.4231,
+      "step": 7821
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.542221090380286e-06,
+      "loss": 0.7109,
+      "step": 7822
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.54003230298902e-06,
+      "loss": 0.7419,
+      "step": 7823
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.537843537680336e-06,
+      "loss": 0.4163,
+      "step": 7824
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.535654794559319e-06,
+      "loss": 0.7307,
+      "step": 7825
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.533466073731047e-06,
+      "loss": 0.7737,
+      "step": 7826
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.531277375300599e-06,
+      "loss": 0.4238,
+      "step": 7827
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.529088699373053e-06,
+      "loss": 0.7628,
+      "step": 7828
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.526900046053486e-06,
+      "loss": 0.7771,
+      "step": 7829
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.524711415446967e-06,
+      "loss": 0.3981,
+      "step": 7830
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.522522807658578e-06,
+      "loss": 0.398,
+      "step": 7831
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.520334222793384e-06,
+      "loss": 0.7785,
+      "step": 7832
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.518145660956462e-06,
+      "loss": 0.7309,
+      "step": 7833
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.515957122252883e-06,
+      "loss": 0.7136,
+      "step": 7834
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.513768606787711e-06,
+      "loss": 0.3949,
+      "step": 7835
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.511580114666022e-06,
+      "loss": 0.788,
+      "step": 7836
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.509391645992877e-06,
+      "loss": 0.7766,
+      "step": 7837
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.507203200873346e-06,
+      "loss": 0.7643,
+      "step": 7838
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.505014779412494e-06,
+      "loss": 0.7442,
+      "step": 7839
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.502826381715382e-06,
+      "loss": 0.7635,
+      "step": 7840
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.500638007887074e-06,
+      "loss": 0.7665,
+      "step": 7841
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.498449658032632e-06,
+      "loss": 0.7365,
+      "step": 7842
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.496261332257118e-06,
+      "loss": 0.8365,
+      "step": 7843
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.494073030665592e-06,
+      "loss": 0.7504,
+      "step": 7844
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.49188475336311e-06,
+      "loss": 0.7399,
+      "step": 7845
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.489696500454725e-06,
+      "loss": 0.7748,
+      "step": 7846
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.487508272045499e-06,
+      "loss": 0.7617,
+      "step": 7847
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.485320068240483e-06,
+      "loss": 0.756,
+      "step": 7848
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.483131889144734e-06,
+      "loss": 0.7418,
+      "step": 7849
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.4809437348633e-06,
+      "loss": 0.7806,
+      "step": 7850
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.478755605501235e-06,
+      "loss": 0.7487,
+      "step": 7851
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.476567501163589e-06,
+      "loss": 0.7693,
+      "step": 7852
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.474379421955408e-06,
+      "loss": 0.7335,
+      "step": 7853
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.472191367981744e-06,
+      "loss": 0.7593,
+      "step": 7854
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.47000333934764e-06,
+      "loss": 0.38,
+      "step": 7855
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.46781533615814e-06,
+      "loss": 0.7747,
+      "step": 7856
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.465627358518291e-06,
+      "loss": 0.7463,
+      "step": 7857
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.463439406533133e-06,
+      "loss": 0.7467,
+      "step": 7858
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.461251480307713e-06,
+      "loss": 0.7766,
+      "step": 7859
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.459063579947062e-06,
+      "loss": 0.7743,
+      "step": 7860
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.456875705556225e-06,
+      "loss": 0.7425,
+      "step": 7861
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.454687857240237e-06,
+      "loss": 0.4092,
+      "step": 7862
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.452500035104137e-06,
+      "loss": 0.4141,
+      "step": 7863
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.450312239252956e-06,
+      "loss": 0.7355,
+      "step": 7864
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.448124469791733e-06,
+      "loss": 0.7171,
+      "step": 7865
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.445936726825498e-06,
+      "loss": 0.7403,
+      "step": 7866
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.44374901045928e-06,
+      "loss": 0.383,
+      "step": 7867
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.441561320798114e-06,
+      "loss": 0.7992,
+      "step": 7868
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.439373657947024e-06,
+      "loss": 0.4354,
+      "step": 7869
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.437186022011042e-06,
+      "loss": 0.6909,
+      "step": 7870
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.434998413095193e-06,
+      "loss": 0.4387,
+      "step": 7871
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.432810831304499e-06,
+      "loss": 0.4193,
+      "step": 7872
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.430623276743987e-06,
+      "loss": 0.7845,
+      "step": 7873
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.428435749518679e-06,
+      "loss": 0.7248,
+      "step": 7874
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.426248249733591e-06,
+      "loss": 0.4403,
+      "step": 7875
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.424060777493749e-06,
+      "loss": 0.7732,
+      "step": 7876
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.421873332904168e-06,
+      "loss": 0.7911,
+      "step": 7877
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.419685916069866e-06,
+      "loss": 0.7963,
+      "step": 7878
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.417498527095857e-06,
+      "loss": 0.7617,
+      "step": 7879
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.415311166087157e-06,
+      "loss": 0.4458,
+      "step": 7880
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.41312383314878e-06,
+      "loss": 0.7341,
+      "step": 7881
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.410936528385737e-06,
+      "loss": 0.7643,
+      "step": 7882
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.408749251903036e-06,
+      "loss": 0.7836,
+      "step": 7883
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.406562003805689e-06,
+      "loss": 0.7241,
+      "step": 7884
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.4043747841987e-06,
+      "loss": 0.7754,
+      "step": 7885
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.40218759318708e-06,
+      "loss": 0.7739,
+      "step": 7886
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.400000430875831e-06,
+      "loss": 0.7597,
+      "step": 7887
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.397813297369955e-06,
+      "loss": 0.7611,
+      "step": 7888
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.395626192774461e-06,
+      "loss": 0.7372,
+      "step": 7889
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.393439117194342e-06,
+      "loss": 0.7715,
+      "step": 7890
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.391252070734596e-06,
+      "loss": 0.7542,
+      "step": 7891
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.38906505350023e-06,
+      "loss": 0.7836,
+      "step": 7892
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.386878065596232e-06,
+      "loss": 0.7473,
+      "step": 7893
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.3846911071276e-06,
+      "loss": 0.4077,
+      "step": 7894
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.38250417819933e-06,
+      "loss": 0.7962,
+      "step": 7895
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.38031727891641e-06,
+      "loss": 0.4391,
+      "step": 7896
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.378130409383834e-06,
+      "loss": 0.8309,
+      "step": 7897
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.375943569706592e-06,
+      "loss": 0.7839,
+      "step": 7898
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.373756759989668e-06,
+      "loss": 0.4306,
+      "step": 7899
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.371569980338053e-06,
+      "loss": 0.7315,
+      "step": 7900
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.369383230856726e-06,
+      "loss": 0.4102,
+      "step": 7901
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.36719651165068e-06,
+      "loss": 0.38,
+      "step": 7902
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.36500982282489e-06,
+      "loss": 0.7395,
+      "step": 7903
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.362823164484341e-06,
+      "loss": 0.7817,
+      "step": 7904
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.360636536734008e-06,
+      "loss": 0.7376,
+      "step": 7905
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.358449939678869e-06,
+      "loss": 0.4497,
+      "step": 7906
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.356263373423902e-06,
+      "loss": 0.7948,
+      "step": 7907
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.354076838074083e-06,
+      "loss": 0.7872,
+      "step": 7908
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.351890333734385e-06,
+      "loss": 0.812,
+      "step": 7909
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.349703860509776e-06,
+      "loss": 0.7462,
+      "step": 7910
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.347517418505233e-06,
+      "loss": 0.4092,
+      "step": 7911
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.345331007825719e-06,
+      "loss": 0.7305,
+      "step": 7912
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.343144628576204e-06,
+      "loss": 0.4558,
+      "step": 7913
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.340958280861656e-06,
+      "loss": 0.4405,
+      "step": 7914
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.338771964787032e-06,
+      "loss": 0.7658,
+      "step": 7915
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.336585680457305e-06,
+      "loss": 0.7557,
+      "step": 7916
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.334399427977426e-06,
+      "loss": 0.7656,
+      "step": 7917
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.33221320745237e-06,
+      "loss": 0.7522,
+      "step": 7918
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.330027018987075e-06,
+      "loss": 0.7309,
+      "step": 7919
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.327840862686511e-06,
+      "loss": 0.7775,
+      "step": 7920
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.325654738655627e-06,
+      "loss": 0.8023,
+      "step": 7921
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.323468646999383e-06,
+      "loss": 0.7792,
+      "step": 7922
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.321282587822726e-06,
+      "loss": 0.6962,
+      "step": 7923
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.31909656123061e-06,
+      "loss": 0.688,
+      "step": 7924
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.31691056732798e-06,
+      "loss": 0.7176,
+      "step": 7925
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.314724606219784e-06,
+      "loss": 0.7697,
+      "step": 7926
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.312538678010973e-06,
+      "loss": 0.7236,
+      "step": 7927
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.310352782806484e-06,
+      "loss": 0.7479,
+      "step": 7928
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.308166920711265e-06,
+      "loss": 0.7378,
+      "step": 7929
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.305981091830255e-06,
+      "loss": 0.777,
+      "step": 7930
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.303795296268393e-06,
+      "loss": 0.7467,
+      "step": 7931
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.301609534130618e-06,
+      "loss": 0.6763,
+      "step": 7932
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.299423805521869e-06,
+      "loss": 0.7971,
+      "step": 7933
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.297238110547075e-06,
+      "loss": 0.7574,
+      "step": 7934
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.295052449311167e-06,
+      "loss": 0.7441,
+      "step": 7935
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.292866821919084e-06,
+      "loss": 0.4362,
+      "step": 7936
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.29068122847575e-06,
+      "loss": 0.6777,
+      "step": 7937
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.2884956690861e-06,
+      "loss": 0.4182,
+      "step": 7938
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.286310143855051e-06,
+      "loss": 0.7669,
+      "step": 7939
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.284124652887536e-06,
+      "loss": 0.7594,
+      "step": 7940
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.281939196288476e-06,
+      "loss": 0.787,
+      "step": 7941
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.279753774162788e-06,
+      "loss": 0.6938,
+      "step": 7942
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.277568386615398e-06,
+      "loss": 0.7013,
+      "step": 7943
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.27538303375122e-06,
+      "loss": 0.7249,
+      "step": 7944
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.273197715675176e-06,
+      "loss": 0.73,
+      "step": 7945
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.271012432492174e-06,
+      "loss": 0.7629,
+      "step": 7946
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.268827184307131e-06,
+      "loss": 0.7463,
+      "step": 7947
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.266641971224963e-06,
+      "loss": 0.7435,
+      "step": 7948
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.264456793350571e-06,
+      "loss": 0.7355,
+      "step": 7949
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.262271650788866e-06,
+      "loss": 0.7753,
+      "step": 7950
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.260086543644758e-06,
+      "loss": 0.4304,
+      "step": 7951
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.257901472023151e-06,
+      "loss": 0.4001,
+      "step": 7952
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.255716436028941e-06,
+      "loss": 0.7642,
+      "step": 7953
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.25353143576704e-06,
+      "loss": 0.7698,
+      "step": 7954
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.251346471342339e-06,
+      "loss": 0.7319,
+      "step": 7955
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.249161542859744e-06,
+      "loss": 0.7296,
+      "step": 7956
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.246976650424144e-06,
+      "loss": 0.4437,
+      "step": 7957
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.244791794140436e-06,
+      "loss": 0.7685,
+      "step": 7958
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.242606974113515e-06,
+      "loss": 0.8148,
+      "step": 7959
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.240422190448268e-06,
+      "loss": 0.3902,
+      "step": 7960
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.238237443249587e-06,
+      "loss": 0.7184,
+      "step": 7961
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.236052732622359e-06,
+      "loss": 0.7693,
+      "step": 7962
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.233868058671472e-06,
+      "loss": 0.7466,
+      "step": 7963
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.231683421501804e-06,
+      "loss": 0.7644,
+      "step": 7964
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.229498821218244e-06,
+      "loss": 0.4424,
+      "step": 7965
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.227314257925665e-06,
+      "loss": 0.7504,
+      "step": 7966
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.225129731728953e-06,
+      "loss": 0.7074,
+      "step": 7967
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.22294524273298e-06,
+      "loss": 0.7822,
+      "step": 7968
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.220760791042625e-06,
+      "loss": 0.4811,
+      "step": 7969
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.218576376762758e-06,
+      "loss": 0.4115,
+      "step": 7970
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.21639199999825e-06,
+      "loss": 0.7887,
+      "step": 7971
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.214207660853975e-06,
+      "loss": 0.7394,
+      "step": 7972
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.2120233594348e-06,
+      "loss": 0.7759,
+      "step": 7973
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.209839095845586e-06,
+      "loss": 0.7513,
+      "step": 7974
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.207654870191203e-06,
+      "loss": 0.4522,
+      "step": 7975
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.20547068257651e-06,
+      "loss": 0.7383,
+      "step": 7976
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.203286533106371e-06,
+      "loss": 0.3897,
+      "step": 7977
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.201102421885643e-06,
+      "loss": 0.8078,
+      "step": 7978
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.198918349019184e-06,
+      "loss": 0.7685,
+      "step": 7979
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.196734314611844e-06,
+      "loss": 0.7775,
+      "step": 7980
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.194550318768481e-06,
+      "loss": 0.7713,
+      "step": 7981
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.192366361593944e-06,
+      "loss": 0.7913,
+      "step": 7982
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.19018244319309e-06,
+      "loss": 0.7478,
+      "step": 7983
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.187998563670757e-06,
+      "loss": 0.4609,
+      "step": 7984
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.185814723131794e-06,
+      "loss": 0.7515,
+      "step": 7985
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.183630921681048e-06,
+      "loss": 0.7531,
+      "step": 7986
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.181447159423356e-06,
+      "loss": 0.7558,
+      "step": 7987
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.179263436463563e-06,
+      "loss": 0.761,
+      "step": 7988
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.177079752906507e-06,
+      "loss": 0.7849,
+      "step": 7989
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.17489610885702e-06,
+      "loss": 0.7383,
+      "step": 7990
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.172712504419942e-06,
+      "loss": 0.3948,
+      "step": 7991
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.170528939700101e-06,
+      "loss": 0.7931,
+      "step": 7992
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.168345414802336e-06,
+      "loss": 0.7525,
+      "step": 7993
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.166161929831464e-06,
+      "loss": 0.7506,
+      "step": 7994
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.163978484892317e-06,
+      "loss": 0.4012,
+      "step": 7995
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.161795080089722e-06,
+      "loss": 0.411,
+      "step": 7996
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.1596117155285e-06,
+      "loss": 0.7391,
+      "step": 7997
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.157428391313472e-06,
+      "loss": 0.7795,
+      "step": 7998
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.15524510754946e-06,
+      "loss": 0.404,
+      "step": 7999
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.15306186434128e-06,
+      "loss": 0.71,
+      "step": 8000
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.150878661793742e-06,
+      "loss": 0.8218,
+      "step": 8001
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.148695500011669e-06,
+      "loss": 0.7792,
+      "step": 8002
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.146512379099864e-06,
+      "loss": 0.6912,
+      "step": 8003
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.144329299163142e-06,
+      "loss": 0.7986,
+      "step": 8004
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.142146260306309e-06,
+      "loss": 0.7511,
+      "step": 8005
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.139963262634166e-06,
+      "loss": 0.7856,
+      "step": 8006
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.137780306251523e-06,
+      "loss": 0.7856,
+      "step": 8007
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.13559739126318e-06,
+      "loss": 0.7781,
+      "step": 8008
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.133414517773934e-06,
+      "loss": 0.7616,
+      "step": 8009
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.131231685888581e-06,
+      "loss": 0.4748,
+      "step": 8010
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.129048895711922e-06,
+      "loss": 0.7638,
+      "step": 8011
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.126866147348748e-06,
+      "loss": 0.43,
+      "step": 8012
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.12468344090385e-06,
+      "loss": 0.42,
+      "step": 8013
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.122500776482016e-06,
+      "loss": 0.4333,
+      "step": 8014
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.120318154188039e-06,
+      "loss": 0.4036,
+      "step": 8015
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.118135574126701e-06,
+      "loss": 0.7365,
+      "step": 8016
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.115953036402783e-06,
+      "loss": 0.7755,
+      "step": 8017
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.11377054112107e-06,
+      "loss": 0.4622,
+      "step": 8018
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.11158808838634e-06,
+      "loss": 0.4671,
+      "step": 8019
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.10940567830337e-06,
+      "loss": 0.7666,
+      "step": 8020
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.10722331097694e-06,
+      "loss": 0.7587,
+      "step": 8021
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.105040986511816e-06,
+      "loss": 0.7678,
+      "step": 8022
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.102858705012775e-06,
+      "loss": 0.7538,
+      "step": 8023
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.100676466584584e-06,
+      "loss": 0.4003,
+      "step": 8024
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.098494271332005e-06,
+      "loss": 0.4222,
+      "step": 8025
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.096312119359811e-06,
+      "loss": 0.4214,
+      "step": 8026
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.094130010772762e-06,
+      "loss": 0.7376,
+      "step": 8027
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.091947945675616e-06,
+      "loss": 0.7241,
+      "step": 8028
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.089765924173137e-06,
+      "loss": 0.7835,
+      "step": 8029
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.087583946370076e-06,
+      "loss": 0.7856,
+      "step": 8030
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.085402012371192e-06,
+      "loss": 0.7516,
+      "step": 8031
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.083220122281238e-06,
+      "loss": 0.7902,
+      "step": 8032
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.081038276204958e-06,
+      "loss": 0.4309,
+      "step": 8033
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.078856474247108e-06,
+      "loss": 0.7625,
+      "step": 8034
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.076674716512427e-06,
+      "loss": 0.7344,
+      "step": 8035
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.074493003105665e-06,
+      "loss": 0.7207,
+      "step": 8036
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.072311334131563e-06,
+      "loss": 0.7625,
+      "step": 8037
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.07012970969486e-06,
+      "loss": 0.7783,
+      "step": 8038
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.067948129900288e-06,
+      "loss": 0.7376,
+      "step": 8039
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.06576659485259e-06,
+      "loss": 0.7451,
+      "step": 8040
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.063585104656494e-06,
+      "loss": 0.6912,
+      "step": 8041
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.061403659416736e-06,
+      "loss": 0.4289,
+      "step": 8042
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.059222259238042e-06,
+      "loss": 0.727,
+      "step": 8043
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.057040904225138e-06,
+      "loss": 0.7605,
+      "step": 8044
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.054859594482752e-06,
+      "loss": 0.7761,
+      "step": 8045
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.052678330115603e-06,
+      "loss": 0.7627,
+      "step": 8046
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.050497111228414e-06,
+      "loss": 0.7963,
+      "step": 8047
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.048315937925902e-06,
+      "loss": 0.7275,
+      "step": 8048
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.046134810312783e-06,
+      "loss": 0.7568,
+      "step": 8049
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.043953728493771e-06,
+      "loss": 0.7519,
+      "step": 8050
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.041772692573574e-06,
+      "loss": 0.7695,
+      "step": 8051
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.039591702656913e-06,
+      "loss": 0.4313,
+      "step": 8052
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.03741075884848e-06,
+      "loss": 0.7207,
+      "step": 8053
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.035229861252988e-06,
+      "loss": 0.4411,
+      "step": 8054
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.033049009975135e-06,
+      "loss": 0.7725,
+      "step": 8055
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.030868205119629e-06,
+      "loss": 0.7306,
+      "step": 8056
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.028687446791162e-06,
+      "loss": 0.7443,
+      "step": 8057
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.026506735094432e-06,
+      "loss": 0.7708,
+      "step": 8058
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.024326070134135e-06,
+      "loss": 0.4185,
+      "step": 8059
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.022145452014957e-06,
+      "loss": 0.7665,
+      "step": 8060
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.019964880841593e-06,
+      "loss": 0.7338,
+      "step": 8061
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.017784356718727e-06,
+      "loss": 0.4322,
+      "step": 8062
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.015603879751044e-06,
+      "loss": 0.7053,
+      "step": 8063
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.013423450043228e-06,
+      "loss": 0.7397,
+      "step": 8064
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.011243067699956e-06,
+      "loss": 0.7124,
+      "step": 8065
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.00906273282591e-06,
+      "loss": 0.7101,
+      "step": 8066
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.006882445525767e-06,
+      "loss": 0.7511,
+      "step": 8067
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.004702205904193e-06,
+      "loss": 0.8028,
+      "step": 8068
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.002522014065862e-06,
+      "loss": 0.762,
+      "step": 8069
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.000341870115447e-06,
+      "loss": 0.741,
+      "step": 8070
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.998161774157608e-06,
+      "loss": 0.7701,
+      "step": 8071
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.995981726297016e-06,
+      "loss": 0.7735,
+      "step": 8072
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.993801726638328e-06,
+      "loss": 0.7472,
+      "step": 8073
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.991621775286207e-06,
+      "loss": 0.4023,
+      "step": 8074
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.989441872345308e-06,
+      "loss": 0.7768,
+      "step": 8075
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.987262017920285e-06,
+      "loss": 0.7847,
+      "step": 8076
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.985082212115794e-06,
+      "loss": 0.7787,
+      "step": 8077
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.982902455036483e-06,
+      "loss": 0.819,
+      "step": 8078
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.980722746787e-06,
+      "loss": 0.7802,
+      "step": 8079
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.978543087471991e-06,
+      "loss": 0.4783,
+      "step": 8080
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.976363477196099e-06,
+      "loss": 0.7285,
+      "step": 8081
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.974183916063967e-06,
+      "loss": 0.8293,
+      "step": 8082
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.972004404180232e-06,
+      "loss": 0.6965,
+      "step": 8083
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.969824941649528e-06,
+      "loss": 0.7583,
+      "step": 8084
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.96764552857649e-06,
+      "loss": 0.4423,
+      "step": 8085
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.96546616506575e-06,
+      "loss": 0.7837,
+      "step": 8086
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.963286851221935e-06,
+      "loss": 0.4364,
+      "step": 8087
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.961107587149677e-06,
+      "loss": 0.7637,
+      "step": 8088
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.958928372953596e-06,
+      "loss": 0.8277,
+      "step": 8089
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.956749208738315e-06,
+      "loss": 0.7289,
+      "step": 8090
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.954570094608454e-06,
+      "loss": 0.7409,
+      "step": 8091
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.952391030668627e-06,
+      "loss": 0.7625,
+      "step": 8092
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.950212017023454e-06,
+      "loss": 0.7308,
+      "step": 8093
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.94803305377754e-06,
+      "loss": 0.7483,
+      "step": 8094
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.945854141035503e-06,
+      "loss": 0.7861,
+      "step": 8095
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.943675278901948e-06,
+      "loss": 0.3885,
+      "step": 8096
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.941496467481478e-06,
+      "loss": 0.7439,
+      "step": 8097
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.939317706878694e-06,
+      "loss": 0.4266,
+      "step": 8098
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.9371389971982e-06,
+      "loss": 0.446,
+      "step": 8099
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.934960338544589e-06,
+      "loss": 0.7453,
+      "step": 8100
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.932781731022461e-06,
+      "loss": 0.43,
+      "step": 8101
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.930603174736407e-06,
+      "loss": 0.7198,
+      "step": 8102
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.928424669791015e-06,
+      "loss": 0.4072,
+      "step": 8103
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.926246216290876e-06,
+      "loss": 0.7567,
+      "step": 8104
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.924067814340575e-06,
+      "loss": 0.7463,
+      "step": 8105
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.921889464044695e-06,
+      "loss": 0.753,
+      "step": 8106
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.919711165507814e-06,
+      "loss": 0.7358,
+      "step": 8107
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.91753291883451e-06,
+      "loss": 0.7621,
+      "step": 8108
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.915354724129363e-06,
+      "loss": 0.3932,
+      "step": 8109
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.91317658149694e-06,
+      "loss": 0.7805,
+      "step": 8110
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.910998491041818e-06,
+      "loss": 0.7452,
+      "step": 8111
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.908820452868562e-06,
+      "loss": 0.4546,
+      "step": 8112
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.906642467081735e-06,
+      "loss": 0.7438,
+      "step": 8113
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.904464533785898e-06,
+      "loss": 0.7401,
+      "step": 8114
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.902286653085617e-06,
+      "loss": 0.7421,
+      "step": 8115
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.900108825085448e-06,
+      "loss": 0.7724,
+      "step": 8116
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.897931049889946e-06,
+      "loss": 0.7849,
+      "step": 8117
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.895753327603665e-06,
+      "loss": 0.7738,
+      "step": 8118
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.893575658331151e-06,
+      "loss": 0.4451,
+      "step": 8119
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.89139804217696e-06,
+      "loss": 0.4002,
+      "step": 8120
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.889220479245627e-06,
+      "loss": 0.7379,
+      "step": 8121
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.887042969641701e-06,
+      "loss": 0.7301,
+      "step": 8122
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.884865513469722e-06,
+      "loss": 0.4154,
+      "step": 8123
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.882688110834225e-06,
+      "loss": 0.7581,
+      "step": 8124
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.880510761839747e-06,
+      "loss": 0.7921,
+      "step": 8125
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.878333466590818e-06,
+      "loss": 0.7388,
+      "step": 8126
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.876156225191974e-06,
+      "loss": 0.765,
+      "step": 8127
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.873979037747731e-06,
+      "loss": 0.7568,
+      "step": 8128
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.871801904362625e-06,
+      "loss": 0.7496,
+      "step": 8129
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.869624825141167e-06,
+      "loss": 0.4033,
+      "step": 8130
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.867447800187886e-06,
+      "loss": 0.7642,
+      "step": 8131
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.865270829607293e-06,
+      "loss": 0.8118,
+      "step": 8132
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.863093913503906e-06,
+      "loss": 0.7955,
+      "step": 8133
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.860917051982236e-06,
+      "loss": 0.7667,
+      "step": 8134
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.858740245146786e-06,
+      "loss": 0.4674,
+      "step": 8135
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.856563493102071e-06,
+      "loss": 0.7505,
+      "step": 8136
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.854386795952588e-06,
+      "loss": 0.7336,
+      "step": 8137
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.852210153802842e-06,
+      "loss": 0.7031,
+      "step": 8138
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.85003356675733e-06,
+      "loss": 0.7487,
+      "step": 8139
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.847857034920547e-06,
+      "loss": 0.7903,
+      "step": 8140
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.845680558396988e-06,
+      "loss": 0.8003,
+      "step": 8141
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.843504137291146e-06,
+      "loss": 0.7715,
+      "step": 8142
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.8413277717075e-06,
+      "loss": 0.8417,
+      "step": 8143
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.83915146175054e-06,
+      "loss": 0.7478,
+      "step": 8144
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.836975207524751e-06,
+      "loss": 0.8175,
+      "step": 8145
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.83479900913461e-06,
+      "loss": 0.7239,
+      "step": 8146
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.832622866684594e-06,
+      "loss": 0.7976,
+      "step": 8147
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.830446780279175e-06,
+      "loss": 0.4183,
+      "step": 8148
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.828270750022832e-06,
+      "loss": 0.7531,
+      "step": 8149
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.82609477602003e-06,
+      "loss": 0.7687,
+      "step": 8150
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.823918858375231e-06,
+      "loss": 0.7473,
+      "step": 8151
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.821742997192906e-06,
+      "loss": 0.7528,
+      "step": 8152
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.81956719257751e-06,
+      "loss": 0.7827,
+      "step": 8153
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.817391444633506e-06,
+      "loss": 0.7619,
+      "step": 8154
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.815215753465349e-06,
+      "loss": 0.7925,
+      "step": 8155
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.81304011917749e-06,
+      "loss": 0.7928,
+      "step": 8156
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.810864541874375e-06,
+      "loss": 0.7913,
+      "step": 8157
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.808689021660456e-06,
+      "loss": 0.7035,
+      "step": 8158
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.806513558640179e-06,
+      "loss": 0.7079,
+      "step": 8159
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.804338152917979e-06,
+      "loss": 0.4332,
+      "step": 8160
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.802162804598304e-06,
+      "loss": 0.4506,
+      "step": 8161
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.799987513785583e-06,
+      "loss": 0.7356,
+      "step": 8162
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.797812280584255e-06,
+      "loss": 0.736,
+      "step": 8163
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.795637105098744e-06,
+      "loss": 0.7811,
+      "step": 8164
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.793461987433488e-06,
+      "loss": 0.7099,
+      "step": 8165
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.791286927692903e-06,
+      "loss": 0.8172,
+      "step": 8166
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.789111925981412e-06,
+      "loss": 0.7664,
+      "step": 8167
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.786936982403443e-06,
+      "loss": 0.3798,
+      "step": 8168
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.784762097063401e-06,
+      "loss": 0.7655,
+      "step": 8169
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.782587270065712e-06,
+      "loss": 0.4872,
+      "step": 8170
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.780412501514782e-06,
+      "loss": 0.3929,
+      "step": 8171
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.778237791515018e-06,
+      "loss": 0.7578,
+      "step": 8172
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.776063140170823e-06,
+      "loss": 0.7897,
+      "step": 8173
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.773888547586607e-06,
+      "loss": 0.7788,
+      "step": 8174
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.771714013866764e-06,
+      "loss": 0.426,
+      "step": 8175
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.769539539115695e-06,
+      "loss": 0.7501,
+      "step": 8176
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.767365123437794e-06,
+      "loss": 0.3997,
+      "step": 8177
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.765190766937447e-06,
+      "loss": 0.4139,
+      "step": 8178
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.76301646971905e-06,
+      "loss": 0.7577,
+      "step": 8179
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.760842231886986e-06,
+      "loss": 0.7876,
+      "step": 8180
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.758668053545639e-06,
+      "loss": 0.7996,
+      "step": 8181
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.756493934799387e-06,
+      "loss": 0.7196,
+      "step": 8182
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.754319875752608e-06,
+      "loss": 0.7778,
+      "step": 8183
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.752145876509677e-06,
+      "loss": 0.8241,
+      "step": 8184
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.749971937174964e-06,
+      "loss": 0.4516,
+      "step": 8185
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.747798057852846e-06,
+      "loss": 0.7548,
+      "step": 8186
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.745624238647673e-06,
+      "loss": 0.7719,
+      "step": 8187
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.74345047966382e-06,
+      "loss": 0.7754,
+      "step": 8188
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.741276781005641e-06,
+      "loss": 0.7258,
+      "step": 8189
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.739103142777497e-06,
+      "loss": 0.7563,
+      "step": 8190
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.73692956508374e-06,
+      "loss": 0.764,
+      "step": 8191
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.734756048028722e-06,
+      "loss": 0.7504,
+      "step": 8192
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.732582591716794e-06,
+      "loss": 0.7223,
+      "step": 8193
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.730409196252296e-06,
+      "loss": 0.4436,
+      "step": 8194
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.728235861739573e-06,
+      "loss": 0.7568,
+      "step": 8195
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.726062588282964e-06,
+      "loss": 0.7124,
+      "step": 8196
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.72388937598681e-06,
+      "loss": 0.4324,
+      "step": 8197
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.72171622495544e-06,
+      "loss": 0.7115,
+      "step": 8198
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.719543135293184e-06,
+      "loss": 0.7569,
+      "step": 8199
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.717370107104375e-06,
+      "loss": 0.4289,
+      "step": 8200
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.715197140493336e-06,
+      "loss": 0.7205,
+      "step": 8201
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.713024235564388e-06,
+      "loss": 0.735,
+      "step": 8202
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.710851392421846e-06,
+      "loss": 0.7771,
+      "step": 8203
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.70867861117003e-06,
+      "loss": 0.7975,
+      "step": 8204
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.706505891913253e-06,
+      "loss": 0.3868,
+      "step": 8205
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.704333234755826e-06,
+      "loss": 0.7953,
+      "step": 8206
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.702160639802052e-06,
+      "loss": 0.7072,
+      "step": 8207
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.699988107156243e-06,
+      "loss": 0.7286,
+      "step": 8208
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.697815636922694e-06,
+      "loss": 0.7735,
+      "step": 8209
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.6956432292057e-06,
+      "loss": 0.4078,
+      "step": 8210
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.693470884109563e-06,
+      "loss": 0.7599,
+      "step": 8211
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.69129860173857e-06,
+      "loss": 0.7715,
+      "step": 8212
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.689126382197018e-06,
+      "loss": 0.4182,
+      "step": 8213
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.686954225589186e-06,
+      "loss": 0.4117,
+      "step": 8214
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.684782132019356e-06,
+      "loss": 0.7805,
+      "step": 8215
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.682610101591813e-06,
+      "loss": 0.7348,
+      "step": 8216
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.680438134410833e-06,
+      "loss": 0.7831,
+      "step": 8217
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.678266230580688e-06,
+      "loss": 0.7808,
+      "step": 8218
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.676094390205644e-06,
+      "loss": 0.7794,
+      "step": 8219
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.67392261338998e-06,
+      "loss": 0.4241,
+      "step": 8220
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.671750900237953e-06,
+      "loss": 0.7374,
+      "step": 8221
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.669579250853829e-06,
+      "loss": 0.7652,
+      "step": 8222
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.667407665341861e-06,
+      "loss": 0.3813,
+      "step": 8223
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.665236143806311e-06,
+      "loss": 0.4684,
+      "step": 8224
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.663064686351429e-06,
+      "loss": 0.3953,
+      "step": 8225
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.660893293081463e-06,
+      "loss": 0.4078,
+      "step": 8226
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.65872196410066e-06,
+      "loss": 0.7446,
+      "step": 8227
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.656550699513267e-06,
+      "loss": 0.7823,
+      "step": 8228
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.65437949942352e-06,
+      "loss": 0.4195,
+      "step": 8229
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.652208363935658e-06,
+      "loss": 0.7594,
+      "step": 8230
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.650037293153917e-06,
+      "loss": 0.7406,
+      "step": 8231
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.64786628718252e-06,
+      "loss": 0.7371,
+      "step": 8232
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.645695346125705e-06,
+      "loss": 0.7945,
+      "step": 8233
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.643524470087689e-06,
+      "loss": 0.7607,
+      "step": 8234
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.641353659172696e-06,
+      "loss": 0.4264,
+      "step": 8235
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.639182913484945e-06,
+      "loss": 0.7293,
+      "step": 8236
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.637012233128651e-06,
+      "loss": 0.797,
+      "step": 8237
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.634841618208028e-06,
+      "loss": 0.4169,
+      "step": 8238
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.63267106882728e-06,
+      "loss": 0.7989,
+      "step": 8239
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.63050058509062e-06,
+      "loss": 0.7679,
+      "step": 8240
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.628330167102249e-06,
+      "loss": 0.7409,
+      "step": 8241
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.62615981496636e-06,
+      "loss": 0.781,
+      "step": 8242
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.623989528787155e-06,
+      "loss": 0.7696,
+      "step": 8243
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.621819308668826e-06,
+      "loss": 0.7759,
+      "step": 8244
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.619649154715566e-06,
+      "loss": 0.7795,
+      "step": 8245
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.61747906703156e-06,
+      "loss": 0.7409,
+      "step": 8246
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.61530904572099e-06,
+      "loss": 0.7175,
+      "step": 8247
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.613139090888033e-06,
+      "loss": 0.4531,
+      "step": 8248
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.610969202636876e-06,
+      "loss": 0.7213,
+      "step": 8249
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.608799381071687e-06,
+      "loss": 0.7636,
+      "step": 8250
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.606629626296636e-06,
+      "loss": 0.7912,
+      "step": 8251
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.604459938415895e-06,
+      "loss": 0.4379,
+      "step": 8252
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.602290317533624e-06,
+      "loss": 0.7788,
+      "step": 8253
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.600120763753988e-06,
+      "loss": 0.7502,
+      "step": 8254
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.597951277181143e-06,
+      "loss": 0.7959,
+      "step": 8255
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.595781857919245e-06,
+      "loss": 0.449,
+      "step": 8256
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.593612506072446e-06,
+      "loss": 0.7372,
+      "step": 8257
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.591443221744892e-06,
+      "loss": 0.7118,
+      "step": 8258
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.58927400504073e-06,
+      "loss": 0.7084,
+      "step": 8259
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.587104856064106e-06,
+      "loss": 0.3918,
+      "step": 8260
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.584935774919151e-06,
+      "loss": 0.3898,
+      "step": 8261
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.582766761710001e-06,
+      "loss": 0.4102,
+      "step": 8262
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.580597816540795e-06,
+      "loss": 0.7379,
+      "step": 8263
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.578428939515652e-06,
+      "loss": 0.7272,
+      "step": 8264
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.576260130738709e-06,
+      "loss": 0.4425,
+      "step": 8265
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.574091390314078e-06,
+      "loss": 0.7923,
+      "step": 8266
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.571922718345886e-06,
+      "loss": 0.4264,
+      "step": 8267
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.569754114938247e-06,
+      "loss": 0.6978,
+      "step": 8268
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.567585580195268e-06,
+      "loss": 0.733,
+      "step": 8269
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.565417114221064e-06,
+      "loss": 0.4346,
+      "step": 8270
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.563248717119736e-06,
+      "loss": 0.7673,
+      "step": 8271
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.561080388995394e-06,
+      "loss": 0.7671,
+      "step": 8272
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.55891212995213e-06,
+      "loss": 0.756,
+      "step": 8273
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.556743940094043e-06,
+      "loss": 0.7241,
+      "step": 8274
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.554575819525228e-06,
+      "loss": 0.4337,
+      "step": 8275
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.55240776834977e-06,
+      "loss": 0.7641,
+      "step": 8276
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.550239786671756e-06,
+      "loss": 0.7269,
+      "step": 8277
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.548071874595267e-06,
+      "loss": 0.7424,
+      "step": 8278
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.545904032224388e-06,
+      "loss": 0.7697,
+      "step": 8279
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.543736259663187e-06,
+      "loss": 0.4235,
+      "step": 8280
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.541568557015743e-06,
+      "loss": 0.7438,
+      "step": 8281
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.539400924386124e-06,
+      "loss": 0.7194,
+      "step": 8282
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.537233361878395e-06,
+      "loss": 0.7563,
+      "step": 8283
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.535065869596617e-06,
+      "loss": 0.7543,
+      "step": 8284
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.532898447644851e-06,
+      "loss": 0.7759,
+      "step": 8285
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.530731096127153e-06,
+      "loss": 0.6748,
+      "step": 8286
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.528563815147573e-06,
+      "loss": 0.7514,
+      "step": 8287
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.526396604810163e-06,
+      "loss": 0.7817,
+      "step": 8288
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.52422946521897e-06,
+      "loss": 0.7464,
+      "step": 8289
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.522062396478032e-06,
+      "loss": 0.7817,
+      "step": 8290
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.519895398691387e-06,
+      "loss": 0.4449,
+      "step": 8291
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.517728471963075e-06,
+      "loss": 0.7296,
+      "step": 8292
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.515561616397126e-06,
+      "loss": 0.6759,
+      "step": 8293
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.513394832097565e-06,
+      "loss": 0.7488,
+      "step": 8294
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.511228119168422e-06,
+      "loss": 0.7584,
+      "step": 8295
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.509061477713715e-06,
+      "loss": 0.7276,
+      "step": 8296
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.506894907837467e-06,
+      "loss": 0.7041,
+      "step": 8297
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.504728409643687e-06,
+      "loss": 0.7396,
+      "step": 8298
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.502561983236393e-06,
+      "loss": 0.7887,
+      "step": 8299
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.500395628719587e-06,
+      "loss": 0.7605,
+      "step": 8300
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.498229346197277e-06,
+      "loss": 0.44,
+      "step": 8301
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.496063135773463e-06,
+      "loss": 0.701,
+      "step": 8302
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.49389699755214e-06,
+      "loss": 0.77,
+      "step": 8303
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.491730931637308e-06,
+      "loss": 0.724,
+      "step": 8304
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.489564938132956e-06,
+      "loss": 0.8013,
+      "step": 8305
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.487399017143068e-06,
+      "loss": 0.4022,
+      "step": 8306
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.485233168771626e-06,
+      "loss": 0.4018,
+      "step": 8307
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.483067393122616e-06,
+      "loss": 0.7437,
+      "step": 8308
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.480901690300012e-06,
+      "loss": 0.4554,
+      "step": 8309
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.478736060407785e-06,
+      "loss": 0.7838,
+      "step": 8310
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.476570503549909e-06,
+      "loss": 0.695,
+      "step": 8311
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.474405019830346e-06,
+      "loss": 0.8183,
+      "step": 8312
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.472239609353063e-06,
+      "loss": 0.4333,
+      "step": 8313
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.470074272222016e-06,
+      "loss": 0.7607,
+      "step": 8314
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.467909008541164e-06,
+      "loss": 0.4963,
+      "step": 8315
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.465743818414455e-06,
+      "loss": 0.3988,
+      "step": 8316
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.463578701945839e-06,
+      "loss": 0.7557,
+      "step": 8317
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.461413659239262e-06,
+      "loss": 0.4225,
+      "step": 8318
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.459248690398662e-06,
+      "loss": 0.7689,
+      "step": 8319
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.45708379552799e-06,
+      "loss": 0.7519,
+      "step": 8320
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.45491897473116e-06,
+      "loss": 0.7377,
+      "step": 8321
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.45275422811212e-06,
+      "loss": 0.727,
+      "step": 8322
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.450589555774785e-06,
+      "loss": 0.4046,
+      "step": 8323
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.448424957823087e-06,
+      "loss": 0.7122,
+      "step": 8324
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.446260434360943e-06,
+      "loss": 0.4041,
+      "step": 8325
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.444095985492269e-06,
+      "loss": 0.7736,
+      "step": 8326
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.44193161132098e-06,
+      "loss": 0.4452,
+      "step": 8327
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.439767311950984e-06,
+      "loss": 0.774,
+      "step": 8328
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.437603087486188e-06,
+      "loss": 0.4238,
+      "step": 8329
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.435438938030493e-06,
+      "loss": 0.7678,
+      "step": 8330
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.433274863687799e-06,
+      "loss": 0.4088,
+      "step": 8331
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.431110864562e-06,
+      "loss": 0.7804,
+      "step": 8332
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.428946940756986e-06,
+      "loss": 0.7293,
+      "step": 8333
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.42678309237665e-06,
+      "loss": 0.7496,
+      "step": 8334
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.424619319524873e-06,
+      "loss": 0.8018,
+      "step": 8335
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.422455622305533e-06,
+      "loss": 0.7898,
+      "step": 8336
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.420292000822507e-06,
+      "loss": 0.7654,
+      "step": 8337
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.418128455179673e-06,
+      "loss": 0.7474,
+      "step": 8338
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.415964985480896e-06,
+      "loss": 0.7907,
+      "step": 8339
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.413801591830047e-06,
+      "loss": 0.451,
+      "step": 8340
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.411638274330981e-06,
+      "loss": 0.4334,
+      "step": 8341
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.409475033087565e-06,
+      "loss": 0.7795,
+      "step": 8342
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.40731186820365e-06,
+      "loss": 0.4162,
+      "step": 8343
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.405148779783085e-06,
+      "loss": 0.7472,
+      "step": 8344
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.40298576792972e-06,
+      "loss": 0.4418,
+      "step": 8345
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.4008228327474e-06,
+      "loss": 0.7707,
+      "step": 8346
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.398659974339967e-06,
+      "loss": 0.7968,
+      "step": 8347
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.396497192811253e-06,
+      "loss": 0.7915,
+      "step": 8348
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.394334488265092e-06,
+      "loss": 0.7333,
+      "step": 8349
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.39217186080532e-06,
+      "loss": 0.7146,
+      "step": 8350
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.390009310535754e-06,
+      "loss": 0.722,
+      "step": 8351
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.387846837560219e-06,
+      "loss": 0.7101,
+      "step": 8352
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.385684441982528e-06,
+      "loss": 0.7252,
+      "step": 8353
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.383522123906508e-06,
+      "loss": 0.7592,
+      "step": 8354
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.381359883435955e-06,
+      "loss": 0.4199,
+      "step": 8355
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.379197720674688e-06,
+      "loss": 0.7208,
+      "step": 8356
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.377035635726501e-06,
+      "loss": 0.7638,
+      "step": 8357
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.374873628695202e-06,
+      "loss": 0.7628,
+      "step": 8358
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.372711699684582e-06,
+      "loss": 0.393,
+      "step": 8359
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.370549848798431e-06,
+      "loss": 0.3927,
+      "step": 8360
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.368388076140544e-06,
+      "loss": 0.7476,
+      "step": 8361
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.366226381814698e-06,
+      "loss": 0.4062,
+      "step": 8362
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.364064765924679e-06,
+      "loss": 0.7782,
+      "step": 8363
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.361903228574264e-06,
+      "loss": 0.7472,
+      "step": 8364
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.359741769867225e-06,
+      "loss": 0.4228,
+      "step": 8365
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.357580389907327e-06,
+      "loss": 0.4165,
+      "step": 8366
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.355419088798342e-06,
+      "loss": 0.4563,
+      "step": 8367
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.35325786664403e-06,
+      "loss": 0.7269,
+      "step": 8368
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.351096723548145e-06,
+      "loss": 0.7779,
+      "step": 8369
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.348935659614448e-06,
+      "loss": 0.7593,
+      "step": 8370
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.346774674946684e-06,
+      "loss": 0.6835,
+      "step": 8371
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.344613769648605e-06,
+      "loss": 0.4326,
+      "step": 8372
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.342452943823948e-06,
+      "loss": 0.3991,
+      "step": 8373
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.340292197576458e-06,
+      "loss": 0.6965,
+      "step": 8374
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.338131531009868e-06,
+      "loss": 0.7371,
+      "step": 8375
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.335970944227905e-06,
+      "loss": 0.4408,
+      "step": 8376
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.333810437334304e-06,
+      "loss": 0.7549,
+      "step": 8377
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.331650010432781e-06,
+      "loss": 0.7695,
+      "step": 8378
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.329489663627069e-06,
+      "loss": 0.4211,
+      "step": 8379
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.327329397020866e-06,
+      "loss": 0.7991,
+      "step": 8380
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.325169210717897e-06,
+      "loss": 0.4364,
+      "step": 8381
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.323009104821865e-06,
+      "loss": 0.4535,
+      "step": 8382
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.320849079436477e-06,
+      "loss": 0.7224,
+      "step": 8383
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.318689134665434e-06,
+      "loss": 0.7776,
+      "step": 8384
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.316529270612427e-06,
+      "loss": 0.7302,
+      "step": 8385
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.314369487381158e-06,
+      "loss": 0.4148,
+      "step": 8386
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.312209785075309e-06,
+      "loss": 0.4169,
+      "step": 8387
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.31005016379857e-06,
+      "loss": 0.7672,
+      "step": 8388
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.307890623654618e-06,
+      "loss": 0.7759,
+      "step": 8389
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.305731164747135e-06,
+      "loss": 0.735,
+      "step": 8390
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.30357178717979e-06,
+      "loss": 0.7522,
+      "step": 8391
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.301412491056253e-06,
+      "loss": 0.774,
+      "step": 8392
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.299253276480192e-06,
+      "loss": 0.7188,
+      "step": 8393
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.297094143555272e-06,
+      "loss": 0.3816,
+      "step": 8394
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.294935092385141e-06,
+      "loss": 0.7962,
+      "step": 8395
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.292776123073458e-06,
+      "loss": 0.7563,
+      "step": 8396
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.290617235723875e-06,
+      "loss": 0.4606,
+      "step": 8397
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.288458430440035e-06,
+      "loss": 0.7571,
+      "step": 8398
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.286299707325582e-06,
+      "loss": 0.7624,
+      "step": 8399
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.284141066484154e-06,
+      "loss": 0.4364,
+      "step": 8400
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.281982508019384e-06,
+      "loss": 0.7259,
+      "step": 8401
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.279824032034904e-06,
+      "loss": 0.3997,
+      "step": 8402
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.277665638634336e-06,
+      "loss": 0.7742,
+      "step": 8403
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.275507327921307e-06,
+      "loss": 0.7309,
+      "step": 8404
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.273349099999432e-06,
+      "loss": 0.6834,
+      "step": 8405
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.27119095497233e-06,
+      "loss": 0.8253,
+      "step": 8406
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.269032892943608e-06,
+      "loss": 0.4054,
+      "step": 8407
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.266874914016871e-06,
+      "loss": 0.4042,
+      "step": 8408
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.264717018295726e-06,
+      "loss": 0.7185,
+      "step": 8409
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.262559205883768e-06,
+      "loss": 0.7424,
+      "step": 8410
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.260401476884592e-06,
+      "loss": 0.8352,
+      "step": 8411
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.258243831401784e-06,
+      "loss": 0.7642,
+      "step": 8412
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.256086269538939e-06,
+      "loss": 0.7224,
+      "step": 8413
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.25392879139963e-06,
+      "loss": 0.79,
+      "step": 8414
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.251771397087447e-06,
+      "loss": 0.7768,
+      "step": 8415
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.249614086705953e-06,
+      "loss": 0.7974,
+      "step": 8416
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.247456860358725e-06,
+      "loss": 0.768,
+      "step": 8417
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.245299718149328e-06,
+      "loss": 0.7073,
+      "step": 8418
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.243142660181322e-06,
+      "loss": 0.4062,
+      "step": 8419
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.240985686558268e-06,
+      "loss": 0.7946,
+      "step": 8420
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.238828797383718e-06,
+      "loss": 0.768,
+      "step": 8421
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.236671992761225e-06,
+      "loss": 0.3796,
+      "step": 8422
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.23451527279433e-06,
+      "loss": 0.7359,
+      "step": 8423
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.232358637586583e-06,
+      "loss": 0.7465,
+      "step": 8424
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.23020208724151e-06,
+      "loss": 0.3818,
+      "step": 8425
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.228045621862657e-06,
+      "loss": 0.7776,
+      "step": 8426
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.225889241553546e-06,
+      "loss": 0.7655,
+      "step": 8427
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.223732946417705e-06,
+      "loss": 0.3687,
+      "step": 8428
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.221576736558657e-06,
+      "loss": 0.4662,
+      "step": 8429
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.219420612079914e-06,
+      "loss": 0.7348,
+      "step": 8430
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.217264573084996e-06,
+      "loss": 0.7461,
+      "step": 8431
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.215108619677409e-06,
+      "loss": 0.7246,
+      "step": 8432
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.21295275196066e-06,
+      "loss": 0.8313,
+      "step": 8433
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.210796970038249e-06,
+      "loss": 0.7696,
+      "step": 8434
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.20864127401367e-06,
+      "loss": 0.7515,
+      "step": 8435
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.20648566399042e-06,
+      "loss": 0.79,
+      "step": 8436
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.204330140071986e-06,
+      "loss": 0.7748,
+      "step": 8437
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.202174702361854e-06,
+      "loss": 0.757,
+      "step": 8438
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.200019350963505e-06,
+      "loss": 0.7114,
+      "step": 8439
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.197864085980412e-06,
+      "loss": 0.6738,
+      "step": 8440
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.195708907516047e-06,
+      "loss": 0.4297,
+      "step": 8441
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.19355381567388e-06,
+      "loss": 0.7031,
+      "step": 8442
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.191398810557375e-06,
+      "loss": 0.7877,
+      "step": 8443
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.189243892269987e-06,
+      "loss": 0.4389,
+      "step": 8444
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.18708906091518e-06,
+      "loss": 0.7779,
+      "step": 8445
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.184934316596397e-06,
+      "loss": 0.7494,
+      "step": 8446
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.18277965941709e-06,
+      "loss": 0.7592,
+      "step": 8447
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.1806250894807e-06,
+      "loss": 0.7876,
+      "step": 8448
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.178470606890668e-06,
+      "loss": 0.7164,
+      "step": 8449
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.176316211750428e-06,
+      "loss": 0.7464,
+      "step": 8450
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.174161904163404e-06,
+      "loss": 0.756,
+      "step": 8451
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.172007684233032e-06,
+      "loss": 0.7839,
+      "step": 8452
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.169853552062728e-06,
+      "loss": 0.7206,
+      "step": 8453
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.167699507755915e-06,
+      "loss": 0.3816,
+      "step": 8454
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.165545551415997e-06,
+      "loss": 0.705,
+      "step": 8455
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.163391683146391e-06,
+      "loss": 0.7666,
+      "step": 8456
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.161237903050497e-06,
+      "loss": 0.7825,
+      "step": 8457
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.159084211231722e-06,
+      "loss": 0.7209,
+      "step": 8458
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.156930607793457e-06,
+      "loss": 0.6969,
+      "step": 8459
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.154777092839096e-06,
+      "loss": 0.4047,
+      "step": 8460
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.15262366647203e-06,
+      "loss": 0.7768,
+      "step": 8461
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.150470328795638e-06,
+      "loss": 0.7344,
+      "step": 8462
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.148317079913304e-06,
+      "loss": 0.7337,
+      "step": 8463
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.146163919928398e-06,
+      "loss": 0.74,
+      "step": 8464
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.144010848944298e-06,
+      "loss": 0.7721,
+      "step": 8465
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.141857867064367e-06,
+      "loss": 0.4023,
+      "step": 8466
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.139704974391966e-06,
+      "loss": 0.7596,
+      "step": 8467
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.137552171030455e-06,
+      "loss": 0.7474,
+      "step": 8468
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.13539945708319e-06,
+      "loss": 0.7306,
+      "step": 8469
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.133246832653519e-06,
+      "loss": 0.704,
+      "step": 8470
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.131094297844781e-06,
+      "loss": 0.7691,
+      "step": 8471
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.128941852760328e-06,
+      "loss": 0.7363,
+      "step": 8472
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.126789497503488e-06,
+      "loss": 0.7509,
+      "step": 8473
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.124637232177599e-06,
+      "loss": 0.4198,
+      "step": 8474
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.122485056885987e-06,
+      "loss": 0.3867,
+      "step": 8475
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.120332971731972e-06,
+      "loss": 0.7919,
+      "step": 8476
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.118180976818881e-06,
+      "loss": 0.7518,
+      "step": 8477
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.116029072250024e-06,
+      "loss": 0.7381,
+      "step": 8478
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.113877258128712e-06,
+      "loss": 0.3809,
+      "step": 8479
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.111725534558253e-06,
+      "loss": 0.7691,
+      "step": 8480
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.109573901641951e-06,
+      "loss": 0.7851,
+      "step": 8481
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.1074223594831e-06,
+      "loss": 0.4501,
+      "step": 8482
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.105270908184992e-06,
+      "loss": 0.4019,
+      "step": 8483
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.103119547850924e-06,
+      "loss": 0.7156,
+      "step": 8484
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.100968278584174e-06,
+      "loss": 0.7469,
+      "step": 8485
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.09881710048802e-06,
+      "loss": 0.4089,
+      "step": 8486
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.096666013665744e-06,
+      "loss": 0.7603,
+      "step": 8487
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.094515018220615e-06,
+      "loss": 0.7549,
+      "step": 8488
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.092364114255897e-06,
+      "loss": 0.7425,
+      "step": 8489
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.090213301874858e-06,
+      "loss": 0.743,
+      "step": 8490
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.088062581180753e-06,
+      "loss": 0.745,
+      "step": 8491
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.085911952276838e-06,
+      "loss": 0.8029,
+      "step": 8492
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.083761415266363e-06,
+      "loss": 0.7916,
+      "step": 8493
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.08161097025257e-06,
+      "loss": 0.7017,
+      "step": 8494
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.079460617338701e-06,
+      "loss": 0.7406,
+      "step": 8495
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.077310356627993e-06,
+      "loss": 0.7597,
+      "step": 8496
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.075160188223678e-06,
+      "loss": 0.7349,
+      "step": 8497
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.073010112228987e-06,
+      "loss": 0.7834,
+      "step": 8498
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.070860128747136e-06,
+      "loss": 0.4274,
+      "step": 8499
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.068710237881345e-06,
+      "loss": 0.7697,
+      "step": 8500
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.06656043973483e-06,
+      "loss": 0.7869,
+      "step": 8501
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.064410734410802e-06,
+      "loss": 0.7313,
+      "step": 8502
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.06226112201246e-06,
+      "loss": 0.7982,
+      "step": 8503
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.060111602643013e-06,
+      "loss": 0.768,
+      "step": 8504
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.057962176405652e-06,
+      "loss": 0.4075,
+      "step": 8505
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.05581284340357e-06,
+      "loss": 0.4126,
+      "step": 8506
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.053663603739954e-06,
+      "loss": 0.7614,
+      "step": 8507
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.051514457517988e-06,
+      "loss": 0.4174,
+      "step": 8508
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.049365404840853e-06,
+      "loss": 0.6929,
+      "step": 8509
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.047216445811714e-06,
+      "loss": 0.7466,
+      "step": 8510
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.04506758053375e-06,
+      "loss": 0.6965,
+      "step": 8511
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.042918809110118e-06,
+      "loss": 0.4489,
+      "step": 8512
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.04077013164399e-06,
+      "loss": 0.7604,
+      "step": 8513
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.038621548238506e-06,
+      "loss": 0.3904,
+      "step": 8514
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.036473058996827e-06,
+      "loss": 0.752,
+      "step": 8515
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.034324664022097e-06,
+      "loss": 0.7973,
+      "step": 8516
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.032176363417463e-06,
+      "loss": 0.77,
+      "step": 8517
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.030028157286057e-06,
+      "loss": 0.7839,
+      "step": 8518
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.027880045731014e-06,
+      "loss": 0.81,
+      "step": 8519
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.025732028855462e-06,
+      "loss": 0.416,
+      "step": 8520
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.023584106762528e-06,
+      "loss": 0.7606,
+      "step": 8521
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.021436279555327e-06,
+      "loss": 0.7522,
+      "step": 8522
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.019288547336978e-06,
+      "loss": 0.8336,
+      "step": 8523
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.017140910210593e-06,
+      "loss": 0.7503,
+      "step": 8524
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.014993368279274e-06,
+      "loss": 0.7735,
+      "step": 8525
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.01284592164612e-06,
+      "loss": 0.4128,
+      "step": 8526
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.010698570414234e-06,
+      "loss": 0.7057,
+      "step": 8527
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.008551314686708e-06,
+      "loss": 0.7368,
+      "step": 8528
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.006404154566624e-06,
+      "loss": 0.7086,
+      "step": 8529
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.004257090157067e-06,
+      "loss": 0.7667,
+      "step": 8530
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.002110121561117e-06,
+      "loss": 0.7284,
+      "step": 8531
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.999963248881844e-06,
+      "loss": 0.4397,
+      "step": 8532
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.997816472222324e-06,
+      "loss": 0.7364,
+      "step": 8533
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.995669791685618e-06,
+      "loss": 0.7592,
+      "step": 8534
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.993523207374782e-06,
+      "loss": 0.7445,
+      "step": 8535
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.991376719392879e-06,
+      "loss": 0.7221,
+      "step": 8536
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.989230327842954e-06,
+      "loss": 0.4496,
+      "step": 8537
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.987084032828056e-06,
+      "loss": 0.735,
+      "step": 8538
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.984937834451225e-06,
+      "loss": 0.7308,
+      "step": 8539
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.982791732815497e-06,
+      "loss": 0.7705,
+      "step": 8540
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.98064572802391e-06,
+      "loss": 0.7048,
+      "step": 8541
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.978499820179481e-06,
+      "loss": 0.7584,
+      "step": 8542
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.976354009385245e-06,
+      "loss": 0.7974,
+      "step": 8543
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.97420829574421e-06,
+      "loss": 0.7161,
+      "step": 8544
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.972062679359396e-06,
+      "loss": 0.754,
+      "step": 8545
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.969917160333807e-06,
+      "loss": 0.7597,
+      "step": 8546
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.967771738770452e-06,
+      "loss": 0.7373,
+      "step": 8547
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.965626414772325e-06,
+      "loss": 0.7093,
+      "step": 8548
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.963481188442426e-06,
+      "loss": 0.7764,
+      "step": 8549
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.961336059883743e-06,
+      "loss": 0.4177,
+      "step": 8550
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.95919102919926e-06,
+      "loss": 0.7926,
+      "step": 8551
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.957046096491963e-06,
+      "loss": 0.7549,
+      "step": 8552
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.954901261864822e-06,
+      "loss": 0.7226,
+      "step": 8553
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.952756525420811e-06,
+      "loss": 0.7726,
+      "step": 8554
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.950611887262895e-06,
+      "loss": 0.4087,
+      "step": 8555
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.948467347494042e-06,
+      "loss": 0.8222,
+      "step": 8556
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.946322906217202e-06,
+      "loss": 0.7068,
+      "step": 8557
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.944178563535334e-06,
+      "loss": 0.7894,
+      "step": 8558
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.942034319551374e-06,
+      "loss": 0.7827,
+      "step": 8559
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.939890174368279e-06,
+      "loss": 0.8243,
+      "step": 8560
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.937746128088979e-06,
+      "loss": 0.4295,
+      "step": 8561
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.935602180816409e-06,
+      "loss": 0.422,
+      "step": 8562
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.933458332653499e-06,
+      "loss": 0.7714,
+      "step": 8563
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.93131458370317e-06,
+      "loss": 0.7648,
+      "step": 8564
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.929170934068347e-06,
+      "loss": 0.7823,
+      "step": 8565
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.927027383851938e-06,
+      "loss": 0.7794,
+      "step": 8566
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.924883933156858e-06,
+      "loss": 0.7663,
+      "step": 8567
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.922740582086007e-06,
+      "loss": 0.6897,
+      "step": 8568
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.920597330742288e-06,
+      "loss": 0.7643,
+      "step": 8569
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.918454179228599e-06,
+      "loss": 0.7503,
+      "step": 8570
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.916311127647824e-06,
+      "loss": 0.7803,
+      "step": 8571
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.914168176102855e-06,
+      "loss": 0.701,
+      "step": 8572
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.912025324696573e-06,
+      "loss": 0.7164,
+      "step": 8573
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.909882573531847e-06,
+      "loss": 0.4238,
+      "step": 8574
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.907739922711552e-06,
+      "loss": 0.7386,
+      "step": 8575
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.905597372338558e-06,
+      "loss": 0.7638,
+      "step": 8576
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.903454922515723e-06,
+      "loss": 0.7443,
+      "step": 8577
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.901312573345904e-06,
+      "loss": 0.7036,
+      "step": 8578
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.899170324931956e-06,
+      "loss": 0.8026,
+      "step": 8579
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.89702817737672e-06,
+      "loss": 0.7507,
+      "step": 8580
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.894886130783045e-06,
+      "loss": 0.7035,
+      "step": 8581
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.892744185253763e-06,
+      "loss": 0.7261,
+      "step": 8582
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.890602340891714e-06,
+      "loss": 0.4325,
+      "step": 8583
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.88846059779972e-06,
+      "loss": 0.7143,
+      "step": 8584
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.886318956080603e-06,
+      "loss": 0.7811,
+      "step": 8585
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.884177415837183e-06,
+      "loss": 0.7766,
+      "step": 8586
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.882035977172274e-06,
+      "loss": 0.6951,
+      "step": 8587
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.87989464018869e-06,
+      "loss": 0.7044,
+      "step": 8588
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.877753404989223e-06,
+      "loss": 0.7695,
+      "step": 8589
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.875612271676676e-06,
+      "loss": 0.7491,
+      "step": 8590
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.873471240353843e-06,
+      "loss": 0.424,
+      "step": 8591
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.871330311123515e-06,
+      "loss": 0.3927,
+      "step": 8592
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.869189484088475e-06,
+      "loss": 0.7163,
+      "step": 8593
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.867048759351497e-06,
+      "loss": 0.7748,
+      "step": 8594
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.864908137015362e-06,
+      "loss": 0.4401,
+      "step": 8595
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.862767617182834e-06,
+      "loss": 0.763,
+      "step": 8596
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.86062719995668e-06,
+      "loss": 0.7625,
+      "step": 8597
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.858486885439658e-06,
+      "loss": 0.7884,
+      "step": 8598
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.856346673734525e-06,
+      "loss": 0.7667,
+      "step": 8599
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.854206564944028e-06,
+      "loss": 0.7813,
+      "step": 8600
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.85206655917091e-06,
+      "loss": 0.794,
+      "step": 8601
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.849926656517914e-06,
+      "loss": 0.6788,
+      "step": 8602
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.847786857087775e-06,
+      "loss": 0.3955,
+      "step": 8603
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.845647160983218e-06,
+      "loss": 0.7263,
+      "step": 8604
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.843507568306969e-06,
+      "loss": 0.7797,
+      "step": 8605
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.841368079161751e-06,
+      "loss": 0.4379,
+      "step": 8606
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.839228693650274e-06,
+      "loss": 0.7665,
+      "step": 8607
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.837089411875252e-06,
+      "loss": 0.7334,
+      "step": 8608
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.834950233939389e-06,
+      "loss": 0.8197,
+      "step": 8609
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.832811159945381e-06,
+      "loss": 0.7388,
+      "step": 8610
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.830672189995929e-06,
+      "loss": 0.7367,
+      "step": 8611
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.828533324193715e-06,
+      "loss": 0.4113,
+      "step": 8612
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.826394562641432e-06,
+      "loss": 0.7315,
+      "step": 8613
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.824255905441753e-06,
+      "loss": 0.7386,
+      "step": 8614
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.822117352697359e-06,
+      "loss": 0.4005,
+      "step": 8615
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.819978904510917e-06,
+      "loss": 0.7454,
+      "step": 8616
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.817840560985092e-06,
+      "loss": 0.7967,
+      "step": 8617
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.815702322222539e-06,
+      "loss": 0.7296,
+      "step": 8618
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.813564188325918e-06,
+      "loss": 0.742,
+      "step": 8619
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.81142615939788e-06,
+      "loss": 0.7423,
+      "step": 8620
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.809288235541062e-06,
+      "loss": 0.7205,
+      "step": 8621
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.80715041685811e-06,
+      "loss": 0.7763,
+      "step": 8622
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.805012703451659e-06,
+      "loss": 0.4491,
+      "step": 8623
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.802875095424336e-06,
+      "loss": 0.7287,
+      "step": 8624
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.800737592878766e-06,
+      "loss": 0.7289,
+      "step": 8625
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.798600195917567e-06,
+      "loss": 0.7692,
+      "step": 8626
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.796462904643356e-06,
+      "loss": 0.6636,
+      "step": 8627
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.794325719158739e-06,
+      "loss": 0.4124,
+      "step": 8628
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.792188639566323e-06,
+      "loss": 0.7313,
+      "step": 8629
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.790051665968705e-06,
+      "loss": 0.7505,
+      "step": 8630
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.787914798468481e-06,
+      "loss": 0.7364,
+      "step": 8631
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.785778037168242e-06,
+      "loss": 0.7343,
+      "step": 8632
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.783641382170566e-06,
+      "loss": 0.7477,
+      "step": 8633
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.781504833578031e-06,
+      "loss": 0.7495,
+      "step": 8634
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.779368391493217e-06,
+      "loss": 0.7315,
+      "step": 8635
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.777232056018688e-06,
+      "loss": 0.7465,
+      "step": 8636
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.775095827257007e-06,
+      "loss": 0.8102,
+      "step": 8637
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.772959705310736e-06,
+      "loss": 0.7657,
+      "step": 8638
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.770823690282422e-06,
+      "loss": 0.4242,
+      "step": 8639
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.76868778227462e-06,
+      "loss": 0.7688,
+      "step": 8640
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.766551981389868e-06,
+      "loss": 0.7423,
+      "step": 8641
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.764416287730706e-06,
+      "loss": 0.399,
+      "step": 8642
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.762280701399666e-06,
+      "loss": 0.7446,
+      "step": 8643
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.760145222499272e-06,
+      "loss": 0.7195,
+      "step": 8644
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.758009851132052e-06,
+      "loss": 0.7193,
+      "step": 8645
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.755874587400519e-06,
+      "loss": 0.3665,
+      "step": 8646
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.75373943140719e-06,
+      "loss": 0.738,
+      "step": 8647
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.751604383254563e-06,
+      "loss": 0.7363,
+      "step": 8648
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.749469443045146e-06,
+      "loss": 0.7481,
+      "step": 8649
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.747334610881431e-06,
+      "loss": 0.4131,
+      "step": 8650
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.745199886865912e-06,
+      "loss": 0.7162,
+      "step": 8651
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.743065271101079e-06,
+      "loss": 0.7646,
+      "step": 8652
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.740930763689402e-06,
+      "loss": 0.7715,
+      "step": 8653
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.738796364733368e-06,
+      "loss": 0.7725,
+      "step": 8654
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.736662074335436e-06,
+      "loss": 0.6786,
+      "step": 8655
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.734527892598084e-06,
+      "loss": 0.7383,
+      "step": 8656
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.73239381962376e-06,
+      "loss": 0.3959,
+      "step": 8657
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.730259855514928e-06,
+      "loss": 0.7816,
+      "step": 8658
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.728126000374034e-06,
+      "loss": 0.405,
+      "step": 8659
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.725992254303518e-06,
+      "loss": 0.7798,
+      "step": 8660
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.723858617405824e-06,
+      "loss": 0.7989,
+      "step": 8661
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.721725089783388e-06,
+      "loss": 0.7855,
+      "step": 8662
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.719591671538634e-06,
+      "loss": 0.4248,
+      "step": 8663
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.717458362773981e-06,
+      "loss": 0.7246,
+      "step": 8664
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.715325163591856e-06,
+      "loss": 0.7493,
+      "step": 8665
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.713192074094667e-06,
+      "loss": 0.7043,
+      "step": 8666
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.711059094384823e-06,
+      "loss": 0.7287,
+      "step": 8667
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.708926224564726e-06,
+      "loss": 0.7518,
+      "step": 8668
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.706793464736771e-06,
+      "loss": 0.7529,
+      "step": 8669
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.704660815003352e-06,
+      "loss": 0.4003,
+      "step": 8670
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.702528275466852e-06,
+      "loss": 0.7764,
+      "step": 8671
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.70039584622966e-06,
+      "loss": 0.7591,
+      "step": 8672
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.69826352739414e-06,
+      "loss": 0.7524,
+      "step": 8673
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.696131319062673e-06,
+      "loss": 0.7038,
+      "step": 8674
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.69399922133762e-06,
+      "loss": 0.7555,
+      "step": 8675
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.691867234321341e-06,
+      "loss": 0.7536,
+      "step": 8676
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.689735358116193e-06,
+      "loss": 0.4134,
+      "step": 8677
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.68760359282452e-06,
+      "loss": 0.7607,
+      "step": 8678
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.685471938548669e-06,
+      "loss": 0.7882,
+      "step": 8679
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.683340395390978e-06,
+      "loss": 0.7682,
+      "step": 8680
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.681208963453779e-06,
+      "loss": 0.7401,
+      "step": 8681
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.6790776428394e-06,
+      "loss": 0.4317,
+      "step": 8682
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.67694643365017e-06,
+      "loss": 0.341,
+      "step": 8683
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.674815335988398e-06,
+      "loss": 0.7557,
+      "step": 8684
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.672684349956399e-06,
+      "loss": 0.7405,
+      "step": 8685
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.67055347565648e-06,
+      "loss": 0.7267,
+      "step": 8686
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.66842271319094e-06,
+      "loss": 0.7469,
+      "step": 8687
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.666292062662079e-06,
+      "loss": 0.787,
+      "step": 8688
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.66416152417218e-06,
+      "loss": 0.7688,
+      "step": 8689
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.662031097823537e-06,
+      "loss": 0.7166,
+      "step": 8690
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.659900783718426e-06,
+      "loss": 0.7072,
+      "step": 8691
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.657770581959122e-06,
+      "loss": 0.7682,
+      "step": 8692
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.655640492647889e-06,
+      "loss": 0.7041,
+      "step": 8693
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.653510515886995e-06,
+      "loss": 0.4064,
+      "step": 8694
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.651380651778696e-06,
+      "loss": 0.7283,
+      "step": 8695
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.649250900425245e-06,
+      "loss": 0.7174,
+      "step": 8696
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.647121261928892e-06,
+      "loss": 0.733,
+      "step": 8697
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.644991736391873e-06,
+      "loss": 0.7362,
+      "step": 8698
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.64286232391643e-06,
+      "loss": 0.3991,
+      "step": 8699
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.640733024604792e-06,
+      "loss": 0.3858,
+      "step": 8700
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.638603838559184e-06,
+      "loss": 0.4027,
+      "step": 8701
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.636474765881827e-06,
+      "loss": 0.727,
+      "step": 8702
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.634345806674933e-06,
+      "loss": 0.7535,
+      "step": 8703
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.632216961040716e-06,
+      "loss": 0.8376,
+      "step": 8704
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.630088229081376e-06,
+      "loss": 0.8103,
+      "step": 8705
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.627959610899111e-06,
+      "loss": 0.745,
+      "step": 8706
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.62583110659612e-06,
+      "loss": 0.7361,
+      "step": 8707
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.623702716274582e-06,
+      "loss": 0.7247,
+      "step": 8708
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.621574440036681e-06,
+      "loss": 0.4011,
+      "step": 8709
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.6194462779845965e-06,
+      "loss": 0.7581,
+      "step": 8710
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.617318230220498e-06,
+      "loss": 0.7351,
+      "step": 8711
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.615190296846548e-06,
+      "loss": 0.7697,
+      "step": 8712
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.613062477964912e-06,
+      "loss": 0.7887,
+      "step": 8713
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.610934773677738e-06,
+      "loss": 0.7687,
+      "step": 8714
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.608807184087181e-06,
+      "loss": 0.7404,
+      "step": 8715
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.60667970929538e-06,
+      "loss": 0.7361,
+      "step": 8716
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.604552349404477e-06,
+      "loss": 0.4051,
+      "step": 8717
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.602425104516602e-06,
+      "loss": 0.7309,
+      "step": 8718
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.60029797473388e-06,
+      "loss": 0.7325,
+      "step": 8719
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.598170960158437e-06,
+      "loss": 0.7302,
+      "step": 8720
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.596044060892382e-06,
+      "loss": 0.7668,
+      "step": 8721
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.593917277037838e-06,
+      "loss": 0.784,
+      "step": 8722
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.591790608696894e-06,
+      "loss": 0.7309,
+      "step": 8723
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5896640559716595e-06,
+      "loss": 0.7699,
+      "step": 8724
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.587537618964221e-06,
+      "loss": 0.7478,
+      "step": 8725
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.585411297776675e-06,
+      "loss": 0.7166,
+      "step": 8726
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.583285092511099e-06,
+      "loss": 0.775,
+      "step": 8727
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.581159003269568e-06,
+      "loss": 0.7953,
+      "step": 8728
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.579033030154158e-06,
+      "loss": 0.8117,
+      "step": 8729
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.576907173266932e-06,
+      "loss": 0.759,
+      "step": 8730
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.574781432709952e-06,
+      "loss": 0.6995,
+      "step": 8731
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.572655808585271e-06,
+      "loss": 0.7653,
+      "step": 8732
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.57053030099494e-06,
+      "loss": 0.416,
+      "step": 8733
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.568404910041003e-06,
+      "loss": 0.7529,
+      "step": 8734
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.566279635825493e-06,
+      "loss": 0.7191,
+      "step": 8735
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.564154478450449e-06,
+      "loss": 0.7635,
+      "step": 8736
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.562029438017894e-06,
+      "loss": 0.8096,
+      "step": 8737
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.559904514629848e-06,
+      "loss": 0.7947,
+      "step": 8738
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.557779708388326e-06,
+      "loss": 0.7026,
+      "step": 8739
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5556550193953415e-06,
+      "loss": 0.7152,
+      "step": 8740
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.553530447752895e-06,
+      "loss": 0.7815,
+      "step": 8741
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.551405993562988e-06,
+      "loss": 0.4435,
+      "step": 8742
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.549281656927613e-06,
+      "loss": 0.7343,
+      "step": 8743
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.547157437948753e-06,
+      "loss": 0.7048,
+      "step": 8744
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.545033336728397e-06,
+      "loss": 0.7463,
+      "step": 8745
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5429093533685124e-06,
+      "loss": 0.7245,
+      "step": 8746
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.540785487971079e-06,
+      "loss": 0.7547,
+      "step": 8747
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.538661740638052e-06,
+      "loss": 0.7468,
+      "step": 8748
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.536538111471399e-06,
+      "loss": 0.7465,
+      "step": 8749
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.534414600573068e-06,
+      "loss": 0.7452,
+      "step": 8750
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.532291208045011e-06,
+      "loss": 0.738,
+      "step": 8751
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.530167933989161e-06,
+      "loss": 0.4092,
+      "step": 8752
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.528044778507462e-06,
+      "loss": 0.7223,
+      "step": 8753
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.525921741701844e-06,
+      "loss": 0.7464,
+      "step": 8754
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.523798823674227e-06,
+      "loss": 0.7909,
+      "step": 8755
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.521676024526537e-06,
+      "loss": 0.7332,
+      "step": 8756
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.51955334436068e-06,
+      "loss": 0.7677,
+      "step": 8757
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.517430783278571e-06,
+      "loss": 0.7827,
+      "step": 8758
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.51530834138211e-06,
+      "loss": 0.7354,
+      "step": 8759
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.513186018773187e-06,
+      "loss": 0.7524,
+      "step": 8760
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.511063815553702e-06,
+      "loss": 0.7337,
+      "step": 8761
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.508941731825532e-06,
+      "loss": 0.7566,
+      "step": 8762
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.506819767690563e-06,
+      "loss": 0.3788,
+      "step": 8763
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.504697923250662e-06,
+      "loss": 0.7266,
+      "step": 8764
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5025761986077015e-06,
+      "loss": 0.7611,
+      "step": 8765
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5004545938635445e-06,
+      "loss": 0.752,
+      "step": 8766
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.498333109120042e-06,
+      "loss": 0.7828,
+      "step": 8767
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.496211744479043e-06,
+      "loss": 0.4045,
+      "step": 8768
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.494090500042397e-06,
+      "loss": 0.7351,
+      "step": 8769
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.491969375911943e-06,
+      "loss": 0.4316,
+      "step": 8770
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.489848372189509e-06,
+      "loss": 0.7615,
+      "step": 8771
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.487727488976928e-06,
+      "loss": 0.7695,
+      "step": 8772
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.4856067263760166e-06,
+      "loss": 0.7292,
+      "step": 8773
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.483486084488596e-06,
+      "loss": 0.7694,
+      "step": 8774
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.4813655634164715e-06,
+      "loss": 0.7483,
+      "step": 8775
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.4792451632614485e-06,
+      "loss": 0.7342,
+      "step": 8776
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.477124884125325e-06,
+      "loss": 0.7532,
+      "step": 8777
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.475004726109893e-06,
+      "loss": 0.4336,
+      "step": 8778
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.472884689316942e-06,
+      "loss": 0.4042,
+      "step": 8779
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.470764773848249e-06,
+      "loss": 0.3916,
+      "step": 8780
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.4686449798055975e-06,
+      "loss": 0.4297,
+      "step": 8781
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.466525307290743e-06,
+      "loss": 0.7422,
+      "step": 8782
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.464405756405457e-06,
+      "loss": 0.4175,
+      "step": 8783
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.462286327251496e-06,
+      "loss": 0.762,
+      "step": 8784
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.460167019930612e-06,
+      "loss": 0.7162,
+      "step": 8785
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.458047834544551e-06,
+      "loss": 0.4367,
+      "step": 8786
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.45592877119505e-06,
+      "loss": 0.7519,
+      "step": 8787
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.453809829983847e-06,
+      "loss": 0.7431,
+      "step": 8788
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.451691011012667e-06,
+      "loss": 0.739,
+      "step": 8789
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.449572314383237e-06,
+      "loss": 0.4646,
+      "step": 8790
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.447453740197268e-06,
+      "loss": 0.7367,
+      "step": 8791
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.445335288556474e-06,
+      "loss": 0.4337,
+      "step": 8792
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.443216959562561e-06,
+      "loss": 0.7374,
+      "step": 8793
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.4410987533172204e-06,
+      "loss": 0.7977,
+      "step": 8794
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.438980669922154e-06,
+      "loss": 0.8036,
+      "step": 8795
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.436862709479048e-06,
+      "loss": 0.686,
+      "step": 8796
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.434744872089579e-06,
+      "loss": 0.3957,
+      "step": 8797
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.432627157855421e-06,
+      "loss": 0.7145,
+      "step": 8798
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.430509566878248e-06,
+      "loss": 0.7633,
+      "step": 8799
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.42839209925972e-06,
+      "loss": 0.4216,
+      "step": 8800
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.4262747551014995e-06,
+      "loss": 0.6902,
+      "step": 8801
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.424157534505234e-06,
+      "loss": 0.3759,
+      "step": 8802
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.422040437572567e-06,
+      "loss": 0.7589,
+      "step": 8803
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.419923464405145e-06,
+      "loss": 0.7756,
+      "step": 8804
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.4178066151045945e-06,
+      "loss": 0.7795,
+      "step": 8805
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.415689889772548e-06,
+      "loss": 0.7186,
+      "step": 8806
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.413573288510625e-06,
+      "loss": 0.732,
+      "step": 8807
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.411456811420445e-06,
+      "loss": 0.7614,
+      "step": 8808
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.4093404586036155e-06,
+      "loss": 0.4322,
+      "step": 8809
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.407224230161738e-06,
+      "loss": 0.7654,
+      "step": 8810
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.4051081261964164e-06,
+      "loss": 0.7104,
+      "step": 8811
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.402992146809237e-06,
+      "loss": 0.7356,
+      "step": 8812
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.4008762921017885e-06,
+      "loss": 0.7572,
+      "step": 8813
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3987605621756485e-06,
+      "loss": 0.7029,
+      "step": 8814
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3966449571323955e-06,
+      "loss": 0.7681,
+      "step": 8815
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.394529477073591e-06,
+      "loss": 0.432,
+      "step": 8816
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.392414122100805e-06,
+      "loss": 0.699,
+      "step": 8817
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.390298892315588e-06,
+      "loss": 0.7402,
+      "step": 8818
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.388183787819492e-06,
+      "loss": 0.7689,
+      "step": 8819
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.386068808714062e-06,
+      "loss": 0.3835,
+      "step": 8820
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.383953955100832e-06,
+      "loss": 0.8168,
+      "step": 8821
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.38183922708134e-06,
+      "loss": 0.7419,
+      "step": 8822
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.379724624757105e-06,
+      "loss": 0.7308,
+      "step": 8823
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.377610148229654e-06,
+      "loss": 0.7949,
+      "step": 8824
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.375495797600497e-06,
+      "loss": 0.7612,
+      "step": 8825
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.373381572971145e-06,
+      "loss": 0.7318,
+      "step": 8826
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3712674744430935e-06,
+      "loss": 0.7506,
+      "step": 8827
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.369153502117843e-06,
+      "loss": 0.7219,
+      "step": 8828
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.367039656096882e-06,
+      "loss": 0.7542,
+      "step": 8829
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.364925936481693e-06,
+      "loss": 0.7751,
+      "step": 8830
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3628123433737554e-06,
+      "loss": 0.7513,
+      "step": 8831
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.360698876874539e-06,
+      "loss": 0.4203,
+      "step": 8832
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.358585537085513e-06,
+      "loss": 0.4237,
+      "step": 8833
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.356472324108134e-06,
+      "loss": 0.7237,
+      "step": 8834
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.354359238043853e-06,
+      "loss": 0.6901,
+      "step": 8835
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.352246278994119e-06,
+      "loss": 0.7057,
+      "step": 8836
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.350133447060375e-06,
+      "loss": 0.7542,
+      "step": 8837
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.348020742344053e-06,
+      "loss": 0.7194,
+      "step": 8838
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.345908164946584e-06,
+      "loss": 0.4287,
+      "step": 8839
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3437957149693905e-06,
+      "loss": 0.4084,
+      "step": 8840
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.34168339251389e-06,
+      "loss": 0.7055,
+      "step": 8841
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.33957119768149e-06,
+      "loss": 0.7189,
+      "step": 8842
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.337459130573594e-06,
+      "loss": 0.3701,
+      "step": 8843
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.335347191291604e-06,
+      "loss": 0.6957,
+      "step": 8844
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.333235379936912e-06,
+      "loss": 0.4169,
+      "step": 8845
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.331123696610899e-06,
+      "loss": 0.7318,
+      "step": 8846
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.329012141414951e-06,
+      "loss": 0.4238,
+      "step": 8847
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.326900714450437e-06,
+      "loss": 0.7477,
+      "step": 8848
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3247894158187285e-06,
+      "loss": 0.6688,
+      "step": 8849
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.322678245621184e-06,
+      "loss": 0.4178,
+      "step": 8850
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.320567203959157e-06,
+      "loss": 0.7896,
+      "step": 8851
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.318456290934002e-06,
+      "loss": 0.721,
+      "step": 8852
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3163455066470555e-06,
+      "loss": 0.7897,
+      "step": 8853
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.314234851199661e-06,
+      "loss": 0.754,
+      "step": 8854
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.312124324693146e-06,
+      "loss": 0.7213,
+      "step": 8855
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3100139272288316e-06,
+      "loss": 0.7049,
+      "step": 8856
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.307903658908037e-06,
+      "loss": 0.7797,
+      "step": 8857
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.305793519832078e-06,
+      "loss": 0.7315,
+      "step": 8858
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.303683510102254e-06,
+      "loss": 0.7287,
+      "step": 8859
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.301573629819872e-06,
+      "loss": 0.7422,
+      "step": 8860
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.29946387908622e-06,
+      "loss": 0.4127,
+      "step": 8861
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.297354258002582e-06,
+      "loss": 0.7099,
+      "step": 8862
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.295244766670248e-06,
+      "loss": 0.7728,
+      "step": 8863
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.293135405190483e-06,
+      "loss": 0.7519,
+      "step": 8864
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2910261736645635e-06,
+      "loss": 0.7377,
+      "step": 8865
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.288917072193746e-06,
+      "loss": 0.7527,
+      "step": 8866
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.286808100879291e-06,
+      "loss": 0.7996,
+      "step": 8867
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.284699259822445e-06,
+      "loss": 0.7703,
+      "step": 8868
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2825905491244485e-06,
+      "loss": 0.7485,
+      "step": 8869
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.280481968886547e-06,
+      "loss": 0.72,
+      "step": 8870
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.278373519209963e-06,
+      "loss": 0.4348,
+      "step": 8871
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.276265200195925e-06,
+      "loss": 0.7626,
+      "step": 8872
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2741570119456485e-06,
+      "loss": 0.7779,
+      "step": 8873
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.272048954560349e-06,
+      "loss": 0.423,
+      "step": 8874
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.269941028141227e-06,
+      "loss": 0.7434,
+      "step": 8875
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.267833232789491e-06,
+      "loss": 0.4239,
+      "step": 8876
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.265725568606326e-06,
+      "loss": 0.7971,
+      "step": 8877
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.263618035692921e-06,
+      "loss": 0.4174,
+      "step": 8878
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2615106341504586e-06,
+      "loss": 0.7822,
+      "step": 8879
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.259403364080108e-06,
+      "loss": 0.7227,
+      "step": 8880
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.257296225583043e-06,
+      "loss": 0.7483,
+      "step": 8881
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.255189218760423e-06,
+      "loss": 0.721,
+      "step": 8882
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.253082343713403e-06,
+      "loss": 0.8162,
+      "step": 8883
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.250975600543133e-06,
+      "loss": 0.4308,
+      "step": 8884
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2488689893507566e-06,
+      "loss": 0.7425,
+      "step": 8885
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.246762510237404e-06,
+      "loss": 0.7128,
+      "step": 8886
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.24465616330421e-06,
+      "loss": 0.763,
+      "step": 8887
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.242549948652298e-06,
+      "loss": 0.4239,
+      "step": 8888
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.240443866382783e-06,
+      "loss": 0.7337,
+      "step": 8889
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.23833791659678e-06,
+      "loss": 0.7285,
+      "step": 8890
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.236232099395388e-06,
+      "loss": 0.76,
+      "step": 8891
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.234126414879712e-06,
+      "loss": 0.7415,
+      "step": 8892
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.232020863150838e-06,
+      "loss": 0.7152,
+      "step": 8893
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.229915444309852e-06,
+      "loss": 0.7473,
+      "step": 8894
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.227810158457838e-06,
+      "loss": 0.7853,
+      "step": 8895
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.225705005695861e-06,
+      "loss": 0.7725,
+      "step": 8896
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.223599986124994e-06,
+      "loss": 0.7598,
+      "step": 8897
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.221495099846293e-06,
+      "loss": 0.6899,
+      "step": 8898
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.219390346960815e-06,
+      "loss": 0.7515,
+      "step": 8899
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.217285727569606e-06,
+      "loss": 0.7974,
+      "step": 8900
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.215181241773704e-06,
+      "loss": 0.4339,
+      "step": 8901
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2130768896741424e-06,
+      "loss": 0.7166,
+      "step": 8902
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.210972671371953e-06,
+      "loss": 0.7256,
+      "step": 8903
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.208868586968156e-06,
+      "loss": 0.4414,
+      "step": 8904
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.206764636563763e-06,
+      "loss": 0.7607,
+      "step": 8905
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.204660820259788e-06,
+      "loss": 0.4017,
+      "step": 8906
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.202557138157229e-06,
+      "loss": 0.7891,
+      "step": 8907
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.200453590357085e-06,
+      "loss": 0.7464,
+      "step": 8908
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.198350176960342e-06,
+      "loss": 0.6879,
+      "step": 8909
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.196246898067985e-06,
+      "loss": 0.74,
+      "step": 8910
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.194143753780989e-06,
+      "loss": 0.8236,
+      "step": 8911
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.192040744200324e-06,
+      "loss": 0.4117,
+      "step": 8912
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.189937869426955e-06,
+      "loss": 0.8118,
+      "step": 8913
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.187835129561835e-06,
+      "loss": 0.7303,
+      "step": 8914
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.185732524705924e-06,
+      "loss": 0.763,
+      "step": 8915
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.183630054960153e-06,
+      "loss": 0.4356,
+      "step": 8916
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.181527720425465e-06,
+      "loss": 0.7783,
+      "step": 8917
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1794255212027915e-06,
+      "loss": 0.7647,
+      "step": 8918
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1773234573930594e-06,
+      "loss": 0.7622,
+      "step": 8919
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.175221529097183e-06,
+      "loss": 0.7719,
+      "step": 8920
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.173119736416073e-06,
+      "loss": 0.7354,
+      "step": 8921
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.171018079450638e-06,
+      "loss": 0.7579,
+      "step": 8922
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.168916558301773e-06,
+      "loss": 0.6889,
+      "step": 8923
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.166815173070374e-06,
+      "loss": 0.6547,
+      "step": 8924
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.164713923857323e-06,
+      "loss": 0.7957,
+      "step": 8925
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.162612810763498e-06,
+      "loss": 0.7864,
+      "step": 8926
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.160511833889775e-06,
+      "loss": 0.7588,
+      "step": 8927
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.158410993337016e-06,
+      "loss": 0.7182,
+      "step": 8928
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.156310289206084e-06,
+      "loss": 0.7357,
+      "step": 8929
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.154209721597833e-06,
+      "loss": 0.7424,
+      "step": 8930
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.152109290613103e-06,
+      "loss": 0.7314,
+      "step": 8931
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.150008996352734e-06,
+      "loss": 0.7818,
+      "step": 8932
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.147908838917564e-06,
+      "loss": 0.745,
+      "step": 8933
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.145808818408415e-06,
+      "loss": 0.7741,
+      "step": 8934
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.143708934926111e-06,
+      "loss": 0.707,
+      "step": 8935
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.141609188571464e-06,
+      "loss": 0.7391,
+      "step": 8936
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.139509579445276e-06,
+      "loss": 0.815,
+      "step": 8937
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1374101076483546e-06,
+      "loss": 0.7686,
+      "step": 8938
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.135310773281488e-06,
+      "loss": 0.4568,
+      "step": 8939
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.133211576445464e-06,
+      "loss": 0.7476,
+      "step": 8940
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.131112517241064e-06,
+      "loss": 0.6656,
+      "step": 8941
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.129013595769063e-06,
+      "loss": 0.7362,
+      "step": 8942
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1269148121302265e-06,
+      "loss": 0.7273,
+      "step": 8943
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.124816166425313e-06,
+      "loss": 0.7557,
+      "step": 8944
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.122717658755083e-06,
+      "loss": 0.7952,
+      "step": 8945
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.120619289220276e-06,
+      "loss": 0.4015,
+      "step": 8946
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.118521057921637e-06,
+      "loss": 0.7603,
+      "step": 8947
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.116422964959897e-06,
+      "loss": 0.7586,
+      "step": 8948
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.114325010435786e-06,
+      "loss": 0.7791,
+      "step": 8949
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.112227194450022e-06,
+      "loss": 0.7952,
+      "step": 8950
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.110129517103324e-06,
+      "loss": 0.7829,
+      "step": 8951
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.1080319784963945e-06,
+      "loss": 0.4481,
+      "step": 8952
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.105934578729934e-06,
+      "loss": 0.7617,
+      "step": 8953
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.103837317904641e-06,
+      "loss": 0.7661,
+      "step": 8954
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.101740196121198e-06,
+      "loss": 0.7039,
+      "step": 8955
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.099643213480289e-06,
+      "loss": 0.7367,
+      "step": 8956
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.097546370082588e-06,
+      "loss": 0.7637,
+      "step": 8957
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.095449666028761e-06,
+      "loss": 0.7664,
+      "step": 8958
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.093353101419469e-06,
+      "loss": 0.7451,
+      "step": 8959
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0912566763553694e-06,
+      "loss": 0.7671,
+      "step": 8960
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.089160390937101e-06,
+      "loss": 0.7527,
+      "step": 8961
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.087064245265312e-06,
+      "loss": 0.4265,
+      "step": 8962
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.084968239440633e-06,
+      "loss": 0.7408,
+      "step": 8963
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0828723735636894e-06,
+      "loss": 0.73,
+      "step": 8964
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.080776647735106e-06,
+      "loss": 0.399,
+      "step": 8965
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.078681062055493e-06,
+      "loss": 0.4386,
+      "step": 8966
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.076585616625462e-06,
+      "loss": 0.7823,
+      "step": 8967
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.074490311545608e-06,
+      "loss": 0.7553,
+      "step": 8968
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0723951469165254e-06,
+      "loss": 0.7023,
+      "step": 8969
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.070300122838803e-06,
+      "loss": 0.8049,
+      "step": 8970
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.06820523941302e-06,
+      "loss": 0.7567,
+      "step": 8971
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.066110496739749e-06,
+      "loss": 0.7623,
+      "step": 8972
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.064015894919557e-06,
+      "loss": 0.7466,
+      "step": 8973
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0619214340530096e-06,
+      "loss": 0.7543,
+      "step": 8974
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.059827114240647e-06,
+      "loss": 0.7793,
+      "step": 8975
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.057732935583024e-06,
+      "loss": 0.6941,
+      "step": 8976
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0556388981806765e-06,
+      "loss": 0.7176,
+      "step": 8977
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.053545002134141e-06,
+      "loss": 0.4304,
+      "step": 8978
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.051451247543941e-06,
+      "loss": 0.4041,
+      "step": 8979
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.049357634510594e-06,
+      "loss": 0.7817,
+      "step": 8980
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.047264163134617e-06,
+      "loss": 0.7118,
+      "step": 8981
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.045170833516508e-06,
+      "loss": 0.4036,
+      "step": 8982
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.043077645756775e-06,
+      "loss": 0.7734,
+      "step": 8983
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.040984599955903e-06,
+      "loss": 0.6904,
+      "step": 8984
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.038891696214378e-06,
+      "loss": 0.7767,
+      "step": 8985
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.03679893463268e-06,
+      "loss": 0.7284,
+      "step": 8986
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0347063153112786e-06,
+      "loss": 0.7412,
+      "step": 8987
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.032613838350642e-06,
+      "loss": 0.7555,
+      "step": 8988
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0305215038512245e-06,
+      "loss": 0.7628,
+      "step": 8989
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.028429311913478e-06,
+      "loss": 0.7692,
+      "step": 8990
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.026337262637844e-06,
+      "loss": 0.4238,
+      "step": 8991
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.024245356124762e-06,
+      "loss": 0.4161,
+      "step": 8992
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.022153592474662e-06,
+      "loss": 0.7591,
+      "step": 8993
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0200619717879695e-06,
+      "loss": 0.7521,
+      "step": 8994
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.017970494165098e-06,
+      "loss": 0.75,
+      "step": 8995
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.015879159706458e-06,
+      "loss": 0.7243,
+      "step": 8996
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.013787968512454e-06,
+      "loss": 0.6663,
+      "step": 8997
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.011696920683479e-06,
+      "loss": 0.3624,
+      "step": 8998
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0096060163199265e-06,
+      "loss": 0.731,
+      "step": 8999
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.007515255522175e-06,
+      "loss": 0.7983,
+      "step": 9000
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0054246383906e-06,
+      "loss": 0.7499,
+      "step": 9001
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0033341650255724e-06,
+      "loss": 0.7318,
+      "step": 9002
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0012438355274515e-06,
+      "loss": 0.7523,
+      "step": 9003
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.999153649996595e-06,
+      "loss": 0.434,
+      "step": 9004
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.997063608533347e-06,
+      "loss": 0.7404,
+      "step": 9005
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.994973711238049e-06,
+      "loss": 0.7122,
+      "step": 9006
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.992883958211034e-06,
+      "loss": 0.7817,
+      "step": 9007
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.990794349552632e-06,
+      "loss": 0.7625,
+      "step": 9008
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.988704885363158e-06,
+      "loss": 0.8135,
+      "step": 9009
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.986615565742932e-06,
+      "loss": 0.7273,
+      "step": 9010
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.984526390792257e-06,
+      "loss": 0.4575,
+      "step": 9011
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9824373606114285e-06,
+      "loss": 0.731,
+      "step": 9012
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.980348475300745e-06,
+      "loss": 0.7259,
+      "step": 9013
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.978259734960484e-06,
+      "loss": 0.7523,
+      "step": 9014
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.976171139690932e-06,
+      "loss": 0.7282,
+      "step": 9015
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.974082689592358e-06,
+      "loss": 0.4503,
+      "step": 9016
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.971994384765024e-06,
+      "loss": 0.7428,
+      "step": 9017
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.969906225309189e-06,
+      "loss": 0.7256,
+      "step": 9018
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.967818211325106e-06,
+      "loss": 0.7543,
+      "step": 9019
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.965730342913011e-06,
+      "loss": 0.788,
+      "step": 9020
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.963642620173147e-06,
+      "loss": 0.7411,
+      "step": 9021
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.961555043205741e-06,
+      "loss": 0.388,
+      "step": 9022
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.959467612111015e-06,
+      "loss": 0.7181,
+      "step": 9023
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.957380326989187e-06,
+      "loss": 0.7695,
+      "step": 9024
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.955293187940462e-06,
+      "loss": 0.7266,
+      "step": 9025
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.953206195065044e-06,
+      "loss": 0.763,
+      "step": 9026
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.951119348463129e-06,
+      "loss": 0.3914,
+      "step": 9027
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.949032648234899e-06,
+      "loss": 0.399,
+      "step": 9028
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.946946094480539e-06,
+      "loss": 0.3999,
+      "step": 9029
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.944859687300219e-06,
+      "loss": 0.7505,
+      "step": 9030
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.942773426794109e-06,
+      "loss": 0.4364,
+      "step": 9031
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9406873130623665e-06,
+      "loss": 0.421,
+      "step": 9032
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9386013462051424e-06,
+      "loss": 0.6792,
+      "step": 9033
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9365155263225885e-06,
+      "loss": 0.7424,
+      "step": 9034
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.934429853514833e-06,
+      "loss": 0.7488,
+      "step": 9035
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.93234432788201e-06,
+      "loss": 0.4083,
+      "step": 9036
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.930258949524248e-06,
+      "loss": 0.3989,
+      "step": 9037
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9281737185416595e-06,
+      "loss": 0.4664,
+      "step": 9038
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.926088635034356e-06,
+      "loss": 0.7471,
+      "step": 9039
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.924003699102439e-06,
+      "loss": 0.7036,
+      "step": 9040
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.921918910846005e-06,
+      "loss": 0.3792,
+      "step": 9041
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.919834270365144e-06,
+      "loss": 0.42,
+      "step": 9042
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.917749777759937e-06,
+      "loss": 0.7499,
+      "step": 9043
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.915665433130455e-06,
+      "loss": 0.7439,
+      "step": 9044
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.91358123657677e-06,
+      "loss": 0.7165,
+      "step": 9045
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.911497188198938e-06,
+      "loss": 0.7628,
+      "step": 9046
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.909413288097017e-06,
+      "loss": 0.7558,
+      "step": 9047
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9073295363710465e-06,
+      "loss": 0.7548,
+      "step": 9048
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.905245933121077e-06,
+      "loss": 0.8366,
+      "step": 9049
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.903162478447124e-06,
+      "loss": 0.7354,
+      "step": 9050
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.901079172449223e-06,
+      "loss": 0.7523,
+      "step": 9051
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.898996015227386e-06,
+      "loss": 0.7494,
+      "step": 9052
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.896913006881629e-06,
+      "loss": 0.7353,
+      "step": 9053
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.894830147511951e-06,
+      "loss": 0.6785,
+      "step": 9054
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.892747437218347e-06,
+      "loss": 0.4258,
+      "step": 9055
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.89066487610081e-06,
+      "loss": 0.75,
+      "step": 9056
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.888582464259318e-06,
+      "loss": 0.7204,
+      "step": 9057
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.886500201793849e-06,
+      "loss": 0.3967,
+      "step": 9058
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.884418088804368e-06,
+      "loss": 0.3975,
+      "step": 9059
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.882336125390833e-06,
+      "loss": 0.7583,
+      "step": 9060
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8802543116532004e-06,
+      "loss": 0.74,
+      "step": 9061
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.878172647691414e-06,
+      "loss": 0.4246,
+      "step": 9062
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.876091133605416e-06,
+      "loss": 0.7753,
+      "step": 9063
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.874009769495138e-06,
+      "loss": 0.417,
+      "step": 9064
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.871928555460497e-06,
+      "loss": 0.7857,
+      "step": 9065
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.869847491601413e-06,
+      "loss": 0.7482,
+      "step": 9066
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.867766578017801e-06,
+      "loss": 0.6948,
+      "step": 9067
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.865685814809555e-06,
+      "loss": 0.697,
+      "step": 9068
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8636052020765775e-06,
+      "loss": 0.4062,
+      "step": 9069
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.861524739918755e-06,
+      "loss": 0.449,
+      "step": 9070
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.859444428435966e-06,
+      "loss": 0.3793,
+      "step": 9071
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.857364267728086e-06,
+      "loss": 0.4556,
+      "step": 9072
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.855284257894981e-06,
+      "loss": 0.4245,
+      "step": 9073
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.853204399036511e-06,
+      "loss": 0.7774,
+      "step": 9074
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.851124691252528e-06,
+      "loss": 0.3973,
+      "step": 9075
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.849045134642873e-06,
+      "loss": 0.7415,
+      "step": 9076
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.846965729307389e-06,
+      "loss": 0.7725,
+      "step": 9077
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8448864753459e-06,
+      "loss": 0.7338,
+      "step": 9078
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8428073728582375e-06,
+      "loss": 0.7677,
+      "step": 9079
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.840728421944211e-06,
+      "loss": 0.7869,
+      "step": 9080
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8386496227036284e-06,
+      "loss": 0.7293,
+      "step": 9081
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.836570975236291e-06,
+      "loss": 0.7512,
+      "step": 9082
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.834492479641995e-06,
+      "loss": 0.7672,
+      "step": 9083
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.832414136020524e-06,
+      "loss": 0.747,
+      "step": 9084
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.83033594447166e-06,
+      "loss": 0.6902,
+      "step": 9085
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.828257905095175e-06,
+      "loss": 0.4099,
+      "step": 9086
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8261800179908284e-06,
+      "loss": 0.7168,
+      "step": 9087
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.824102283258386e-06,
+      "loss": 0.7212,
+      "step": 9088
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.822024700997589e-06,
+      "loss": 0.7213,
+      "step": 9089
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.819947271308187e-06,
+      "loss": 0.3664,
+      "step": 9090
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.817869994289912e-06,
+      "loss": 0.7125,
+      "step": 9091
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.81579287004249e-06,
+      "loss": 0.7579,
+      "step": 9092
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.81371589866565e-06,
+      "loss": 0.7347,
+      "step": 9093
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.811639080259095e-06,
+      "loss": 0.7275,
+      "step": 9094
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.809562414922535e-06,
+      "loss": 0.757,
+      "step": 9095
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.807485902755668e-06,
+      "loss": 0.7635,
+      "step": 9096
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.805409543858189e-06,
+      "loss": 0.7081,
+      "step": 9097
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.803333338329776e-06,
+      "loss": 0.7502,
+      "step": 9098
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.801257286270111e-06,
+      "loss": 0.7319,
+      "step": 9099
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.799181387778859e-06,
+      "loss": 0.4386,
+      "step": 9100
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.797105642955685e-06,
+      "loss": 0.7992,
+      "step": 9101
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.795030051900242e-06,
+      "loss": 0.7642,
+      "step": 9102
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.792954614712175e-06,
+      "loss": 0.722,
+      "step": 9103
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.790879331491127e-06,
+      "loss": 0.7561,
+      "step": 9104
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7888042023367285e-06,
+      "loss": 0.813,
+      "step": 9105
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.786729227348605e-06,
+      "loss": 0.7144,
+      "step": 9106
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.784654406626373e-06,
+      "loss": 0.4105,
+      "step": 9107
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.782579740269649e-06,
+      "loss": 0.7444,
+      "step": 9108
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7805052283780245e-06,
+      "loss": 0.7529,
+      "step": 9109
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.778430871051101e-06,
+      "loss": 0.7853,
+      "step": 9110
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.776356668388464e-06,
+      "loss": 0.3946,
+      "step": 9111
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7742826204896974e-06,
+      "loss": 0.6968,
+      "step": 9112
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.772208727454372e-06,
+      "loss": 0.7428,
+      "step": 9113
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.770134989382051e-06,
+      "loss": 0.7622,
+      "step": 9114
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.768061406372297e-06,
+      "loss": 0.7489,
+      "step": 9115
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.765987978524658e-06,
+      "loss": 0.4143,
+      "step": 9116
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.763914705938677e-06,
+      "loss": 0.7555,
+      "step": 9117
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.761841588713892e-06,
+      "loss": 0.4546,
+      "step": 9118
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7597686269498276e-06,
+      "loss": 0.7192,
+      "step": 9119
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.757695820746008e-06,
+      "loss": 0.7539,
+      "step": 9120
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.755623170201944e-06,
+      "loss": 0.7099,
+      "step": 9121
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.753550675417144e-06,
+      "loss": 0.7495,
+      "step": 9122
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.751478336491109e-06,
+      "loss": 0.3854,
+      "step": 9123
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.749406153523321e-06,
+      "loss": 0.767,
+      "step": 9124
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.747334126613269e-06,
+      "loss": 0.7101,
+      "step": 9125
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.745262255860428e-06,
+      "loss": 0.4324,
+      "step": 9126
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.743190541364266e-06,
+      "loss": 0.416,
+      "step": 9127
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.741118983224247e-06,
+      "loss": 0.7287,
+      "step": 9128
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.739047581539822e-06,
+      "loss": 0.721,
+      "step": 9129
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.736976336410435e-06,
+      "loss": 0.7399,
+      "step": 9130
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.734905247935528e-06,
+      "loss": 0.7353,
+      "step": 9131
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7328343162145286e-06,
+      "loss": 0.7487,
+      "step": 9132
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.730763541346863e-06,
+      "loss": 0.7385,
+      "step": 9133
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7286929234319465e-06,
+      "loss": 0.7185,
+      "step": 9134
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.726622462569185e-06,
+      "loss": 0.7329,
+      "step": 9135
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.724552158857982e-06,
+      "loss": 0.4051,
+      "step": 9136
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7224820123977285e-06,
+      "loss": 0.7496,
+      "step": 9137
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.720412023287814e-06,
+      "loss": 0.7688,
+      "step": 9138
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.718342191627614e-06,
+      "loss": 0.7258,
+      "step": 9139
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7162725175164975e-06,
+      "loss": 0.445,
+      "step": 9140
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7142030010538275e-06,
+      "loss": 0.7692,
+      "step": 9141
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.712133642338963e-06,
+      "loss": 0.7255,
+      "step": 9142
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.710064441471247e-06,
+      "loss": 0.7376,
+      "step": 9143
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.707995398550024e-06,
+      "loss": 0.7647,
+      "step": 9144
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.705926513674626e-06,
+      "loss": 0.771,
+      "step": 9145
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.703857786944373e-06,
+      "loss": 0.787,
+      "step": 9146
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.70178921845859e-06,
+      "loss": 0.7388,
+      "step": 9147
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.699720808316581e-06,
+      "loss": 0.7326,
+      "step": 9148
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.697652556617654e-06,
+      "loss": 0.4477,
+      "step": 9149
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6955844634611e-06,
+      "loss": 0.7845,
+      "step": 9150
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.693516528946205e-06,
+      "loss": 0.6949,
+      "step": 9151
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.691448753172249e-06,
+      "loss": 0.7608,
+      "step": 9152
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.689381136238511e-06,
+      "loss": 0.7224,
+      "step": 9153
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.687313678244243e-06,
+      "loss": 0.3944,
+      "step": 9154
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6852463792887075e-06,
+      "loss": 0.7379,
+      "step": 9155
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.683179239471157e-06,
+      "loss": 0.809,
+      "step": 9156
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.681112258890825e-06,
+      "loss": 0.7469,
+      "step": 9157
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.679045437646953e-06,
+      "loss": 0.7874,
+      "step": 9158
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.67697877583876e-06,
+      "loss": 0.3853,
+      "step": 9159
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.674912273565472e-06,
+      "loss": 0.7573,
+      "step": 9160
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.672845930926293e-06,
+      "loss": 0.4248,
+      "step": 9161
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.670779748020428e-06,
+      "loss": 0.7698,
+      "step": 9162
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.668713724947075e-06,
+      "loss": 0.7752,
+      "step": 9163
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.666647861805416e-06,
+      "loss": 0.7295,
+      "step": 9164
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.664582158694639e-06,
+      "loss": 0.4116,
+      "step": 9165
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.66251661571391e-06,
+      "loss": 0.4316,
+      "step": 9166
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.660451232962395e-06,
+      "loss": 0.7526,
+      "step": 9167
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.658386010539254e-06,
+      "loss": 0.7545,
+      "step": 9168
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.656320948543633e-06,
+      "loss": 0.6986,
+      "step": 9169
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.654256047074672e-06,
+      "loss": 0.4257,
+      "step": 9170
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.652191306231508e-06,
+      "loss": 0.8157,
+      "step": 9171
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.650126726113268e-06,
+      "loss": 0.7813,
+      "step": 9172
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.648062306819065e-06,
+      "loss": 0.7012,
+      "step": 9173
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.645998048448017e-06,
+      "loss": 0.6783,
+      "step": 9174
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.64393395109922e-06,
+      "loss": 0.753,
+      "step": 9175
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.641870014871775e-06,
+      "loss": 0.4051,
+      "step": 9176
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.639806239864768e-06,
+      "loss": 0.7859,
+      "step": 9177
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.637742626177275e-06,
+      "loss": 0.7774,
+      "step": 9178
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.635679173908371e-06,
+      "loss": 0.7775,
+      "step": 9179
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.633615883157121e-06,
+      "loss": 0.765,
+      "step": 9180
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.631552754022581e-06,
+      "loss": 0.4559,
+      "step": 9181
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.629489786603798e-06,
+      "loss": 0.7814,
+      "step": 9182
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.62742698099982e-06,
+      "loss": 0.7199,
+      "step": 9183
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.62536433730967e-06,
+      "loss": 0.416,
+      "step": 9184
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.623301855632378e-06,
+      "loss": 0.7622,
+      "step": 9185
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.621239536066961e-06,
+      "loss": 0.7425,
+      "step": 9186
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.619177378712432e-06,
+      "loss": 0.698,
+      "step": 9187
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.617115383667789e-06,
+      "loss": 0.4628,
+      "step": 9188
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6150535510320265e-06,
+      "loss": 0.7854,
+      "step": 9189
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.612991880904134e-06,
+      "loss": 0.7534,
+      "step": 9190
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.610930373383087e-06,
+      "loss": 0.739,
+      "step": 9191
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.608869028567861e-06,
+      "loss": 0.7405,
+      "step": 9192
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6068078465574145e-06,
+      "loss": 0.4174,
+      "step": 9193
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.604746827450704e-06,
+      "loss": 0.4267,
+      "step": 9194
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.60268597134668e-06,
+      "loss": 0.7448,
+      "step": 9195
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.600625278344276e-06,
+      "loss": 0.743,
+      "step": 9196
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.598564748542428e-06,
+      "loss": 0.7207,
+      "step": 9197
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.596504382040065e-06,
+      "loss": 0.7514,
+      "step": 9198
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.594444178936092e-06,
+      "loss": 0.7277,
+      "step": 9199
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.592384139329423e-06,
+      "loss": 0.6683,
+      "step": 9200
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.590324263318959e-06,
+      "loss": 0.3831,
+      "step": 9201
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.588264551003589e-06,
+      "loss": 0.7366,
+      "step": 9202
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.586205002482205e-06,
+      "loss": 0.3871,
+      "step": 9203
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.584145617853677e-06,
+      "loss": 0.7447,
+      "step": 9204
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.582086397216873e-06,
+      "loss": 0.72,
+      "step": 9205
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.580027340670661e-06,
+      "loss": 0.7096,
+      "step": 9206
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5779684483138895e-06,
+      "loss": 0.7382,
+      "step": 9207
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5759097202454055e-06,
+      "loss": 0.7061,
+      "step": 9208
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.573851156564046e-06,
+      "loss": 0.7213,
+      "step": 9209
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.571792757368638e-06,
+      "loss": 0.7159,
+      "step": 9210
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5697345227580086e-06,
+      "loss": 0.3852,
+      "step": 9211
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.56767645283097e-06,
+      "loss": 0.8351,
+      "step": 9212
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.565618547686324e-06,
+      "loss": 0.7547,
+      "step": 9213
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.563560807422871e-06,
+      "loss": 0.7464,
+      "step": 9214
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.561503232139401e-06,
+      "loss": 0.7201,
+      "step": 9215
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.559445821934696e-06,
+      "loss": 0.7848,
+      "step": 9216
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.557388576907533e-06,
+      "loss": 0.7411,
+      "step": 9217
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.555331497156671e-06,
+      "loss": 0.3747,
+      "step": 9218
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.553274582780877e-06,
+      "loss": 0.7409,
+      "step": 9219
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.551217833878898e-06,
+      "loss": 0.7136,
+      "step": 9220
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.549161250549474e-06,
+      "loss": 0.7636,
+      "step": 9221
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.547104832891342e-06,
+      "loss": 0.4166,
+      "step": 9222
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.545048581003226e-06,
+      "loss": 0.7717,
+      "step": 9223
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.54299249498385e-06,
+      "loss": 0.752,
+      "step": 9224
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5409365749319194e-06,
+      "loss": 0.7304,
+      "step": 9225
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.538880820946139e-06,
+      "loss": 0.7325,
+      "step": 9226
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.536825233125206e-06,
+      "loss": 0.8182,
+      "step": 9227
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5347698115678004e-06,
+      "loss": 0.7408,
+      "step": 9228
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.532714556372605e-06,
+      "loss": 0.7365,
+      "step": 9229
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5306594676382915e-06,
+      "loss": 0.7893,
+      "step": 9230
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.528604545463521e-06,
+      "loss": 0.3963,
+      "step": 9231
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.526549789946947e-06,
+      "loss": 0.7248,
+      "step": 9232
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5244952011872195e-06,
+      "loss": 0.4124,
+      "step": 9233
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.522440779282974e-06,
+      "loss": 0.7114,
+      "step": 9234
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.520386524332844e-06,
+      "loss": 0.7549,
+      "step": 9235
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.518332436435453e-06,
+      "loss": 0.4578,
+      "step": 9236
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.516278515689409e-06,
+      "loss": 0.7006,
+      "step": 9237
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5142247621933286e-06,
+      "loss": 0.7191,
+      "step": 9238
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.5121711760458005e-06,
+      "loss": 0.4174,
+      "step": 9239
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.510117757345423e-06,
+      "loss": 0.7338,
+      "step": 9240
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.508064506190777e-06,
+      "loss": 0.707,
+      "step": 9241
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.506011422680436e-06,
+      "loss": 0.7447,
+      "step": 9242
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.503958506912962e-06,
+      "loss": 0.715,
+      "step": 9243
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.501905758986921e-06,
+      "loss": 0.7724,
+      "step": 9244
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.499853179000857e-06,
+      "loss": 0.4412,
+      "step": 9245
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.497800767053317e-06,
+      "loss": 0.7623,
+      "step": 9246
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4957485232428345e-06,
+      "loss": 0.4538,
+      "step": 9247
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.493696447667932e-06,
+      "loss": 0.7153,
+      "step": 9248
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4916445404271325e-06,
+      "loss": 0.7134,
+      "step": 9249
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4895928016189415e-06,
+      "loss": 0.7138,
+      "step": 9250
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.487541231341865e-06,
+      "loss": 0.7665,
+      "step": 9251
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.485489829694396e-06,
+      "loss": 0.7521,
+      "step": 9252
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.483438596775016e-06,
+      "loss": 0.725,
+      "step": 9253
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4813875326822075e-06,
+      "loss": 0.7505,
+      "step": 9254
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.479336637514438e-06,
+      "loss": 0.4496,
+      "step": 9255
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4772859113701695e-06,
+      "loss": 0.6905,
+      "step": 9256
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.475235354347857e-06,
+      "loss": 0.748,
+      "step": 9257
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.473184966545943e-06,
+      "loss": 0.7792,
+      "step": 9258
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.471134748062864e-06,
+      "loss": 0.7001,
+      "step": 9259
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.46908469899705e-06,
+      "loss": 0.7869,
+      "step": 9260
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.467034819446921e-06,
+      "loss": 0.4179,
+      "step": 9261
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.464985109510893e-06,
+      "loss": 0.7293,
+      "step": 9262
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.462935569287367e-06,
+      "loss": 0.7417,
+      "step": 9263
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.46088619887474e-06,
+      "loss": 0.7768,
+      "step": 9264
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4588369983714016e-06,
+      "loss": 0.7277,
+      "step": 9265
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.456787967875729e-06,
+      "loss": 0.4245,
+      "step": 9266
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4547391074860985e-06,
+      "loss": 0.6832,
+      "step": 9267
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.452690417300871e-06,
+      "loss": 0.6971,
+      "step": 9268
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.450641897418402e-06,
+      "loss": 0.3994,
+      "step": 9269
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.448593547937039e-06,
+      "loss": 0.7655,
+      "step": 9270
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.446545368955121e-06,
+      "loss": 0.7705,
+      "step": 9271
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.444497360570983e-06,
+      "loss": 0.7902,
+      "step": 9272
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.442449522882943e-06,
+      "loss": 0.7331,
+      "step": 9273
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.440401855989316e-06,
+      "loss": 0.4076,
+      "step": 9274
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.438354359988408e-06,
+      "loss": 0.7524,
+      "step": 9275
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.436307034978519e-06,
+      "loss": 0.731,
+      "step": 9276
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.434259881057938e-06,
+      "loss": 0.7005,
+      "step": 9277
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.43221289832495e-06,
+      "loss": 0.767,
+      "step": 9278
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.430166086877825e-06,
+      "loss": 0.4256,
+      "step": 9279
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.428119446814828e-06,
+      "loss": 0.7446,
+      "step": 9280
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.426072978234219e-06,
+      "loss": 0.3705,
+      "step": 9281
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.424026681234243e-06,
+      "loss": 0.7354,
+      "step": 9282
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.421980555913145e-06,
+      "loss": 0.7322,
+      "step": 9283
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4199346023691565e-06,
+      "loss": 0.7654,
+      "step": 9284
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4178888207004985e-06,
+      "loss": 0.4126,
+      "step": 9285
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.41584321100539e-06,
+      "loss": 0.4465,
+      "step": 9286
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.41379777338204e-06,
+      "loss": 0.6997,
+      "step": 9287
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.411752507928643e-06,
+      "loss": 0.7714,
+      "step": 9288
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.409707414743393e-06,
+      "loss": 0.7488,
+      "step": 9289
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.407662493924474e-06,
+      "loss": 0.4406,
+      "step": 9290
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.405617745570055e-06,
+      "loss": 0.7166,
+      "step": 9291
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.40357316977831e-06,
+      "loss": 0.6817,
+      "step": 9292
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4015287666473915e-06,
+      "loss": 0.7634,
+      "step": 9293
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3994845362754535e-06,
+      "loss": 0.7239,
+      "step": 9294
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3974404787606335e-06,
+      "loss": 0.725,
+      "step": 9295
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.395396594201067e-06,
+      "loss": 0.6838,
+      "step": 9296
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.393352882694877e-06,
+      "loss": 0.7426,
+      "step": 9297
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.39130934434018e-06,
+      "loss": 0.429,
+      "step": 9298
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.389265979235088e-06,
+      "loss": 0.7297,
+      "step": 9299
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3872227874777005e-06,
+      "loss": 0.7353,
+      "step": 9300
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.385179769166102e-06,
+      "loss": 0.7676,
+      "step": 9301
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3831369243983855e-06,
+      "loss": 0.7556,
+      "step": 9302
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.381094253272619e-06,
+      "loss": 0.4031,
+      "step": 9303
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.379051755886869e-06,
+      "loss": 0.698,
+      "step": 9304
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.377009432339198e-06,
+      "loss": 0.4187,
+      "step": 9305
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3749672827276535e-06,
+      "loss": 0.7544,
+      "step": 9306
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.372925307150276e-06,
+      "loss": 0.7732,
+      "step": 9307
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.370883505705101e-06,
+      "loss": 0.7291,
+      "step": 9308
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.368841878490152e-06,
+      "loss": 0.7613,
+      "step": 9309
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.366800425603446e-06,
+      "loss": 0.7623,
+      "step": 9310
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.364759147142992e-06,
+      "loss": 0.7795,
+      "step": 9311
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.362718043206786e-06,
+      "loss": 0.77,
+      "step": 9312
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.360677113892824e-06,
+      "loss": 0.357,
+      "step": 9313
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3586363592990855e-06,
+      "loss": 0.7515,
+      "step": 9314
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.356595779523548e-06,
+      "loss": 0.747,
+      "step": 9315
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3545553746641765e-06,
+      "loss": 0.4364,
+      "step": 9316
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.352515144818929e-06,
+      "loss": 0.7456,
+      "step": 9317
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.350475090085752e-06,
+      "loss": 0.7525,
+      "step": 9318
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.348435210562591e-06,
+      "loss": 0.7258,
+      "step": 9319
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.346395506347374e-06,
+      "loss": 0.7759,
+      "step": 9320
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.344355977538028e-06,
+      "loss": 0.3829,
+      "step": 9321
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3423166242324694e-06,
+      "loss": 0.7627,
+      "step": 9322
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3402774465286034e-06,
+      "loss": 0.7598,
+      "step": 9323
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.33823844452433e-06,
+      "loss": 0.8027,
+      "step": 9324
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.336199618317538e-06,
+      "loss": 0.7133,
+      "step": 9325
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.334160968006113e-06,
+      "loss": 0.7582,
+      "step": 9326
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.332122493687926e-06,
+      "loss": 0.7563,
+      "step": 9327
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.330084195460842e-06,
+      "loss": 0.3922,
+      "step": 9328
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3280460734227185e-06,
+      "loss": 0.7826,
+      "step": 9329
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.326008127671401e-06,
+      "loss": 0.4322,
+      "step": 9330
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.323970358304737e-06,
+      "loss": 0.7813,
+      "step": 9331
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.32193276542055e-06,
+      "loss": 0.751,
+      "step": 9332
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.319895349116665e-06,
+      "loss": 0.718,
+      "step": 9333
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.317858109490893e-06,
+      "loss": 0.7221,
+      "step": 9334
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3158210466410465e-06,
+      "loss": 0.7332,
+      "step": 9335
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.313784160664917e-06,
+      "loss": 0.7302,
+      "step": 9336
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.311747451660298e-06,
+      "loss": 0.7797,
+      "step": 9337
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.309710919724967e-06,
+      "loss": 0.6997,
+      "step": 9338
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.307674564956696e-06,
+      "loss": 0.6978,
+      "step": 9339
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.30563838745325e-06,
+      "loss": 0.702,
+      "step": 9340
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.303602387312381e-06,
+      "loss": 0.7246,
+      "step": 9341
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.301566564631838e-06,
+      "loss": 0.4322,
+      "step": 9342
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.29953091950936e-06,
+      "loss": 0.4321,
+      "step": 9343
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.29749545204267e-06,
+      "loss": 0.7607,
+      "step": 9344
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.295460162329496e-06,
+      "loss": 0.7547,
+      "step": 9345
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.293425050467548e-06,
+      "loss": 0.7051,
+      "step": 9346
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.291390116554525e-06,
+      "loss": 0.7992,
+      "step": 9347
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.28935536068813e-06,
+      "loss": 0.7605,
+      "step": 9348
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.287320782966043e-06,
+      "loss": 0.7598,
+      "step": 9349
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.285286383485944e-06,
+      "loss": 0.7019,
+      "step": 9350
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.283252162345503e-06,
+      "loss": 0.7344,
+      "step": 9351
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.281218119642381e-06,
+      "loss": 0.7551,
+      "step": 9352
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.279184255474231e-06,
+      "loss": 0.7957,
+      "step": 9353
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.277150569938695e-06,
+      "loss": 0.7177,
+      "step": 9354
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.275117063133409e-06,
+      "loss": 0.4044,
+      "step": 9355
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.273083735156001e-06,
+      "loss": 0.4381,
+      "step": 9356
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.271050586104085e-06,
+      "loss": 0.7617,
+      "step": 9357
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.269017616075276e-06,
+      "loss": 0.7659,
+      "step": 9358
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.266984825167172e-06,
+      "loss": 0.7513,
+      "step": 9359
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2649522134773635e-06,
+      "loss": 0.6793,
+      "step": 9360
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.262919781103439e-06,
+      "loss": 0.7236,
+      "step": 9361
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.260887528142969e-06,
+      "loss": 0.7691,
+      "step": 9362
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.25885545469352e-06,
+      "loss": 0.3853,
+      "step": 9363
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.256823560852653e-06,
+      "loss": 0.7053,
+      "step": 9364
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.254791846717915e-06,
+      "loss": 0.6831,
+      "step": 9365
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2527603123868456e-06,
+      "loss": 0.7216,
+      "step": 9366
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.250728957956979e-06,
+      "loss": 0.7353,
+      "step": 9367
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.248697783525836e-06,
+      "loss": 0.4365,
+      "step": 9368
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2466667891909335e-06,
+      "loss": 0.7702,
+      "step": 9369
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2446359750497785e-06,
+      "loss": 0.3695,
+      "step": 9370
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.242605341199864e-06,
+      "loss": 0.7944,
+      "step": 9371
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.240574887738684e-06,
+      "loss": 0.7101,
+      "step": 9372
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.238544614763713e-06,
+      "loss": 0.7285,
+      "step": 9373
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.236514522372426e-06,
+      "loss": 0.7446,
+      "step": 9374
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.234484610662287e-06,
+      "loss": 0.7264,
+      "step": 9375
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.232454879730749e-06,
+      "loss": 0.6897,
+      "step": 9376
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.230425329675252e-06,
+      "loss": 0.7683,
+      "step": 9377
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.228395960593239e-06,
+      "loss": 0.4582,
+      "step": 9378
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.226366772582133e-06,
+      "loss": 0.4491,
+      "step": 9379
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2243377657393585e-06,
+      "loss": 0.3859,
+      "step": 9380
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.222308940162325e-06,
+      "loss": 0.7241,
+      "step": 9381
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2202802959484285e-06,
+      "loss": 0.7817,
+      "step": 9382
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.21825183319507e-06,
+      "loss": 0.7574,
+      "step": 9383
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.21622355199963e-06,
+      "loss": 0.757,
+      "step": 9384
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.214195452459486e-06,
+      "loss": 0.6868,
+      "step": 9385
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2121675346720035e-06,
+      "loss": 0.7604,
+      "step": 9386
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.210139798734539e-06,
+      "loss": 0.3999,
+      "step": 9387
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.208112244744446e-06,
+      "loss": 0.7628,
+      "step": 9388
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.206084872799063e-06,
+      "loss": 0.7604,
+      "step": 9389
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.204057682995724e-06,
+      "loss": 0.4642,
+      "step": 9390
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.202030675431752e-06,
+      "loss": 0.7629,
+      "step": 9391
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.200003850204459e-06,
+      "loss": 0.7163,
+      "step": 9392
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.197977207411151e-06,
+      "loss": 0.7217,
+      "step": 9393
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.195950747149128e-06,
+      "loss": 0.4491,
+      "step": 9394
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.193924469515675e-06,
+      "loss": 0.6912,
+      "step": 9395
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.191898374608077e-06,
+      "loss": 0.4277,
+      "step": 9396
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1898724625235986e-06,
+      "loss": 0.7329,
+      "step": 9397
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.187846733359503e-06,
+      "loss": 0.7077,
+      "step": 9398
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.185821187213047e-06,
+      "loss": 0.7651,
+      "step": 9399
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.183795824181471e-06,
+      "loss": 0.4161,
+      "step": 9400
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.181770644362013e-06,
+      "loss": 0.4278,
+      "step": 9401
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.179745647851901e-06,
+      "loss": 0.7786,
+      "step": 9402
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.177720834748349e-06,
+      "loss": 0.4247,
+      "step": 9403
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.17569620514857e-06,
+      "loss": 0.4043,
+      "step": 9404
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.173671759149762e-06,
+      "loss": 0.7319,
+      "step": 9405
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.17164749684912e-06,
+      "loss": 0.3732,
+      "step": 9406
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1696234183438244e-06,
+      "loss": 0.7438,
+      "step": 9407
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.167599523731047e-06,
+      "loss": 0.7396,
+      "step": 9408
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.165575813107954e-06,
+      "loss": 0.7764,
+      "step": 9409
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.163552286571705e-06,
+      "loss": 0.7112,
+      "step": 9410
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1615289442194435e-06,
+      "loss": 0.6815,
+      "step": 9411
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.159505786148311e-06,
+      "loss": 0.7017,
+      "step": 9412
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.157482812455438e-06,
+      "loss": 0.8014,
+      "step": 9413
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.155460023237939e-06,
+      "loss": 0.7471,
+      "step": 9414
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1534374185929335e-06,
+      "loss": 0.72,
+      "step": 9415
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1514149986175206e-06,
+      "loss": 0.4047,
+      "step": 9416
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.149392763408798e-06,
+      "loss": 0.788,
+      "step": 9417
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.147370713063849e-06,
+      "loss": 0.7595,
+      "step": 9418
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.145348847679748e-06,
+      "loss": 0.7657,
+      "step": 9419
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.143327167353567e-06,
+      "loss": 0.4177,
+      "step": 9420
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.141305672182365e-06,
+      "loss": 0.7366,
+      "step": 9421
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.139284362263185e-06,
+      "loss": 0.4222,
+      "step": 9422
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.137263237693076e-06,
+      "loss": 0.7625,
+      "step": 9423
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1352422985690664e-06,
+      "loss": 0.7685,
+      "step": 9424
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.133221544988179e-06,
+      "loss": 0.7025,
+      "step": 9425
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.13120097704743e-06,
+      "loss": 0.7361,
+      "step": 9426
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.129180594843822e-06,
+      "loss": 0.7348,
+      "step": 9427
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.127160398474356e-06,
+      "loss": 0.74,
+      "step": 9428
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.125140388036017e-06,
+      "loss": 0.7213,
+      "step": 9429
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.123120563625783e-06,
+      "loss": 0.7566,
+      "step": 9430
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.121100925340626e-06,
+      "loss": 0.407,
+      "step": 9431
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.119081473277502e-06,
+      "loss": 0.356,
+      "step": 9432
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.11706220753337e-06,
+      "loss": 0.7555,
+      "step": 9433
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.115043128205167e-06,
+      "loss": 0.7062,
+      "step": 9434
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.113024235389833e-06,
+      "loss": 0.7896,
+      "step": 9435
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.111005529184284e-06,
+      "loss": 0.7338,
+      "step": 9436
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1089870096854445e-06,
+      "loss": 0.4567,
+      "step": 9437
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.106968676990215e-06,
+      "loss": 0.7545,
+      "step": 9438
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1049505311955e-06,
+      "loss": 0.7291,
+      "step": 9439
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.102932572398184e-06,
+      "loss": 0.7675,
+      "step": 9440
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.100914800695147e-06,
+      "loss": 0.4277,
+      "step": 9441
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0988972161832634e-06,
+      "loss": 0.7485,
+      "step": 9442
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.096879818959394e-06,
+      "loss": 0.7434,
+      "step": 9443
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.094862609120391e-06,
+      "loss": 0.6797,
+      "step": 9444
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.092845586763101e-06,
+      "loss": 0.3814,
+      "step": 9445
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.090828751984356e-06,
+      "loss": 0.7208,
+      "step": 9446
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.088812104880985e-06,
+      "loss": 0.7775,
+      "step": 9447
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.086795645549803e-06,
+      "loss": 0.7202,
+      "step": 9448
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.08477937408762e-06,
+      "loss": 0.724,
+      "step": 9449
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.082763290591238e-06,
+      "loss": 0.7486,
+      "step": 9450
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.08074739515744e-06,
+      "loss": 0.3983,
+      "step": 9451
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.078731687883011e-06,
+      "loss": 0.7506,
+      "step": 9452
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.076716168864722e-06,
+      "loss": 0.3953,
+      "step": 9453
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.074700838199337e-06,
+      "loss": 0.6927,
+      "step": 9454
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.07268569598361e-06,
+      "loss": 0.7328,
+      "step": 9455
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.070670742314287e-06,
+      "loss": 0.7916,
+      "step": 9456
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0686559772880995e-06,
+      "loss": 0.7333,
+      "step": 9457
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.066641401001781e-06,
+      "loss": 0.3966,
+      "step": 9458
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.064627013552043e-06,
+      "loss": 0.8008,
+      "step": 9459
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.062612815035598e-06,
+      "loss": 0.7924,
+      "step": 9460
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.060598805549145e-06,
+      "loss": 0.4243,
+      "step": 9461
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.058584985189373e-06,
+      "loss": 0.7761,
+      "step": 9462
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.056571354052965e-06,
+      "loss": 0.7309,
+      "step": 9463
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.054557912236593e-06,
+      "loss": 0.6883,
+      "step": 9464
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.052544659836923e-06,
+      "loss": 0.7353,
+      "step": 9465
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.050531596950605e-06,
+      "loss": 0.7245,
+      "step": 9466
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.048518723674286e-06,
+      "loss": 0.6914,
+      "step": 9467
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0465060401046e-06,
+      "loss": 0.7724,
+      "step": 9468
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0444935463381775e-06,
+      "loss": 0.743,
+      "step": 9469
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.042481242471632e-06,
+      "loss": 0.7431,
+      "step": 9470
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.040469128601576e-06,
+      "loss": 0.8029,
+      "step": 9471
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.038457204824608e-06,
+      "loss": 0.7582,
+      "step": 9472
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.036445471237316e-06,
+      "loss": 0.7027,
+      "step": 9473
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.034433927936287e-06,
+      "loss": 0.7628,
+      "step": 9474
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.032422575018085e-06,
+      "loss": 0.7408,
+      "step": 9475
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.030411412579283e-06,
+      "loss": 0.7808,
+      "step": 9476
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.028400440716427e-06,
+      "loss": 0.7188,
+      "step": 9477
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.026389659526062e-06,
+      "loss": 0.4143,
+      "step": 9478
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0243790691047285e-06,
+      "loss": 0.7688,
+      "step": 9479
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.022368669548953e-06,
+      "loss": 0.7733,
+      "step": 9480
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0203584609552444e-06,
+      "loss": 0.7334,
+      "step": 9481
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.018348443420119e-06,
+      "loss": 0.7316,
+      "step": 9482
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.016338617040073e-06,
+      "loss": 0.7806,
+      "step": 9483
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.014328981911595e-06,
+      "loss": 0.8093,
+      "step": 9484
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.012319538131167e-06,
+      "loss": 0.4428,
+      "step": 9485
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.01031028579526e-06,
+      "loss": 0.3931,
+      "step": 9486
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.008301225000337e-06,
+      "loss": 0.7333,
+      "step": 9487
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.006292355842851e-06,
+      "loss": 0.4134,
+      "step": 9488
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.004283678419243e-06,
+      "loss": 0.7143,
+      "step": 9489
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.00227519282595e-06,
+      "loss": 0.7317,
+      "step": 9490
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0002668991593975e-06,
+      "loss": 0.7766,
+      "step": 9491
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.998258797516003e-06,
+      "loss": 0.6773,
+      "step": 9492
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.99625088799217e-06,
+      "loss": 0.4369,
+      "step": 9493
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.994243170684297e-06,
+      "loss": 0.4372,
+      "step": 9494
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.992235645688778e-06,
+      "loss": 0.808,
+      "step": 9495
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.990228313101985e-06,
+      "loss": 0.4662,
+      "step": 9496
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9882211730202874e-06,
+      "loss": 0.745,
+      "step": 9497
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.986214225540053e-06,
+      "loss": 0.731,
+      "step": 9498
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.984207470757628e-06,
+      "loss": 0.7419,
+      "step": 9499
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9822009087693545e-06,
+      "loss": 0.7396,
+      "step": 9500
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.980194539671569e-06,
+      "loss": 0.7294,
+      "step": 9501
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.978188363560593e-06,
+      "loss": 0.6943,
+      "step": 9502
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.976182380532743e-06,
+      "loss": 0.7761,
+      "step": 9503
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.974176590684321e-06,
+      "loss": 0.3977,
+      "step": 9504
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9721709941116235e-06,
+      "loss": 0.7555,
+      "step": 9505
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.97016559091094e-06,
+      "loss": 0.7452,
+      "step": 9506
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.968160381178545e-06,
+      "loss": 0.7191,
+      "step": 9507
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9661553650107085e-06,
+      "loss": 0.7735,
+      "step": 9508
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.964150542503691e-06,
+      "loss": 0.4323,
+      "step": 9509
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.962145913753738e-06,
+      "loss": 0.7584,
+      "step": 9510
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.960141478857091e-06,
+      "loss": 0.7034,
+      "step": 9511
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.95813723790998e-06,
+      "loss": 0.7725,
+      "step": 9512
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.956133191008628e-06,
+      "loss": 0.3911,
+      "step": 9513
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9541293382492486e-06,
+      "loss": 0.6984,
+      "step": 9514
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.952125679728042e-06,
+      "loss": 0.7595,
+      "step": 9515
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.950122215541203e-06,
+      "loss": 0.718,
+      "step": 9516
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.948118945784916e-06,
+      "loss": 0.7246,
+      "step": 9517
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.946115870555355e-06,
+      "loss": 0.7484,
+      "step": 9518
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.94411298994869e-06,
+      "loss": 0.733,
+      "step": 9519
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.942110304061071e-06,
+      "loss": 0.8031,
+      "step": 9520
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.940107812988647e-06,
+      "loss": 0.7208,
+      "step": 9521
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.938105516827557e-06,
+      "loss": 0.6923,
+      "step": 9522
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.936103415673928e-06,
+      "loss": 0.4104,
+      "step": 9523
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.934101509623882e-06,
+      "loss": 0.755,
+      "step": 9524
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.932099798773526e-06,
+      "loss": 0.7517,
+      "step": 9525
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.930098283218959e-06,
+      "loss": 0.7077,
+      "step": 9526
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.928096963056271e-06,
+      "loss": 0.7037,
+      "step": 9527
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.926095838381547e-06,
+      "loss": 0.7793,
+      "step": 9528
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.924094909290854e-06,
+      "loss": 0.6687,
+      "step": 9529
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9220941758802605e-06,
+      "loss": 0.8065,
+      "step": 9530
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.920093638245818e-06,
+      "loss": 0.7286,
+      "step": 9531
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.918093296483565e-06,
+      "loss": 0.6652,
+      "step": 9532
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.916093150689543e-06,
+      "loss": 0.4121,
+      "step": 9533
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.914093200959772e-06,
+      "loss": 0.7598,
+      "step": 9534
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.912093447390272e-06,
+      "loss": 0.7506,
+      "step": 9535
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.910093890077047e-06,
+      "loss": 0.395,
+      "step": 9536
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.908094529116091e-06,
+      "loss": 0.7669,
+      "step": 9537
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.9060953646033956e-06,
+      "loss": 0.7483,
+      "step": 9538
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.904096396634935e-06,
+      "loss": 0.753,
+      "step": 9539
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.902097625306684e-06,
+      "loss": 0.7416,
+      "step": 9540
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.900099050714594e-06,
+      "loss": 0.7759,
+      "step": 9541
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.898100672954618e-06,
+      "loss": 0.4462,
+      "step": 9542
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.896102492122695e-06,
+      "loss": 0.7947,
+      "step": 9543
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.894104508314757e-06,
+      "loss": 0.4708,
+      "step": 9544
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.892106721626725e-06,
+      "loss": 0.7171,
+      "step": 9545
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.89010913215451e-06,
+      "loss": 0.7405,
+      "step": 9546
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.888111739994016e-06,
+      "loss": 0.7012,
+      "step": 9547
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.886114545241131e-06,
+      "loss": 0.7325,
+      "step": 9548
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.884117547991744e-06,
+      "loss": 0.426,
+      "step": 9549
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.882120748341724e-06,
+      "loss": 0.7329,
+      "step": 9550
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.880124146386941e-06,
+      "loss": 0.786,
+      "step": 9551
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.878127742223247e-06,
+      "loss": 0.7052,
+      "step": 9552
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.876131535946483e-06,
+      "loss": 0.6889,
+      "step": 9553
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.874135527652496e-06,
+      "loss": 0.7609,
+      "step": 9554
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.872139717437101e-06,
+      "loss": 0.3808,
+      "step": 9555
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8701441053961185e-06,
+      "loss": 0.3956,
+      "step": 9556
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.868148691625357e-06,
+      "loss": 0.7196,
+      "step": 9557
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.866153476220614e-06,
+      "loss": 0.7348,
+      "step": 9558
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.864158459277676e-06,
+      "loss": 0.396,
+      "step": 9559
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8621636408923244e-06,
+      "loss": 0.4405,
+      "step": 9560
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.860169021160327e-06,
+      "loss": 0.7018,
+      "step": 9561
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.858174600177444e-06,
+      "loss": 0.705,
+      "step": 9562
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.856180378039426e-06,
+      "loss": 0.7073,
+      "step": 9563
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.854186354842011e-06,
+      "loss": 0.7187,
+      "step": 9564
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8521925306809335e-06,
+      "loss": 0.7036,
+      "step": 9565
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.850198905651911e-06,
+      "loss": 0.7649,
+      "step": 9566
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.848205479850662e-06,
+      "loss": 0.4002,
+      "step": 9567
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.846212253372883e-06,
+      "loss": 0.7253,
+      "step": 9568
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.84421922631427e-06,
+      "loss": 0.7627,
+      "step": 9569
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.842226398770502e-06,
+      "loss": 0.7601,
+      "step": 9570
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.840233770837257e-06,
+      "loss": 0.7772,
+      "step": 9571
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.838241342610197e-06,
+      "loss": 0.7884,
+      "step": 9572
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.836249114184978e-06,
+      "loss": 0.4343,
+      "step": 9573
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.834257085657238e-06,
+      "loss": 0.4613,
+      "step": 9574
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.832265257122624e-06,
+      "loss": 0.7239,
+      "step": 9575
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.830273628676755e-06,
+      "loss": 0.4123,
+      "step": 9576
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.828282200415247e-06,
+      "loss": 0.7507,
+      "step": 9577
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.826290972433703e-06,
+      "loss": 0.7086,
+      "step": 9578
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.82429994482773e-06,
+      "loss": 0.7571,
+      "step": 9579
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.822309117692908e-06,
+      "loss": 0.7353,
+      "step": 9580
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8203184911248165e-06,
+      "loss": 0.7457,
+      "step": 9581
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8183280652190236e-06,
+      "loss": 0.439,
+      "step": 9582
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.816337840071083e-06,
+      "loss": 0.7241,
+      "step": 9583
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.814347815776555e-06,
+      "loss": 0.7394,
+      "step": 9584
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.812357992430966e-06,
+      "loss": 0.7606,
+      "step": 9585
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.810368370129851e-06,
+      "loss": 0.7223,
+      "step": 9586
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.808378948968726e-06,
+      "loss": 0.6852,
+      "step": 9587
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.806389729043108e-06,
+      "loss": 0.7351,
+      "step": 9588
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.804400710448493e-06,
+      "loss": 0.766,
+      "step": 9589
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.802411893280373e-06,
+      "loss": 0.7074,
+      "step": 9590
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8004232776342295e-06,
+      "loss": 0.3976,
+      "step": 9591
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.798434863605529e-06,
+      "loss": 0.7949,
+      "step": 9592
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.79644665128974e-06,
+      "loss": 0.7469,
+      "step": 9593
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.794458640782314e-06,
+      "loss": 0.7295,
+      "step": 9594
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.792470832178689e-06,
+      "loss": 0.6724,
+      "step": 9595
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.790483225574301e-06,
+      "loss": 0.3664,
+      "step": 9596
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.788495821064568e-06,
+      "loss": 0.7178,
+      "step": 9597
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.786508618744912e-06,
+      "loss": 0.7572,
+      "step": 9598
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.784521618710735e-06,
+      "loss": 0.7126,
+      "step": 9599
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.782534821057424e-06,
+      "loss": 0.7402,
+      "step": 9600
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.780548225880363e-06,
+      "loss": 0.7283,
+      "step": 9601
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.778561833274934e-06,
+      "loss": 0.4502,
+      "step": 9602
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.776575643336498e-06,
+      "loss": 0.7577,
+      "step": 9603
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.77458965616041e-06,
+      "loss": 0.7256,
+      "step": 9604
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.772603871842015e-06,
+      "loss": 0.7626,
+      "step": 9605
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.770618290476646e-06,
+      "loss": 0.6777,
+      "step": 9606
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.768632912159635e-06,
+      "loss": 0.779,
+      "step": 9607
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.766647736986294e-06,
+      "loss": 0.7615,
+      "step": 9608
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7646627650519294e-06,
+      "loss": 0.6878,
+      "step": 9609
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7626779964518355e-06,
+      "loss": 0.7344,
+      "step": 9610
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.760693431281305e-06,
+      "loss": 0.4006,
+      "step": 9611
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.75870906963561e-06,
+      "loss": 0.7633,
+      "step": 9612
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.756724911610021e-06,
+      "loss": 0.709,
+      "step": 9613
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.754740957299793e-06,
+      "loss": 0.4062,
+      "step": 9614
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7527572068001744e-06,
+      "loss": 0.7579,
+      "step": 9615
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.750773660206403e-06,
+      "loss": 0.7524,
+      "step": 9616
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.748790317613707e-06,
+      "loss": 0.7639,
+      "step": 9617
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.746807179117304e-06,
+      "loss": 0.7493,
+      "step": 9618
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.744824244812399e-06,
+      "loss": 0.4051,
+      "step": 9619
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.742841514794199e-06,
+      "loss": 0.7133,
+      "step": 9620
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.740858989157889e-06,
+      "loss": 0.7699,
+      "step": 9621
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.738876667998647e-06,
+      "loss": 0.7145,
+      "step": 9622
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.736894551411642e-06,
+      "loss": 0.7054,
+      "step": 9623
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.734912639492032e-06,
+      "loss": 0.7283,
+      "step": 9624
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7329309323349725e-06,
+      "loss": 0.7982,
+      "step": 9625
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.730949430035599e-06,
+      "loss": 0.7398,
+      "step": 9626
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7289681326890415e-06,
+      "loss": 0.6943,
+      "step": 9627
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.726987040390421e-06,
+      "loss": 0.7467,
+      "step": 9628
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7250061532348475e-06,
+      "loss": 0.3736,
+      "step": 9629
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.723025471317421e-06,
+      "loss": 0.7348,
+      "step": 9630
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7210449947332335e-06,
+      "loss": 0.7565,
+      "step": 9631
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.719064723577363e-06,
+      "loss": 0.7993,
+      "step": 9632
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.717084657944878e-06,
+      "loss": 0.7839,
+      "step": 9633
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.715104797930849e-06,
+      "loss": 0.7745,
+      "step": 9634
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.713125143630319e-06,
+      "loss": 0.6839,
+      "step": 9635
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.711145695138333e-06,
+      "loss": 0.7582,
+      "step": 9636
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.70916645254992e-06,
+      "loss": 0.4061,
+      "step": 9637
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.707187415960099e-06,
+      "loss": 0.7234,
+      "step": 9638
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.705208585463889e-06,
+      "loss": 0.7469,
+      "step": 9639
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7032299611562865e-06,
+      "loss": 0.7463,
+      "step": 9640
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.701251543132285e-06,
+      "loss": 0.7283,
+      "step": 9641
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.699273331486862e-06,
+      "loss": 0.7405,
+      "step": 9642
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.697295326314998e-06,
+      "loss": 0.7348,
+      "step": 9643
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.695317527711653e-06,
+      "loss": 0.7479,
+      "step": 9644
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6933399357717714e-06,
+      "loss": 0.7815,
+      "step": 9645
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6913625505902966e-06,
+      "loss": 0.6911,
+      "step": 9646
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6893853722621685e-06,
+      "loss": 0.7188,
+      "step": 9647
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.687408400882306e-06,
+      "loss": 0.7329,
+      "step": 9648
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.685431636545622e-06,
+      "loss": 0.7447,
+      "step": 9649
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.683455079347016e-06,
+      "loss": 0.7139,
+      "step": 9650
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.68147872938138e-06,
+      "loss": 0.7506,
+      "step": 9651
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.679502586743601e-06,
+      "loss": 0.4279,
+      "step": 9652
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6775266515285524e-06,
+      "loss": 0.731,
+      "step": 9653
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.675550923831092e-06,
+      "loss": 0.7707,
+      "step": 9654
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.673575403746075e-06,
+      "loss": 0.7717,
+      "step": 9655
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.671600091368341e-06,
+      "loss": 0.4068,
+      "step": 9656
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.669624986792728e-06,
+      "loss": 0.4102,
+      "step": 9657
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.667650090114056e-06,
+      "loss": 0.4424,
+      "step": 9658
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.665675401427144e-06,
+      "loss": 0.8519,
+      "step": 9659
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.663700920826779e-06,
+      "loss": 0.4197,
+      "step": 9660
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.661726648407768e-06,
+      "loss": 0.72,
+      "step": 9661
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6597525842648904e-06,
+      "loss": 0.3749,
+      "step": 9662
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6577787284929175e-06,
+      "loss": 0.6948,
+      "step": 9663
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6558050811866135e-06,
+      "loss": 0.7591,
+      "step": 9664
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.653831642440726e-06,
+      "loss": 0.662,
+      "step": 9665
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.651858412350006e-06,
+      "loss": 0.7149,
+      "step": 9666
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.649885391009182e-06,
+      "loss": 0.764,
+      "step": 9667
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6479125785129795e-06,
+      "loss": 0.703,
+      "step": 9668
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.645939974956103e-06,
+      "loss": 0.7702,
+      "step": 9669
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.643967580433267e-06,
+      "loss": 0.417,
+      "step": 9670
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.641995395039157e-06,
+      "loss": 0.7536,
+      "step": 9671
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.640023418868458e-06,
+      "loss": 0.6954,
+      "step": 9672
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.638051652015841e-06,
+      "loss": 0.723,
+      "step": 9673
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.636080094575971e-06,
+      "loss": 0.4377,
+      "step": 9674
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.634108746643496e-06,
+      "loss": 0.3988,
+      "step": 9675
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.632137608313063e-06,
+      "loss": 0.705,
+      "step": 9676
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.630166679679303e-06,
+      "loss": 0.6821,
+      "step": 9677
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.628195960836833e-06,
+      "loss": 0.4025,
+      "step": 9678
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.626225451880274e-06,
+      "loss": 0.7134,
+      "step": 9679
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.624255152904223e-06,
+      "loss": 0.7918,
+      "step": 9680
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.622285064003276e-06,
+      "loss": 0.7394,
+      "step": 9681
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.620315185272009e-06,
+      "loss": 0.7222,
+      "step": 9682
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.618345516804996e-06,
+      "loss": 0.7163,
+      "step": 9683
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.616376058696803e-06,
+      "loss": 0.7151,
+      "step": 9684
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.614406811041979e-06,
+      "loss": 0.7276,
+      "step": 9685
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.612437773935065e-06,
+      "loss": 0.7079,
+      "step": 9686
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.610468947470592e-06,
+      "loss": 0.7233,
+      "step": 9687
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.608500331743083e-06,
+      "loss": 0.7306,
+      "step": 9688
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.606531926847048e-06,
+      "loss": 0.3852,
+      "step": 9689
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.604563732876989e-06,
+      "loss": 0.7847,
+      "step": 9690
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.602595749927395e-06,
+      "loss": 0.7189,
+      "step": 9691
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.600627978092745e-06,
+      "loss": 0.7118,
+      "step": 9692
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.598660417467518e-06,
+      "loss": 0.4055,
+      "step": 9693
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.596693068146168e-06,
+      "loss": 0.7337,
+      "step": 9694
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.594725930223151e-06,
+      "loss": 0.7045,
+      "step": 9695
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5927590037929004e-06,
+      "loss": 0.4059,
+      "step": 9696
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.590792288949845e-06,
+      "loss": 0.7441,
+      "step": 9697
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5888257857884145e-06,
+      "loss": 0.746,
+      "step": 9698
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5868594944030145e-06,
+      "loss": 0.7056,
+      "step": 9699
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.584893414888042e-06,
+      "loss": 0.4307,
+      "step": 9700
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5829275473378844e-06,
+      "loss": 0.7295,
+      "step": 9701
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5809618918469274e-06,
+      "loss": 0.817,
+      "step": 9702
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.578996448509544e-06,
+      "loss": 0.7193,
+      "step": 9703
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.577031217420081e-06,
+      "loss": 0.4149,
+      "step": 9704
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.57506619867289e-06,
+      "loss": 0.7061,
+      "step": 9705
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.573101392362317e-06,
+      "loss": 0.7517,
+      "step": 9706
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.571136798582684e-06,
+      "loss": 0.7303,
+      "step": 9707
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5691724174283126e-06,
+      "loss": 0.7901,
+      "step": 9708
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.567208248993508e-06,
+      "loss": 0.8439,
+      "step": 9709
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.565244293372565e-06,
+      "loss": 0.7059,
+      "step": 9710
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.56328055065978e-06,
+      "loss": 0.7015,
+      "step": 9711
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.561317020949424e-06,
+      "loss": 0.756,
+      "step": 9712
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.559353704335767e-06,
+      "loss": 0.7485,
+      "step": 9713
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.557390600913064e-06,
+      "loss": 0.7373,
+      "step": 9714
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.555427710775558e-06,
+      "loss": 0.7312,
+      "step": 9715
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.553465034017493e-06,
+      "loss": 0.6575,
+      "step": 9716
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.551502570733092e-06,
+      "loss": 0.7199,
+      "step": 9717
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.549540321016576e-06,
+      "loss": 0.6911,
+      "step": 9718
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.547578284962134e-06,
+      "loss": 0.7358,
+      "step": 9719
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.545616462663979e-06,
+      "loss": 0.6977,
+      "step": 9720
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.543654854216287e-06,
+      "loss": 0.7587,
+      "step": 9721
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.541693459713238e-06,
+      "loss": 0.7061,
+      "step": 9722
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5397322792489925e-06,
+      "loss": 0.4157,
+      "step": 9723
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5377713129177025e-06,
+      "loss": 0.7682,
+      "step": 9724
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.535810560813519e-06,
+      "loss": 0.7095,
+      "step": 9725
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.533850023030575e-06,
+      "loss": 0.7706,
+      "step": 9726
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.531889699662989e-06,
+      "loss": 0.7468,
+      "step": 9727
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.529929590804879e-06,
+      "loss": 0.7656,
+      "step": 9728
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.527969696550339e-06,
+      "loss": 0.7623,
+      "step": 9729
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.526010016993473e-06,
+      "loss": 0.7339,
+      "step": 9730
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.52405055222836e-06,
+      "loss": 0.773,
+      "step": 9731
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.522091302349067e-06,
+      "loss": 0.4266,
+      "step": 9732
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.52013226744966e-06,
+      "loss": 0.6827,
+      "step": 9733
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.51817344762419e-06,
+      "loss": 0.7117,
+      "step": 9734
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.516214842966696e-06,
+      "loss": 0.4092,
+      "step": 9735
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.51425645357121e-06,
+      "loss": 0.7276,
+      "step": 9736
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5122982795317494e-06,
+      "loss": 0.4521,
+      "step": 9737
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.510340320942329e-06,
+      "loss": 0.7211,
+      "step": 9738
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.508382577896947e-06,
+      "loss": 0.7299,
+      "step": 9739
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.506425050489591e-06,
+      "loss": 0.7046,
+      "step": 9740
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.504467738814241e-06,
+      "loss": 0.7424,
+      "step": 9741
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5025106429648625e-06,
+      "loss": 0.3761,
+      "step": 9742
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.500553763035419e-06,
+      "loss": 0.7653,
+      "step": 9743
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.498597099119858e-06,
+      "loss": 0.7452,
+      "step": 9744
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.496640651312113e-06,
+      "loss": 0.7269,
+      "step": 9745
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.494684419706113e-06,
+      "loss": 0.7737,
+      "step": 9746
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.492728404395771e-06,
+      "loss": 0.7273,
+      "step": 9747
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4907726054750055e-06,
+      "loss": 0.7506,
+      "step": 9748
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.488817023037698e-06,
+      "loss": 0.7493,
+      "step": 9749
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.486861657177742e-06,
+      "loss": 0.7903,
+      "step": 9750
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.484906507989005e-06,
+      "loss": 0.405,
+      "step": 9751
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.482951575565363e-06,
+      "loss": 0.6947,
+      "step": 9752
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.480996860000664e-06,
+      "loss": 0.7204,
+      "step": 9753
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.479042361388751e-06,
+      "loss": 0.7455,
+      "step": 9754
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.47708807982346e-06,
+      "loss": 0.7349,
+      "step": 9755
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.475134015398611e-06,
+      "loss": 0.7596,
+      "step": 9756
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.473180168208021e-06,
+      "loss": 0.7529,
+      "step": 9757
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.471226538345491e-06,
+      "loss": 0.7602,
+      "step": 9758
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.46927312590481e-06,
+      "loss": 0.7493,
+      "step": 9759
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.46731993097976e-06,
+      "loss": 0.7619,
+      "step": 9760
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4653669536641154e-06,
+      "loss": 0.7628,
+      "step": 9761
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.463414194051635e-06,
+      "loss": 0.705,
+      "step": 9762
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.461461652236072e-06,
+      "loss": 0.7377,
+      "step": 9763
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.459509328311154e-06,
+      "loss": 0.7404,
+      "step": 9764
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.457557222370622e-06,
+      "loss": 0.3998,
+      "step": 9765
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.455605334508192e-06,
+      "loss": 0.7501,
+      "step": 9766
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.453653664817571e-06,
+      "loss": 0.4219,
+      "step": 9767
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.451702213392456e-06,
+      "loss": 0.7747,
+      "step": 9768
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.449750980326532e-06,
+      "loss": 0.76,
+      "step": 9769
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4477999657134825e-06,
+      "loss": 0.7505,
+      "step": 9770
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.445849169646969e-06,
+      "loss": 0.7699,
+      "step": 9771
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.443898592220649e-06,
+      "loss": 0.7258,
+      "step": 9772
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.441948233528167e-06,
+      "loss": 0.7334,
+      "step": 9773
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.439998093663155e-06,
+      "loss": 0.6952,
+      "step": 9774
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.438048172719242e-06,
+      "loss": 0.3701,
+      "step": 9775
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.436098470790041e-06,
+      "loss": 0.78,
+      "step": 9776
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.434148987969154e-06,
+      "loss": 0.7313,
+      "step": 9777
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.432199724350173e-06,
+      "loss": 0.7645,
+      "step": 9778
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.430250680026682e-06,
+      "loss": 0.7105,
+      "step": 9779
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.428301855092251e-06,
+      "loss": 0.7254,
+      "step": 9780
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.426353249640441e-06,
+      "loss": 0.7089,
+      "step": 9781
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.424404863764805e-06,
+      "loss": 0.7488,
+      "step": 9782
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.422456697558876e-06,
+      "loss": 0.7055,
+      "step": 9783
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.420508751116193e-06,
+      "loss": 0.7495,
+      "step": 9784
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.41856102453027e-06,
+      "loss": 0.714,
+      "step": 9785
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4166135178946165e-06,
+      "loss": 0.737,
+      "step": 9786
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.41466623130273e-06,
+      "loss": 0.7653,
+      "step": 9787
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.412719164848095e-06,
+      "loss": 0.7302,
+      "step": 9788
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4107723186241935e-06,
+      "loss": 0.4237,
+      "step": 9789
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.408825692724489e-06,
+      "loss": 0.4243,
+      "step": 9790
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.406879287242438e-06,
+      "loss": 0.7446,
+      "step": 9791
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.404933102271483e-06,
+      "loss": 0.41,
+      "step": 9792
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.402987137905062e-06,
+      "loss": 0.7673,
+      "step": 9793
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.401041394236595e-06,
+      "loss": 0.7006,
+      "step": 9794
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.399095871359499e-06,
+      "loss": 0.7143,
+      "step": 9795
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.397150569367169e-06,
+      "loss": 0.7178,
+      "step": 9796
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.395205488353006e-06,
+      "loss": 0.731,
+      "step": 9797
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.393260628410389e-06,
+      "loss": 0.6955,
+      "step": 9798
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.391315989632687e-06,
+      "loss": 0.693,
+      "step": 9799
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.389371572113262e-06,
+      "loss": 0.7603,
+      "step": 9800
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3874273759454575e-06,
+      "loss": 0.7228,
+      "step": 9801
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3854834012226195e-06,
+      "loss": 0.4117,
+      "step": 9802
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.383539648038077e-06,
+      "loss": 0.4017,
+      "step": 9803
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3815961164851425e-06,
+      "loss": 0.7076,
+      "step": 9804
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.379652806657126e-06,
+      "loss": 0.7485,
+      "step": 9805
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.37770971864732e-06,
+      "loss": 0.7146,
+      "step": 9806
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3757668525490195e-06,
+      "loss": 0.7106,
+      "step": 9807
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.373824208455489e-06,
+      "loss": 0.7712,
+      "step": 9808
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.371881786459998e-06,
+      "loss": 0.7306,
+      "step": 9809
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.369939586655794e-06,
+      "loss": 0.3728,
+      "step": 9810
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.367997609136132e-06,
+      "loss": 0.7816,
+      "step": 9811
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.366055853994236e-06,
+      "loss": 0.7754,
+      "step": 9812
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.36411432132333e-06,
+      "loss": 0.4001,
+      "step": 9813
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.362173011216623e-06,
+      "loss": 0.7694,
+      "step": 9814
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.360231923767315e-06,
+      "loss": 0.7366,
+      "step": 9815
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.358291059068601e-06,
+      "loss": 0.7142,
+      "step": 9816
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.356350417213656e-06,
+      "loss": 0.7569,
+      "step": 9817
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.354409998295648e-06,
+      "loss": 0.7429,
+      "step": 9818
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3524698024077315e-06,
+      "loss": 0.7323,
+      "step": 9819
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.350529829643062e-06,
+      "loss": 0.771,
+      "step": 9820
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.348590080094771e-06,
+      "loss": 0.7224,
+      "step": 9821
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.346650553855987e-06,
+      "loss": 0.7176,
+      "step": 9822
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.344711251019813e-06,
+      "loss": 0.7394,
+      "step": 9823
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.342772171679364e-06,
+      "loss": 0.4247,
+      "step": 9824
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.340833315927734e-06,
+      "loss": 0.7,
+      "step": 9825
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3388946838580005e-06,
+      "loss": 0.7576,
+      "step": 9826
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.336956275563237e-06,
+      "loss": 0.3909,
+      "step": 9827
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.335018091136499e-06,
+      "loss": 0.7302,
+      "step": 9828
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.333080130670848e-06,
+      "loss": 0.708,
+      "step": 9829
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3311423942593165e-06,
+      "loss": 0.7638,
+      "step": 9830
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.329204881994936e-06,
+      "loss": 0.7843,
+      "step": 9831
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.327267593970722e-06,
+      "loss": 0.3946,
+      "step": 9832
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.325330530279678e-06,
+      "loss": 0.7192,
+      "step": 9833
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.323393691014811e-06,
+      "loss": 0.7906,
+      "step": 9834
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.3214570762690985e-06,
+      "loss": 0.7026,
+      "step": 9835
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.31952068613552e-06,
+      "loss": 0.7741,
+      "step": 9836
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.317584520707037e-06,
+      "loss": 0.7637,
+      "step": 9837
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.315648580076602e-06,
+      "loss": 0.7326,
+      "step": 9838
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.313712864337162e-06,
+      "loss": 0.7326,
+      "step": 9839
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.311777373581643e-06,
+      "loss": 0.6898,
+      "step": 9840
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.309842107902969e-06,
+      "loss": 0.7197,
+      "step": 9841
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.307907067394046e-06,
+      "loss": 0.7543,
+      "step": 9842
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.30597225214778e-06,
+      "loss": 0.4405,
+      "step": 9843
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.304037662257059e-06,
+      "loss": 0.361,
+      "step": 9844
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.302103297814757e-06,
+      "loss": 0.7776,
+      "step": 9845
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.300169158913742e-06,
+      "loss": 0.7928,
+      "step": 9846
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.298235245646867e-06,
+      "loss": 0.7391,
+      "step": 9847
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.296301558106984e-06,
+      "loss": 0.7469,
+      "step": 9848
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.294368096386925e-06,
+      "loss": 0.6644,
+      "step": 9849
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.292434860579511e-06,
+      "loss": 0.7234,
+      "step": 9850
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.290501850777554e-06,
+      "loss": 0.7196,
+      "step": 9851
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.288569067073864e-06,
+      "loss": 0.7268,
+      "step": 9852
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2866365095612225e-06,
+      "loss": 0.4249,
+      "step": 9853
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.284704178332414e-06,
+      "loss": 0.7893,
+      "step": 9854
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.282772073480203e-06,
+      "loss": 0.7419,
+      "step": 9855
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.280840195097355e-06,
+      "loss": 0.7481,
+      "step": 9856
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.278908543276615e-06,
+      "loss": 0.3721,
+      "step": 9857
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.276977118110719e-06,
+      "loss": 0.7278,
+      "step": 9858
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.275045919692392e-06,
+      "loss": 0.7634,
+      "step": 9859
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.273114948114346e-06,
+      "loss": 0.6974,
+      "step": 9860
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2711842034692935e-06,
+      "loss": 0.7072,
+      "step": 9861
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.269253685849922e-06,
+      "loss": 0.7582,
+      "step": 9862
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2673233953489155e-06,
+      "loss": 0.777,
+      "step": 9863
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.265393332058944e-06,
+      "loss": 0.3975,
+      "step": 9864
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.263463496072664e-06,
+      "loss": 0.7165,
+      "step": 9865
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.261533887482733e-06,
+      "loss": 0.7626,
+      "step": 9866
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.25960450638179e-06,
+      "loss": 0.7766,
+      "step": 9867
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.257675352862455e-06,
+      "loss": 0.7369,
+      "step": 9868
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.255746427017343e-06,
+      "loss": 0.4214,
+      "step": 9869
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.253817728939069e-06,
+      "loss": 0.7289,
+      "step": 9870
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.251889258720224e-06,
+      "loss": 0.7314,
+      "step": 9871
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.249961016453391e-06,
+      "loss": 0.7531,
+      "step": 9872
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.248033002231146e-06,
+      "loss": 0.7457,
+      "step": 9873
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.246105216146043e-06,
+      "loss": 0.7013,
+      "step": 9874
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.244177658290642e-06,
+      "loss": 0.7504,
+      "step": 9875
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.242250328757481e-06,
+      "loss": 0.779,
+      "step": 9876
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.240323227639088e-06,
+      "loss": 0.7509,
+      "step": 9877
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.23839635502798e-06,
+      "loss": 0.7254,
+      "step": 9878
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2364697110166625e-06,
+      "loss": 0.7266,
+      "step": 9879
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.234543295697638e-06,
+      "loss": 0.7223,
+      "step": 9880
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.232617109163389e-06,
+      "loss": 0.3813,
+      "step": 9881
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.230691151506389e-06,
+      "loss": 0.7558,
+      "step": 9882
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.228765422819101e-06,
+      "loss": 0.7393,
+      "step": 9883
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.226839923193978e-06,
+      "loss": 0.7321,
+      "step": 9884
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.224914652723461e-06,
+      "loss": 0.7707,
+      "step": 9885
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.22298961149998e-06,
+      "loss": 0.7361,
+      "step": 9886
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.221064799615951e-06,
+      "loss": 0.7334,
+      "step": 9887
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.219140217163791e-06,
+      "loss": 0.7236,
+      "step": 9888
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.217215864235891e-06,
+      "loss": 0.7289,
+      "step": 9889
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.215291740924637e-06,
+      "loss": 0.7261,
+      "step": 9890
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.213367847322408e-06,
+      "loss": 0.7458,
+      "step": 9891
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.211444183521561e-06,
+      "loss": 0.429,
+      "step": 9892
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2095207496144585e-06,
+      "loss": 0.7199,
+      "step": 9893
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.207597545693437e-06,
+      "loss": 0.7195,
+      "step": 9894
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2056745718508295e-06,
+      "loss": 0.3878,
+      "step": 9895
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2037518281789546e-06,
+      "loss": 0.7263,
+      "step": 9896
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.201829314770122e-06,
+      "loss": 0.7438,
+      "step": 9897
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.19990703171663e-06,
+      "loss": 0.6734,
+      "step": 9898
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.197984979110764e-06,
+      "loss": 0.772,
+      "step": 9899
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1960631570448015e-06,
+      "loss": 0.6998,
+      "step": 9900
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.194141565611004e-06,
+      "loss": 0.7496,
+      "step": 9901
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.19222020490163e-06,
+      "loss": 0.6904,
+      "step": 9902
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.19029907500892e-06,
+      "loss": 0.7274,
+      "step": 9903
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.188378176025106e-06,
+      "loss": 0.7566,
+      "step": 9904
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.186457508042407e-06,
+      "loss": 0.7228,
+      "step": 9905
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.18453707115303e-06,
+      "loss": 0.7528,
+      "step": 9906
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.18261686544918e-06,
+      "loss": 0.8153,
+      "step": 9907
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.18069689102304e-06,
+      "loss": 0.7369,
+      "step": 9908
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.178777147966787e-06,
+      "loss": 0.7437,
+      "step": 9909
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1768576363725806e-06,
+      "loss": 0.749,
+      "step": 9910
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.174938356332588e-06,
+      "loss": 0.7546,
+      "step": 9911
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.17301930793894e-06,
+      "loss": 0.3855,
+      "step": 9912
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.17110049128377e-06,
+      "loss": 0.7682,
+      "step": 9913
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1691819064591975e-06,
+      "loss": 0.4046,
+      "step": 9914
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.167263553557338e-06,
+      "loss": 0.3798,
+      "step": 9915
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.165345432670286e-06,
+      "loss": 0.758,
+      "step": 9916
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.163427543890128e-06,
+      "loss": 0.7068,
+      "step": 9917
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1615098873089405e-06,
+      "loss": 0.7668,
+      "step": 9918
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1595924630187855e-06,
+      "loss": 0.3832,
+      "step": 9919
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1576752711117215e-06,
+      "loss": 0.7097,
+      "step": 9920
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1557583116797904e-06,
+      "loss": 0.7443,
+      "step": 9921
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.15384158481502e-06,
+      "loss": 0.4091,
+      "step": 9922
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.151925090609434e-06,
+      "loss": 0.712,
+      "step": 9923
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.150008829155035e-06,
+      "loss": 0.6961,
+      "step": 9924
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.14809280054383e-06,
+      "loss": 0.721,
+      "step": 9925
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.146177004867805e-06,
+      "loss": 0.712,
+      "step": 9926
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.144261442218927e-06,
+      "loss": 0.7427,
+      "step": 9927
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.142346112689161e-06,
+      "loss": 0.3846,
+      "step": 9928
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1404310163704674e-06,
+      "loss": 0.7156,
+      "step": 9929
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1385161533547845e-06,
+      "loss": 0.6916,
+      "step": 9930
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.136601523734043e-06,
+      "loss": 0.7644,
+      "step": 9931
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.134687127600162e-06,
+      "loss": 0.7572,
+      "step": 9932
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1327729650450455e-06,
+      "loss": 0.7562,
+      "step": 9933
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1308590361606e-06,
+      "loss": 0.7665,
+      "step": 9934
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1289453410387045e-06,
+      "loss": 0.7171,
+      "step": 9935
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.127031879771237e-06,
+      "loss": 0.4162,
+      "step": 9936
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.125118652450059e-06,
+      "loss": 0.7488,
+      "step": 9937
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.123205659167019e-06,
+      "loss": 0.7084,
+      "step": 9938
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.121292900013963e-06,
+      "loss": 0.7146,
+      "step": 9939
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.119380375082722e-06,
+      "loss": 0.3887,
+      "step": 9940
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.117468084465111e-06,
+      "loss": 0.4277,
+      "step": 9941
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.115556028252937e-06,
+      "loss": 0.8097,
+      "step": 9942
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.113644206537997e-06,
+      "loss": 0.7105,
+      "step": 9943
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.111732619412077e-06,
+      "loss": 0.7556,
+      "step": 9944
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.109821266966947e-06,
+      "loss": 0.416,
+      "step": 9945
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.107910149294367e-06,
+      "loss": 0.7309,
+      "step": 9946
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.105999266486096e-06,
+      "loss": 0.4223,
+      "step": 9947
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.104088618633869e-06,
+      "loss": 0.7589,
+      "step": 9948
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.102178205829414e-06,
+      "loss": 0.7123,
+      "step": 9949
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.10026802816445e-06,
+      "loss": 0.7573,
+      "step": 9950
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.098358085730676e-06,
+      "loss": 0.721,
+      "step": 9951
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.096448378619796e-06,
+      "loss": 0.7208,
+      "step": 9952
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.094538906923487e-06,
+      "loss": 0.4017,
+      "step": 9953
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.092629670733424e-06,
+      "loss": 0.7634,
+      "step": 9954
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.090720670141267e-06,
+      "loss": 0.7191,
+      "step": 9955
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.088811905238663e-06,
+      "loss": 0.7148,
+      "step": 9956
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.086903376117251e-06,
+      "loss": 0.6492,
+      "step": 9957
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.084995082868658e-06,
+      "loss": 0.6952,
+      "step": 9958
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.083087025584499e-06,
+      "loss": 0.7552,
+      "step": 9959
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.081179204356375e-06,
+      "loss": 0.697,
+      "step": 9960
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.079271619275883e-06,
+      "loss": 0.3918,
+      "step": 9961
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.077364270434604e-06,
+      "loss": 0.4301,
+      "step": 9962
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0754571579241064e-06,
+      "loss": 0.3552,
+      "step": 9963
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.073550281835948e-06,
+      "loss": 0.4635,
+      "step": 9964
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.071643642261674e-06,
+      "loss": 0.781,
+      "step": 9965
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.069737239292827e-06,
+      "loss": 0.7439,
+      "step": 9966
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.067831073020928e-06,
+      "loss": 0.7465,
+      "step": 9967
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.065925143537489e-06,
+      "loss": 0.3767,
+      "step": 9968
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.064019450934008e-06,
+      "loss": 0.6781,
+      "step": 9969
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.062113995301985e-06,
+      "loss": 0.77,
+      "step": 9970
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.060208776732897e-06,
+      "loss": 0.6864,
+      "step": 9971
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.058303795318205e-06,
+      "loss": 0.7574,
+      "step": 9972
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.056399051149363e-06,
+      "loss": 0.7071,
+      "step": 9973
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.054494544317827e-06,
+      "loss": 0.7779,
+      "step": 9974
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.052590274915026e-06,
+      "loss": 0.4188,
+      "step": 9975
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.050686243032379e-06,
+      "loss": 0.7649,
+      "step": 9976
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0487824487613e-06,
+      "loss": 0.7669,
+      "step": 9977
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.046878892193184e-06,
+      "loss": 0.4081,
+      "step": 9978
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.044975573419424e-06,
+      "loss": 0.7381,
+      "step": 9979
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.043072492531394e-06,
+      "loss": 0.7293,
+      "step": 9980
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0411696496204606e-06,
+      "loss": 0.4392,
+      "step": 9981
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.039267044777976e-06,
+      "loss": 0.7418,
+      "step": 9982
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.037364678095278e-06,
+      "loss": 0.7145,
+      "step": 9983
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.035462549663705e-06,
+      "loss": 0.7193,
+      "step": 9984
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.033560659574573e-06,
+      "loss": 0.6402,
+      "step": 9985
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.031659007919194e-06,
+      "loss": 0.7947,
+      "step": 9986
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.029757594788852e-06,
+      "loss": 0.7265,
+      "step": 9987
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.027856420274845e-06,
+      "loss": 0.7785,
+      "step": 9988
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.025955484468441e-06,
+      "loss": 0.7398,
+      "step": 9989
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0240547874609014e-06,
+      "loss": 0.7321,
+      "step": 9990
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.022154329343478e-06,
+      "loss": 0.7366,
+      "step": 9991
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.020254110207408e-06,
+      "loss": 0.7482,
+      "step": 9992
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.018354130143923e-06,
+      "loss": 0.6704,
+      "step": 9993
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.016454389244237e-06,
+      "loss": 0.7526,
+      "step": 9994
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.014554887599554e-06,
+      "loss": 0.7615,
+      "step": 9995
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.012655625301068e-06,
+      "loss": 0.7172,
+      "step": 9996
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.010756602439958e-06,
+      "loss": 0.7284,
+      "step": 9997
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0088578191074e-06,
+      "loss": 0.7298,
+      "step": 9998
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.006959275394547e-06,
+      "loss": 0.7593,
+      "step": 9999
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.00506097139255e-06,
+      "loss": 0.6757,
+      "step": 10000
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.003162907192543e-06,
+      "loss": 0.4502,
+      "step": 10001
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.001265082885651e-06,
+      "loss": 0.7348,
+      "step": 10002
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.999367498562985e-06,
+      "loss": 0.6993,
+      "step": 10003
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.997470154315647e-06,
+      "loss": 0.436,
+      "step": 10004
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9955730502347225e-06,
+      "loss": 0.4371,
+      "step": 10005
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.993676186411298e-06,
+      "loss": 0.7457,
+      "step": 10006
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.991779562936435e-06,
+      "loss": 0.7188,
+      "step": 10007
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.989883179901189e-06,
+      "loss": 0.4052,
+      "step": 10008
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.987987037396604e-06,
+      "loss": 0.7779,
+      "step": 10009
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.986091135513708e-06,
+      "loss": 0.7441,
+      "step": 10010
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.984195474343527e-06,
+      "loss": 0.7046,
+      "step": 10011
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.982300053977068e-06,
+      "loss": 0.6991,
+      "step": 10012
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.980404874505329e-06,
+      "loss": 0.7503,
+      "step": 10013
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.978509936019293e-06,
+      "loss": 0.7299,
+      "step": 10014
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9766152386099335e-06,
+      "loss": 0.7192,
+      "step": 10015
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.974720782368221e-06,
+      "loss": 0.7244,
+      "step": 10016
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.972826567385097e-06,
+      "loss": 0.7295,
+      "step": 10017
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9709325937515055e-06,
+      "loss": 0.7397,
+      "step": 10018
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.969038861558367e-06,
+      "loss": 0.7136,
+      "step": 10019
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.96714537089661e-06,
+      "loss": 0.4154,
+      "step": 10020
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.965252121857131e-06,
+      "loss": 0.7447,
+      "step": 10021
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.963359114530826e-06,
+      "loss": 0.6895,
+      "step": 10022
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.961466349008575e-06,
+      "loss": 0.726,
+      "step": 10023
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.959573825381243e-06,
+      "loss": 0.3959,
+      "step": 10024
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.957681543739697e-06,
+      "loss": 0.3938,
+      "step": 10025
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.955789504174782e-06,
+      "loss": 0.4209,
+      "step": 10026
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.953897706777329e-06,
+      "loss": 0.7256,
+      "step": 10027
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.952006151638162e-06,
+      "loss": 0.7388,
+      "step": 10028
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.95011483884809e-06,
+      "loss": 0.7218,
+      "step": 10029
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.948223768497925e-06,
+      "loss": 0.74,
+      "step": 10030
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.946332940678442e-06,
+      "loss": 0.7643,
+      "step": 10031
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.94444235548042e-06,
+      "loss": 0.7594,
+      "step": 10032
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.94255201299463e-06,
+      "loss": 0.7321,
+      "step": 10033
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.940661913311822e-06,
+      "loss": 0.7697,
+      "step": 10034
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.938772056522738e-06,
+      "loss": 0.7647,
+      "step": 10035
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.936882442718108e-06,
+      "loss": 0.7091,
+      "step": 10036
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.934993071988648e-06,
+      "loss": 0.7242,
+      "step": 10037
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.933103944425069e-06,
+      "loss": 0.7538,
+      "step": 10038
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.931215060118066e-06,
+      "loss": 0.4214,
+      "step": 10039
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.92932641915832e-06,
+      "loss": 0.7166,
+      "step": 10040
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.927438021636504e-06,
+      "loss": 0.7195,
+      "step": 10041
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9255498676432735e-06,
+      "loss": 0.723,
+      "step": 10042
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.923661957269284e-06,
+      "loss": 0.6995,
+      "step": 10043
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.921774290605171e-06,
+      "loss": 0.7127,
+      "step": 10044
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.919886867741561e-06,
+      "loss": 0.7573,
+      "step": 10045
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.917999688769056e-06,
+      "loss": 0.7522,
+      "step": 10046
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.916112753778269e-06,
+      "loss": 0.3954,
+      "step": 10047
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.914226062859787e-06,
+      "loss": 0.7361,
+      "step": 10048
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.912339616104187e-06,
+      "loss": 0.7189,
+      "step": 10049
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.910453413602036e-06,
+      "loss": 0.6998,
+      "step": 10050
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.908567455443885e-06,
+      "loss": 0.736,
+      "step": 10051
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.906681741720285e-06,
+      "loss": 0.6909,
+      "step": 10052
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.904796272521761e-06,
+      "loss": 0.7881,
+      "step": 10053
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.902911047938836e-06,
+      "loss": 0.7551,
+      "step": 10054
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.901026068062015e-06,
+      "loss": 0.6937,
+      "step": 10055
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8991413329817915e-06,
+      "loss": 0.7347,
+      "step": 10056
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.897256842788657e-06,
+      "loss": 0.4182,
+      "step": 10057
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.895372597573079e-06,
+      "loss": 0.7098,
+      "step": 10058
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.893488597425521e-06,
+      "loss": 0.7121,
+      "step": 10059
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.891604842436429e-06,
+      "loss": 0.6866,
+      "step": 10060
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.889721332696242e-06,
+      "loss": 0.7486,
+      "step": 10061
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8878380682953854e-06,
+      "loss": 0.745,
+      "step": 10062
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.885955049324272e-06,
+      "loss": 0.7373,
+      "step": 10063
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8840722758733e-06,
+      "loss": 0.3958,
+      "step": 10064
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8821897480328675e-06,
+      "loss": 0.7411,
+      "step": 10065
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.880307465893348e-06,
+      "loss": 0.7102,
+      "step": 10066
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.878425429545109e-06,
+      "loss": 0.6607,
+      "step": 10067
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.876543639078506e-06,
+      "loss": 0.7146,
+      "step": 10068
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.874662094583875e-06,
+      "loss": 0.7353,
+      "step": 10069
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.872780796151558e-06,
+      "loss": 0.4354,
+      "step": 10070
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.870899743871869e-06,
+      "loss": 0.7205,
+      "step": 10071
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.869018937835114e-06,
+      "loss": 0.7797,
+      "step": 10072
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8671383781315904e-06,
+      "loss": 0.7632,
+      "step": 10073
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.865258064851579e-06,
+      "loss": 0.775,
+      "step": 10074
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8633779980853605e-06,
+      "loss": 0.4604,
+      "step": 10075
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.861498177923186e-06,
+      "loss": 0.4081,
+      "step": 10076
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.859618604455306e-06,
+      "loss": 0.76,
+      "step": 10077
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.857739277771955e-06,
+      "loss": 0.7623,
+      "step": 10078
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.855860197963363e-06,
+      "loss": 0.7718,
+      "step": 10079
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.853981365119739e-06,
+      "loss": 0.7495,
+      "step": 10080
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.852102779331286e-06,
+      "loss": 0.7482,
+      "step": 10081
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.85022444068819e-06,
+      "loss": 0.7246,
+      "step": 10082
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.848346349280628e-06,
+      "loss": 0.7355,
+      "step": 10083
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.84646850519877e-06,
+      "loss": 0.4212,
+      "step": 10084
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.844590908532767e-06,
+      "loss": 0.7249,
+      "step": 10085
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8427135593727605e-06,
+      "loss": 0.7524,
+      "step": 10086
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.84083645780888e-06,
+      "loss": 0.3837,
+      "step": 10087
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.838959603931237e-06,
+      "loss": 0.4204,
+      "step": 10088
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.837082997829949e-06,
+      "loss": 0.4378,
+      "step": 10089
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.835206639595108e-06,
+      "loss": 0.7367,
+      "step": 10090
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8333305293167844e-06,
+      "loss": 0.4164,
+      "step": 10091
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.831454667085059e-06,
+      "loss": 0.7813,
+      "step": 10092
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8295790529899885e-06,
+      "loss": 0.7847,
+      "step": 10093
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8277036871216186e-06,
+      "loss": 0.7292,
+      "step": 10094
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.825828569569984e-06,
+      "loss": 0.7723,
+      "step": 10095
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.823953700425101e-06,
+      "loss": 0.7544,
+      "step": 10096
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.82207907977699e-06,
+      "loss": 0.7331,
+      "step": 10097
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.820204707715646e-06,
+      "loss": 0.4068,
+      "step": 10098
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.818330584331054e-06,
+      "loss": 0.7758,
+      "step": 10099
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.81645670971319e-06,
+      "loss": 0.6953,
+      "step": 10100
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.814583083952013e-06,
+      "loss": 0.7172,
+      "step": 10101
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8127097071374805e-06,
+      "loss": 0.713,
+      "step": 10102
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.81083657935953e-06,
+      "loss": 0.7049,
+      "step": 10103
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.808963700708086e-06,
+      "loss": 0.7054,
+      "step": 10104
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.807091071273064e-06,
+      "loss": 0.7344,
+      "step": 10105
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.805218691144369e-06,
+      "loss": 0.7968,
+      "step": 10106
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.803346560411889e-06,
+      "loss": 0.432,
+      "step": 10107
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.801474679165505e-06,
+      "loss": 0.759,
+      "step": 10108
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.799603047495085e-06,
+      "loss": 0.7524,
+      "step": 10109
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7977316654904785e-06,
+      "loss": 0.3907,
+      "step": 10110
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7958605332415365e-06,
+      "loss": 0.7026,
+      "step": 10111
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.793989650838087e-06,
+      "loss": 0.3819,
+      "step": 10112
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.792119018369949e-06,
+      "loss": 0.7561,
+      "step": 10113
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.790248635926929e-06,
+      "loss": 0.6957,
+      "step": 10114
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.78837850359882e-06,
+      "loss": 0.7286,
+      "step": 10115
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.786508621475412e-06,
+      "loss": 0.7512,
+      "step": 10116
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.784638989646471e-06,
+      "loss": 0.415,
+      "step": 10117
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.782769608201758e-06,
+      "loss": 0.4201,
+      "step": 10118
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7809004772310166e-06,
+      "loss": 0.7692,
+      "step": 10119
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.779031596823991e-06,
+      "loss": 0.6935,
+      "step": 10120
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.777162967070394e-06,
+      "loss": 0.7004,
+      "step": 10121
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.775294588059942e-06,
+      "loss": 0.4087,
+      "step": 10122
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.773426459882327e-06,
+      "loss": 0.7135,
+      "step": 10123
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.771558582627247e-06,
+      "loss": 0.7331,
+      "step": 10124
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.769690956384371e-06,
+      "loss": 0.7489,
+      "step": 10125
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.767823581243363e-06,
+      "loss": 0.7057,
+      "step": 10126
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.765956457293871e-06,
+      "loss": 0.3996,
+      "step": 10127
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.764089584625534e-06,
+      "loss": 0.7624,
+      "step": 10128
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.762222963327983e-06,
+      "loss": 0.706,
+      "step": 10129
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.760356593490832e-06,
+      "loss": 0.7344,
+      "step": 10130
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.75849047520368e-06,
+      "loss": 0.7699,
+      "step": 10131
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.756624608556119e-06,
+      "loss": 0.4098,
+      "step": 10132
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.754758993637725e-06,
+      "loss": 0.7335,
+      "step": 10133
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.75289363053807e-06,
+      "loss": 0.7024,
+      "step": 10134
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.751028519346709e-06,
+      "loss": 0.7475,
+      "step": 10135
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.749163660153176e-06,
+      "loss": 0.7214,
+      "step": 10136
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.747299053047001e-06,
+      "loss": 0.3811,
+      "step": 10137
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.74543469811771e-06,
+      "loss": 0.7318,
+      "step": 10138
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.743570595454805e-06,
+      "loss": 0.4089,
+      "step": 10139
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.74170674514778e-06,
+      "loss": 0.7386,
+      "step": 10140
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7398431472861166e-06,
+      "loss": 0.6975,
+      "step": 10141
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7379798019592785e-06,
+      "loss": 0.7562,
+      "step": 10142
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.736116709256733e-06,
+      "loss": 0.4202,
+      "step": 10143
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.73425386926792e-06,
+      "loss": 0.718,
+      "step": 10144
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.732391282082274e-06,
+      "loss": 0.3759,
+      "step": 10145
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7305289477892136e-06,
+      "loss": 0.7416,
+      "step": 10146
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.728666866478147e-06,
+      "loss": 0.7468,
+      "step": 10147
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.726805038238475e-06,
+      "loss": 0.413,
+      "step": 10148
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.724943463159586e-06,
+      "loss": 0.7157,
+      "step": 10149
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.723082141330837e-06,
+      "loss": 0.8013,
+      "step": 10150
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.721221072841602e-06,
+      "loss": 0.718,
+      "step": 10151
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.719360257781225e-06,
+      "loss": 0.7735,
+      "step": 10152
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.717499696239041e-06,
+      "loss": 0.7242,
+      "step": 10153
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.715639388304374e-06,
+      "loss": 0.6943,
+      "step": 10154
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.713779334066533e-06,
+      "loss": 0.7378,
+      "step": 10155
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.711919533614822e-06,
+      "loss": 0.7588,
+      "step": 10156
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.710059987038526e-06,
+      "loss": 0.3951,
+      "step": 10157
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.708200694426922e-06,
+      "loss": 0.4199,
+      "step": 10158
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7063416558692694e-06,
+      "loss": 0.4082,
+      "step": 10159
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7044828714548176e-06,
+      "loss": 0.7149,
+      "step": 10160
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7026243412728104e-06,
+      "loss": 0.7342,
+      "step": 10161
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.700766065412472e-06,
+      "loss": 0.7403,
+      "step": 10162
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.698908043963014e-06,
+      "loss": 0.7457,
+      "step": 10163
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.697050277013641e-06,
+      "loss": 0.7207,
+      "step": 10164
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.695192764653541e-06,
+      "loss": 0.7399,
+      "step": 10165
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.693335506971891e-06,
+      "loss": 0.357,
+      "step": 10166
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.691478504057857e-06,
+      "loss": 0.392,
+      "step": 10167
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.689621756000591e-06,
+      "loss": 0.735,
+      "step": 10168
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.68776526288923e-06,
+      "loss": 0.4282,
+      "step": 10169
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.68590902481291e-06,
+      "loss": 0.6708,
+      "step": 10170
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.684053041860743e-06,
+      "loss": 0.408,
+      "step": 10171
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.682197314121835e-06,
+      "loss": 0.3991,
+      "step": 10172
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6803418416852754e-06,
+      "loss": 0.7156,
+      "step": 10173
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.678486624640139e-06,
+      "loss": 0.3947,
+      "step": 10174
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.676631663075502e-06,
+      "loss": 0.7622,
+      "step": 10175
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.674776957080417e-06,
+      "loss": 0.4555,
+      "step": 10176
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6729225067439235e-06,
+      "loss": 0.3881,
+      "step": 10177
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.671068312155054e-06,
+      "loss": 0.7491,
+      "step": 10178
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.669214373402824e-06,
+      "loss": 0.4276,
+      "step": 10179
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6673606905762415e-06,
+      "loss": 0.7518,
+      "step": 10180
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.665507263764299e-06,
+      "loss": 0.758,
+      "step": 10181
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.663654093055976e-06,
+      "loss": 0.77,
+      "step": 10182
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.661801178540245e-06,
+      "loss": 0.7611,
+      "step": 10183
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6599485203060625e-06,
+      "loss": 0.715,
+      "step": 10184
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.65809611844237e-06,
+      "loss": 0.4393,
+      "step": 10185
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6562439730381e-06,
+      "loss": 0.731,
+      "step": 10186
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.65439208418217e-06,
+      "loss": 0.4084,
+      "step": 10187
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6525404519634934e-06,
+      "loss": 0.428,
+      "step": 10188
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.650689076470962e-06,
+      "loss": 0.716,
+      "step": 10189
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.648837957793459e-06,
+      "loss": 0.7504,
+      "step": 10190
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.646987096019852e-06,
+      "loss": 0.6638,
+      "step": 10191
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.645136491238998e-06,
+      "loss": 0.6761,
+      "step": 10192
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.64328614353975e-06,
+      "loss": 0.399,
+      "step": 10193
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6414360530109415e-06,
+      "loss": 0.6452,
+      "step": 10194
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.639586219741385e-06,
+      "loss": 0.7267,
+      "step": 10195
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.637736643819888e-06,
+      "loss": 0.7175,
+      "step": 10196
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.635887325335257e-06,
+      "loss": 0.7307,
+      "step": 10197
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.63403826437627e-06,
+      "loss": 0.7117,
+      "step": 10198
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6321894610316994e-06,
+      "loss": 0.4324,
+      "step": 10199
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.630340915390304e-06,
+      "loss": 0.7293,
+      "step": 10200
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.628492627540828e-06,
+      "loss": 0.7626,
+      "step": 10201
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.626644597572013e-06,
+      "loss": 0.6875,
+      "step": 10202
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6247968255725765e-06,
+      "loss": 0.7667,
+      "step": 10203
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6229493116312285e-06,
+      "loss": 0.3786,
+      "step": 10204
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.621102055836666e-06,
+      "loss": 0.4066,
+      "step": 10205
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.619255058277571e-06,
+      "loss": 0.457,
+      "step": 10206
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.617408319042622e-06,
+      "loss": 0.7189,
+      "step": 10207
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.615561838220477e-06,
+      "loss": 0.416,
+      "step": 10208
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.613715615899783e-06,
+      "loss": 0.7706,
+      "step": 10209
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.611869652169176e-06,
+      "loss": 0.7624,
+      "step": 10210
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.610023947117277e-06,
+      "loss": 0.3657,
+      "step": 10211
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6081785008326975e-06,
+      "loss": 0.6704,
+      "step": 10212
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.606333313404037e-06,
+      "loss": 0.6836,
+      "step": 10213
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.604488384919874e-06,
+      "loss": 0.3816,
+      "step": 10214
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.602643715468792e-06,
+      "loss": 0.4221,
+      "step": 10215
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6007993051393475e-06,
+      "loss": 0.7587,
+      "step": 10216
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5989551540200875e-06,
+      "loss": 0.7486,
+      "step": 10217
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.597111262199549e-06,
+      "loss": 0.7429,
+      "step": 10218
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5952676297662515e-06,
+      "loss": 0.4347,
+      "step": 10219
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.593424256808713e-06,
+      "loss": 0.7545,
+      "step": 10220
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.591581143415429e-06,
+      "loss": 0.7274,
+      "step": 10221
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.589738289674884e-06,
+      "loss": 0.7435,
+      "step": 10222
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5878956956755536e-06,
+      "loss": 0.4137,
+      "step": 10223
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.586053361505898e-06,
+      "loss": 0.4313,
+      "step": 10224
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.584211287254364e-06,
+      "loss": 0.7283,
+      "step": 10225
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.58236947300939e-06,
+      "loss": 0.74,
+      "step": 10226
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5805279188594e-06,
+      "loss": 0.7594,
+      "step": 10227
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5786866248927995e-06,
+      "loss": 0.7516,
+      "step": 10228
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5768455911979955e-06,
+      "loss": 0.6993,
+      "step": 10229
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.575004817863371e-06,
+      "loss": 0.7471,
+      "step": 10230
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.573164304977299e-06,
+      "loss": 0.427,
+      "step": 10231
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.571324052628141e-06,
+      "loss": 0.4099,
+      "step": 10232
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.569484060904241e-06,
+      "loss": 0.8015,
+      "step": 10233
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5676443298939445e-06,
+      "loss": 0.6828,
+      "step": 10234
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.565804859685569e-06,
+      "loss": 0.4658,
+      "step": 10235
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.563965650367428e-06,
+      "loss": 0.6942,
+      "step": 10236
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5621267020278195e-06,
+      "loss": 0.7357,
+      "step": 10237
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.560288014755024e-06,
+      "loss": 0.7173,
+      "step": 10238
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.558449588637328e-06,
+      "loss": 0.7391,
+      "step": 10239
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.556611423762981e-06,
+      "loss": 0.4167,
+      "step": 10240
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.55477352022023e-06,
+      "loss": 0.7333,
+      "step": 10241
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.552935878097321e-06,
+      "loss": 0.647,
+      "step": 10242
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.551098497482472e-06,
+      "loss": 0.7567,
+      "step": 10243
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.549261378463894e-06,
+      "loss": 0.734,
+      "step": 10244
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.547424521129785e-06,
+      "loss": 0.7463,
+      "step": 10245
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.545587925568328e-06,
+      "loss": 0.7122,
+      "step": 10246
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.543751591867702e-06,
+      "loss": 0.7395,
+      "step": 10247
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.541915520116065e-06,
+      "loss": 0.7355,
+      "step": 10248
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.540079710401563e-06,
+      "loss": 0.7619,
+      "step": 10249
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.538244162812335e-06,
+      "loss": 0.6636,
+      "step": 10250
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5364088774364975e-06,
+      "loss": 0.4199,
+      "step": 10251
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.534573854362168e-06,
+      "loss": 0.7225,
+      "step": 10252
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.532739093677442e-06,
+      "loss": 0.7089,
+      "step": 10253
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.530904595470406e-06,
+      "loss": 0.3797,
+      "step": 10254
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.529070359829123e-06,
+      "loss": 0.7622,
+      "step": 10255
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.527236386841664e-06,
+      "loss": 0.7636,
+      "step": 10256
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5254026765960705e-06,
+      "loss": 0.756,
+      "step": 10257
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.523569229180379e-06,
+      "loss": 0.7299,
+      "step": 10258
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.521736044682612e-06,
+      "loss": 0.3887,
+      "step": 10259
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.519903123190773e-06,
+      "loss": 0.3993,
+      "step": 10260
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.518070464792867e-06,
+      "loss": 0.7229,
+      "step": 10261
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.516238069576876e-06,
+      "loss": 0.4209,
+      "step": 10262
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5144059376307695e-06,
+      "loss": 0.4425,
+      "step": 10263
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5125740690425054e-06,
+      "loss": 0.379,
+      "step": 10264
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5107424639000295e-06,
+      "loss": 0.6917,
+      "step": 10265
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.50891112229128e-06,
+      "loss": 0.699,
+      "step": 10266
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.507080044304175e-06,
+      "loss": 0.3694,
+      "step": 10267
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.505249230026623e-06,
+      "loss": 0.7271,
+      "step": 10268
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.503418679546519e-06,
+      "loss": 0.7244,
+      "step": 10269
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.501588392951745e-06,
+      "loss": 0.7259,
+      "step": 10270
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.4997583703301735e-06,
+      "loss": 0.3732,
+      "step": 10271
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.497928611769661e-06,
+      "loss": 0.727,
+      "step": 10272
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.496099117358047e-06,
+      "loss": 0.7309,
+      "step": 10273
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4942698871831735e-06,
+      "loss": 0.7231,
+      "step": 10274
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.492440921332856e-06,
+      "loss": 0.781,
+      "step": 10275
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4906122198949e-06,
+      "loss": 0.7643,
+      "step": 10276
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4887837829571e-06,
+      "loss": 0.3771,
+      "step": 10277
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.486955610607235e-06,
+      "loss": 0.7339,
+      "step": 10278
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.485127702933078e-06,
+      "loss": 0.7348,
+      "step": 10279
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.483300060022385e-06,
+      "loss": 0.7365,
+      "step": 10280
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.481472681962897e-06,
+      "loss": 0.7636,
+      "step": 10281
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.479645568842347e-06,
+      "loss": 0.7539,
+      "step": 10282
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.477818720748449e-06,
+      "loss": 0.6972,
+      "step": 10283
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.475992137768912e-06,
+      "loss": 0.7408,
+      "step": 10284
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.474165819991425e-06,
+      "loss": 0.3988,
+      "step": 10285
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.472339767503671e-06,
+      "loss": 0.437,
+      "step": 10286
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4705139803933105e-06,
+      "loss": 0.4378,
+      "step": 10287
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.468688458748006e-06,
+      "loss": 0.696,
+      "step": 10288
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.466863202655396e-06,
+      "loss": 0.7495,
+      "step": 10289
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.465038212203109e-06,
+      "loss": 0.7798,
+      "step": 10290
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.46321348747876e-06,
+      "loss": 0.3969,
+      "step": 10291
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4613890285699504e-06,
+      "loss": 0.3658,
+      "step": 10292
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.459564835564275e-06,
+      "loss": 0.7336,
+      "step": 10293
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4577409085493105e-06,
+      "loss": 0.7357,
+      "step": 10294
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.45591724761262e-06,
+      "loss": 0.6779,
+      "step": 10295
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.454093852841756e-06,
+      "loss": 0.6751,
+      "step": 10296
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.452270724324257e-06,
+      "loss": 0.3798,
+      "step": 10297
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4504478621476564e-06,
+      "loss": 0.77,
+      "step": 10298
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.448625266399459e-06,
+      "loss": 0.737,
+      "step": 10299
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4468029371671665e-06,
+      "loss": 0.7782,
+      "step": 10300
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.444980874538271e-06,
+      "loss": 0.3691,
+      "step": 10301
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.443159078600249e-06,
+      "loss": 0.7367,
+      "step": 10302
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4413375494405595e-06,
+      "loss": 0.686,
+      "step": 10303
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.439516287146654e-06,
+      "loss": 0.753,
+      "step": 10304
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.437695291805966e-06,
+      "loss": 0.7105,
+      "step": 10305
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.435874563505926e-06,
+      "loss": 0.7325,
+      "step": 10306
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.434054102333942e-06,
+      "loss": 0.4091,
+      "step": 10307
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.432233908377413e-06,
+      "loss": 0.7478,
+      "step": 10308
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4304139817237225e-06,
+      "loss": 0.7504,
+      "step": 10309
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.428594322460243e-06,
+      "loss": 0.6914,
+      "step": 10310
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.426774930674339e-06,
+      "loss": 0.7199,
+      "step": 10311
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.424955806453356e-06,
+      "loss": 0.417,
+      "step": 10312
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.423136949884631e-06,
+      "loss": 0.7279,
+      "step": 10313
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.421318361055474e-06,
+      "loss": 0.664,
+      "step": 10314
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.419500040053206e-06,
+      "loss": 0.4265,
+      "step": 10315
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.417681986965119e-06,
+      "loss": 0.4161,
+      "step": 10316
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.415864201878495e-06,
+      "loss": 0.7009,
+      "step": 10317
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.414046684880603e-06,
+      "loss": 0.7075,
+      "step": 10318
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.412229436058699e-06,
+      "loss": 0.7359,
+      "step": 10319
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4104124555000325e-06,
+      "loss": 0.4209,
+      "step": 10320
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.408595743291833e-06,
+      "loss": 0.7565,
+      "step": 10321
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.406779299521318e-06,
+      "loss": 0.7315,
+      "step": 10322
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.404963124275693e-06,
+      "loss": 0.7357,
+      "step": 10323
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.403147217642148e-06,
+      "loss": 0.7041,
+      "step": 10324
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.40133157970787e-06,
+      "loss": 0.734,
+      "step": 10325
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.399516210560021e-06,
+      "loss": 0.7556,
+      "step": 10326
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.397701110285756e-06,
+      "loss": 0.7224,
+      "step": 10327
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.395886278972216e-06,
+      "loss": 0.7068,
+      "step": 10328
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.394071716706528e-06,
+      "loss": 0.423,
+      "step": 10329
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.392257423575811e-06,
+      "loss": 0.6848,
+      "step": 10330
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.390443399667164e-06,
+      "loss": 0.7557,
+      "step": 10331
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3886296450676725e-06,
+      "loss": 0.8028,
+      "step": 10332
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.386816159864423e-06,
+      "loss": 0.4148,
+      "step": 10333
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.385002944144473e-06,
+      "loss": 0.7859,
+      "step": 10334
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.383189997994876e-06,
+      "loss": 0.7575,
+      "step": 10335
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.381377321502665e-06,
+      "loss": 0.7552,
+      "step": 10336
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.379564914754867e-06,
+      "loss": 0.74,
+      "step": 10337
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3777527778384965e-06,
+      "loss": 0.7274,
+      "step": 10338
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.37594091084055e-06,
+      "loss": 0.7548,
+      "step": 10339
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.374129313848016e-06,
+      "loss": 0.4021,
+      "step": 10340
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.372317986947862e-06,
+      "loss": 0.4324,
+      "step": 10341
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.370506930227051e-06,
+      "loss": 0.7501,
+      "step": 10342
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.368696143772535e-06,
+      "loss": 0.7538,
+      "step": 10343
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3668856276712416e-06,
+      "loss": 0.6949,
+      "step": 10344
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.365075382010092e-06,
+      "loss": 0.7181,
+      "step": 10345
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3632654068759925e-06,
+      "loss": 0.7111,
+      "step": 10346
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.361455702355845e-06,
+      "loss": 0.7474,
+      "step": 10347
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.359646268536528e-06,
+      "loss": 0.384,
+      "step": 10348
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.357837105504911e-06,
+      "loss": 0.3865,
+      "step": 10349
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.356028213347849e-06,
+      "loss": 0.7547,
+      "step": 10350
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3542195921521835e-06,
+      "loss": 0.7234,
+      "step": 10351
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.352411242004749e-06,
+      "loss": 0.7813,
+      "step": 10352
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3506031629923615e-06,
+      "loss": 0.7294,
+      "step": 10353
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.348795355201823e-06,
+      "loss": 0.7048,
+      "step": 10354
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.346987818719928e-06,
+      "loss": 0.7551,
+      "step": 10355
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3451805536334455e-06,
+      "loss": 0.7489,
+      "step": 10356
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.343373560029153e-06,
+      "loss": 0.4379,
+      "step": 10357
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.341566837993798e-06,
+      "loss": 0.6801,
+      "step": 10358
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.339760387614112e-06,
+      "loss": 0.6779,
+      "step": 10359
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.33795420897683e-06,
+      "loss": 0.7584,
+      "step": 10360
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.336148302168661e-06,
+      "loss": 0.3454,
+      "step": 10361
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.334342667276306e-06,
+      "loss": 0.4389,
+      "step": 10362
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.332537304386451e-06,
+      "loss": 0.4173,
+      "step": 10363
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.330732213585766e-06,
+      "loss": 0.4221,
+      "step": 10364
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.32892739496092e-06,
+      "loss": 0.7475,
+      "step": 10365
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.327122848598554e-06,
+      "loss": 0.727,
+      "step": 10366
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3253185745853054e-06,
+      "loss": 0.6769,
+      "step": 10367
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.323514573007794e-06,
+      "loss": 0.703,
+      "step": 10368
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.321710843952624e-06,
+      "loss": 0.6996,
+      "step": 10369
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.319907387506401e-06,
+      "loss": 0.4311,
+      "step": 10370
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.318104203755699e-06,
+      "loss": 0.7489,
+      "step": 10371
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.316301292787091e-06,
+      "loss": 0.4602,
+      "step": 10372
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.31449865468713e-06,
+      "loss": 0.6824,
+      "step": 10373
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.312696289542362e-06,
+      "loss": 0.7006,
+      "step": 10374
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3108941974393125e-06,
+      "loss": 0.7025,
+      "step": 10375
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.309092378464502e-06,
+      "loss": 0.7321,
+      "step": 10376
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.307290832704432e-06,
+      "loss": 0.7297,
+      "step": 10377
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3054895602455895e-06,
+      "loss": 0.7327,
+      "step": 10378
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3036885611744594e-06,
+      "loss": 0.7385,
+      "step": 10379
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.301887835577503e-06,
+      "loss": 0.7362,
+      "step": 10380
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3000873835411696e-06,
+      "loss": 0.7169,
+      "step": 10381
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.298287205151898e-06,
+      "loss": 0.753,
+      "step": 10382
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.296487300496108e-06,
+      "loss": 0.7253,
+      "step": 10383
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.294687669660221e-06,
+      "loss": 0.4006,
+      "step": 10384
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.292888312730631e-06,
+      "loss": 0.7555,
+      "step": 10385
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.291089229793722e-06,
+      "loss": 0.3825,
+      "step": 10386
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2892904209358665e-06,
+      "loss": 0.7082,
+      "step": 10387
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.287491886243424e-06,
+      "loss": 0.4078,
+      "step": 10388
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.285693625802741e-06,
+      "loss": 0.7179,
+      "step": 10389
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.28389563970015e-06,
+      "loss": 0.7209,
+      "step": 10390
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.282097928021966e-06,
+      "loss": 0.3911,
+      "step": 10391
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.280300490854503e-06,
+      "loss": 0.7614,
+      "step": 10392
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.278503328284051e-06,
+      "loss": 0.7376,
+      "step": 10393
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.276706440396888e-06,
+      "loss": 0.7039,
+      "step": 10394
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.274909827279283e-06,
+      "loss": 0.7291,
+      "step": 10395
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.273113489017485e-06,
+      "loss": 0.4772,
+      "step": 10396
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.271317425697742e-06,
+      "loss": 0.7366,
+      "step": 10397
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.269521637406278e-06,
+      "loss": 0.7421,
+      "step": 10398
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2677261242293055e-06,
+      "loss": 0.735,
+      "step": 10399
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.265930886253025e-06,
+      "loss": 0.3954,
+      "step": 10400
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.264135923563624e-06,
+      "loss": 0.7516,
+      "step": 10401
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2623412362472836e-06,
+      "loss": 0.4143,
+      "step": 10402
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2605468243901575e-06,
+      "loss": 0.4145,
+      "step": 10403
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.258752688078394e-06,
+      "loss": 0.7201,
+      "step": 10404
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2569588273981266e-06,
+      "loss": 0.7235,
+      "step": 10405
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.255165242435483e-06,
+      "loss": 0.392,
+      "step": 10406
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.253371933276567e-06,
+      "loss": 0.4426,
+      "step": 10407
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.251578900007474e-06,
+      "loss": 0.6826,
+      "step": 10408
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.249786142714287e-06,
+      "loss": 0.7614,
+      "step": 10409
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.247993661483071e-06,
+      "loss": 0.3771,
+      "step": 10410
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.246201456399885e-06,
+      "loss": 0.7746,
+      "step": 10411
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.244409527550772e-06,
+      "loss": 0.4343,
+      "step": 10412
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.242617875021757e-06,
+      "loss": 0.7038,
+      "step": 10413
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.240826498898858e-06,
+      "loss": 0.7314,
+      "step": 10414
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.239035399268072e-06,
+      "loss": 0.7368,
+      "step": 10415
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.237244576215396e-06,
+      "loss": 0.7242,
+      "step": 10416
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2354540298268045e-06,
+      "loss": 0.379,
+      "step": 10417
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.233663760188249e-06,
+      "loss": 0.4104,
+      "step": 10418
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.231873767385692e-06,
+      "loss": 0.7572,
+      "step": 10419
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.230084051505062e-06,
+      "loss": 0.737,
+      "step": 10420
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.228294612632283e-06,
+      "loss": 0.7873,
+      "step": 10421
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.226505450853266e-06,
+      "loss": 0.4213,
+      "step": 10422
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2247165662539e-06,
+      "loss": 0.7546,
+      "step": 10423
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.222927958920077e-06,
+      "loss": 0.7748,
+      "step": 10424
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2211396289376615e-06,
+      "loss": 0.7544,
+      "step": 10425
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.219351576392511e-06,
+      "loss": 0.3692,
+      "step": 10426
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2175638013704655e-06,
+      "loss": 0.6743,
+      "step": 10427
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.215776303957353e-06,
+      "loss": 0.7568,
+      "step": 10428
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.213989084238995e-06,
+      "loss": 0.4193,
+      "step": 10429
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.212202142301192e-06,
+      "loss": 0.7069,
+      "step": 10430
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.210415478229732e-06,
+      "loss": 0.6564,
+      "step": 10431
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.208629092110392e-06,
+      "loss": 0.7271,
+      "step": 10432
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.206842984028934e-06,
+      "loss": 0.7607,
+      "step": 10433
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.205057154071108e-06,
+      "loss": 0.7785,
+      "step": 10434
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2032716023226486e-06,
+      "loss": 0.4606,
+      "step": 10435
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2014863288692785e-06,
+      "loss": 0.3967,
+      "step": 10436
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.199701333796705e-06,
+      "loss": 0.7306,
+      "step": 10437
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1979166171906295e-06,
+      "loss": 0.688,
+      "step": 10438
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.19613217913673e-06,
+      "loss": 0.3591,
+      "step": 10439
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.194348019720679e-06,
+      "loss": 0.7422,
+      "step": 10440
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.192564139028128e-06,
+      "loss": 0.7287,
+      "step": 10441
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.190780537144719e-06,
+      "loss": 0.6704,
+      "step": 10442
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1889972141560865e-06,
+      "loss": 0.7213,
+      "step": 10443
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.187214170147843e-06,
+      "loss": 0.7369,
+      "step": 10444
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.18543140520559e-06,
+      "loss": 0.7036,
+      "step": 10445
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.183648919414917e-06,
+      "loss": 0.7227,
+      "step": 10446
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.181866712861399e-06,
+      "loss": 0.7198,
+      "step": 10447
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1800847856305984e-06,
+      "loss": 0.7077,
+      "step": 10448
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.178303137808062e-06,
+      "loss": 0.6653,
+      "step": 10449
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1765217694793235e-06,
+      "loss": 0.75,
+      "step": 10450
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1747406807299105e-06,
+      "loss": 0.7857,
+      "step": 10451
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.172959871645329e-06,
+      "loss": 0.4218,
+      "step": 10452
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.171179342311073e-06,
+      "loss": 0.7272,
+      "step": 10453
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.169399092812623e-06,
+      "loss": 0.8038,
+      "step": 10454
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.167619123235444e-06,
+      "loss": 0.7184,
+      "step": 10455
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.165839433664999e-06,
+      "loss": 0.703,
+      "step": 10456
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1640600241867245e-06,
+      "loss": 0.705,
+      "step": 10457
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.162280894886047e-06,
+      "loss": 0.6978,
+      "step": 10458
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.160502045848384e-06,
+      "loss": 0.7338,
+      "step": 10459
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.158723477159129e-06,
+      "loss": 0.4032,
+      "step": 10460
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.156945188903678e-06,
+      "loss": 0.4337,
+      "step": 10461
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.155167181167404e-06,
+      "loss": 0.6997,
+      "step": 10462
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1533894540356635e-06,
+      "loss": 0.7044,
+      "step": 10463
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.151612007593799e-06,
+      "loss": 0.7092,
+      "step": 10464
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.149834841927153e-06,
+      "loss": 0.4099,
+      "step": 10465
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.148057957121042e-06,
+      "loss": 0.7332,
+      "step": 10466
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.146281353260773e-06,
+      "loss": 0.6836,
+      "step": 10467
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.144505030431638e-06,
+      "loss": 0.418,
+      "step": 10468
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.142728988718912e-06,
+      "loss": 0.7317,
+      "step": 10469
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.140953228207871e-06,
+      "loss": 0.7236,
+      "step": 10470
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.139177748983761e-06,
+      "loss": 0.4305,
+      "step": 10471
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.137402551131823e-06,
+      "loss": 0.6752,
+      "step": 10472
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.135627634737281e-06,
+      "loss": 0.4284,
+      "step": 10473
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.133852999885344e-06,
+      "loss": 0.7024,
+      "step": 10474
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.132078646661217e-06,
+      "loss": 0.7218,
+      "step": 10475
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1303045751500835e-06,
+      "loss": 0.7838,
+      "step": 10476
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.128530785437112e-06,
+      "loss": 0.3756,
+      "step": 10477
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1267572776074615e-06,
+      "loss": 0.7351,
+      "step": 10478
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.124984051746277e-06,
+      "loss": 0.7681,
+      "step": 10479
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.123211107938688e-06,
+      "loss": 0.7438,
+      "step": 10480
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.121438446269815e-06,
+      "loss": 0.421,
+      "step": 10481
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.119666066824753e-06,
+      "loss": 0.7977,
+      "step": 10482
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.117893969688602e-06,
+      "loss": 0.7155,
+      "step": 10483
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.116122154946436e-06,
+      "loss": 0.7274,
+      "step": 10484
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1143506226833165e-06,
+      "loss": 0.7348,
+      "step": 10485
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1125793729842925e-06,
+      "loss": 0.7288,
+      "step": 10486
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.110808405934398e-06,
+      "loss": 0.6786,
+      "step": 10487
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.109037721618663e-06,
+      "loss": 0.728,
+      "step": 10488
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.107267320122089e-06,
+      "loss": 0.3927,
+      "step": 10489
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.105497201529675e-06,
+      "loss": 0.4328,
+      "step": 10490
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.103727365926402e-06,
+      "loss": 0.6518,
+      "step": 10491
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.101957813397237e-06,
+      "loss": 0.3894,
+      "step": 10492
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.100188544027134e-06,
+      "loss": 0.4579,
+      "step": 10493
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.098419557901036e-06,
+      "loss": 0.7644,
+      "step": 10494
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.096650855103868e-06,
+      "loss": 0.7064,
+      "step": 10495
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.09488243572054e-06,
+      "loss": 0.7321,
+      "step": 10496
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.093114299835961e-06,
+      "loss": 0.7161,
+      "step": 10497
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.091346447535012e-06,
+      "loss": 0.7441,
+      "step": 10498
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.089578878902567e-06,
+      "loss": 0.7314,
+      "step": 10499
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.087811594023485e-06,
+      "loss": 0.4173,
+      "step": 10500
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0860445929826085e-06,
+      "loss": 0.6677,
+      "step": 10501
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.084277875864776e-06,
+      "loss": 0.7083,
+      "step": 10502
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.082511442754802e-06,
+      "loss": 0.6986,
+      "step": 10503
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.080745293737492e-06,
+      "loss": 0.6864,
+      "step": 10504
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0789794288976344e-06,
+      "loss": 0.7247,
+      "step": 10505
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.077213848320011e-06,
+      "loss": 0.4223,
+      "step": 10506
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.075448552089381e-06,
+      "loss": 0.7558,
+      "step": 10507
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.073683540290498e-06,
+      "loss": 0.7225,
+      "step": 10508
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.071918813008094e-06,
+      "loss": 0.4162,
+      "step": 10509
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.070154370326898e-06,
+      "loss": 0.7234,
+      "step": 10510
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0683902123316155e-06,
+      "loss": 0.4029,
+      "step": 10511
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.066626339106943e-06,
+      "loss": 0.4056,
+      "step": 10512
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.06486275073756e-06,
+      "loss": 0.7474,
+      "step": 10513
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.063099447308134e-06,
+      "loss": 0.7062,
+      "step": 10514
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.061336428903327e-06,
+      "loss": 0.7519,
+      "step": 10515
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0595736956077724e-06,
+      "loss": 0.7712,
+      "step": 10516
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.057811247506099e-06,
+      "loss": 0.4075,
+      "step": 10517
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0560490846829216e-06,
+      "loss": 0.7531,
+      "step": 10518
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.054287207222836e-06,
+      "loss": 0.7148,
+      "step": 10519
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0525256152104335e-06,
+      "loss": 0.6815,
+      "step": 10520
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.050764308730286e-06,
+      "loss": 0.7456,
+      "step": 10521
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.049003287866949e-06,
+      "loss": 0.7344,
+      "step": 10522
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.047242552704962e-06,
+      "loss": 0.7625,
+      "step": 10523
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.045482103328867e-06,
+      "loss": 0.4039,
+      "step": 10524
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.043721939823176e-06,
+      "loss": 0.7126,
+      "step": 10525
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.041962062272394e-06,
+      "loss": 0.4371,
+      "step": 10526
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.04020247076101e-06,
+      "loss": 0.7301,
+      "step": 10527
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.038443165373498e-06,
+      "loss": 0.3891,
+      "step": 10528
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.036684146194325e-06,
+      "loss": 0.7512,
+      "step": 10529
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.034925413307939e-06,
+      "loss": 0.8076,
+      "step": 10530
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.033166966798773e-06,
+      "loss": 0.7329,
+      "step": 10531
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.031408806751249e-06,
+      "loss": 0.7417,
+      "step": 10532
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.029650933249771e-06,
+      "loss": 0.7314,
+      "step": 10533
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.027893346378739e-06,
+      "loss": 0.3927,
+      "step": 10534
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.026136046222532e-06,
+      "loss": 0.7095,
+      "step": 10535
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.024379032865516e-06,
+      "loss": 0.7283,
+      "step": 10536
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.022622306392037e-06,
+      "loss": 0.7013,
+      "step": 10537
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.02086586688644e-06,
+      "loss": 0.6936,
+      "step": 10538
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0191097144330495e-06,
+      "loss": 0.3969,
+      "step": 10539
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0173538491161744e-06,
+      "loss": 0.7555,
+      "step": 10540
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.015598271020111e-06,
+      "loss": 0.7029,
+      "step": 10541
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.013842980229147e-06,
+      "loss": 0.7919,
+      "step": 10542
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.012087976827551e-06,
+      "loss": 0.7393,
+      "step": 10543
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.010333260899578e-06,
+      "loss": 0.7842,
+      "step": 10544
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.00857883252947e-06,
+      "loss": 0.7823,
+      "step": 10545
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.006824691801452e-06,
+      "loss": 0.6724,
+      "step": 10546
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.005070838799745e-06,
+      "loss": 0.4047,
+      "step": 10547
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.003317273608548e-06,
+      "loss": 0.6912,
+      "step": 10548
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0015639963120454e-06,
+      "loss": 0.7311,
+      "step": 10549
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.999811006994413e-06,
+      "loss": 0.6851,
+      "step": 10550
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.998058305739807e-06,
+      "loss": 0.7431,
+      "step": 10551
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.996305892632375e-06,
+      "loss": 0.7601,
+      "step": 10552
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.994553767756248e-06,
+      "loss": 0.7261,
+      "step": 10553
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.992801931195545e-06,
+      "loss": 0.7155,
+      "step": 10554
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.991050383034364e-06,
+      "loss": 0.4305,
+      "step": 10555
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.989299123356802e-06,
+      "loss": 0.6485,
+      "step": 10556
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.987548152246935e-06,
+      "loss": 0.7064,
+      "step": 10557
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.985797469788822e-06,
+      "loss": 0.6939,
+      "step": 10558
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.984047076066514e-06,
+      "loss": 0.7157,
+      "step": 10559
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.98229697116404e-06,
+      "loss": 0.7692,
+      "step": 10560
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.980547155165429e-06,
+      "loss": 0.7368,
+      "step": 10561
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9787976281546845e-06,
+      "loss": 0.4258,
+      "step": 10562
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.977048390215798e-06,
+      "loss": 0.4179,
+      "step": 10563
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.975299441432751e-06,
+      "loss": 0.7687,
+      "step": 10564
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.973550781889504e-06,
+      "loss": 0.7292,
+      "step": 10565
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9718024116700185e-06,
+      "loss": 0.7314,
+      "step": 10566
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.970054330858222e-06,
+      "loss": 0.771,
+      "step": 10567
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.968306539538037e-06,
+      "loss": 0.3986,
+      "step": 10568
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9665590377933825e-06,
+      "loss": 0.6802,
+      "step": 10569
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.964811825708149e-06,
+      "loss": 0.776,
+      "step": 10570
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.963064903366219e-06,
+      "loss": 0.7272,
+      "step": 10571
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9613182708514605e-06,
+      "loss": 0.7008,
+      "step": 10572
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.959571928247724e-06,
+      "loss": 0.7327,
+      "step": 10573
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.957825875638856e-06,
+      "loss": 0.4121,
+      "step": 10574
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.956080113108681e-06,
+      "loss": 0.4225,
+      "step": 10575
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.95433464074101e-06,
+      "loss": 0.6919,
+      "step": 10576
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.952589458619641e-06,
+      "loss": 0.7233,
+      "step": 10577
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9508445668283555e-06,
+      "loss": 0.4215,
+      "step": 10578
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.949099965450931e-06,
+      "loss": 0.4121,
+      "step": 10579
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.947355654571121e-06,
+      "loss": 0.7186,
+      "step": 10580
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9456116342726705e-06,
+      "loss": 0.4074,
+      "step": 10581
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.943867904639298e-06,
+      "loss": 0.3626,
+      "step": 10582
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.94212446575473e-06,
+      "loss": 0.762,
+      "step": 10583
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9403813177026616e-06,
+      "loss": 0.6831,
+      "step": 10584
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.938638460566782e-06,
+      "loss": 0.7321,
+      "step": 10585
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.936895894430761e-06,
+      "loss": 0.7574,
+      "step": 10586
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9351536193782566e-06,
+      "loss": 0.7376,
+      "step": 10587
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.93341163549292e-06,
+      "loss": 0.7609,
+      "step": 10588
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9316699428583785e-06,
+      "loss": 0.7043,
+      "step": 10589
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.929928541558248e-06,
+      "loss": 0.7125,
+      "step": 10590
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.928187431676133e-06,
+      "loss": 0.7639,
+      "step": 10591
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.926446613295617e-06,
+      "loss": 0.4221,
+      "step": 10592
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.924706086500284e-06,
+      "loss": 0.6978,
+      "step": 10593
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.92296585137369e-06,
+      "loss": 0.7475,
+      "step": 10594
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.921225907999383e-06,
+      "loss": 0.7637,
+      "step": 10595
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.919486256460895e-06,
+      "loss": 0.4319,
+      "step": 10596
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.917746896841745e-06,
+      "loss": 0.7427,
+      "step": 10597
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.91600782922544e-06,
+      "loss": 0.7573,
+      "step": 10598
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.914269053695468e-06,
+      "loss": 0.4211,
+      "step": 10599
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.912530570335303e-06,
+      "loss": 0.7146,
+      "step": 10600
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.910792379228416e-06,
+      "loss": 0.4133,
+      "step": 10601
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.909054480458251e-06,
+      "loss": 0.7104,
+      "step": 10602
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.907316874108245e-06,
+      "loss": 0.7623,
+      "step": 10603
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.905579560261815e-06,
+      "loss": 0.7261,
+      "step": 10604
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9038425390023675e-06,
+      "loss": 0.7331,
+      "step": 10605
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9021058104133e-06,
+      "loss": 0.7091,
+      "step": 10606
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.900369374577989e-06,
+      "loss": 0.7123,
+      "step": 10607
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.898633231579799e-06,
+      "loss": 0.7072,
+      "step": 10608
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.896897381502081e-06,
+      "loss": 0.7763,
+      "step": 10609
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.895161824428168e-06,
+      "loss": 0.6749,
+      "step": 10610
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.893426560441387e-06,
+      "loss": 0.7216,
+      "step": 10611
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.891691589625043e-06,
+      "loss": 0.7184,
+      "step": 10612
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.889956912062432e-06,
+      "loss": 0.7042,
+      "step": 10613
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.888222527836828e-06,
+      "loss": 0.7546,
+      "step": 10614
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.886488437031508e-06,
+      "loss": 0.7429,
+      "step": 10615
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.884754639729718e-06,
+      "loss": 0.7069,
+      "step": 10616
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8830211360146955e-06,
+      "loss": 0.7351,
+      "step": 10617
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.881287925969665e-06,
+      "loss": 0.7445,
+      "step": 10618
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.879555009677834e-06,
+      "loss": 0.73,
+      "step": 10619
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.877822387222402e-06,
+      "loss": 0.74,
+      "step": 10620
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.87609005868655e-06,
+      "loss": 0.7417,
+      "step": 10621
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.874358024153444e-06,
+      "loss": 0.3908,
+      "step": 10622
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.872626283706236e-06,
+      "loss": 0.7512,
+      "step": 10623
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8708948374280645e-06,
+      "loss": 0.717,
+      "step": 10624
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.869163685402063e-06,
+      "loss": 0.7595,
+      "step": 10625
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.867432827711332e-06,
+      "loss": 0.3748,
+      "step": 10626
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.865702264438969e-06,
+      "loss": 0.681,
+      "step": 10627
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8639719956680624e-06,
+      "loss": 0.7372,
+      "step": 10628
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.862242021481678e-06,
+      "loss": 0.7433,
+      "step": 10629
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.86051234196287e-06,
+      "loss": 0.4057,
+      "step": 10630
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.858782957194679e-06,
+      "loss": 0.4084,
+      "step": 10631
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.857053867260127e-06,
+      "loss": 0.7433,
+      "step": 10632
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.855325072242232e-06,
+      "loss": 0.7679,
+      "step": 10633
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.85359657222399e-06,
+      "loss": 0.7619,
+      "step": 10634
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.851868367288383e-06,
+      "loss": 0.4055,
+      "step": 10635
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.850140457518382e-06,
+      "loss": 0.4399,
+      "step": 10636
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.848412842996937e-06,
+      "loss": 0.7517,
+      "step": 10637
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.846685523806996e-06,
+      "loss": 0.6928,
+      "step": 10638
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.844958500031486e-06,
+      "loss": 0.7232,
+      "step": 10639
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.843231771753318e-06,
+      "loss": 0.7404,
+      "step": 10640
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.841505339055383e-06,
+      "loss": 0.6881,
+      "step": 10641
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.839779202020575e-06,
+      "loss": 0.6996,
+      "step": 10642
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.83805336073176e-06,
+      "loss": 0.7043,
+      "step": 10643
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.836327815271797e-06,
+      "loss": 0.7201,
+      "step": 10644
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.834602565723523e-06,
+      "loss": 0.4206,
+      "step": 10645
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.832877612169764e-06,
+      "loss": 0.6907,
+      "step": 10646
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.831152954693342e-06,
+      "loss": 0.7631,
+      "step": 10647
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.82942859337705e-06,
+      "loss": 0.7141,
+      "step": 10648
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.827704528303675e-06,
+      "loss": 0.7357,
+      "step": 10649
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.825980759555986e-06,
+      "loss": 0.7666,
+      "step": 10650
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.824257287216737e-06,
+      "loss": 0.7409,
+      "step": 10651
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.822534111368676e-06,
+      "loss": 0.7227,
+      "step": 10652
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.82081123209453e-06,
+      "loss": 0.721,
+      "step": 10653
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8190886494770085e-06,
+      "loss": 0.7106,
+      "step": 10654
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.817366363598814e-06,
+      "loss": 0.6851,
+      "step": 10655
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.815644374542632e-06,
+      "loss": 0.7494,
+      "step": 10656
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8139226823911314e-06,
+      "loss": 0.4657,
+      "step": 10657
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8122012872269708e-06,
+      "loss": 0.7591,
+      "step": 10658
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.810480189132788e-06,
+      "loss": 0.7589,
+      "step": 10659
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8087593881912176e-06,
+      "loss": 0.7307,
+      "step": 10660
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8070388844848715e-06,
+      "loss": 0.7208,
+      "step": 10661
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.805318678096349e-06,
+      "loss": 0.8106,
+      "step": 10662
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8035987691082355e-06,
+      "loss": 0.7092,
+      "step": 10663
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8018791576030977e-06,
+      "loss": 0.779,
+      "step": 10664
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8001598436635e-06,
+      "loss": 0.7005,
+      "step": 10665
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.798440827371982e-06,
+      "loss": 0.703,
+      "step": 10666
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.796722108811072e-06,
+      "loss": 0.7234,
+      "step": 10667
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7950036880632813e-06,
+      "loss": 0.737,
+      "step": 10668
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.79328556521111e-06,
+      "loss": 0.7503,
+      "step": 10669
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.791567740337051e-06,
+      "loss": 0.7409,
+      "step": 10670
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.789850213523567e-06,
+      "loss": 0.7878,
+      "step": 10671
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7881329848531155e-06,
+      "loss": 0.7284,
+      "step": 10672
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7864160544081385e-06,
+      "loss": 0.7239,
+      "step": 10673
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7846994222710698e-06,
+      "loss": 0.422,
+      "step": 10674
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.782983088524318e-06,
+      "loss": 0.7111,
+      "step": 10675
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.781267053250285e-06,
+      "loss": 0.7415,
+      "step": 10676
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.779551316531356e-06,
+      "loss": 0.7471,
+      "step": 10677
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.777835878449896e-06,
+      "loss": 0.735,
+      "step": 10678
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7761207390882703e-06,
+      "loss": 0.7136,
+      "step": 10679
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.774405898528817e-06,
+      "loss": 0.4144,
+      "step": 10680
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7726913568538625e-06,
+      "loss": 0.7146,
+      "step": 10681
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.770977114145723e-06,
+      "loss": 0.7438,
+      "step": 10682
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.769263170486692e-06,
+      "loss": 0.742,
+      "step": 10683
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7675495259590623e-06,
+      "loss": 0.3831,
+      "step": 10684
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7658361806451036e-06,
+      "loss": 0.42,
+      "step": 10685
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7641231346270656e-06,
+      "loss": 0.7298,
+      "step": 10686
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.762410387987189e-06,
+      "loss": 0.4071,
+      "step": 10687
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.760697940807708e-06,
+      "loss": 0.7306,
+      "step": 10688
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7589857931708317e-06,
+      "loss": 0.7191,
+      "step": 10689
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7572739451587605e-06,
+      "loss": 0.7132,
+      "step": 10690
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7555623968536737e-06,
+      "loss": 0.7182,
+      "step": 10691
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.753851148337748e-06,
+      "loss": 0.7728,
+      "step": 10692
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.752140199693135e-06,
+      "loss": 0.3994,
+      "step": 10693
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.750429551001976e-06,
+      "loss": 0.4047,
+      "step": 10694
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.748719202346397e-06,
+      "loss": 0.6769,
+      "step": 10695
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7470091538085073e-06,
+      "loss": 0.7274,
+      "step": 10696
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.745299405470412e-06,
+      "loss": 0.6984,
+      "step": 10697
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7435899574141897e-06,
+      "loss": 0.3605,
+      "step": 10698
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7418808097219095e-06,
+      "loss": 0.7079,
+      "step": 10699
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.740171962475627e-06,
+      "loss": 0.7543,
+      "step": 10700
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7384634157573807e-06,
+      "loss": 0.7397,
+      "step": 10701
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7367551696491977e-06,
+      "loss": 0.7006,
+      "step": 10702
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7350472242330882e-06,
+      "loss": 0.7518,
+      "step": 10703
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7333395795910486e-06,
+      "loss": 0.7093,
+      "step": 10704
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7316322358050584e-06,
+      "loss": 0.7029,
+      "step": 10705
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.729925192957092e-06,
+      "loss": 0.7415,
+      "step": 10706
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7282184511291e-06,
+      "loss": 0.427,
+      "step": 10707
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7265120104030208e-06,
+      "loss": 0.7076,
+      "step": 10708
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7248058708607792e-06,
+      "loss": 0.4038,
+      "step": 10709
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7231000325842813e-06,
+      "loss": 0.7256,
+      "step": 10710
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7213944956554304e-06,
+      "loss": 0.7218,
+      "step": 10711
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.719689260156104e-06,
+      "loss": 0.4058,
+      "step": 10712
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.717984326168168e-06,
+      "loss": 0.4323,
+      "step": 10713
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7162796937734747e-06,
+      "loss": 0.7004,
+      "step": 10714
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7145753630538616e-06,
+      "loss": 0.6845,
+      "step": 10715
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.712871334091154e-06,
+      "loss": 0.3886,
+      "step": 10716
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7111676069671576e-06,
+      "loss": 0.4255,
+      "step": 10717
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7094641817636644e-06,
+      "loss": 0.7648,
+      "step": 10718
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7077610585624624e-06,
+      "loss": 0.7473,
+      "step": 10719
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7060582374453115e-06,
+      "loss": 0.7857,
+      "step": 10720
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7043557184939627e-06,
+      "loss": 0.4288,
+      "step": 10721
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7026535017901533e-06,
+      "loss": 0.7368,
+      "step": 10722
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7009515874155997e-06,
+      "loss": 0.7387,
+      "step": 10723
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6992499754520185e-06,
+      "loss": 0.7178,
+      "step": 10724
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.697548665981098e-06,
+      "loss": 0.7605,
+      "step": 10725
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.695847659084514e-06,
+      "loss": 0.7663,
+      "step": 10726
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6941469548439334e-06,
+      "loss": 0.4253,
+      "step": 10727
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6924465533410004e-06,
+      "loss": 0.6859,
+      "step": 10728
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6907464546573592e-06,
+      "loss": 0.6834,
+      "step": 10729
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.689046658874621e-06,
+      "loss": 0.6985,
+      "step": 10730
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.687347166074394e-06,
+      "loss": 0.7107,
+      "step": 10731
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.685647976338266e-06,
+      "loss": 0.7301,
+      "step": 10732
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6839490897478192e-06,
+      "loss": 0.6796,
+      "step": 10733
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6822505063846136e-06,
+      "loss": 0.6865,
+      "step": 10734
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.680552226330195e-06,
+      "loss": 0.4377,
+      "step": 10735
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6788542496660974e-06,
+      "loss": 0.7459,
+      "step": 10736
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.677156576473835e-06,
+      "loss": 0.7104,
+      "step": 10737
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6754592068349172e-06,
+      "loss": 0.7402,
+      "step": 10738
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.673762140830831e-06,
+      "loss": 0.7417,
+      "step": 10739
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6720653785430506e-06,
+      "loss": 0.7869,
+      "step": 10740
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6703689200530357e-06,
+      "loss": 0.77,
+      "step": 10741
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6686727654422273e-06,
+      "loss": 0.7491,
+      "step": 10742
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.666976914792064e-06,
+      "loss": 0.7114,
+      "step": 10743
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6652813681839616e-06,
+      "loss": 0.3711,
+      "step": 10744
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6635861256993146e-06,
+      "loss": 0.7367,
+      "step": 10745
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6618911874195095e-06,
+      "loss": 0.736,
+      "step": 10746
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6601965534259265e-06,
+      "loss": 0.7342,
+      "step": 10747
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.658502223799919e-06,
+      "loss": 0.7163,
+      "step": 10748
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.656808198622831e-06,
+      "loss": 0.6903,
+      "step": 10749
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6551144779759864e-06,
+      "loss": 0.6932,
+      "step": 10750
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.653421061940705e-06,
+      "loss": 0.7191,
+      "step": 10751
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.651727950598285e-06,
+      "loss": 0.752,
+      "step": 10752
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6500351440300097e-06,
+      "loss": 0.7132,
+      "step": 10753
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6483426423171496e-06,
+      "loss": 0.769,
+      "step": 10754
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6466504455409558e-06,
+      "loss": 0.7696,
+      "step": 10755
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.644958553782676e-06,
+      "loss": 0.6919,
+      "step": 10756
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.643266967123533e-06,
+      "loss": 0.7766,
+      "step": 10757
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6415756856447384e-06,
+      "loss": 0.7509,
+      "step": 10758
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.639884709427488e-06,
+      "loss": 0.4448,
+      "step": 10759
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6381940385529646e-06,
+      "loss": 0.6981,
+      "step": 10760
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.636503673102335e-06,
+      "loss": 0.7565,
+      "step": 10761
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6348136131567537e-06,
+      "loss": 0.7643,
+      "step": 10762
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6331238587973574e-06,
+      "loss": 0.7007,
+      "step": 10763
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6314344101052656e-06,
+      "loss": 0.6919,
+      "step": 10764
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.629745267161594e-06,
+      "loss": 0.7495,
+      "step": 10765
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.628056430047434e-06,
+      "loss": 0.8004,
+      "step": 10766
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6263678988438654e-06,
+      "loss": 0.6934,
+      "step": 10767
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.624679673631951e-06,
+      "loss": 0.7965,
+      "step": 10768
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6229917544927396e-06,
+      "loss": 0.7165,
+      "step": 10769
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6213041415072724e-06,
+      "loss": 0.4458,
+      "step": 10770
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.619616834756565e-06,
+      "loss": 0.7577,
+      "step": 10771
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.617929834321626e-06,
+      "loss": 0.7352,
+      "step": 10772
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.616243140283445e-06,
+      "loss": 0.7056,
+      "step": 10773
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6145567527229984e-06,
+      "loss": 0.73,
+      "step": 10774
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.612870671721249e-06,
+      "loss": 0.7553,
+      "step": 10775
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6111848973591437e-06,
+      "loss": 0.746,
+      "step": 10776
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.60949942971761e-06,
+      "loss": 0.7536,
+      "step": 10777
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6078142688775732e-06,
+      "loss": 0.405,
+      "step": 10778
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.606129414919933e-06,
+      "loss": 0.7662,
+      "step": 10779
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6044448679255774e-06,
+      "loss": 0.393,
+      "step": 10780
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.60276062797538e-06,
+      "loss": 0.7877,
+      "step": 10781
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.601076695150195e-06,
+      "loss": 0.4353,
+      "step": 10782
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5993930695308753e-06,
+      "loss": 0.8078,
+      "step": 10783
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.597709751198244e-06,
+      "loss": 0.7739,
+      "step": 10784
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5960267402331183e-06,
+      "loss": 0.7731,
+      "step": 10785
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5943440367162952e-06,
+      "loss": 0.7661,
+      "step": 10786
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5926616407285573e-06,
+      "loss": 0.7155,
+      "step": 10787
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.590979552350683e-06,
+      "loss": 0.4133,
+      "step": 10788
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5892977716634247e-06,
+      "loss": 0.7303,
+      "step": 10789
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5876162987475195e-06,
+      "loss": 0.3998,
+      "step": 10790
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.58593513368369e-06,
+      "loss": 0.3915,
+      "step": 10791
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5842542765526554e-06,
+      "loss": 0.77,
+      "step": 10792
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.58257372743511e-06,
+      "loss": 0.7508,
+      "step": 10793
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.580893486411734e-06,
+      "loss": 0.708,
+      "step": 10794
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5792135535631934e-06,
+      "loss": 0.707,
+      "step": 10795
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.577533928970137e-06,
+      "loss": 0.7359,
+      "step": 10796
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.57585461271321e-06,
+      "loss": 0.7181,
+      "step": 10797
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5741756048730293e-06,
+      "loss": 0.4103,
+      "step": 10798
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.572496905530204e-06,
+      "loss": 0.7358,
+      "step": 10799
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5708185147653262e-06,
+      "loss": 0.7131,
+      "step": 10800
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5691404326589696e-06,
+      "loss": 0.7518,
+      "step": 10801
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.567462659291705e-06,
+      "loss": 0.403,
+      "step": 10802
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.565785194744078e-06,
+      "loss": 0.7173,
+      "step": 10803
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.564108039096623e-06,
+      "loss": 0.6725,
+      "step": 10804
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5624311924298506e-06,
+      "loss": 0.4005,
+      "step": 10805
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5607546548242735e-06,
+      "loss": 0.4278,
+      "step": 10806
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.559078426360377e-06,
+      "loss": 0.7939,
+      "step": 10807
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5574025071186358e-06,
+      "loss": 0.7441,
+      "step": 10808
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5557268971795068e-06,
+      "loss": 0.7212,
+      "step": 10809
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.554051596623438e-06,
+      "loss": 0.7168,
+      "step": 10810
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5523766055308582e-06,
+      "loss": 0.7255,
+      "step": 10811
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5507019239821818e-06,
+      "loss": 0.4436,
+      "step": 10812
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5490275520578064e-06,
+      "loss": 0.7035,
+      "step": 10813
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.547353489838117e-06,
+      "loss": 0.7518,
+      "step": 10814
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.545679737403487e-06,
+      "loss": 0.7295,
+      "step": 10815
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5440062948342703e-06,
+      "loss": 0.6846,
+      "step": 10816
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.542333162210806e-06,
+      "loss": 0.7665,
+      "step": 10817
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.540660339613421e-06,
+      "loss": 0.7175,
+      "step": 10818
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.538987827122424e-06,
+      "loss": 0.4459,
+      "step": 10819
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.537315624818113e-06,
+      "loss": 0.4579,
+      "step": 10820
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5356437327807656e-06,
+      "loss": 0.4088,
+      "step": 10821
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.53397215109065e-06,
+      "loss": 0.7108,
+      "step": 10822
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.532300879828013e-06,
+      "loss": 0.7348,
+      "step": 10823
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5306299190730984e-06,
+      "loss": 0.7283,
+      "step": 10824
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5289592689061224e-06,
+      "loss": 0.7305,
+      "step": 10825
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.527288929407292e-06,
+      "loss": 0.7242,
+      "step": 10826
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5256189006567975e-06,
+      "loss": 0.752,
+      "step": 10827
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.523949182734814e-06,
+      "loss": 0.7466,
+      "step": 10828
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5222797757215087e-06,
+      "loss": 0.4087,
+      "step": 10829
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.520610679697024e-06,
+      "loss": 0.7653,
+      "step": 10830
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5189418947414922e-06,
+      "loss": 0.7128,
+      "step": 10831
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5172734209350303e-06,
+      "loss": 0.4286,
+      "step": 10832
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.515605258357736e-06,
+      "loss": 0.7662,
+      "step": 10833
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.513937407089707e-06,
+      "loss": 0.6841,
+      "step": 10834
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5122698672110047e-06,
+      "loss": 0.7135,
+      "step": 10835
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.510602638801689e-06,
+      "loss": 0.6744,
+      "step": 10836
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5089357219417984e-06,
+      "loss": 0.7404,
+      "step": 10837
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.507269116711368e-06,
+      "loss": 0.7262,
+      "step": 10838
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.505602823190405e-06,
+      "loss": 0.7432,
+      "step": 10839
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.503936841458907e-06,
+      "loss": 0.709,
+      "step": 10840
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5022711715968516e-06,
+      "loss": 0.7669,
+      "step": 10841
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.500605813684215e-06,
+      "loss": 0.7268,
+      "step": 10842
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4989407678009444e-06,
+      "loss": 0.3535,
+      "step": 10843
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.497276034026977e-06,
+      "loss": 0.7612,
+      "step": 10844
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4956116124422357e-06,
+      "loss": 0.7333,
+      "step": 10845
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4939475031266244e-06,
+      "loss": 0.7298,
+      "step": 10846
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4922837061600413e-06,
+      "loss": 0.7242,
+      "step": 10847
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4906202216223626e-06,
+      "loss": 0.4115,
+      "step": 10848
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4889570495934464e-06,
+      "loss": 0.7705,
+      "step": 10849
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.487294190153139e-06,
+      "loss": 0.7492,
+      "step": 10850
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4856316433812777e-06,
+      "loss": 0.7027,
+      "step": 10851
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.483969409357678e-06,
+      "loss": 0.7743,
+      "step": 10852
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.482307488162142e-06,
+      "loss": 0.6912,
+      "step": 10853
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4806458798744568e-06,
+      "loss": 0.7096,
+      "step": 10854
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.478984584574391e-06,
+      "loss": 0.7178,
+      "step": 10855
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4773236023417066e-06,
+      "loss": 0.7331,
+      "step": 10856
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4756629332561454e-06,
+      "loss": 0.7131,
+      "step": 10857
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.474002577397433e-06,
+      "loss": 0.7714,
+      "step": 10858
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.472342534845281e-06,
+      "loss": 0.7575,
+      "step": 10859
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.470682805679384e-06,
+      "loss": 0.7175,
+      "step": 10860
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4690233899794303e-06,
+      "loss": 0.762,
+      "step": 10861
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4673642878250836e-06,
+      "loss": 0.4526,
+      "step": 10862
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.465705499295998e-06,
+      "loss": 0.7403,
+      "step": 10863
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4640470244718017e-06,
+      "loss": 0.6907,
+      "step": 10864
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4623888634321245e-06,
+      "loss": 0.7209,
+      "step": 10865
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4607310162565722e-06,
+      "loss": 0.7379,
+      "step": 10866
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4590734830247343e-06,
+      "loss": 0.7377,
+      "step": 10867
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4574162638161844e-06,
+      "loss": 0.7026,
+      "step": 10868
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4557593587104923e-06,
+      "loss": 0.7103,
+      "step": 10869
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4541027677871986e-06,
+      "loss": 0.6864,
+      "step": 10870
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.452446491125836e-06,
+      "loss": 0.3832,
+      "step": 10871
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4507905288059205e-06,
+      "loss": 0.4422,
+      "step": 10872
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.449134880906949e-06,
+      "loss": 0.7741,
+      "step": 10873
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4474795475084155e-06,
+      "loss": 0.7298,
+      "step": 10874
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.445824528689786e-06,
+      "loss": 0.6955,
+      "step": 10875
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4441698245305177e-06,
+      "loss": 0.7215,
+      "step": 10876
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.442515435110051e-06,
+      "loss": 0.7342,
+      "step": 10877
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.440861360507811e-06,
+      "loss": 0.7193,
+      "step": 10878
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4392076008032084e-06,
+      "loss": 0.4106,
+      "step": 10879
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4375541560756376e-06,
+      "loss": 0.7865,
+      "step": 10880
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4359010264044812e-06,
+      "loss": 0.4174,
+      "step": 10881
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.434248211869098e-06,
+      "loss": 0.771,
+      "step": 10882
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4325957125488474e-06,
+      "loss": 0.3898,
+      "step": 10883
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4309435285230585e-06,
+      "loss": 0.7301,
+      "step": 10884
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4292916598710525e-06,
+      "loss": 0.7121,
+      "step": 10885
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4276401066721334e-06,
+      "loss": 0.6889,
+      "step": 10886
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4259888690055875e-06,
+      "loss": 0.7004,
+      "step": 10887
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4243379469506956e-06,
+      "loss": 0.7083,
+      "step": 10888
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4226873405867136e-06,
+      "loss": 0.4098,
+      "step": 10889
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.421037049992885e-06,
+      "loss": 0.7211,
+      "step": 10890
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4193870752484395e-06,
+      "loss": 0.7744,
+      "step": 10891
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.417737416432586e-06,
+      "loss": 0.7315,
+      "step": 10892
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.416088073624534e-06,
+      "loss": 0.7534,
+      "step": 10893
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.414439046903457e-06,
+      "loss": 0.4286,
+      "step": 10894
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.412790336348526e-06,
+      "loss": 0.6967,
+      "step": 10895
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4111419420388904e-06,
+      "loss": 0.7645,
+      "step": 10896
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4094938640536946e-06,
+      "loss": 0.7243,
+      "step": 10897
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4078461024720578e-06,
+      "loss": 0.7019,
+      "step": 10898
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4061986573730877e-06,
+      "loss": 0.7559,
+      "step": 10899
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.404551528835873e-06,
+      "loss": 0.3892,
+      "step": 10900
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.402904716939498e-06,
+      "loss": 0.7729,
+      "step": 10901
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4012582217630207e-06,
+      "loss": 0.7263,
+      "step": 10902
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3996120433854874e-06,
+      "loss": 0.7488,
+      "step": 10903
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.397966181885929e-06,
+      "loss": 0.7647,
+      "step": 10904
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.396320637343359e-06,
+      "loss": 0.729,
+      "step": 10905
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.394675409836785e-06,
+      "loss": 0.7243,
+      "step": 10906
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.39303049944519e-06,
+      "loss": 0.7085,
+      "step": 10907
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3913859062475464e-06,
+      "loss": 0.3977,
+      "step": 10908
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3897416303228004e-06,
+      "loss": 0.7518,
+      "step": 10909
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3880976717499015e-06,
+      "loss": 0.3977,
+      "step": 10910
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.386454030607772e-06,
+      "loss": 0.785,
+      "step": 10911
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.384810706975321e-06,
+      "loss": 0.4168,
+      "step": 10912
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.383167700931442e-06,
+      "loss": 0.4255,
+      "step": 10913
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3815250125550127e-06,
+      "loss": 0.6884,
+      "step": 10914
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3798826419249007e-06,
+      "loss": 0.7542,
+      "step": 10915
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3782405891199545e-06,
+      "loss": 0.7364,
+      "step": 10916
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.376598854219004e-06,
+      "loss": 0.4706,
+      "step": 10917
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3749574373008707e-06,
+      "loss": 0.4682,
+      "step": 10918
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.373316338444351e-06,
+      "loss": 0.3836,
+      "step": 10919
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.371675557728242e-06,
+      "loss": 0.372,
+      "step": 10920
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3700350952313097e-06,
+      "loss": 0.4175,
+      "step": 10921
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3683949510323133e-06,
+      "loss": 0.7394,
+      "step": 10922
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.366755125209993e-06,
+      "loss": 0.7046,
+      "step": 10923
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3651156178430756e-06,
+      "loss": 0.7031,
+      "step": 10924
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.363476429010273e-06,
+      "loss": 0.4014,
+      "step": 10925
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3618375587902794e-06,
+      "loss": 0.4488,
+      "step": 10926
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3601990072617776e-06,
+      "loss": 0.7108,
+      "step": 10927
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3585607745034265e-06,
+      "loss": 0.4386,
+      "step": 10928
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.356922860593885e-06,
+      "loss": 0.6883,
+      "step": 10929
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.355285265611784e-06,
+      "loss": 0.7167,
+      "step": 10930
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.353647989635742e-06,
+      "loss": 0.7264,
+      "step": 10931
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.35201103274436e-06,
+      "loss": 0.7371,
+      "step": 10932
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.350374395016234e-06,
+      "loss": 0.7485,
+      "step": 10933
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3487380765299325e-06,
+      "loss": 0.7366,
+      "step": 10934
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.347102077364015e-06,
+      "loss": 0.6909,
+      "step": 10935
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3454663975970235e-06,
+      "loss": 0.7211,
+      "step": 10936
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.343831037307482e-06,
+      "loss": 0.6687,
+      "step": 10937
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3421959965739125e-06,
+      "loss": 0.3884,
+      "step": 10938
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.340561275474802e-06,
+      "loss": 0.7137,
+      "step": 10939
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.338926874088635e-06,
+      "loss": 0.7359,
+      "step": 10940
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3372927924938735e-06,
+      "loss": 0.7104,
+      "step": 10941
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.335659030768975e-06,
+      "loss": 0.3748,
+      "step": 10942
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.334025588992372e-06,
+      "loss": 0.6746,
+      "step": 10943
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3323924672424847e-06,
+      "loss": 0.7527,
+      "step": 10944
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.330759665597716e-06,
+      "loss": 0.731,
+      "step": 10945
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3291271841364536e-06,
+      "loss": 0.4082,
+      "step": 10946
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3274950229370775e-06,
+      "loss": 0.3804,
+      "step": 10947
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.325863182077941e-06,
+      "loss": 0.7178,
+      "step": 10948
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.324231661637389e-06,
+      "loss": 0.4129,
+      "step": 10949
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3226004616937493e-06,
+      "loss": 0.4419,
+      "step": 10950
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3209695823253297e-06,
+      "loss": 0.7103,
+      "step": 10951
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.319339023610434e-06,
+      "loss": 0.7283,
+      "step": 10952
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.317708785627344e-06,
+      "loss": 0.7106,
+      "step": 10953
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.31607886845432e-06,
+      "loss": 0.7612,
+      "step": 10954
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3144492721696107e-06,
+      "loss": 0.7162,
+      "step": 10955
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3128199968514595e-06,
+      "loss": 0.7049,
+      "step": 10956
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3111910425780823e-06,
+      "loss": 0.7651,
+      "step": 10957
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3095624094276834e-06,
+      "loss": 0.3742,
+      "step": 10958
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.307934097478449e-06,
+      "loss": 0.7061,
+      "step": 10959
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.30630610680856e-06,
+      "loss": 0.704,
+      "step": 10960
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3046784374961706e-06,
+      "loss": 0.818,
+      "step": 10961
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.303051089619422e-06,
+      "loss": 0.7247,
+      "step": 10962
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3014240632564443e-06,
+      "loss": 0.707,
+      "step": 10963
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2997973584853436e-06,
+      "loss": 0.4489,
+      "step": 10964
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2981709753842263e-06,
+      "loss": 0.6948,
+      "step": 10965
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2965449140311655e-06,
+      "loss": 0.7205,
+      "step": 10966
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2949191745042344e-06,
+      "loss": 0.7188,
+      "step": 10967
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2932937568814704e-06,
+      "loss": 0.7107,
+      "step": 10968
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.291668661240919e-06,
+      "loss": 0.7467,
+      "step": 10969
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2900438876605967e-06,
+      "loss": 0.4022,
+      "step": 10970
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.288419436218506e-06,
+      "loss": 0.4214,
+      "step": 10971
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2867953069926364e-06,
+      "loss": 0.7191,
+      "step": 10972
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.285171500060956e-06,
+      "loss": 0.6818,
+      "step": 10973
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2835480155014297e-06,
+      "loss": 0.7174,
+      "step": 10974
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.281924853391997e-06,
+      "loss": 0.4003,
+      "step": 10975
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2803020138105823e-06,
+      "loss": 0.7182,
+      "step": 10976
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2786794968350976e-06,
+      "loss": 0.3847,
+      "step": 10977
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.277057302543435e-06,
+      "loss": 0.4065,
+      "step": 10978
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2754354310134804e-06,
+      "loss": 0.7323,
+      "step": 10979
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2738138823230957e-06,
+      "loss": 0.7616,
+      "step": 10980
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2721926565501295e-06,
+      "loss": 0.6885,
+      "step": 10981
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.270571753772415e-06,
+      "loss": 0.4266,
+      "step": 10982
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2689511740677703e-06,
+      "loss": 0.4124,
+      "step": 10983
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2673309175139977e-06,
+      "loss": 0.4317,
+      "step": 10984
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2657109841888845e-06,
+      "loss": 0.7697,
+      "step": 10985
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.264091374170202e-06,
+      "loss": 0.777,
+      "step": 10986
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.262472087535702e-06,
+      "loss": 0.7294,
+      "step": 10987
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2608531243631313e-06,
+      "loss": 0.3886,
+      "step": 10988
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2592344847302125e-06,
+      "loss": 0.7428,
+      "step": 10989
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2576161687146546e-06,
+      "loss": 0.7408,
+      "step": 10990
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2559981763941473e-06,
+      "loss": 0.7147,
+      "step": 10991
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2543805078463763e-06,
+      "loss": 0.7341,
+      "step": 10992
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.252763163149001e-06,
+      "loss": 0.7247,
+      "step": 10993
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2511461423796677e-06,
+      "loss": 0.7255,
+      "step": 10994
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2495294456160076e-06,
+      "loss": 0.7117,
+      "step": 10995
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2479130729356333e-06,
+      "loss": 0.7185,
+      "step": 10996
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2462970244161574e-06,
+      "loss": 0.726,
+      "step": 10997
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.244681300135153e-06,
+      "loss": 0.7174,
+      "step": 10998
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2430659001701925e-06,
+      "loss": 0.7114,
+      "step": 10999
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.241450824598826e-06,
+      "loss": 0.4096,
+      "step": 11000
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2398360734986e-06,
+      "loss": 0.7121,
+      "step": 11001
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.238221646947033e-06,
+      "loss": 0.6873,
+      "step": 11002
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2366075450216317e-06,
+      "loss": 0.728,
+      "step": 11003
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.234993767799888e-06,
+      "loss": 0.7102,
+      "step": 11004
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2333803153592734e-06,
+      "loss": 0.4015,
+      "step": 11005
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.231767187777256e-06,
+      "loss": 0.7063,
+      "step": 11006
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2301543851312757e-06,
+      "loss": 0.7294,
+      "step": 11007
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2285419074987624e-06,
+      "loss": 0.7456,
+      "step": 11008
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.22692975495713e-06,
+      "loss": 0.7597,
+      "step": 11009
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.225317927583772e-06,
+      "loss": 0.4003,
+      "step": 11010
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2237064254560778e-06,
+      "loss": 0.4289,
+      "step": 11011
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2220952486514147e-06,
+      "loss": 0.6574,
+      "step": 11012
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2204843972471256e-06,
+      "loss": 0.7433,
+      "step": 11013
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.218873871320546e-06,
+      "loss": 0.6912,
+      "step": 11014
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2172636709490037e-06,
+      "loss": 0.7131,
+      "step": 11015
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.215653796209798e-06,
+      "loss": 0.6474,
+      "step": 11016
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.214044247180218e-06,
+      "loss": 0.7451,
+      "step": 11017
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2124350239375325e-06,
+      "loss": 0.7017,
+      "step": 11018
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.210826126559007e-06,
+      "loss": 0.704,
+      "step": 11019
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2092175551218774e-06,
+      "loss": 0.7149,
+      "step": 11020
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.207609309703372e-06,
+      "loss": 0.3706,
+      "step": 11021
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2060013903807e-06,
+      "loss": 0.7289,
+      "step": 11022
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.204393797231051e-06,
+      "loss": 0.7499,
+      "step": 11023
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.202786530331613e-06,
+      "loss": 0.3813,
+      "step": 11024
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2011795897595454e-06,
+      "loss": 0.3818,
+      "step": 11025
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1995729755919957e-06,
+      "loss": 0.7076,
+      "step": 11026
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1979666879060945e-06,
+      "loss": 0.3656,
+      "step": 11027
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.19636072677896e-06,
+      "loss": 0.3827,
+      "step": 11028
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1947550922876914e-06,
+      "loss": 0.7253,
+      "step": 11029
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1931497845093753e-06,
+      "loss": 0.6989,
+      "step": 11030
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1915448035210784e-06,
+      "loss": 0.7615,
+      "step": 11031
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.189940149399853e-06,
+      "loss": 0.7024,
+      "step": 11032
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1883358222227423e-06,
+      "loss": 0.6917,
+      "step": 11033
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.186731822066765e-06,
+      "loss": 0.7235,
+      "step": 11034
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.185128149008929e-06,
+      "loss": 0.7107,
+      "step": 11035
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1835248031262244e-06,
+      "loss": 0.7412,
+      "step": 11036
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1819217844956216e-06,
+      "loss": 0.7089,
+      "step": 11037
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1803190931940887e-06,
+      "loss": 0.4304,
+      "step": 11038
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1787167292985645e-06,
+      "loss": 0.7355,
+      "step": 11039
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1771146928859774e-06,
+      "loss": 0.7203,
+      "step": 11040
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.175512984033239e-06,
+      "loss": 0.7105,
+      "step": 11041
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1739116028172468e-06,
+      "loss": 0.7077,
+      "step": 11042
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.17231054931488e-06,
+      "loss": 0.4038,
+      "step": 11043
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1707098236030054e-06,
+      "loss": 0.3947,
+      "step": 11044
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.16910942575847e-06,
+      "loss": 0.4357,
+      "step": 11045
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.167509355858106e-06,
+      "loss": 0.7197,
+      "step": 11046
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.165909613978737e-06,
+      "loss": 0.7079,
+      "step": 11047
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1643102001971614e-06,
+      "loss": 0.718,
+      "step": 11048
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1627111145901656e-06,
+      "loss": 0.6708,
+      "step": 11049
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.161112357234517e-06,
+      "loss": 0.732,
+      "step": 11050
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1595139282069777e-06,
+      "loss": 0.4239,
+      "step": 11051
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1579158275842814e-06,
+      "loss": 0.7269,
+      "step": 11052
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1563180554431525e-06,
+      "loss": 0.3731,
+      "step": 11053
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.154720611860299e-06,
+      "loss": 0.7124,
+      "step": 11054
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.153123496912408e-06,
+      "loss": 0.3529,
+      "step": 11055
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1515267106761626e-06,
+      "loss": 0.7218,
+      "step": 11056
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.149930253228223e-06,
+      "loss": 0.7425,
+      "step": 11057
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.148334124645227e-06,
+      "loss": 0.7127,
+      "step": 11058
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1467383250038018e-06,
+      "loss": 0.7315,
+      "step": 11059
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1451428543805693e-06,
+      "loss": 0.3503,
+      "step": 11060
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.143547712852121e-06,
+      "loss": 0.7084,
+      "step": 11061
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.141952900495039e-06,
+      "loss": 0.7326,
+      "step": 11062
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.140358417385888e-06,
+      "loss": 0.4031,
+      "step": 11063
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.138764263601214e-06,
+      "loss": 0.7505,
+      "step": 11064
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1371704392175585e-06,
+      "loss": 0.7371,
+      "step": 11065
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1355769443114357e-06,
+      "loss": 0.7517,
+      "step": 11066
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1339837789593476e-06,
+      "loss": 0.7175,
+      "step": 11067
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1323909432377797e-06,
+      "loss": 0.415,
+      "step": 11068
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1307984372232003e-06,
+      "loss": 0.699,
+      "step": 11069
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1292062609920703e-06,
+      "loss": 0.4259,
+      "step": 11070
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1276144146208242e-06,
+      "loss": 0.7133,
+      "step": 11071
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1260228981858886e-06,
+      "loss": 0.7845,
+      "step": 11072
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1244317117636625e-06,
+      "loss": 0.7661,
+      "step": 11073
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1228408554305444e-06,
+      "loss": 0.7203,
+      "step": 11074
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1212503292629082e-06,
+      "loss": 0.7056,
+      "step": 11075
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1196601333371137e-06,
+      "loss": 0.7865,
+      "step": 11076
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1180702677295026e-06,
+      "loss": 0.7334,
+      "step": 11077
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1164807325164014e-06,
+      "loss": 0.6945,
+      "step": 11078
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1148915277741274e-06,
+      "loss": 0.7,
+      "step": 11079
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1133026535789747e-06,
+      "loss": 0.6624,
+      "step": 11080
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.111714110007222e-06,
+      "loss": 0.7343,
+      "step": 11081
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.110125897135132e-06,
+      "loss": 0.7282,
+      "step": 11082
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1085380150389588e-06,
+      "loss": 0.7158,
+      "step": 11083
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.106950463794932e-06,
+      "loss": 0.4264,
+      "step": 11084
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.105363243479268e-06,
+      "loss": 0.7318,
+      "step": 11085
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1037763541681676e-06,
+      "loss": 0.4178,
+      "step": 11086
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1021897959378167e-06,
+      "loss": 0.7099,
+      "step": 11087
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1006035688643834e-06,
+      "loss": 0.7082,
+      "step": 11088
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0990176730240205e-06,
+      "loss": 0.4148,
+      "step": 11089
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.097432108492867e-06,
+      "loss": 0.7078,
+      "step": 11090
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0958468753470395e-06,
+      "loss": 0.7294,
+      "step": 11091
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.094261973662651e-06,
+      "loss": 0.75,
+      "step": 11092
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.092677403515787e-06,
+      "loss": 0.7136,
+      "step": 11093
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0910931649825202e-06,
+      "loss": 0.3985,
+      "step": 11094
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0895092581389106e-06,
+      "loss": 0.4203,
+      "step": 11095
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.087925683060995e-06,
+      "loss": 0.7247,
+      "step": 11096
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0863424398248067e-06,
+      "loss": 0.3674,
+      "step": 11097
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0847595285063515e-06,
+      "loss": 0.6867,
+      "step": 11098
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0831769491816235e-06,
+      "loss": 0.7125,
+      "step": 11099
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.081594701926601e-06,
+      "loss": 0.7145,
+      "step": 11100
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0800127868172468e-06,
+      "loss": 0.71,
+      "step": 11101
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0784312039295063e-06,
+      "loss": 0.7535,
+      "step": 11102
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0768499533393106e-06,
+      "loss": 0.4233,
+      "step": 11103
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.075269035122571e-06,
+      "loss": 0.6816,
+      "step": 11104
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0736884493551866e-06,
+      "loss": 0.4083,
+      "step": 11105
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0721081961130428e-06,
+      "loss": 0.762,
+      "step": 11106
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0705282754720056e-06,
+      "loss": 0.7198,
+      "step": 11107
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0689486875079234e-06,
+      "loss": 0.7226,
+      "step": 11108
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0673694322966285e-06,
+      "loss": 0.3845,
+      "step": 11109
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.065790509913944e-06,
+      "loss": 0.7658,
+      "step": 11110
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0642119204356713e-06,
+      "loss": 0.7316,
+      "step": 11111
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0626336639375964e-06,
+      "loss": 0.4193,
+      "step": 11112
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0610557404954886e-06,
+      "loss": 0.4693,
+      "step": 11113
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.059478150185101e-06,
+      "loss": 0.7142,
+      "step": 11114
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0579008930821776e-06,
+      "loss": 0.7031,
+      "step": 11115
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0563239692624403e-06,
+      "loss": 0.7373,
+      "step": 11116
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0547473788015903e-06,
+      "loss": 0.4546,
+      "step": 11117
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.053171121775318e-06,
+      "loss": 0.7956,
+      "step": 11118
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.051595198259303e-06,
+      "loss": 0.7566,
+      "step": 11119
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0500196083292023e-06,
+      "loss": 0.6599,
+      "step": 11120
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0484443520606576e-06,
+      "loss": 0.6923,
+      "step": 11121
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0468694295292956e-06,
+      "loss": 0.711,
+      "step": 11122
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.045294840810723e-06,
+      "loss": 0.7341,
+      "step": 11123
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.043720585980541e-06,
+      "loss": 0.7059,
+      "step": 11124
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.042146665114324e-06,
+      "loss": 0.6964,
+      "step": 11125
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0405730782876364e-06,
+      "loss": 0.7394,
+      "step": 11126
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0389998255760224e-06,
+      "loss": 0.6981,
+      "step": 11127
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.037426907055011e-06,
+      "loss": 0.7378,
+      "step": 11128
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0358543228001214e-06,
+      "loss": 0.7072,
+      "step": 11129
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0342820728868482e-06,
+      "loss": 0.3995,
+      "step": 11130
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.032710157390678e-06,
+      "loss": 0.7332,
+      "step": 11131
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.031138576387067e-06,
+      "loss": 0.7258,
+      "step": 11132
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.029567329951474e-06,
+      "loss": 0.4046,
+      "step": 11133
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0279964181593303e-06,
+      "loss": 0.7441,
+      "step": 11134
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0264258410860547e-06,
+      "loss": 0.3625,
+      "step": 11135
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.024855598807047e-06,
+      "loss": 0.7231,
+      "step": 11136
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0232856913976915e-06,
+      "loss": 0.7022,
+      "step": 11137
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0217161189333643e-06,
+      "loss": 0.4096,
+      "step": 11138
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.020146881489414e-06,
+      "loss": 0.7506,
+      "step": 11139
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0185779791411805e-06,
+      "loss": 0.4274,
+      "step": 11140
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0170094119639803e-06,
+      "loss": 0.7264,
+      "step": 11141
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.015441180033125e-06,
+      "loss": 0.4164,
+      "step": 11142
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0138732834239025e-06,
+      "loss": 0.7418,
+      "step": 11143
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0123057222115835e-06,
+      "loss": 0.3809,
+      "step": 11144
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.010738496471427e-06,
+      "loss": 0.6634,
+      "step": 11145
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0091716062786735e-06,
+      "loss": 0.4076,
+      "step": 11146
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0076050517085465e-06,
+      "loss": 0.7243,
+      "step": 11147
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0060388328362565e-06,
+      "loss": 0.7275,
+      "step": 11148
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.004472949736995e-06,
+      "loss": 0.7359,
+      "step": 11149
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0029074024859352e-06,
+      "loss": 0.4006,
+      "step": 11150
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.001342191158244e-06,
+      "loss": 0.4046,
+      "step": 11151
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.999777315829063e-06,
+      "loss": 0.3911,
+      "step": 11152
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.99821277657352e-06,
+      "loss": 0.714,
+      "step": 11153
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.996648573466727e-06,
+      "loss": 0.7537,
+      "step": 11154
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.995084706583776e-06,
+      "loss": 0.7577,
+      "step": 11155
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.993521175999755e-06,
+      "loss": 0.7489,
+      "step": 11156
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9919579817897217e-06,
+      "loss": 0.67,
+      "step": 11157
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9903951240287245e-06,
+      "loss": 0.6797,
+      "step": 11158
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9888326027917945e-06,
+      "loss": 0.7585,
+      "step": 11159
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9872704181539447e-06,
+      "loss": 0.3579,
+      "step": 11160
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.985708570190182e-06,
+      "loss": 0.7045,
+      "step": 11161
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.984147058975482e-06,
+      "loss": 0.375,
+      "step": 11162
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.982585884584811e-06,
+      "loss": 0.7538,
+      "step": 11163
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.981025047093118e-06,
+      "loss": 0.4169,
+      "step": 11164
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9794645465753445e-06,
+      "loss": 0.6728,
+      "step": 11165
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.977904383106405e-06,
+      "loss": 0.7337,
+      "step": 11166
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9763445567611993e-06,
+      "loss": 0.7023,
+      "step": 11167
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.974785067614612e-06,
+      "loss": 0.6921,
+      "step": 11168
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9732259157415187e-06,
+      "loss": 0.7521,
+      "step": 11169
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9716671012167695e-06,
+      "loss": 0.7471,
+      "step": 11170
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9701086241152012e-06,
+      "loss": 0.6928,
+      "step": 11171
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9685504845116344e-06,
+      "loss": 0.4078,
+      "step": 11172
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9669926824808717e-06,
+      "loss": 0.7244,
+      "step": 11173
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9654352180977066e-06,
+      "loss": 0.7086,
+      "step": 11174
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9638780914369093e-06,
+      "loss": 0.73,
+      "step": 11175
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.962321302573238e-06,
+      "loss": 0.443,
+      "step": 11176
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9607648515814248e-06,
+      "loss": 0.3485,
+      "step": 11177
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.959208738536201e-06,
+      "loss": 0.4242,
+      "step": 11178
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9576529635122718e-06,
+      "loss": 0.6912,
+      "step": 11179
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9560975265843283e-06,
+      "loss": 0.7539,
+      "step": 11180
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.954542427827045e-06,
+      "loss": 0.4184,
+      "step": 11181
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9529876673150782e-06,
+      "loss": 0.6981,
+      "step": 11182
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9514332451230765e-06,
+      "loss": 0.7277,
+      "step": 11183
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9498791613256626e-06,
+      "loss": 0.7669,
+      "step": 11184
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.948325415997447e-06,
+      "loss": 0.7383,
+      "step": 11185
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9467720092130236e-06,
+      "loss": 0.7002,
+      "step": 11186
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9452189410469656e-06,
+      "loss": 0.4013,
+      "step": 11187
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9436662115738413e-06,
+      "loss": 0.7347,
+      "step": 11188
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.942113820868192e-06,
+      "loss": 0.4045,
+      "step": 11189
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9405617690045485e-06,
+      "loss": 0.7569,
+      "step": 11190
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9390100560574197e-06,
+      "loss": 0.6874,
+      "step": 11191
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9374586821013052e-06,
+      "loss": 0.7097,
+      "step": 11192
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.935907647210683e-06,
+      "loss": 0.7455,
+      "step": 11193
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.934356951460017e-06,
+      "loss": 0.6913,
+      "step": 11194
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.932806594923755e-06,
+      "loss": 0.3842,
+      "step": 11195
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9312565776763238e-06,
+      "loss": 0.7489,
+      "step": 11196
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9297068997921463e-06,
+      "loss": 0.6871,
+      "step": 11197
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.928157561345616e-06,
+      "loss": 0.6717,
+      "step": 11198
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9266085624111163e-06,
+      "loss": 0.7176,
+      "step": 11199
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9250599030630088e-06,
+      "loss": 0.7129,
+      "step": 11200
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9235115833756513e-06,
+      "loss": 0.7732,
+      "step": 11201
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.921963603423371e-06,
+      "loss": 0.7542,
+      "step": 11202
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9204159632804874e-06,
+      "loss": 0.7571,
+      "step": 11203
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9188686630213004e-06,
+      "loss": 0.7527,
+      "step": 11204
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.917321702720095e-06,
+      "loss": 0.6789,
+      "step": 11205
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.915775082451138e-06,
+      "loss": 0.6998,
+      "step": 11206
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.91422880228868e-06,
+      "loss": 0.7229,
+      "step": 11207
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.912682862306959e-06,
+      "loss": 0.7045,
+      "step": 11208
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9111372625801903e-06,
+      "loss": 0.7426,
+      "step": 11209
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.909592003182581e-06,
+      "loss": 0.385,
+      "step": 11210
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9080470841883167e-06,
+      "loss": 0.7342,
+      "step": 11211
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9065025056715655e-06,
+      "loss": 0.7113,
+      "step": 11212
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9049582677064823e-06,
+      "loss": 0.7345,
+      "step": 11213
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9034143703672013e-06,
+      "loss": 0.4429,
+      "step": 11214
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9018708137278482e-06,
+      "loss": 0.7611,
+      "step": 11215
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9003275978625265e-06,
+      "loss": 0.7231,
+      "step": 11216
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8987847228453227e-06,
+      "loss": 0.7376,
+      "step": 11217
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.89724218875031e-06,
+      "loss": 0.3903,
+      "step": 11218
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8956999956515396e-06,
+      "loss": 0.7074,
+      "step": 11219
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8941581436230614e-06,
+      "loss": 0.683,
+      "step": 11220
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.892616632738887e-06,
+      "loss": 0.6792,
+      "step": 11221
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8910754630730276e-06,
+      "loss": 0.7144,
+      "step": 11222
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.88953463469947e-06,
+      "loss": 0.7062,
+      "step": 11223
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8879941476921925e-06,
+      "loss": 0.7019,
+      "step": 11224
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.886454002125151e-06,
+      "loss": 0.7055,
+      "step": 11225
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.884914198072285e-06,
+      "loss": 0.7215,
+      "step": 11226
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8833747356075193e-06,
+      "loss": 0.7608,
+      "step": 11227
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8818356148047587e-06,
+      "loss": 0.7644,
+      "step": 11228
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8802968357379023e-06,
+      "loss": 0.4212,
+      "step": 11229
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.878758398480821e-06,
+      "loss": 0.4011,
+      "step": 11230
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.877220303107373e-06,
+      "loss": 0.3782,
+      "step": 11231
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8756825496914e-06,
+      "loss": 0.6966,
+      "step": 11232
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.874145138306732e-06,
+      "loss": 0.7224,
+      "step": 11233
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8726080690271764e-06,
+      "loss": 0.6797,
+      "step": 11234
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.871071341926528e-06,
+      "loss": 0.7328,
+      "step": 11235
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.869534957078557e-06,
+      "loss": 0.6997,
+      "step": 11236
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.867998914557031e-06,
+      "loss": 0.6917,
+      "step": 11237
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8664632144356906e-06,
+      "loss": 0.7198,
+      "step": 11238
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8649278567882642e-06,
+      "loss": 0.7621,
+      "step": 11239
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8633928416884626e-06,
+      "loss": 0.3638,
+      "step": 11240
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8618581692099768e-06,
+      "loss": 0.7298,
+      "step": 11241
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.860323839426491e-06,
+      "loss": 0.7314,
+      "step": 11242
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.858789852411664e-06,
+      "loss": 0.4369,
+      "step": 11243
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.857256208239141e-06,
+      "loss": 0.7224,
+      "step": 11244
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.855722906982551e-06,
+      "loss": 0.4068,
+      "step": 11245
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8541899487155024e-06,
+      "loss": 0.694,
+      "step": 11246
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.852657333511598e-06,
+      "loss": 0.6622,
+      "step": 11247
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8511250614444143e-06,
+      "loss": 0.6967,
+      "step": 11248
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8495931325875125e-06,
+      "loss": 0.4221,
+      "step": 11249
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.84806154701444e-06,
+      "loss": 0.3868,
+      "step": 11250
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.846530304798727e-06,
+      "loss": 0.3924,
+      "step": 11251
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8449994060138855e-06,
+      "loss": 0.764,
+      "step": 11252
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.843468850733414e-06,
+      "loss": 0.7859,
+      "step": 11253
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8419386390307925e-06,
+      "loss": 0.7398,
+      "step": 11254
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8404087709794816e-06,
+      "loss": 0.6952,
+      "step": 11255
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8388792466529357e-06,
+      "loss": 0.6768,
+      "step": 11256
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8373500661245812e-06,
+      "loss": 0.7217,
+      "step": 11257
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8358212294678322e-06,
+      "loss": 0.7397,
+      "step": 11258
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8342927367560858e-06,
+      "loss": 0.3746,
+      "step": 11259
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8327645880627284e-06,
+      "loss": 0.7616,
+      "step": 11260
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.831236783461121e-06,
+      "loss": 0.4511,
+      "step": 11261
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.829709323024613e-06,
+      "loss": 0.7006,
+      "step": 11262
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8281822068265364e-06,
+      "loss": 0.421,
+      "step": 11263
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.826655434940202e-06,
+      "loss": 0.7803,
+      "step": 11264
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.825129007438919e-06,
+      "loss": 0.7296,
+      "step": 11265
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8236029243959608e-06,
+      "loss": 0.6768,
+      "step": 11266
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.822077185884594e-06,
+      "loss": 0.7053,
+      "step": 11267
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.820551791978068e-06,
+      "loss": 0.3618,
+      "step": 11268
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.819026742749619e-06,
+      "loss": 0.7758,
+      "step": 11269
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8175020382724605e-06,
+      "loss": 0.7304,
+      "step": 11270
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.815977678619792e-06,
+      "loss": 0.6814,
+      "step": 11271
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.814453663864798e-06,
+      "loss": 0.7598,
+      "step": 11272
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.812929994080641e-06,
+      "loss": 0.7843,
+      "step": 11273
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8114066693404753e-06,
+      "loss": 0.7535,
+      "step": 11274
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.809883689717433e-06,
+      "loss": 0.6572,
+      "step": 11275
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8083610552846297e-06,
+      "loss": 0.7445,
+      "step": 11276
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.806838766115166e-06,
+      "loss": 0.7644,
+      "step": 11277
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8053168222821237e-06,
+      "loss": 0.7485,
+      "step": 11278
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.803795223858574e-06,
+      "loss": 0.3904,
+      "step": 11279
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8022739709175684e-06,
+      "loss": 0.3862,
+      "step": 11280
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8007530635321345e-06,
+      "loss": 0.7301,
+      "step": 11281
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7992325017752887e-06,
+      "loss": 0.7524,
+      "step": 11282
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.797712285720038e-06,
+      "loss": 0.7215,
+      "step": 11283
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7961924154393637e-06,
+      "loss": 0.6709,
+      "step": 11284
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.794672891006234e-06,
+      "loss": 0.7425,
+      "step": 11285
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.793153712493599e-06,
+      "loss": 0.3958,
+      "step": 11286
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.79163487997439e-06,
+      "loss": 0.7553,
+      "step": 11287
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.790116393521529e-06,
+      "loss": 0.7081,
+      "step": 11288
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.788598253207917e-06,
+      "loss": 0.4197,
+      "step": 11289
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7870804591064372e-06,
+      "loss": 0.7612,
+      "step": 11290
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7855630112899536e-06,
+      "loss": 0.7263,
+      "step": 11291
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7840459098313234e-06,
+      "loss": 0.7193,
+      "step": 11292
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7825291548033783e-06,
+      "loss": 0.7566,
+      "step": 11293
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7810127462789373e-06,
+      "loss": 0.6963,
+      "step": 11294
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7794966843308003e-06,
+      "loss": 0.7167,
+      "step": 11295
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7779809690317525e-06,
+      "loss": 0.7316,
+      "step": 11296
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.776465600454562e-06,
+      "loss": 0.7153,
+      "step": 11297
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.77495057867198e-06,
+      "loss": 0.6899,
+      "step": 11298
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7734359037567404e-06,
+      "loss": 0.7859,
+      "step": 11299
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7719215757815587e-06,
+      "loss": 0.6369,
+      "step": 11300
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7704075948191435e-06,
+      "loss": 0.6662,
+      "step": 11301
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7688939609421738e-06,
+      "loss": 0.3617,
+      "step": 11302
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.767380674223321e-06,
+      "loss": 0.7309,
+      "step": 11303
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7658677347352336e-06,
+      "loss": 0.679,
+      "step": 11304
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.764355142550544e-06,
+      "loss": 0.7601,
+      "step": 11305
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7628428977418777e-06,
+      "loss": 0.7417,
+      "step": 11306
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.761331000381832e-06,
+      "loss": 0.6982,
+      "step": 11307
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7598194505429908e-06,
+      "loss": 0.3949,
+      "step": 11308
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.758308248297924e-06,
+      "loss": 0.6957,
+      "step": 11309
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7567973937191807e-06,
+      "loss": 0.7826,
+      "step": 11310
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.755286886879297e-06,
+      "loss": 0.7294,
+      "step": 11311
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7537767278507897e-06,
+      "loss": 0.7134,
+      "step": 11312
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.752266916706162e-06,
+      "loss": 0.735,
+      "step": 11313
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7507574535178926e-06,
+      "loss": 0.7482,
+      "step": 11314
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7492483383584577e-06,
+      "loss": 0.7508,
+      "step": 11315
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.747739571300304e-06,
+      "loss": 0.7162,
+      "step": 11316
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.746231152415867e-06,
+      "loss": 0.7053,
+      "step": 11317
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7447230817775605e-06,
+      "loss": 0.7089,
+      "step": 11318
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7432153594577914e-06,
+      "loss": 0.6958,
+      "step": 11319
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.741707985528942e-06,
+      "loss": 0.7063,
+      "step": 11320
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7402009600633785e-06,
+      "loss": 0.3775,
+      "step": 11321
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7386942831334516e-06,
+      "loss": 0.7609,
+      "step": 11322
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7371879548114944e-06,
+      "loss": 0.6734,
+      "step": 11323
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7356819751698304e-06,
+      "loss": 0.7455,
+      "step": 11324
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7341763442807535e-06,
+      "loss": 0.7702,
+      "step": 11325
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7326710622165486e-06,
+      "loss": 0.7584,
+      "step": 11326
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7311661290494808e-06,
+      "loss": 0.4301,
+      "step": 11327
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.729661544851807e-06,
+      "loss": 0.73,
+      "step": 11328
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.728157309695757e-06,
+      "loss": 0.7252,
+      "step": 11329
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.726653423653548e-06,
+      "loss": 0.3899,
+      "step": 11330
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7251498867973793e-06,
+      "loss": 0.3897,
+      "step": 11331
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7236466991994315e-06,
+      "loss": 0.7408,
+      "step": 11332
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7221438609318773e-06,
+      "loss": 0.7533,
+      "step": 11333
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7206413720668635e-06,
+      "loss": 0.7113,
+      "step": 11334
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.719139232676523e-06,
+      "loss": 0.741,
+      "step": 11335
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7176374428329723e-06,
+      "loss": 0.7359,
+      "step": 11336
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7161360026083074e-06,
+      "loss": 0.7483,
+      "step": 11337
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7146349120746175e-06,
+      "loss": 0.6792,
+      "step": 11338
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.713134171303967e-06,
+      "loss": 0.7181,
+      "step": 11339
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.711633780368401e-06,
+      "loss": 0.4305,
+      "step": 11340
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7101337393399505e-06,
+      "loss": 0.6674,
+      "step": 11341
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.708634048290637e-06,
+      "loss": 0.7057,
+      "step": 11342
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.707134707292456e-06,
+      "loss": 0.7406,
+      "step": 11343
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7056357164173907e-06,
+      "loss": 0.7007,
+      "step": 11344
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.704137075737404e-06,
+      "loss": 0.759,
+      "step": 11345
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.702638785324443e-06,
+      "loss": 0.7906,
+      "step": 11346
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.701140845250444e-06,
+      "loss": 0.757,
+      "step": 11347
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6996432555873185e-06,
+      "loss": 0.7305,
+      "step": 11348
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.698146016406966e-06,
+      "loss": 0.7422,
+      "step": 11349
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6966491277812623e-06,
+      "loss": 0.7781,
+      "step": 11350
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6951525897820786e-06,
+      "loss": 0.3576,
+      "step": 11351
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6936564024812585e-06,
+      "loss": 0.7177,
+      "step": 11352
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6921605659506334e-06,
+      "loss": 0.7274,
+      "step": 11353
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6906650802620172e-06,
+      "loss": 0.6436,
+      "step": 11354
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6891699454872046e-06,
+      "loss": 0.7272,
+      "step": 11355
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.687675161697977e-06,
+      "loss": 0.728,
+      "step": 11356
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.686180728966098e-06,
+      "loss": 0.7359,
+      "step": 11357
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6846866473633126e-06,
+      "loss": 0.7614,
+      "step": 11358
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.683192916961348e-06,
+      "loss": 0.7389,
+      "step": 11359
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.681699537831921e-06,
+      "loss": 0.7263,
+      "step": 11360
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.680206510046727e-06,
+      "loss": 0.7092,
+      "step": 11361
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6787138336774434e-06,
+      "loss": 0.7365,
+      "step": 11362
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6772215087957308e-06,
+      "loss": 0.7203,
+      "step": 11363
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.675729535473233e-06,
+      "loss": 0.736,
+      "step": 11364
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.674237913781583e-06,
+      "loss": 0.7444,
+      "step": 11365
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.672746643792391e-06,
+      "loss": 0.373,
+      "step": 11366
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6712557255772487e-06,
+      "loss": 0.4151,
+      "step": 11367
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6697651592077354e-06,
+      "loss": 0.4351,
+      "step": 11368
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.668274944755411e-06,
+      "loss": 0.7151,
+      "step": 11369
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6667850822918197e-06,
+      "loss": 0.7133,
+      "step": 11370
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.665295571888488e-06,
+      "loss": 0.7374,
+      "step": 11371
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.663806413616925e-06,
+      "loss": 0.6731,
+      "step": 11372
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.662317607548621e-06,
+      "loss": 0.7264,
+      "step": 11373
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.66082915375506e-06,
+      "loss": 0.4194,
+      "step": 11374
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6593410523076957e-06,
+      "loss": 0.7717,
+      "step": 11375
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6578533032779718e-06,
+      "loss": 0.7129,
+      "step": 11376
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.656365906737313e-06,
+      "loss": 0.7007,
+      "step": 11377
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.654878862757124e-06,
+      "loss": 0.7598,
+      "step": 11378
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.653392171408804e-06,
+      "loss": 0.7477,
+      "step": 11379
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.651905832763724e-06,
+      "loss": 0.747,
+      "step": 11380
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6504198468932406e-06,
+      "loss": 0.7248,
+      "step": 11381
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6489342138686925e-06,
+      "loss": 0.7041,
+      "step": 11382
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6474489337614083e-06,
+      "loss": 0.419,
+      "step": 11383
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.645964006642697e-06,
+      "loss": 0.3497,
+      "step": 11384
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.644479432583841e-06,
+      "loss": 0.7185,
+      "step": 11385
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6429952116561118e-06,
+      "loss": 0.7427,
+      "step": 11386
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.641511343930774e-06,
+      "loss": 0.7336,
+      "step": 11387
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6400278294790625e-06,
+      "loss": 0.7339,
+      "step": 11388
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6385446683721995e-06,
+      "loss": 0.4031,
+      "step": 11389
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.637061860681389e-06,
+      "loss": 0.4001,
+      "step": 11390
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6355794064778175e-06,
+      "loss": 0.7864,
+      "step": 11391
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6340973058326615e-06,
+      "loss": 0.3844,
+      "step": 11392
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.632615558817072e-06,
+      "loss": 0.7664,
+      "step": 11393
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.631134165502186e-06,
+      "loss": 0.3837,
+      "step": 11394
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.629653125959125e-06,
+      "loss": 0.7166,
+      "step": 11395
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.628172440258988e-06,
+      "loss": 0.7226,
+      "step": 11396
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.626692108472868e-06,
+      "loss": 0.7559,
+      "step": 11397
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.62521213067183e-06,
+      "loss": 0.7312,
+      "step": 11398
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6237325069269305e-06,
+      "loss": 0.715,
+      "step": 11399
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6222532373091958e-06,
+      "loss": 0.7226,
+      "step": 11400
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.620774321889652e-06,
+      "loss": 0.4046,
+      "step": 11401
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.619295760739298e-06,
+      "loss": 0.7437,
+      "step": 11402
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6178175539291183e-06,
+      "loss": 0.4353,
+      "step": 11403
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6163397015300797e-06,
+      "loss": 0.413,
+      "step": 11404
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.614862203613129e-06,
+      "loss": 0.7149,
+      "step": 11405
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.613385060249207e-06,
+      "loss": 0.7195,
+      "step": 11406
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6119082715092247e-06,
+      "loss": 0.7308,
+      "step": 11407
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6104318374640836e-06,
+      "loss": 0.7199,
+      "step": 11408
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.60895575818466e-06,
+      "loss": 0.379,
+      "step": 11409
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.607480033741827e-06,
+      "loss": 0.7101,
+      "step": 11410
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6060046642064297e-06,
+      "loss": 0.7244,
+      "step": 11411
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6045296496492987e-06,
+      "loss": 0.7182,
+      "step": 11412
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.603054990141247e-06,
+      "loss": 0.7258,
+      "step": 11413
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6015806857530736e-06,
+      "loss": 0.7181,
+      "step": 11414
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6001067365555555e-06,
+      "loss": 0.4207,
+      "step": 11415
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5986331426194578e-06,
+      "loss": 0.7524,
+      "step": 11416
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.597159904015526e-06,
+      "loss": 0.3997,
+      "step": 11417
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5956870208144848e-06,
+      "loss": 0.7206,
+      "step": 11418
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5942144930870526e-06,
+      "loss": 0.7003,
+      "step": 11419
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5927423209039213e-06,
+      "loss": 0.456,
+      "step": 11420
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5912705043357666e-06,
+      "loss": 0.6959,
+      "step": 11421
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5897990434532518e-06,
+      "loss": 0.7301,
+      "step": 11422
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5883279383270155e-06,
+      "loss": 0.7667,
+      "step": 11423
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5868571890276893e-06,
+      "loss": 0.7158,
+      "step": 11424
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5853867956258806e-06,
+      "loss": 0.7528,
+      "step": 11425
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5839167581921807e-06,
+      "loss": 0.7354,
+      "step": 11426
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5824470767971653e-06,
+      "loss": 0.7439,
+      "step": 11427
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.58097775151139e-06,
+      "loss": 0.8033,
+      "step": 11428
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5795087824054034e-06,
+      "loss": 0.7321,
+      "step": 11429
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5780401695497194e-06,
+      "loss": 0.7354,
+      "step": 11430
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5765719130148503e-06,
+      "loss": 0.6728,
+      "step": 11431
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.57510401287128e-06,
+      "loss": 0.6806,
+      "step": 11432
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5736364691894887e-06,
+      "loss": 0.7688,
+      "step": 11433
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.572169282039927e-06,
+      "loss": 0.7019,
+      "step": 11434
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5707024514930358e-06,
+      "loss": 0.7307,
+      "step": 11435
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.569235977619232e-06,
+      "loss": 0.7766,
+      "step": 11436
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.567769860488921e-06,
+      "loss": 0.3913,
+      "step": 11437
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.566304100172493e-06,
+      "loss": 0.7264,
+      "step": 11438
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.564838696740315e-06,
+      "loss": 0.7272,
+      "step": 11439
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5633736502627405e-06,
+      "loss": 0.7566,
+      "step": 11440
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5619089608101e-06,
+      "loss": 0.3663,
+      "step": 11441
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5604446284527194e-06,
+      "loss": 0.6954,
+      "step": 11442
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5589806532608997e-06,
+      "loss": 0.7038,
+      "step": 11443
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5575170353049185e-06,
+      "loss": 0.6967,
+      "step": 11444
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.556053774655043e-06,
+      "loss": 0.4163,
+      "step": 11445
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5545908713815283e-06,
+      "loss": 0.7174,
+      "step": 11446
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5531283255546036e-06,
+      "loss": 0.7141,
+      "step": 11447
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.551666137244485e-06,
+      "loss": 0.4298,
+      "step": 11448
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.550204306521371e-06,
+      "loss": 0.7413,
+      "step": 11449
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5487428334554387e-06,
+      "loss": 0.6863,
+      "step": 11450
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5472817181168585e-06,
+      "loss": 0.7511,
+      "step": 11451
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5458209605757732e-06,
+      "loss": 0.7214,
+      "step": 11452
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5443605609023135e-06,
+      "loss": 0.6933,
+      "step": 11453
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5429005191665924e-06,
+      "loss": 0.7479,
+      "step": 11454
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5414408354387e-06,
+      "loss": 0.6917,
+      "step": 11455
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5399815097887202e-06,
+      "loss": 0.4166,
+      "step": 11456
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.538522542286713e-06,
+      "loss": 0.7352,
+      "step": 11457
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5370639330027225e-06,
+      "loss": 0.7406,
+      "step": 11458
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5356056820067686e-06,
+      "loss": 0.7065,
+      "step": 11459
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5341477893688672e-06,
+      "loss": 0.669,
+      "step": 11460
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5326902551590082e-06,
+      "loss": 0.6931,
+      "step": 11461
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5312330794471663e-06,
+      "loss": 0.6744,
+      "step": 11462
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5297762623032996e-06,
+      "loss": 0.7198,
+      "step": 11463
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5283198037973434e-06,
+      "loss": 0.7131,
+      "step": 11464
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5268637039992296e-06,
+      "loss": 0.6714,
+      "step": 11465
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5254079629788595e-06,
+      "loss": 0.7638,
+      "step": 11466
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5239525808061215e-06,
+      "loss": 0.7125,
+      "step": 11467
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5224975575508847e-06,
+      "loss": 0.7554,
+      "step": 11468
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5210428932830087e-06,
+      "loss": 0.6989,
+      "step": 11469
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.519588588072329e-06,
+      "loss": 0.4194,
+      "step": 11470
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.518134641988663e-06,
+      "loss": 0.3955,
+      "step": 11471
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.516681055101815e-06,
+      "loss": 0.3877,
+      "step": 11472
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.51522782748157e-06,
+      "loss": 0.7016,
+      "step": 11473
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5137749591976957e-06,
+      "loss": 0.3388,
+      "step": 11474
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5123224503199417e-06,
+      "loss": 0.4164,
+      "step": 11475
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5108703009180433e-06,
+      "loss": 0.7596,
+      "step": 11476
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.509418511061713e-06,
+      "loss": 0.7449,
+      "step": 11477
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.507967080820656e-06,
+      "loss": 0.6937,
+      "step": 11478
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.506516010264549e-06,
+      "loss": 0.7102,
+      "step": 11479
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5050652994630597e-06,
+      "loss": 0.7359,
+      "step": 11480
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.503614948485834e-06,
+      "loss": 0.6651,
+      "step": 11481
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.502164957402499e-06,
+      "loss": 0.3671,
+      "step": 11482
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.500715326282671e-06,
+      "loss": 0.7346,
+      "step": 11483
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.499266055195946e-06,
+      "loss": 0.7478,
+      "step": 11484
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4978171442118992e-06,
+      "loss": 0.3716,
+      "step": 11485
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4963685934000924e-06,
+      "loss": 0.7601,
+      "step": 11486
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4949204028300666e-06,
+      "loss": 0.7326,
+      "step": 11487
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.493472572571356e-06,
+      "loss": 0.4607,
+      "step": 11488
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4920251026934606e-06,
+      "loss": 0.6773,
+      "step": 11489
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4905779932658756e-06,
+      "loss": 0.7627,
+      "step": 11490
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.489131244358073e-06,
+      "loss": 0.7268,
+      "step": 11491
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.487684856039513e-06,
+      "loss": 0.3852,
+      "step": 11492
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4862388283796345e-06,
+      "loss": 0.7069,
+      "step": 11493
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4847931614478593e-06,
+      "loss": 0.7073,
+      "step": 11494
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4833478553135924e-06,
+      "loss": 0.704,
+      "step": 11495
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.481902910046219e-06,
+      "loss": 0.6878,
+      "step": 11496
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.480458325715115e-06,
+      "loss": 0.7497,
+      "step": 11497
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4790141023896297e-06,
+      "loss": 0.6866,
+      "step": 11498
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4775702401391e-06,
+      "loss": 0.7172,
+      "step": 11499
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4761267390328403e-06,
+      "loss": 0.6931,
+      "step": 11500
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4746835991401587e-06,
+      "loss": 0.4184,
+      "step": 11501
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.473240820530334e-06,
+      "loss": 0.7559,
+      "step": 11502
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4717984032726383e-06,
+      "loss": 0.7012,
+      "step": 11503
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.470356347436309e-06,
+      "loss": 0.4446,
+      "step": 11504
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4689146530905884e-06,
+      "loss": 0.7309,
+      "step": 11505
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4674733203046874e-06,
+      "loss": 0.7337,
+      "step": 11506
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4660323491478023e-06,
+      "loss": 0.7013,
+      "step": 11507
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.464591739689114e-06,
+      "loss": 0.4155,
+      "step": 11508
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4631514919977793e-06,
+      "loss": 0.4031,
+      "step": 11509
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4617116061429525e-06,
+      "loss": 0.3983,
+      "step": 11510
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4602720821937555e-06,
+      "loss": 0.7659,
+      "step": 11511
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4588329202192986e-06,
+      "loss": 0.6985,
+      "step": 11512
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4573941202886754e-06,
+      "loss": 0.731,
+      "step": 11513
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4559556824709586e-06,
+      "loss": 0.3921,
+      "step": 11514
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4545176068352106e-06,
+      "loss": 0.7452,
+      "step": 11515
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4530798934504696e-06,
+      "loss": 0.6685,
+      "step": 11516
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4516425423857595e-06,
+      "loss": 0.7092,
+      "step": 11517
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4502055537100842e-06,
+      "loss": 0.4479,
+      "step": 11518
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4487689274924343e-06,
+      "loss": 0.7503,
+      "step": 11519
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.44733266380178e-06,
+      "loss": 0.7354,
+      "step": 11520
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4458967627070752e-06,
+      "loss": 0.7257,
+      "step": 11521
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4444612242772546e-06,
+      "loss": 0.712,
+      "step": 11522
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.443026048581235e-06,
+      "loss": 0.706,
+      "step": 11523
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4415912356879247e-06,
+      "loss": 0.7212,
+      "step": 11524
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.440156785666202e-06,
+      "loss": 0.7412,
+      "step": 11525
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4387226985849364e-06,
+      "loss": 0.7625,
+      "step": 11526
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4372889745129746e-06,
+      "loss": 0.7423,
+      "step": 11527
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.435855613519147e-06,
+      "loss": 0.6624,
+      "step": 11528
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4344226156722726e-06,
+      "loss": 0.401,
+      "step": 11529
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.432989981041146e-06,
+      "loss": 0.723,
+      "step": 11530
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.431557709694546e-06,
+      "loss": 0.7183,
+      "step": 11531
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.430125801701232e-06,
+      "loss": 0.7096,
+      "step": 11532
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4286942571299564e-06,
+      "loss": 0.7269,
+      "step": 11533
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4272630760494387e-06,
+      "loss": 0.7241,
+      "step": 11534
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.425832258528391e-06,
+      "loss": 0.7579,
+      "step": 11535
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.424401804635501e-06,
+      "loss": 0.7115,
+      "step": 11536
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4229717144394517e-06,
+      "loss": 0.6913,
+      "step": 11537
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4215419880088954e-06,
+      "loss": 0.7867,
+      "step": 11538
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.420112625412472e-06,
+      "loss": 0.7507,
+      "step": 11539
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.418683626718804e-06,
+      "loss": 0.7059,
+      "step": 11540
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.417254991996495e-06,
+      "loss": 0.6751,
+      "step": 11541
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4158267213141362e-06,
+      "loss": 0.3852,
+      "step": 11542
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.414398814740294e-06,
+      "loss": 0.7207,
+      "step": 11543
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4129712723435227e-06,
+      "loss": 0.7703,
+      "step": 11544
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4115440941923563e-06,
+      "loss": 0.3719,
+      "step": 11545
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4101172803553087e-06,
+      "loss": 0.7391,
+      "step": 11546
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.408690830900886e-06,
+      "loss": 0.7317,
+      "step": 11547
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4072647458975705e-06,
+      "loss": 0.7537,
+      "step": 11548
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.405839025413822e-06,
+      "loss": 0.7205,
+      "step": 11549
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4044136695180874e-06,
+      "loss": 0.7246,
+      "step": 11550
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4029886782788025e-06,
+      "loss": 0.3996,
+      "step": 11551
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4015640517643767e-06,
+      "loss": 0.4407,
+      "step": 11552
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4001397900432044e-06,
+      "loss": 0.6873,
+      "step": 11553
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3987158931836642e-06,
+      "loss": 0.6901,
+      "step": 11554
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.397292361254111e-06,
+      "loss": 0.7123,
+      "step": 11555
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.395869194322895e-06,
+      "loss": 0.4022,
+      "step": 11556
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.394446392458337e-06,
+      "loss": 0.694,
+      "step": 11557
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.393023955728746e-06,
+      "loss": 0.4154,
+      "step": 11558
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.391601884202406e-06,
+      "loss": 0.7296,
+      "step": 11559
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3901801779475973e-06,
+      "loss": 0.689,
+      "step": 11560
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.388758837032571e-06,
+      "loss": 0.7442,
+      "step": 11561
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.387337861525566e-06,
+      "loss": 0.7197,
+      "step": 11562
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3859172514947946e-06,
+      "loss": 0.4064,
+      "step": 11563
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3844970070084683e-06,
+      "loss": 0.7107,
+      "step": 11564
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3830771281347665e-06,
+      "loss": 0.41,
+      "step": 11565
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.381657614941858e-06,
+      "loss": 0.7284,
+      "step": 11566
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3802384674978906e-06,
+      "loss": 0.77,
+      "step": 11567
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3788196858709944e-06,
+      "loss": 0.7288,
+      "step": 11568
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.377401270129289e-06,
+      "loss": 0.7481,
+      "step": 11569
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3759832203408697e-06,
+      "loss": 0.768,
+      "step": 11570
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3745655365738127e-06,
+      "loss": 0.7182,
+      "step": 11571
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.373148218896182e-06,
+      "loss": 0.4106,
+      "step": 11572
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3717312673760175e-06,
+      "loss": 0.3903,
+      "step": 11573
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3703146820813517e-06,
+      "loss": 0.7724,
+      "step": 11574
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3688984630801915e-06,
+      "loss": 0.7225,
+      "step": 11575
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3674826104405258e-06,
+      "loss": 0.804,
+      "step": 11576
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3660671242303314e-06,
+      "loss": 0.7311,
+      "step": 11577
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3646520045175612e-06,
+      "loss": 0.7577,
+      "step": 11578
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.363237251370156e-06,
+      "loss": 0.6788,
+      "step": 11579
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.361822864856036e-06,
+      "loss": 0.6621,
+      "step": 11580
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.360408845043104e-06,
+      "loss": 0.6954,
+      "step": 11581
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3589951919992425e-06,
+      "loss": 0.764,
+      "step": 11582
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3575819057923265e-06,
+      "loss": 0.6566,
+      "step": 11583
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3561689864902037e-06,
+      "loss": 0.7248,
+      "step": 11584
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3547564341607055e-06,
+      "loss": 0.6896,
+      "step": 11585
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3533442488716486e-06,
+      "loss": 0.6911,
+      "step": 11586
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3519324306908252e-06,
+      "loss": 0.3993,
+      "step": 11587
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.350520979686024e-06,
+      "loss": 0.4268,
+      "step": 11588
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3491098959250037e-06,
+      "loss": 0.7399,
+      "step": 11589
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.347699179475508e-06,
+      "loss": 0.7277,
+      "step": 11590
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.346288830405261e-06,
+      "loss": 0.3922,
+      "step": 11591
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.344878848781982e-06,
+      "loss": 0.7357,
+      "step": 11592
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3434692346733536e-06,
+      "loss": 0.6984,
+      "step": 11593
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.342059988147053e-06,
+      "loss": 0.7512,
+      "step": 11594
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.340651109270734e-06,
+      "loss": 0.7456,
+      "step": 11595
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.33924259811204e-06,
+      "loss": 0.6504,
+      "step": 11596
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.33783445473859e-06,
+      "loss": 0.7069,
+      "step": 11597
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.336426679217989e-06,
+      "loss": 0.7285,
+      "step": 11598
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.33501927161782e-06,
+      "loss": 0.7143,
+      "step": 11599
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.333612232005651e-06,
+      "loss": 0.4258,
+      "step": 11600
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.332205560449037e-06,
+      "loss": 0.6718,
+      "step": 11601
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.330799257015508e-06,
+      "loss": 0.4185,
+      "step": 11602
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3293933217725808e-06,
+      "loss": 0.7349,
+      "step": 11603
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3279877547877494e-06,
+      "loss": 0.7263,
+      "step": 11604
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.326582556128495e-06,
+      "loss": 0.7578,
+      "step": 11605
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3251777258622823e-06,
+      "loss": 0.7255,
+      "step": 11606
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.323773264056557e-06,
+      "loss": 0.3508,
+      "step": 11607
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.322369170778741e-06,
+      "loss": 0.6929,
+      "step": 11608
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3209654460962417e-06,
+      "loss": 0.6794,
+      "step": 11609
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.319562090076458e-06,
+      "loss": 0.7506,
+      "step": 11610
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.318159102786759e-06,
+      "loss": 0.8075,
+      "step": 11611
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3167564842945025e-06,
+      "loss": 0.7171,
+      "step": 11612
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.315354234667024e-06,
+      "loss": 0.6857,
+      "step": 11613
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3139523539716447e-06,
+      "loss": 0.7372,
+      "step": 11614
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3125508422756714e-06,
+      "loss": 0.7056,
+      "step": 11615
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3111496996463867e-06,
+      "loss": 0.6986,
+      "step": 11616
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3097489261510576e-06,
+      "loss": 0.3736,
+      "step": 11617
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3083485218569336e-06,
+      "loss": 0.8142,
+      "step": 11618
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3069484868312453e-06,
+      "loss": 0.7358,
+      "step": 11619
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3055488211412116e-06,
+      "loss": 0.7386,
+      "step": 11620
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.304149524854026e-06,
+      "loss": 0.4726,
+      "step": 11621
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.302750598036867e-06,
+      "loss": 0.7289,
+      "step": 11622
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.301352040756897e-06,
+      "loss": 0.7519,
+      "step": 11623
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2999538530812595e-06,
+      "loss": 0.7081,
+      "step": 11624
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2985560350770776e-06,
+      "loss": 0.7333,
+      "step": 11625
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.297158586811462e-06,
+      "loss": 0.7512,
+      "step": 11626
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.295761508351498e-06,
+      "loss": 0.7184,
+      "step": 11627
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.294364799764265e-06,
+      "loss": 0.4021,
+      "step": 11628
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2929684611168136e-06,
+      "loss": 0.7401,
+      "step": 11629
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2915724924761816e-06,
+      "loss": 0.7144,
+      "step": 11630
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2901768939093873e-06,
+      "loss": 0.772,
+      "step": 11631
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2887816654834295e-06,
+      "loss": 0.7409,
+      "step": 11632
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.287386807265297e-06,
+      "loss": 0.7446,
+      "step": 11633
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2859923193219526e-06,
+      "loss": 0.7332,
+      "step": 11634
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.284598201720345e-06,
+      "loss": 0.727,
+      "step": 11635
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2832044545274037e-06,
+      "loss": 0.7597,
+      "step": 11636
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2818110778100423e-06,
+      "loss": 0.6771,
+      "step": 11637
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.280418071635153e-06,
+      "loss": 0.6776,
+      "step": 11638
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.279025436069615e-06,
+      "loss": 0.7542,
+      "step": 11639
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.277633171180287e-06,
+      "loss": 0.4282,
+      "step": 11640
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.276241277034006e-06,
+      "loss": 0.7398,
+      "step": 11641
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.274849753697602e-06,
+      "loss": 0.7682,
+      "step": 11642
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2734586012378778e-06,
+      "loss": 0.7581,
+      "step": 11643
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2720678197216217e-06,
+      "loss": 0.3957,
+      "step": 11644
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2706774092156027e-06,
+      "loss": 0.7268,
+      "step": 11645
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2692873697865703e-06,
+      "loss": 0.6974,
+      "step": 11646
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.267897701501266e-06,
+      "loss": 0.4312,
+      "step": 11647
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2665084044264017e-06,
+      "loss": 0.7547,
+      "step": 11648
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2651194786286766e-06,
+      "loss": 0.746,
+      "step": 11649
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.263730924174771e-06,
+      "loss": 0.655,
+      "step": 11650
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2623427411313505e-06,
+      "loss": 0.7068,
+      "step": 11651
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.260954929565061e-06,
+      "loss": 0.4067,
+      "step": 11652
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2595674895425255e-06,
+      "loss": 0.728,
+      "step": 11653
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2581804211303536e-06,
+      "loss": 0.7272,
+      "step": 11654
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.256793724395141e-06,
+      "loss": 0.4071,
+      "step": 11655
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2554073994034607e-06,
+      "loss": 0.738,
+      "step": 11656
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.254021446221868e-06,
+      "loss": 0.3673,
+      "step": 11657
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2526358649169012e-06,
+      "loss": 0.7305,
+      "step": 11658
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2512506555550774e-06,
+      "loss": 0.7061,
+      "step": 11659
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.249865818202905e-06,
+      "loss": 0.6743,
+      "step": 11660
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.248481352926867e-06,
+      "loss": 0.76,
+      "step": 11661
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.247097259793428e-06,
+      "loss": 0.7412,
+      "step": 11662
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2457135388690377e-06,
+      "loss": 0.7031,
+      "step": 11663
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2443301902201255e-06,
+      "loss": 0.7206,
+      "step": 11664
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2429472139131094e-06,
+      "loss": 0.7016,
+      "step": 11665
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2415646100143818e-06,
+      "loss": 0.7144,
+      "step": 11666
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2401823785903222e-06,
+      "loss": 0.4079,
+      "step": 11667
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.238800519707284e-06,
+      "loss": 0.4213,
+      "step": 11668
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2374190334316135e-06,
+      "loss": 0.7153,
+      "step": 11669
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2360379198296356e-06,
+      "loss": 0.4081,
+      "step": 11670
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.234657178967654e-06,
+      "loss": 0.72,
+      "step": 11671
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2332768109119574e-06,
+      "loss": 0.7038,
+      "step": 11672
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2318968157288136e-06,
+      "loss": 0.7128,
+      "step": 11673
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2305171934844784e-06,
+      "loss": 0.3918,
+      "step": 11674
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2291379442451842e-06,
+      "loss": 0.6734,
+      "step": 11675
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.227759068077149e-06,
+      "loss": 0.7604,
+      "step": 11676
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2263805650465687e-06,
+      "loss": 0.7475,
+      "step": 11677
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2250024352196233e-06,
+      "loss": 0.7134,
+      "step": 11678
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2236246786624794e-06,
+      "loss": 0.7053,
+      "step": 11679
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.22224729544128e-06,
+      "loss": 0.4564,
+      "step": 11680
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.220870285622151e-06,
+      "loss": 0.6938,
+      "step": 11681
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2194936492712016e-06,
+      "loss": 0.7654,
+      "step": 11682
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2181173864545223e-06,
+      "loss": 0.7989,
+      "step": 11683
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2167414972381873e-06,
+      "loss": 0.6973,
+      "step": 11684
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2153659816882513e-06,
+      "loss": 0.7236,
+      "step": 11685
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2139908398707467e-06,
+      "loss": 0.4056,
+      "step": 11686
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.212616071851701e-06,
+      "loss": 0.3917,
+      "step": 11687
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2112416776971115e-06,
+      "loss": 0.7204,
+      "step": 11688
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2098676574729604e-06,
+      "loss": 0.7298,
+      "step": 11689
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.208494011245215e-06,
+      "loss": 0.41,
+      "step": 11690
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.207120739079819e-06,
+      "loss": 0.6989,
+      "step": 11691
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.205747841042706e-06,
+      "loss": 0.7005,
+      "step": 11692
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2043753171997873e-06,
+      "loss": 0.7243,
+      "step": 11693
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.203003167616954e-06,
+      "loss": 0.3893,
+      "step": 11694
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2016313923600828e-06,
+      "loss": 0.6932,
+      "step": 11695
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2002599914950316e-06,
+      "loss": 0.7161,
+      "step": 11696
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.19888896508764e-06,
+      "loss": 0.723,
+      "step": 11697
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.197518313203728e-06,
+      "loss": 0.7233,
+      "step": 11698
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1961480359091013e-06,
+      "loss": 0.7543,
+      "step": 11699
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1947781332695406e-06,
+      "loss": 0.7785,
+      "step": 11700
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.193408605350821e-06,
+      "loss": 0.7411,
+      "step": 11701
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.192039452218688e-06,
+      "loss": 0.7279,
+      "step": 11702
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.190670673938875e-06,
+      "loss": 0.7444,
+      "step": 11703
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.189302270577094e-06,
+      "loss": 0.7062,
+      "step": 11704
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.187934242199038e-06,
+      "loss": 0.3958,
+      "step": 11705
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1865665888703915e-06,
+      "loss": 0.3799,
+      "step": 11706
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.18519931065681e-06,
+      "loss": 0.6871,
+      "step": 11707
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.183832407623936e-06,
+      "loss": 0.7354,
+      "step": 11708
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1824658798373898e-06,
+      "loss": 0.7482,
+      "step": 11709
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.181099727362783e-06,
+      "loss": 0.7567,
+      "step": 11710
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1797339502657034e-06,
+      "loss": 0.6895,
+      "step": 11711
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1783685486117132e-06,
+      "loss": 0.7481,
+      "step": 11712
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1770035224663665e-06,
+      "loss": 0.6731,
+      "step": 11713
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1756388718952016e-06,
+      "loss": 0.7202,
+      "step": 11714
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.17427459696373e-06,
+      "loss": 0.7578,
+      "step": 11715
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1729106977374516e-06,
+      "loss": 0.7072,
+      "step": 11716
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1715471742818427e-06,
+      "loss": 0.7602,
+      "step": 11717
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1701840266623644e-06,
+      "loss": 0.7623,
+      "step": 11718
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1688212549444643e-06,
+      "loss": 0.7509,
+      "step": 11719
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1674588591935665e-06,
+      "loss": 0.716,
+      "step": 11720
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1660968394750758e-06,
+      "loss": 0.7132,
+      "step": 11721
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1647351958543827e-06,
+      "loss": 0.7146,
+      "step": 11722
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1633739283968567e-06,
+      "loss": 0.7528,
+      "step": 11723
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.162013037167854e-06,
+      "loss": 0.4271,
+      "step": 11724
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.160652522232709e-06,
+      "loss": 0.4197,
+      "step": 11725
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1592923836567405e-06,
+      "loss": 0.6952,
+      "step": 11726
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.157932621505239e-06,
+      "loss": 0.7799,
+      "step": 11727
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.156573235843493e-06,
+      "loss": 0.7568,
+      "step": 11728
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1552142267367635e-06,
+      "loss": 0.7501,
+      "step": 11729
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.153855594250296e-06,
+      "loss": 0.415,
+      "step": 11730
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.152497338449314e-06,
+      "loss": 0.7129,
+      "step": 11731
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1511394593990263e-06,
+      "loss": 0.7383,
+      "step": 11732
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1497819571646283e-06,
+      "loss": 0.7037,
+      "step": 11733
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1484248318112886e-06,
+      "loss": 0.78,
+      "step": 11734
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1470680834041615e-06,
+      "loss": 0.7605,
+      "step": 11735
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1457117120083837e-06,
+      "loss": 0.6893,
+      "step": 11736
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1443557176890693e-06,
+      "loss": 0.7192,
+      "step": 11737
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1430001005113254e-06,
+      "loss": 0.4107,
+      "step": 11738
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1416448605402296e-06,
+      "loss": 0.753,
+      "step": 11739
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1402899978408476e-06,
+      "loss": 0.7078,
+      "step": 11740
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1389355124782217e-06,
+      "loss": 0.3855,
+      "step": 11741
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1375814045173825e-06,
+      "loss": 0.7793,
+      "step": 11742
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.136227674023339e-06,
+      "loss": 0.7571,
+      "step": 11743
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.13487432106108e-06,
+      "loss": 0.3738,
+      "step": 11744
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.133521345695578e-06,
+      "loss": 0.7323,
+      "step": 11745
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.132168747991794e-06,
+      "loss": 0.742,
+      "step": 11746
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.13081652801466e-06,
+      "loss": 0.7655,
+      "step": 11747
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1294646858290956e-06,
+      "loss": 0.7089,
+      "step": 11748
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1281132215000023e-06,
+      "loss": 0.6692,
+      "step": 11749
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1267621350922585e-06,
+      "loss": 0.6825,
+      "step": 11750
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.125411426670735e-06,
+      "loss": 0.4163,
+      "step": 11751
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.124061096300275e-06,
+      "loss": 0.7889,
+      "step": 11752
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.122711144045706e-06,
+      "loss": 0.7504,
+      "step": 11753
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.121361569971838e-06,
+      "loss": 0.6663,
+      "step": 11754
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1200123741434597e-06,
+      "loss": 0.7269,
+      "step": 11755
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1186635566253543e-06,
+      "loss": 0.7162,
+      "step": 11756
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1173151174822672e-06,
+      "loss": 0.367,
+      "step": 11757
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.115967056778939e-06,
+      "loss": 0.649,
+      "step": 11758
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1146193745800857e-06,
+      "loss": 0.7756,
+      "step": 11759
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.113272070950414e-06,
+      "loss": 0.7147,
+      "step": 11760
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.111925145954604e-06,
+      "loss": 0.7434,
+      "step": 11761
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.11057859965732e-06,
+      "loss": 0.703,
+      "step": 11762
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1092324321232094e-06,
+      "loss": 0.3875,
+      "step": 11763
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1078866434168953e-06,
+      "loss": 0.7268,
+      "step": 11764
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1065412336029943e-06,
+      "loss": 0.7131,
+      "step": 11765
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.105196202746096e-06,
+      "loss": 0.7245,
+      "step": 11766
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1038515509107735e-06,
+      "loss": 0.7217,
+      "step": 11767
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.102507278161583e-06,
+      "loss": 0.7565,
+      "step": 11768
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1011633845630563e-06,
+      "loss": 0.3967,
+      "step": 11769
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0998198701797214e-06,
+      "loss": 0.695,
+      "step": 11770
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.098476735076076e-06,
+      "loss": 0.6885,
+      "step": 11771
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0971339793165967e-06,
+      "loss": 0.7541,
+      "step": 11772
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0957916029657545e-06,
+      "loss": 0.7656,
+      "step": 11773
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0944496060879936e-06,
+      "loss": 0.7586,
+      "step": 11774
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.093107988747742e-06,
+      "loss": 0.7202,
+      "step": 11775
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.09176675100941e-06,
+      "loss": 0.3932,
+      "step": 11776
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0904258929373845e-06,
+      "loss": 0.7247,
+      "step": 11777
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0890854145960447e-06,
+      "loss": 0.4159,
+      "step": 11778
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.087745316049744e-06,
+      "loss": 0.7251,
+      "step": 11779
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0864055973628193e-06,
+      "loss": 0.7168,
+      "step": 11780
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0850662585995875e-06,
+      "loss": 0.6881,
+      "step": 11781
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0837272998243474e-06,
+      "loss": 0.7841,
+      "step": 11782
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.082388721101386e-06,
+      "loss": 0.3796,
+      "step": 11783
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.081050522494965e-06,
+      "loss": 0.7607,
+      "step": 11784
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0797127040693297e-06,
+      "loss": 0.7252,
+      "step": 11785
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.078375265888707e-06,
+      "loss": 0.7088,
+      "step": 11786
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.077038208017307e-06,
+      "loss": 0.7065,
+      "step": 11787
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0757015305193186e-06,
+      "loss": 0.7141,
+      "step": 11788
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.074365233458916e-06,
+      "loss": 0.3602,
+      "step": 11789
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0730293169002524e-06,
+      "loss": 0.7497,
+      "step": 11790
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0716937809074623e-06,
+      "loss": 0.394,
+      "step": 11791
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0703586255446683e-06,
+      "loss": 0.6925,
+      "step": 11792
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0690238508759675e-06,
+      "loss": 0.7048,
+      "step": 11793
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0676894569654393e-06,
+      "loss": 0.7021,
+      "step": 11794
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0663554438771493e-06,
+      "loss": 0.7429,
+      "step": 11795
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0650218116751375e-06,
+      "loss": 0.7149,
+      "step": 11796
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.063688560423436e-06,
+      "loss": 0.6496,
+      "step": 11797
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0623556901860507e-06,
+      "loss": 0.396,
+      "step": 11798
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0610232010269717e-06,
+      "loss": 0.3729,
+      "step": 11799
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0596910930101688e-06,
+      "loss": 0.7006,
+      "step": 11800
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.058359366199596e-06,
+      "loss": 0.7279,
+      "step": 11801
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.05702802065919e-06,
+      "loss": 0.7105,
+      "step": 11802
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.055697056452863e-06,
+      "loss": 0.7186,
+      "step": 11803
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.054366473644516e-06,
+      "loss": 0.7231,
+      "step": 11804
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.05303627229803e-06,
+      "loss": 0.7419,
+      "step": 11805
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0517064524772657e-06,
+      "loss": 0.6976,
+      "step": 11806
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0503770142460666e-06,
+      "loss": 0.6786,
+      "step": 11807
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0490479576682565e-06,
+      "loss": 0.4278,
+      "step": 11808
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.047719282807641e-06,
+      "loss": 0.7155,
+      "step": 11809
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.046390989728013e-06,
+      "loss": 0.6528,
+      "step": 11810
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.045063078493139e-06,
+      "loss": 0.3905,
+      "step": 11811
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0437355491667733e-06,
+      "loss": 0.7402,
+      "step": 11812
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0424084018126455e-06,
+      "loss": 0.6839,
+      "step": 11813
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.041081636494471e-06,
+      "loss": 0.7071,
+      "step": 11814
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0397552532759525e-06,
+      "loss": 0.416,
+      "step": 11815
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0384292522207614e-06,
+      "loss": 0.7115,
+      "step": 11816
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0371036333925607e-06,
+      "loss": 0.7123,
+      "step": 11817
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.035778396854989e-06,
+      "loss": 0.713,
+      "step": 11818
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.034453542671674e-06,
+      "loss": 0.7447,
+      "step": 11819
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0331290709062192e-06,
+      "loss": 0.7489,
+      "step": 11820
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0318049816222096e-06,
+      "loss": 0.7164,
+      "step": 11821
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.030481274883216e-06,
+      "loss": 0.6865,
+      "step": 11822
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0291579507527824e-06,
+      "loss": 0.6957,
+      "step": 11823
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.027835009294448e-06,
+      "loss": 0.7406,
+      "step": 11824
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0265124505717215e-06,
+      "loss": 0.7124,
+      "step": 11825
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.025190274648099e-06,
+      "loss": 0.7528,
+      "step": 11826
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0238684815870566e-06,
+      "loss": 0.4154,
+      "step": 11827
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.022547071452049e-06,
+      "loss": 0.4126,
+      "step": 11828
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.021226044306521e-06,
+      "loss": 0.7226,
+      "step": 11829
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0199054002138953e-06,
+      "loss": 0.685,
+      "step": 11830
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0185851392375644e-06,
+      "loss": 0.4334,
+      "step": 11831
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0172652614409215e-06,
+      "loss": 0.7363,
+      "step": 11832
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0159457668873316e-06,
+      "loss": 0.7171,
+      "step": 11833
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0146266556401405e-06,
+      "loss": 0.4266,
+      "step": 11834
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0133079277626777e-06,
+      "loss": 0.6684,
+      "step": 11835
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0119895833182523e-06,
+      "loss": 0.7035,
+      "step": 11836
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0106716223701604e-06,
+      "loss": 0.7022,
+      "step": 11837
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.009354044981674e-06,
+      "loss": 0.3974,
+      "step": 11838
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0080368512160498e-06,
+      "loss": 0.4254,
+      "step": 11839
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.006720041136523e-06,
+      "loss": 0.7757,
+      "step": 11840
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0054036148063115e-06,
+      "loss": 0.7627,
+      "step": 11841
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0040875722886187e-06,
+      "loss": 0.6905,
+      "step": 11842
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.002771913646626e-06,
+      "loss": 0.7206,
+      "step": 11843
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0014566389434963e-06,
+      "loss": 0.6858,
+      "step": 11844
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0001417482423746e-06,
+      "loss": 0.6826,
+      "step": 11845
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.998827241606387e-06,
+      "loss": 0.6693,
+      "step": 11846
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.997513119098642e-06,
+      "loss": 0.4272,
+      "step": 11847
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9961993807822276e-06,
+      "loss": 0.6981,
+      "step": 11848
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9948860267202186e-06,
+      "loss": 0.66,
+      "step": 11849
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9935730569756616e-06,
+      "loss": 0.6608,
+      "step": 11850
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.992260471611599e-06,
+      "loss": 0.7069,
+      "step": 11851
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.990948270691042e-06,
+      "loss": 0.7235,
+      "step": 11852
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.98963645427699e-06,
+      "loss": 0.6903,
+      "step": 11853
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.98832502243242e-06,
+      "loss": 0.4142,
+      "step": 11854
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.987013975220291e-06,
+      "loss": 0.7773,
+      "step": 11855
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.98570331270355e-06,
+      "loss": 0.711,
+      "step": 11856
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.984393034945118e-06,
+      "loss": 0.7595,
+      "step": 11857
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9830831420079e-06,
+      "loss": 0.7275,
+      "step": 11858
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9817736339547812e-06,
+      "loss": 0.3851,
+      "step": 11859
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9804645108486353e-06,
+      "loss": 0.7268,
+      "step": 11860
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.979155772752307e-06,
+      "loss": 0.4134,
+      "step": 11861
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9778474197286267e-06,
+      "loss": 0.6712,
+      "step": 11862
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.976539451840408e-06,
+      "loss": 0.3928,
+      "step": 11863
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.975231869150448e-06,
+      "loss": 0.3655,
+      "step": 11864
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.97392467172152e-06,
+      "loss": 0.6787,
+      "step": 11865
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9726178596163825e-06,
+      "loss": 0.4242,
+      "step": 11866
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9713114328977732e-06,
+      "loss": 0.7302,
+      "step": 11867
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.97000539162841e-06,
+      "loss": 0.7567,
+      "step": 11868
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.968699735870999e-06,
+      "loss": 0.7321,
+      "step": 11869
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.967394465688223e-06,
+      "loss": 0.6927,
+      "step": 11870
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.966089581142745e-06,
+      "loss": 0.7361,
+      "step": 11871
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9647850822972113e-06,
+      "loss": 0.7035,
+      "step": 11872
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9634809692142466e-06,
+      "loss": 0.7305,
+      "step": 11873
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9621772419564666e-06,
+      "loss": 0.3996,
+      "step": 11874
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.960873900586462e-06,
+      "loss": 0.4384,
+      "step": 11875
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9595709451667975e-06,
+      "loss": 0.7439,
+      "step": 11876
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9582683757600284e-06,
+      "loss": 0.7281,
+      "step": 11877
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9569661924286955e-06,
+      "loss": 0.4348,
+      "step": 11878
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9556643952353103e-06,
+      "loss": 0.7148,
+      "step": 11879
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.954362984242374e-06,
+      "loss": 0.7637,
+      "step": 11880
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9530619595123636e-06,
+      "loss": 0.7319,
+      "step": 11881
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.951761321107738e-06,
+      "loss": 0.7342,
+      "step": 11882
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9504610690909454e-06,
+      "loss": 0.4077,
+      "step": 11883
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9491612035244056e-06,
+      "loss": 0.7258,
+      "step": 11884
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9478617244705256e-06,
+      "loss": 0.4026,
+      "step": 11885
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9465626319916907e-06,
+      "loss": 0.7012,
+      "step": 11886
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.945263926150267e-06,
+      "loss": 0.736,
+      "step": 11887
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9439656070086086e-06,
+      "loss": 0.6915,
+      "step": 11888
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.942667674629045e-06,
+      "loss": 0.6801,
+      "step": 11889
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9413701290738885e-06,
+      "loss": 0.6988,
+      "step": 11890
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.940072970405432e-06,
+      "loss": 0.6776,
+      "step": 11891
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.938776198685951e-06,
+      "loss": 0.7299,
+      "step": 11892
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.937479813977703e-06,
+      "loss": 0.4084,
+      "step": 11893
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.936183816342926e-06,
+      "loss": 0.7349,
+      "step": 11894
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9348882058438356e-06,
+      "loss": 0.4041,
+      "step": 11895
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9335929825426404e-06,
+      "loss": 0.7199,
+      "step": 11896
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.932298146501518e-06,
+      "loss": 0.715,
+      "step": 11897
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9310036977826317e-06,
+      "loss": 0.7455,
+      "step": 11898
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9297096364481293e-06,
+      "loss": 0.6961,
+      "step": 11899
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.928415962560133e-06,
+      "loss": 0.3811,
+      "step": 11900
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.927122676180756e-06,
+      "loss": 0.3658,
+      "step": 11901
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9258297773720857e-06,
+      "loss": 0.413,
+      "step": 11902
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9245372661961926e-06,
+      "loss": 0.3825,
+      "step": 11903
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9232451427151276e-06,
+      "loss": 0.4149,
+      "step": 11904
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9219534069909275e-06,
+      "loss": 0.6861,
+      "step": 11905
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.920662059085604e-06,
+      "loss": 0.6994,
+      "step": 11906
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9193710990611546e-06,
+      "loss": 0.7233,
+      "step": 11907
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9180805269795565e-06,
+      "loss": 0.4387,
+      "step": 11908
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.916790342902768e-06,
+      "loss": 0.6684,
+      "step": 11909
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9155005468927323e-06,
+      "loss": 0.7224,
+      "step": 11910
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9142111390113707e-06,
+      "loss": 0.7089,
+      "step": 11911
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.912922119320585e-06,
+      "loss": 0.3955,
+      "step": 11912
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9116334878822594e-06,
+      "loss": 0.7449,
+      "step": 11913
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.910345244758259e-06,
+      "loss": 0.7405,
+      "step": 11914
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.909057390010435e-06,
+      "loss": 0.7153,
+      "step": 11915
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.907769923700613e-06,
+      "loss": 0.7289,
+      "step": 11916
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9064828458906037e-06,
+      "loss": 0.7254,
+      "step": 11917
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9051961566421984e-06,
+      "loss": 0.7627,
+      "step": 11918
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9039098560171698e-06,
+      "loss": 0.6863,
+      "step": 11919
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9026239440772708e-06,
+      "loss": 0.6908,
+      "step": 11920
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.901338420884239e-06,
+      "loss": 0.4365,
+      "step": 11921
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9000532864997856e-06,
+      "loss": 0.7365,
+      "step": 11922
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.898768540985616e-06,
+      "loss": 0.7528,
+      "step": 11923
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8974841844034065e-06,
+      "loss": 0.6853,
+      "step": 11924
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8962002168148174e-06,
+      "loss": 0.443,
+      "step": 11925
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.894916638281491e-06,
+      "loss": 0.4101,
+      "step": 11926
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8936334488650476e-06,
+      "loss": 0.6615,
+      "step": 11927
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.892350648627097e-06,
+      "loss": 0.7105,
+      "step": 11928
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8910682376292234e-06,
+      "loss": 0.6941,
+      "step": 11929
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8897862159329928e-06,
+      "loss": 0.7158,
+      "step": 11930
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.888504583599955e-06,
+      "loss": 0.6805,
+      "step": 11931
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8872233406916364e-06,
+      "loss": 0.6867,
+      "step": 11932
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8859424872695541e-06,
+      "loss": 0.7601,
+      "step": 11933
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8846620233951994e-06,
+      "loss": 0.4482,
+      "step": 11934
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8833819491300421e-06,
+      "loss": 0.4296,
+      "step": 11935
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.882102264535537e-06,
+      "loss": 0.7464,
+      "step": 11936
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.880822969673125e-06,
+      "loss": 0.3998,
+      "step": 11937
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8795440646042218e-06,
+      "loss": 0.7046,
+      "step": 11938
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.878265549390227e-06,
+      "loss": 0.7304,
+      "step": 11939
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.876987424092519e-06,
+      "loss": 0.7244,
+      "step": 11940
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.875709688772458e-06,
+      "loss": 0.7216,
+      "step": 11941
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8744323434913915e-06,
+      "loss": 0.6687,
+      "step": 11942
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8731553883106412e-06,
+      "loss": 0.6985,
+      "step": 11943
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8718788232915131e-06,
+      "loss": 0.7285,
+      "step": 11944
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8706026484952932e-06,
+      "loss": 0.6775,
+      "step": 11945
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8693268639832462e-06,
+      "loss": 0.7313,
+      "step": 11946
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8680514698166264e-06,
+      "loss": 0.7405,
+      "step": 11947
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8667764660566623e-06,
+      "loss": 0.7499,
+      "step": 11948
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8655018527645664e-06,
+      "loss": 0.7553,
+      "step": 11949
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8642276300015295e-06,
+      "loss": 0.7189,
+      "step": 11950
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.862953797828727e-06,
+      "loss": 0.7354,
+      "step": 11951
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8616803563073137e-06,
+      "loss": 0.7274,
+      "step": 11952
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8604073054984274e-06,
+      "loss": 0.4286,
+      "step": 11953
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.859134645463182e-06,
+      "loss": 0.4091,
+      "step": 11954
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8578623762626824e-06,
+      "loss": 0.7084,
+      "step": 11955
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8565904979580075e-06,
+      "loss": 0.7088,
+      "step": 11956
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8553190106102171e-06,
+      "loss": 0.4056,
+      "step": 11957
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8540479142803536e-06,
+      "loss": 0.6897,
+      "step": 11958
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8527772090294417e-06,
+      "loss": 0.7441,
+      "step": 11959
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8515068949184878e-06,
+      "loss": 0.7385,
+      "step": 11960
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.850236972008479e-06,
+      "loss": 0.7245,
+      "step": 11961
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8489674403603808e-06,
+      "loss": 0.6805,
+      "step": 11962
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8476983000351445e-06,
+      "loss": 0.6886,
+      "step": 11963
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8464295510936981e-06,
+      "loss": 0.6817,
+      "step": 11964
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8451611935969537e-06,
+      "loss": 0.7385,
+      "step": 11965
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.843893227605804e-06,
+      "loss": 0.7333,
+      "step": 11966
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8426256531811227e-06,
+      "loss": 0.7612,
+      "step": 11967
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8413584703837618e-06,
+      "loss": 0.4053,
+      "step": 11968
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8400916792745627e-06,
+      "loss": 0.704,
+      "step": 11969
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8388252799143413e-06,
+      "loss": 0.7537,
+      "step": 11970
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8375592723638947e-06,
+      "loss": 0.7259,
+      "step": 11971
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8362936566840038e-06,
+      "loss": 0.7326,
+      "step": 11972
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8350284329354261e-06,
+      "loss": 0.6701,
+      "step": 11973
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8337636011789096e-06,
+      "loss": 0.391,
+      "step": 11974
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8324991614751741e-06,
+      "loss": 0.3907,
+      "step": 11975
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.831235113884925e-06,
+      "loss": 0.7137,
+      "step": 11976
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8299714584688466e-06,
+      "loss": 0.7212,
+      "step": 11977
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.828708195287605e-06,
+      "loss": 0.7321,
+      "step": 11978
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8274453244018552e-06,
+      "loss": 0.7107,
+      "step": 11979
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8261828458722185e-06,
+      "loss": 0.7444,
+      "step": 11980
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8249207597593044e-06,
+      "loss": 0.7206,
+      "step": 11981
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8236590661237108e-06,
+      "loss": 0.7119,
+      "step": 11982
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8223977650260073e-06,
+      "loss": 0.4294,
+      "step": 11983
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8211368565267472e-06,
+      "loss": 0.7414,
+      "step": 11984
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.819876340686466e-06,
+      "loss": 0.7311,
+      "step": 11985
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8186162175656775e-06,
+      "loss": 0.677,
+      "step": 11986
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8173564872248838e-06,
+      "loss": 0.6889,
+      "step": 11987
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8160971497245594e-06,
+      "loss": 0.6979,
+      "step": 11988
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8148382051251667e-06,
+      "loss": 0.7276,
+      "step": 11989
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8135796534871442e-06,
+      "loss": 0.722,
+      "step": 11990
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.812321494870911e-06,
+      "loss": 0.7247,
+      "step": 11991
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.811063729336876e-06,
+      "loss": 0.6841,
+      "step": 11992
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8098063569454215e-06,
+      "loss": 0.7599,
+      "step": 11993
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8085493777569129e-06,
+      "loss": 0.7323,
+      "step": 11994
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8072927918316907e-06,
+      "loss": 0.7613,
+      "step": 11995
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.80603659923009e-06,
+      "loss": 0.688,
+      "step": 11996
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8047808000124157e-06,
+      "loss": 0.7249,
+      "step": 11997
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8035253942389575e-06,
+      "loss": 0.7592,
+      "step": 11998
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.802270381969987e-06,
+      "loss": 0.4354,
+      "step": 11999
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8010157632657544e-06,
+      "loss": 0.6975,
+      "step": 12000
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7997615381864952e-06,
+      "loss": 0.403,
+      "step": 12001
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7985077067924228e-06,
+      "loss": 0.7288,
+      "step": 12002
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.797254269143731e-06,
+      "loss": 0.6843,
+      "step": 12003
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7960012253005988e-06,
+      "loss": 0.7415,
+      "step": 12004
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7947485753231774e-06,
+      "loss": 0.7051,
+      "step": 12005
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7934963192716138e-06,
+      "loss": 0.397,
+      "step": 12006
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7922444572060217e-06,
+      "loss": 0.6328,
+      "step": 12007
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7909929891865041e-06,
+      "loss": 0.7643,
+      "step": 12008
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7897419152731422e-06,
+      "loss": 0.6944,
+      "step": 12009
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7884912355259988e-06,
+      "loss": 0.3999,
+      "step": 12010
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7872409500051168e-06,
+      "loss": 0.7384,
+      "step": 12011
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.785991058770522e-06,
+      "loss": 0.756,
+      "step": 12012
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7847415618822184e-06,
+      "loss": 0.4087,
+      "step": 12013
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7834924594001968e-06,
+      "loss": 0.6729,
+      "step": 12014
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.782243751384425e-06,
+      "loss": 0.7007,
+      "step": 12015
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.780995437894849e-06,
+      "loss": 0.7019,
+      "step": 12016
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7797475189914026e-06,
+      "loss": 0.4153,
+      "step": 12017
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7784999947339921e-06,
+      "loss": 0.7229,
+      "step": 12018
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.777252865182515e-06,
+      "loss": 0.6764,
+      "step": 12019
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7760061303968434e-06,
+      "loss": 0.7164,
+      "step": 12020
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7747597904368318e-06,
+      "loss": 0.7242,
+      "step": 12021
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7735138453623147e-06,
+      "loss": 0.4226,
+      "step": 12022
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7722682952331094e-06,
+      "loss": 0.6673,
+      "step": 12023
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7710231401090129e-06,
+      "loss": 0.4083,
+      "step": 12024
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7697783800498048e-06,
+      "loss": 0.7006,
+      "step": 12025
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.768534015115244e-06,
+      "loss": 0.7261,
+      "step": 12026
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7672900453650689e-06,
+      "loss": 0.7193,
+      "step": 12027
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7660464708590063e-06,
+      "loss": 0.7253,
+      "step": 12028
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7648032916567559e-06,
+      "loss": 0.4206,
+      "step": 12029
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.763560507818003e-06,
+      "loss": 0.6884,
+      "step": 12030
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7623181194024108e-06,
+      "loss": 0.7307,
+      "step": 12031
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.761076126469623e-06,
+      "loss": 0.7699,
+      "step": 12032
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7598345290792718e-06,
+      "loss": 0.4015,
+      "step": 12033
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7585933272909628e-06,
+      "loss": 0.7121,
+      "step": 12034
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7573525211642839e-06,
+      "loss": 0.7107,
+      "step": 12035
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7561121107588063e-06,
+      "loss": 0.3984,
+      "step": 12036
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.754872096134077e-06,
+      "loss": 0.7148,
+      "step": 12037
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7536324773496349e-06,
+      "loss": 0.7183,
+      "step": 12038
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7523932544649879e-06,
+      "loss": 0.7272,
+      "step": 12039
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7511544275396276e-06,
+      "loss": 0.6434,
+      "step": 12040
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7499159966330347e-06,
+      "loss": 0.6576,
+      "step": 12041
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7486779618046623e-06,
+      "loss": 0.4179,
+      "step": 12042
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.747440323113947e-06,
+      "loss": 0.675,
+      "step": 12043
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7462030806203066e-06,
+      "loss": 0.7002,
+      "step": 12044
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7449662343831386e-06,
+      "loss": 0.7115,
+      "step": 12045
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.743729784461825e-06,
+      "loss": 0.3792,
+      "step": 12046
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7424937309157275e-06,
+      "loss": 0.7324,
+      "step": 12047
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7412580738041852e-06,
+      "loss": 0.7242,
+      "step": 12048
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7400228131865217e-06,
+      "loss": 0.7513,
+      "step": 12049
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7387879491220384e-06,
+      "loss": 0.7192,
+      "step": 12050
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7375534816700234e-06,
+      "loss": 0.7479,
+      "step": 12051
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.736319410889742e-06,
+      "loss": 0.4293,
+      "step": 12052
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7350857368404416e-06,
+      "loss": 0.6753,
+      "step": 12053
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7338524595813423e-06,
+      "loss": 0.698,
+      "step": 12054
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7326195791716605e-06,
+      "loss": 0.653,
+      "step": 12055
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7313870956705826e-06,
+      "loss": 0.4095,
+      "step": 12056
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7301550091372788e-06,
+      "loss": 0.3717,
+      "step": 12057
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.728923319630902e-06,
+      "loss": 0.4007,
+      "step": 12058
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7276920272105791e-06,
+      "loss": 0.6864,
+      "step": 12059
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7264611319354296e-06,
+      "loss": 0.7595,
+      "step": 12060
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7252306338645463e-06,
+      "loss": 0.3992,
+      "step": 12061
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.724000533057002e-06,
+      "loss": 0.7234,
+      "step": 12062
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7227708295718538e-06,
+      "loss": 0.3865,
+      "step": 12063
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7215415234681353e-06,
+      "loss": 0.7171,
+      "step": 12064
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7203126148048699e-06,
+      "loss": 0.771,
+      "step": 12065
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7190841036410545e-06,
+      "loss": 0.7169,
+      "step": 12066
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7178559900356661e-06,
+      "loss": 0.4089,
+      "step": 12067
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7166282740476681e-06,
+      "loss": 0.3656,
+      "step": 12068
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7154009557359997e-06,
+      "loss": 0.6759,
+      "step": 12069
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7141740351595848e-06,
+      "loss": 0.7548,
+      "step": 12070
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7129475123773264e-06,
+      "loss": 0.7272,
+      "step": 12071
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7117213874481052e-06,
+      "loss": 0.7072,
+      "step": 12072
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.710495660430791e-06,
+      "loss": 0.7002,
+      "step": 12073
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7092703313842295e-06,
+      "loss": 0.358,
+      "step": 12074
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7080454003672453e-06,
+      "loss": 0.7373,
+      "step": 12075
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7068208674386467e-06,
+      "loss": 0.7346,
+      "step": 12076
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7055967326572186e-06,
+      "loss": 0.7007,
+      "step": 12077
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.704372996081739e-06,
+      "loss": 0.7277,
+      "step": 12078
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7031496577709517e-06,
+      "loss": 0.7088,
+      "step": 12079
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7019267177835908e-06,
+      "loss": 0.3903,
+      "step": 12080
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7007041761783671e-06,
+      "loss": 0.7004,
+      "step": 12081
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6994820330139715e-06,
+      "loss": 0.7515,
+      "step": 12082
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.698260288349085e-06,
+      "loss": 0.7276,
+      "step": 12083
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.697038942242355e-06,
+      "loss": 0.69,
+      "step": 12084
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.695817994752419e-06,
+      "loss": 0.3871,
+      "step": 12085
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6945974459378923e-06,
+      "loss": 0.7145,
+      "step": 12086
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6933772958573768e-06,
+      "loss": 0.7108,
+      "step": 12087
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6921575445694471e-06,
+      "loss": 0.6994,
+      "step": 12088
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.690938192132664e-06,
+      "loss": 0.7218,
+      "step": 12089
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.689719238605566e-06,
+      "loss": 0.7368,
+      "step": 12090
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.688500684046671e-06,
+      "loss": 0.6934,
+      "step": 12091
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6872825285144866e-06,
+      "loss": 0.7375,
+      "step": 12092
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6860647720674928e-06,
+      "loss": 0.7331,
+      "step": 12093
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6848474147641525e-06,
+      "loss": 0.7286,
+      "step": 12094
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6836304566629102e-06,
+      "loss": 0.7264,
+      "step": 12095
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.682413897822187e-06,
+      "loss": 0.7548,
+      "step": 12096
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6811977383003953e-06,
+      "loss": 0.3928,
+      "step": 12097
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6799819781559202e-06,
+      "loss": 0.4267,
+      "step": 12098
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.678766617447123e-06,
+      "loss": 0.7018,
+      "step": 12099
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6775516562323591e-06,
+      "loss": 0.7518,
+      "step": 12100
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.676337094569954e-06,
+      "loss": 0.6507,
+      "step": 12101
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6751229325182194e-06,
+      "loss": 0.7054,
+      "step": 12102
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6739091701354437e-06,
+      "loss": 0.7129,
+      "step": 12103
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6726958074798982e-06,
+      "loss": 0.7398,
+      "step": 12104
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6714828446098386e-06,
+      "loss": 0.7195,
+      "step": 12105
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.670270281583497e-06,
+      "loss": 0.71,
+      "step": 12106
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6690581184590859e-06,
+      "loss": 0.7196,
+      "step": 12107
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6678463552948e-06,
+      "loss": 0.6687,
+      "step": 12108
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6666349921488135e-06,
+      "loss": 0.7017,
+      "step": 12109
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6654240290792878e-06,
+      "loss": 0.7322,
+      "step": 12110
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6642134661443554e-06,
+      "loss": 0.7166,
+      "step": 12111
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6630033034021365e-06,
+      "loss": 0.6976,
+      "step": 12112
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.661793540910729e-06,
+      "loss": 0.7063,
+      "step": 12113
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6605841787282118e-06,
+      "loss": 0.7368,
+      "step": 12114
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6593752169126465e-06,
+      "loss": 0.7125,
+      "step": 12115
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6581666555220733e-06,
+      "loss": 0.6937,
+      "step": 12116
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6569584946145135e-06,
+      "loss": 0.686,
+      "step": 12117
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6557507342479684e-06,
+      "loss": 0.4398,
+      "step": 12118
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.654543374480425e-06,
+      "loss": 0.6954,
+      "step": 12119
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6533364153698461e-06,
+      "loss": 0.7524,
+      "step": 12120
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.652129856974176e-06,
+      "loss": 0.7369,
+      "step": 12121
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.65092369935134e-06,
+      "loss": 0.737,
+      "step": 12122
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6497179425592424e-06,
+      "loss": 0.3918,
+      "step": 12123
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.648512586655776e-06,
+      "loss": 0.7717,
+      "step": 12124
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6473076316988057e-06,
+      "loss": 0.7388,
+      "step": 12125
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6461030777461806e-06,
+      "loss": 0.7193,
+      "step": 12126
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.644898924855729e-06,
+      "loss": 0.6754,
+      "step": 12127
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6436951730852613e-06,
+      "loss": 0.4062,
+      "step": 12128
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6424918224925701e-06,
+      "loss": 0.4143,
+      "step": 12129
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.641288873135425e-06,
+      "loss": 0.7098,
+      "step": 12130
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.640086325071577e-06,
+      "loss": 0.4106,
+      "step": 12131
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6388841783587628e-06,
+      "loss": 0.7419,
+      "step": 12132
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6376824330546958e-06,
+      "loss": 0.7153,
+      "step": 12133
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6364810892170702e-06,
+      "loss": 0.7318,
+      "step": 12134
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6352801469035595e-06,
+      "loss": 0.7524,
+      "step": 12135
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.634079606171819e-06,
+      "loss": 0.3637,
+      "step": 12136
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.63287946707949e-06,
+      "loss": 0.4282,
+      "step": 12137
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.631679729684188e-06,
+      "loss": 0.3908,
+      "step": 12138
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6304803940435099e-06,
+      "loss": 0.7199,
+      "step": 12139
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6292814602150353e-06,
+      "loss": 0.7518,
+      "step": 12140
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6280829282563215e-06,
+      "loss": 0.6873,
+      "step": 12141
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6268847982249158e-06,
+      "loss": 0.7455,
+      "step": 12142
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6256870701783323e-06,
+      "loss": 0.7335,
+      "step": 12143
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6244897441740747e-06,
+      "loss": 0.7027,
+      "step": 12144
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.623292820269623e-06,
+      "loss": 0.6783,
+      "step": 12145
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6220962985224453e-06,
+      "loss": 0.7372,
+      "step": 12146
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6209001789899835e-06,
+      "loss": 0.7279,
+      "step": 12147
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.619704461729662e-06,
+      "loss": 0.7244,
+      "step": 12148
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6185091467988844e-06,
+      "loss": 0.6652,
+      "step": 12149
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6173142342550362e-06,
+      "loss": 0.7302,
+      "step": 12150
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6161197241554882e-06,
+      "loss": 0.7118,
+      "step": 12151
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6149256165575844e-06,
+      "loss": 0.7393,
+      "step": 12152
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6137319115186534e-06,
+      "loss": 0.7225,
+      "step": 12153
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6125386090960038e-06,
+      "loss": 0.6566,
+      "step": 12154
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.611345709346922e-06,
+      "loss": 0.7318,
+      "step": 12155
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6101532123286835e-06,
+      "loss": 0.6421,
+      "step": 12156
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6089611180985376e-06,
+      "loss": 0.7281,
+      "step": 12157
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.60776942671371e-06,
+      "loss": 0.7212,
+      "step": 12158
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6065781382314182e-06,
+      "loss": 0.7086,
+      "step": 12159
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6053872527088531e-06,
+      "loss": 0.7432,
+      "step": 12160
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6041967702031892e-06,
+      "loss": 0.4457,
+      "step": 12161
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.603006690771578e-06,
+      "loss": 0.7353,
+      "step": 12162
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6018170144711531e-06,
+      "loss": 0.79,
+      "step": 12163
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6006277413590332e-06,
+      "loss": 0.7122,
+      "step": 12164
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5994388714923137e-06,
+      "loss": 0.6812,
+      "step": 12165
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.598250404928069e-06,
+      "loss": 0.7171,
+      "step": 12166
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5970623417233566e-06,
+      "loss": 0.6826,
+      "step": 12167
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5958746819352133e-06,
+      "loss": 0.6926,
+      "step": 12168
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5946874256206612e-06,
+      "loss": 0.6752,
+      "step": 12169
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.593500572836697e-06,
+      "loss": 0.7366,
+      "step": 12170
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5923141236402994e-06,
+      "loss": 0.4005,
+      "step": 12171
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5911280780884308e-06,
+      "loss": 0.7395,
+      "step": 12172
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.589942436238029e-06,
+      "loss": 0.7468,
+      "step": 12173
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5887571981460182e-06,
+      "loss": 0.7091,
+      "step": 12174
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5875723638692996e-06,
+      "loss": 0.7211,
+      "step": 12175
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5863879334647559e-06,
+      "loss": 0.714,
+      "step": 12176
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5852039069892477e-06,
+      "loss": 0.405,
+      "step": 12177
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5840202844996233e-06,
+      "loss": 0.7597,
+      "step": 12178
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5828370660527059e-06,
+      "loss": 0.7523,
+      "step": 12179
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5816542517053002e-06,
+      "loss": 0.6981,
+      "step": 12180
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5804718415141917e-06,
+      "loss": 0.7018,
+      "step": 12181
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.579289835536144e-06,
+      "loss": 0.7494,
+      "step": 12182
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5781082338279096e-06,
+      "loss": 0.3907,
+      "step": 12183
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5769270364462141e-06,
+      "loss": 0.7246,
+      "step": 12184
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5757462434477644e-06,
+      "loss": 0.3519,
+      "step": 12185
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5745658548892494e-06,
+      "loss": 0.4449,
+      "step": 12186
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5733858708273386e-06,
+      "loss": 0.7017,
+      "step": 12187
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.572206291318682e-06,
+      "loss": 0.7247,
+      "step": 12188
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5710271164199098e-06,
+      "loss": 0.3892,
+      "step": 12189
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5698483461876301e-06,
+      "loss": 0.7386,
+      "step": 12190
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.56866998067844e-06,
+      "loss": 0.7033,
+      "step": 12191
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5674920199489096e-06,
+      "loss": 0.6893,
+      "step": 12192
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5663144640555904e-06,
+      "loss": 0.7379,
+      "step": 12193
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5651373130550163e-06,
+      "loss": 0.6942,
+      "step": 12194
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5639605670036994e-06,
+      "loss": 0.7478,
+      "step": 12195
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5627842259581382e-06,
+      "loss": 0.4148,
+      "step": 12196
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5616082899748042e-06,
+      "loss": 0.3797,
+      "step": 12197
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5604327591101554e-06,
+      "loss": 0.4162,
+      "step": 12198
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5592576334206255e-06,
+      "loss": 0.7376,
+      "step": 12199
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5580829129626307e-06,
+      "loss": 0.7238,
+      "step": 12200
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5569085977925702e-06,
+      "loss": 0.8062,
+      "step": 12201
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5557346879668245e-06,
+      "loss": 0.4194,
+      "step": 12202
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5545611835417462e-06,
+      "loss": 0.6746,
+      "step": 12203
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5533880845736725e-06,
+      "loss": 0.7845,
+      "step": 12204
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5522153911189297e-06,
+      "loss": 0.6805,
+      "step": 12205
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5510431032338146e-06,
+      "loss": 0.3452,
+      "step": 12206
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5498712209746069e-06,
+      "loss": 0.7508,
+      "step": 12207
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5486997443975683e-06,
+      "loss": 0.7389,
+      "step": 12208
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5475286735589369e-06,
+      "loss": 0.7268,
+      "step": 12209
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.54635800851494e-06,
+      "loss": 0.7696,
+      "step": 12210
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.545187749321777e-06,
+      "loss": 0.7088,
+      "step": 12211
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5440178960356322e-06,
+      "loss": 0.7513,
+      "step": 12212
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5428484487126671e-06,
+      "loss": 0.741,
+      "step": 12213
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5416794074090258e-06,
+      "loss": 0.3961,
+      "step": 12214
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5405107721808354e-06,
+      "loss": 0.694,
+      "step": 12215
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5393425430842002e-06,
+      "loss": 0.723,
+      "step": 12216
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.538174720175204e-06,
+      "loss": 0.7472,
+      "step": 12217
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5370073035099142e-06,
+      "loss": 0.7287,
+      "step": 12218
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.535840293144376e-06,
+      "loss": 0.7636,
+      "step": 12219
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5346736891346182e-06,
+      "loss": 0.4415,
+      "step": 12220
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5335074915366466e-06,
+      "loss": 0.6952,
+      "step": 12221
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.532341700406448e-06,
+      "loss": 0.7411,
+      "step": 12222
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5311763157999949e-06,
+      "loss": 0.7181,
+      "step": 12223
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.530011337773234e-06,
+      "loss": 0.4179,
+      "step": 12224
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.528846766382094e-06,
+      "loss": 0.7611,
+      "step": 12225
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5276826016824863e-06,
+      "loss": 0.6959,
+      "step": 12226
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.526518843730298e-06,
+      "loss": 0.755,
+      "step": 12227
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.525355492581405e-06,
+      "loss": 0.7129,
+      "step": 12228
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5241925482916553e-06,
+      "loss": 0.6769,
+      "step": 12229
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5230300109168828e-06,
+      "loss": 0.3983,
+      "step": 12230
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5218678805128972e-06,
+      "loss": 0.3881,
+      "step": 12231
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.520706157135493e-06,
+      "loss": 0.6965,
+      "step": 12232
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5195448408404422e-06,
+      "loss": 0.7054,
+      "step": 12233
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5183839316834992e-06,
+      "loss": 0.7499,
+      "step": 12234
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.517223429720398e-06,
+      "loss": 0.6978,
+      "step": 12235
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.516063335006851e-06,
+      "loss": 0.4104,
+      "step": 12236
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5149036475985568e-06,
+      "loss": 0.6583,
+      "step": 12237
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5137443675511899e-06,
+      "loss": 0.7424,
+      "step": 12238
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5125854949204055e-06,
+      "loss": 0.7176,
+      "step": 12239
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5114270297618394e-06,
+      "loss": 0.7425,
+      "step": 12240
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.510268972131107e-06,
+      "loss": 0.7618,
+      "step": 12241
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5091113220838093e-06,
+      "loss": 0.6892,
+      "step": 12242
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.507954079675522e-06,
+      "loss": 0.3925,
+      "step": 12243
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5067972449618028e-06,
+      "loss": 0.724,
+      "step": 12244
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5056408179981919e-06,
+      "loss": 0.3933,
+      "step": 12245
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5044847988402023e-06,
+      "loss": 0.3869,
+      "step": 12246
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5033291875433443e-06,
+      "loss": 0.6652,
+      "step": 12247
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5021739841630878e-06,
+      "loss": 0.3978,
+      "step": 12248
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.501019188754893e-06,
+      "loss": 0.7405,
+      "step": 12249
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4998648013742079e-06,
+      "loss": 0.715,
+      "step": 12250
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.498710822076448e-06,
+      "loss": 0.7232,
+      "step": 12251
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4975572509170167e-06,
+      "loss": 0.7173,
+      "step": 12252
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4964040879512943e-06,
+      "loss": 0.7006,
+      "step": 12253
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.495251333234642e-06,
+      "loss": 0.3745,
+      "step": 12254
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4940989868224054e-06,
+      "loss": 0.735,
+      "step": 12255
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4929470487699071e-06,
+      "loss": 0.7192,
+      "step": 12256
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4917955191324495e-06,
+      "loss": 0.6879,
+      "step": 12257
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4906443979653173e-06,
+      "loss": 0.4163,
+      "step": 12258
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4894936853237707e-06,
+      "loss": 0.4052,
+      "step": 12259
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.48834338126306e-06,
+      "loss": 0.4529,
+      "step": 12260
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4871934858384086e-06,
+      "loss": 0.3936,
+      "step": 12261
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4860439991050201e-06,
+      "loss": 0.7259,
+      "step": 12262
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4848949211180775e-06,
+      "loss": 0.3697,
+      "step": 12263
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4837462519327527e-06,
+      "loss": 0.7505,
+      "step": 12264
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4825979916041899e-06,
+      "loss": 0.7656,
+      "step": 12265
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4814501401875158e-06,
+      "loss": 0.7484,
+      "step": 12266
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4803026977378377e-06,
+      "loss": 0.4164,
+      "step": 12267
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4791556643102411e-06,
+      "loss": 0.7158,
+      "step": 12268
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.478009039959797e-06,
+      "loss": 0.7078,
+      "step": 12269
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4768628247415528e-06,
+      "loss": 0.7336,
+      "step": 12270
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4757170187105374e-06,
+      "loss": 0.4152,
+      "step": 12271
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4745716219217577e-06,
+      "loss": 0.737,
+      "step": 12272
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.473426634430204e-06,
+      "loss": 0.4236,
+      "step": 12273
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.472282056290848e-06,
+      "loss": 0.7023,
+      "step": 12274
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.471137887558638e-06,
+      "loss": 0.3895,
+      "step": 12275
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4699941282885077e-06,
+      "loss": 0.4554,
+      "step": 12276
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.468850778535359e-06,
+      "loss": 0.7027,
+      "step": 12277
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4677078383540922e-06,
+      "loss": 0.7205,
+      "step": 12278
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4665653077995744e-06,
+      "loss": 0.7288,
+      "step": 12279
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.465423186926659e-06,
+      "loss": 0.7329,
+      "step": 12280
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4642814757901747e-06,
+      "loss": 0.6725,
+      "step": 12281
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4631401744449393e-06,
+      "loss": 0.7445,
+      "step": 12282
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4619992829457418e-06,
+      "loss": 0.4395,
+      "step": 12283
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4608588013473557e-06,
+      "loss": 0.7139,
+      "step": 12284
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4597187297045356e-06,
+      "loss": 0.4163,
+      "step": 12285
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4585790680720124e-06,
+      "loss": 0.6954,
+      "step": 12286
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4574398165045045e-06,
+      "loss": 0.4071,
+      "step": 12287
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4563009750567037e-06,
+      "loss": 0.7782,
+      "step": 12288
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.455162543783284e-06,
+      "loss": 0.7317,
+      "step": 12289
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4540245227389016e-06,
+      "loss": 0.6732,
+      "step": 12290
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.452886911978192e-06,
+      "loss": 0.6998,
+      "step": 12291
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4517497115557688e-06,
+      "loss": 0.3933,
+      "step": 12292
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.45061292152623e-06,
+      "loss": 0.7349,
+      "step": 12293
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4494765419441503e-06,
+      "loss": 0.7285,
+      "step": 12294
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4483405728640854e-06,
+      "loss": 0.7159,
+      "step": 12295
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4472050143405747e-06,
+      "loss": 0.4045,
+      "step": 12296
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4460698664281336e-06,
+      "loss": 0.3917,
+      "step": 12297
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4449351291812597e-06,
+      "loss": 0.7129,
+      "step": 12298
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4438008026544304e-06,
+      "loss": 0.3834,
+      "step": 12299
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.442666886902101e-06,
+      "loss": 0.4285,
+      "step": 12300
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4415333819787158e-06,
+      "loss": 0.7451,
+      "step": 12301
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4404002879386892e-06,
+      "loss": 0.6917,
+      "step": 12302
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4392676048364196e-06,
+      "loss": 0.7356,
+      "step": 12303
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.438135332726288e-06,
+      "loss": 0.383,
+      "step": 12304
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.437003471662649e-06,
+      "loss": 0.7214,
+      "step": 12305
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4358720216998512e-06,
+      "loss": 0.7096,
+      "step": 12306
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4347409828922065e-06,
+      "loss": 0.4222,
+      "step": 12307
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.433610355294014e-06,
+      "loss": 0.713,
+      "step": 12308
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.43248013895956e-06,
+      "loss": 0.7313,
+      "step": 12309
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4313503339431023e-06,
+      "loss": 0.4307,
+      "step": 12310
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.430220940298882e-06,
+      "loss": 0.7538,
+      "step": 12311
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4290919580811191e-06,
+      "loss": 0.4158,
+      "step": 12312
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4279633873440135e-06,
+      "loss": 0.7485,
+      "step": 12313
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.426835228141751e-06,
+      "loss": 0.7579,
+      "step": 12314
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.425707480528491e-06,
+      "loss": 0.4352,
+      "step": 12315
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4245801445583763e-06,
+      "loss": 0.6944,
+      "step": 12316
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4234532202855278e-06,
+      "loss": 0.741,
+      "step": 12317
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.422326707764048e-06,
+      "loss": 0.7142,
+      "step": 12318
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.421200607048021e-06,
+      "loss": 0.6959,
+      "step": 12319
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4200749181915097e-06,
+      "loss": 0.3744,
+      "step": 12320
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4189496412485593e-06,
+      "loss": 0.6871,
+      "step": 12321
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4178247762731866e-06,
+      "loss": 0.7612,
+      "step": 12322
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.416700323319401e-06,
+      "loss": 0.4222,
+      "step": 12323
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4155762824411845e-06,
+      "loss": 0.724,
+      "step": 12324
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4144526536925018e-06,
+      "loss": 0.7806,
+      "step": 12325
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4133294371272966e-06,
+      "loss": 0.6812,
+      "step": 12326
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.412206632799491e-06,
+      "loss": 0.7171,
+      "step": 12327
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4110842407629943e-06,
+      "loss": 0.7117,
+      "step": 12328
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4099622610716902e-06,
+      "loss": 0.7935,
+      "step": 12329
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.408840693779442e-06,
+      "loss": 0.6538,
+      "step": 12330
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4077195389400967e-06,
+      "loss": 0.694,
+      "step": 12331
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4065987966074769e-06,
+      "loss": 0.7608,
+      "step": 12332
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4054784668353915e-06,
+      "loss": 0.3764,
+      "step": 12333
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.404358549677627e-06,
+      "loss": 0.3775,
+      "step": 12334
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4032390451879473e-06,
+      "loss": 0.7313,
+      "step": 12335
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.402119953420099e-06,
+      "loss": 0.4222,
+      "step": 12336
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.401001274427809e-06,
+      "loss": 0.4139,
+      "step": 12337
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.399883008264783e-06,
+      "loss": 0.7436,
+      "step": 12338
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3987651549847104e-06,
+      "loss": 0.7257,
+      "step": 12339
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3976477146412527e-06,
+      "loss": 0.7581,
+      "step": 12340
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3965306872880635e-06,
+      "loss": 0.7339,
+      "step": 12341
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.395414072978767e-06,
+      "loss": 0.6845,
+      "step": 12342
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.394297871766972e-06,
+      "loss": 0.7204,
+      "step": 12343
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.393182083706265e-06,
+      "loss": 0.6393,
+      "step": 12344
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3920667088502126e-06,
+      "loss": 0.745,
+      "step": 12345
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.390951747252366e-06,
+      "loss": 0.4131,
+      "step": 12346
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3898371989662518e-06,
+      "loss": 0.7244,
+      "step": 12347
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3887230640453787e-06,
+      "loss": 0.7296,
+      "step": 12348
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3876093425432347e-06,
+      "loss": 0.72,
+      "step": 12349
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3864960345132871e-06,
+      "loss": 0.3765,
+      "step": 12350
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3853831400089901e-06,
+      "loss": 0.3735,
+      "step": 12351
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3842706590837663e-06,
+      "loss": 0.6813,
+      "step": 12352
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3831585917910284e-06,
+      "loss": 0.3988,
+      "step": 12353
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3820469381841618e-06,
+      "loss": 0.7245,
+      "step": 12354
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3809356983165401e-06,
+      "loss": 0.7436,
+      "step": 12355
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3798248722415119e-06,
+      "loss": 0.4309,
+      "step": 12356
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3787144600124058e-06,
+      "loss": 0.6489,
+      "step": 12357
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3776044616825323e-06,
+      "loss": 0.7266,
+      "step": 12358
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.376494877305179e-06,
+      "loss": 0.371,
+      "step": 12359
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3753857069336196e-06,
+      "loss": 0.7271,
+      "step": 12360
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.374276950621103e-06,
+      "loss": 0.7134,
+      "step": 12361
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3731686084208584e-06,
+      "loss": 0.4111,
+      "step": 12362
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.372060680386097e-06,
+      "loss": 0.3742,
+      "step": 12363
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.370953166570007e-06,
+      "loss": 0.7118,
+      "step": 12364
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3698460670257641e-06,
+      "loss": 0.7561,
+      "step": 12365
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3687393818065176e-06,
+      "loss": 0.7108,
+      "step": 12366
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.367633110965393e-06,
+      "loss": 0.686,
+      "step": 12367
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3665272545555074e-06,
+      "loss": 0.3565,
+      "step": 12368
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.365421812629949e-06,
+      "loss": 0.7102,
+      "step": 12369
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3643167852417894e-06,
+      "loss": 0.3981,
+      "step": 12370
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3632121724440806e-06,
+      "loss": 0.4422,
+      "step": 12371
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3621079742898502e-06,
+      "loss": 0.6563,
+      "step": 12372
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3610041908321158e-06,
+      "loss": 0.7397,
+      "step": 12373
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3599008221238651e-06,
+      "loss": 0.6886,
+      "step": 12374
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3587978682180713e-06,
+      "loss": 0.7545,
+      "step": 12375
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3576953291676842e-06,
+      "loss": 0.6726,
+      "step": 12376
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3565932050256357e-06,
+      "loss": 0.722,
+      "step": 12377
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3554914958448406e-06,
+      "loss": 0.7504,
+      "step": 12378
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3543902016781884e-06,
+      "loss": 0.6908,
+      "step": 12379
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3532893225785536e-06,
+      "loss": 0.4477,
+      "step": 12380
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.352188858598783e-06,
+      "loss": 0.4096,
+      "step": 12381
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.351088809791714e-06,
+      "loss": 0.7138,
+      "step": 12382
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3499891762101558e-06,
+      "loss": 0.7136,
+      "step": 12383
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3488899579069025e-06,
+      "loss": 0.7068,
+      "step": 12384
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3477911549347266e-06,
+      "loss": 0.3938,
+      "step": 12385
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.346692767346377e-06,
+      "loss": 0.7599,
+      "step": 12386
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3455947951945903e-06,
+      "loss": 0.6973,
+      "step": 12387
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3444972385320787e-06,
+      "loss": 0.7805,
+      "step": 12388
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3434000974115335e-06,
+      "loss": 0.3825,
+      "step": 12389
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.342303371885627e-06,
+      "loss": 0.7055,
+      "step": 12390
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3412070620070106e-06,
+      "loss": 0.7278,
+      "step": 12391
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3401111678283218e-06,
+      "loss": 0.6971,
+      "step": 12392
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3390156894021688e-06,
+      "loss": 0.6749,
+      "step": 12393
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3379206267811473e-06,
+      "loss": 0.3994,
+      "step": 12394
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3368259800178284e-06,
+      "loss": 0.7385,
+      "step": 12395
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.335731749164766e-06,
+      "loss": 0.7501,
+      "step": 12396
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3346379342744932e-06,
+      "loss": 0.3761,
+      "step": 12397
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3335445353995214e-06,
+      "loss": 0.6767,
+      "step": 12398
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.332451552592342e-06,
+      "loss": 0.7538,
+      "step": 12399
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3313589859054321e-06,
+      "loss": 0.7343,
+      "step": 12400
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.330266835391244e-06,
+      "loss": 0.4153,
+      "step": 12401
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3291751011022103e-06,
+      "loss": 0.7304,
+      "step": 12402
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3280837830907435e-06,
+      "loss": 0.4187,
+      "step": 12403
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3269928814092337e-06,
+      "loss": 0.7292,
+      "step": 12404
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3259023961100593e-06,
+      "loss": 0.6898,
+      "step": 12405
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3248123272455715e-06,
+      "loss": 0.7542,
+      "step": 12406
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3237226748681031e-06,
+      "loss": 0.3996,
+      "step": 12407
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3226334390299677e-06,
+      "loss": 0.4393,
+      "step": 12408
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3215446197834559e-06,
+      "loss": 0.3684,
+      "step": 12409
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3204562171808456e-06,
+      "loss": 0.3892,
+      "step": 12410
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.319368231274386e-06,
+      "loss": 0.7357,
+      "step": 12411
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3182806621163113e-06,
+      "loss": 0.72,
+      "step": 12412
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3171935097588328e-06,
+      "loss": 0.702,
+      "step": 12413
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3161067742541467e-06,
+      "loss": 0.4042,
+      "step": 12414
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3150204556544256e-06,
+      "loss": 0.7299,
+      "step": 12415
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.313934554011822e-06,
+      "loss": 0.6942,
+      "step": 12416
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3128490693784678e-06,
+      "loss": 0.7646,
+      "step": 12417
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3117640018064747e-06,
+      "loss": 0.4027,
+      "step": 12418
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.310679351347941e-06,
+      "loss": 0.6908,
+      "step": 12419
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.309595118054936e-06,
+      "loss": 0.6701,
+      "step": 12420
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3085113019795136e-06,
+      "loss": 0.7446,
+      "step": 12421
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3074279031737068e-06,
+      "loss": 0.701,
+      "step": 12422
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3063449216895252e-06,
+      "loss": 0.6977,
+      "step": 12423
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3052623575789669e-06,
+      "loss": 0.7164,
+      "step": 12424
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3041802108940038e-06,
+      "loss": 0.7197,
+      "step": 12425
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3030984816865854e-06,
+      "loss": 0.7251,
+      "step": 12426
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3020171700086448e-06,
+      "loss": 0.7466,
+      "step": 12427
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.300936275912098e-06,
+      "loss": 0.6884,
+      "step": 12428
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2998557994488348e-06,
+      "loss": 0.748,
+      "step": 12429
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2987757406707303e-06,
+      "loss": 0.724,
+      "step": 12430
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2976960996296318e-06,
+      "loss": 0.7179,
+      "step": 12431
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.296616876377379e-06,
+      "loss": 0.7421,
+      "step": 12432
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2955380709657795e-06,
+      "loss": 0.3765,
+      "step": 12433
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2944596834466272e-06,
+      "loss": 0.7222,
+      "step": 12434
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2933817138716941e-06,
+      "loss": 0.4185,
+      "step": 12435
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2923041622927312e-06,
+      "loss": 0.7297,
+      "step": 12436
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2912270287614737e-06,
+      "loss": 0.7377,
+      "step": 12437
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2901503133296311e-06,
+      "loss": 0.7132,
+      "step": 12438
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2890740160488967e-06,
+      "loss": 0.6858,
+      "step": 12439
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2879981369709427e-06,
+      "loss": 0.3794,
+      "step": 12440
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2869226761474195e-06,
+      "loss": 0.6788,
+      "step": 12441
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2858476336299597e-06,
+      "loss": 0.7306,
+      "step": 12442
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2847730094701749e-06,
+      "loss": 0.3553,
+      "step": 12443
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2836988037196573e-06,
+      "loss": 0.4152,
+      "step": 12444
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2826250164299759e-06,
+      "loss": 0.6912,
+      "step": 12445
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2815516476526857e-06,
+      "loss": 0.3992,
+      "step": 12446
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2804786974393168e-06,
+      "loss": 0.4292,
+      "step": 12447
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2794061658413792e-06,
+      "loss": 0.3954,
+      "step": 12448
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.278334052910366e-06,
+      "loss": 0.6884,
+      "step": 12449
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2772623586977452e-06,
+      "loss": 0.7772,
+      "step": 12450
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2761910832549719e-06,
+      "loss": 0.7497,
+      "step": 12451
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.275120226633475e-06,
+      "loss": 0.7487,
+      "step": 12452
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2740497888846647e-06,
+      "loss": 0.3791,
+      "step": 12453
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2729797700599323e-06,
+      "loss": 0.7157,
+      "step": 12454
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2719101702106485e-06,
+      "loss": 0.7159,
+      "step": 12455
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2708409893881635e-06,
+      "loss": 0.6594,
+      "step": 12456
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2697722276438073e-06,
+      "loss": 0.7007,
+      "step": 12457
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2687038850288868e-06,
+      "loss": 0.3912,
+      "step": 12458
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2676359615946987e-06,
+      "loss": 0.3719,
+      "step": 12459
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2665684573925097e-06,
+      "loss": 0.4212,
+      "step": 12460
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2655013724735699e-06,
+      "loss": 0.6681,
+      "step": 12461
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2644347068891083e-06,
+      "loss": 0.7168,
+      "step": 12462
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2633684606903318e-06,
+      "loss": 0.6587,
+      "step": 12463
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.262302633928436e-06,
+      "loss": 0.6989,
+      "step": 12464
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2612372266545869e-06,
+      "loss": 0.7501,
+      "step": 12465
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2601722389199334e-06,
+      "loss": 0.6809,
+      "step": 12466
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2591076707756044e-06,
+      "loss": 0.7255,
+      "step": 12467
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2580435222727084e-06,
+      "loss": 0.7358,
+      "step": 12468
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2569797934623352e-06,
+      "loss": 0.7411,
+      "step": 12469
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2559164843955563e-06,
+      "loss": 0.7337,
+      "step": 12470
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2548535951234141e-06,
+      "loss": 0.73,
+      "step": 12471
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2537911256969381e-06,
+      "loss": 0.6876,
+      "step": 12472
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2527290761671396e-06,
+      "loss": 0.4172,
+      "step": 12473
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2516674465850042e-06,
+      "loss": 0.702,
+      "step": 12474
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2506062370015003e-06,
+      "loss": 0.7867,
+      "step": 12475
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.249545447467576e-06,
+      "loss": 0.7011,
+      "step": 12476
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.248485078034155e-06,
+      "loss": 0.7644,
+      "step": 12477
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.247425128752151e-06,
+      "loss": 0.7708,
+      "step": 12478
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2463655996724477e-06,
+      "loss": 0.7553,
+      "step": 12479
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.245306490845911e-06,
+      "loss": 0.4292,
+      "step": 12480
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2442478023233894e-06,
+      "loss": 0.7606,
+      "step": 12481
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2431895341557077e-06,
+      "loss": 0.7249,
+      "step": 12482
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.242131686393675e-06,
+      "loss": 0.7439,
+      "step": 12483
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2410742590880753e-06,
+      "loss": 0.4281,
+      "step": 12484
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2400172522896782e-06,
+      "loss": 0.6607,
+      "step": 12485
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2389606660492215e-06,
+      "loss": 0.7278,
+      "step": 12486
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2379045004174384e-06,
+      "loss": 0.717,
+      "step": 12487
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2368487554450314e-06,
+      "loss": 0.6293,
+      "step": 12488
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2357934311826869e-06,
+      "loss": 0.7542,
+      "step": 12489
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2347385276810675e-06,
+      "loss": 0.3559,
+      "step": 12490
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2336840449908215e-06,
+      "loss": 0.7453,
+      "step": 12491
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2326299831625722e-06,
+      "loss": 0.7567,
+      "step": 12492
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.231576342246923e-06,
+      "loss": 0.6915,
+      "step": 12493
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.230523122294459e-06,
+      "loss": 0.4168,
+      "step": 12494
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2294703233557425e-06,
+      "loss": 0.6836,
+      "step": 12495
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2284179454813205e-06,
+      "loss": 0.714,
+      "step": 12496
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2273659887217159e-06,
+      "loss": 0.7127,
+      "step": 12497
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2263144531274306e-06,
+      "loss": 0.3841,
+      "step": 12498
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2252633387489477e-06,
+      "loss": 0.7165,
+      "step": 12499
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2242126456367321e-06,
+      "loss": 0.4119,
+      "step": 12500
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2231623738412246e-06,
+      "loss": 0.3724,
+      "step": 12501
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2221125234128484e-06,
+      "loss": 0.7313,
+      "step": 12502
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2210630944020052e-06,
+      "loss": 0.6541,
+      "step": 12503
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2200140868590759e-06,
+      "loss": 0.6996,
+      "step": 12504
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2189655008344258e-06,
+      "loss": 0.6848,
+      "step": 12505
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2179173363783959e-06,
+      "loss": 0.7758,
+      "step": 12506
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2168695935413043e-06,
+      "loss": 0.7099,
+      "step": 12507
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2158222723734547e-06,
+      "loss": 0.7443,
+      "step": 12508
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2147753729251254e-06,
+      "loss": 0.7404,
+      "step": 12509
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2137288952465808e-06,
+      "loss": 0.6827,
+      "step": 12510
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2126828393880586e-06,
+      "loss": 0.7099,
+      "step": 12511
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2116372053997804e-06,
+      "loss": 0.7009,
+      "step": 12512
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2105919933319443e-06,
+      "loss": 0.7484,
+      "step": 12513
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.209547203234731e-06,
+      "loss": 0.7896,
+      "step": 12514
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2085028351582994e-06,
+      "loss": 0.3803,
+      "step": 12515
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2074588891527894e-06,
+      "loss": 0.6729,
+      "step": 12516
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2064153652683165e-06,
+      "loss": 0.7583,
+      "step": 12517
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2053722635549836e-06,
+      "loss": 0.7024,
+      "step": 12518
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2043295840628676e-06,
+      "loss": 0.6793,
+      "step": 12519
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2032873268420265e-06,
+      "loss": 0.7163,
+      "step": 12520
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2022454919424974e-06,
+      "loss": 0.7619,
+      "step": 12521
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2012040794142954e-06,
+      "loss": 0.4167,
+      "step": 12522
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2001630893074224e-06,
+      "loss": 0.3644,
+      "step": 12523
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1991225216718538e-06,
+      "loss": 0.7178,
+      "step": 12524
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1980823765575444e-06,
+      "loss": 0.6971,
+      "step": 12525
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.197042654014432e-06,
+      "loss": 0.3724,
+      "step": 12526
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.196003354092431e-06,
+      "loss": 0.7531,
+      "step": 12527
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1949644768414392e-06,
+      "loss": 0.4093,
+      "step": 12528
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1939260223113347e-06,
+      "loss": 0.714,
+      "step": 12529
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1928879905519664e-06,
+      "loss": 0.7929,
+      "step": 12530
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.19185038161317e-06,
+      "loss": 0.7056,
+      "step": 12531
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1908131955447633e-06,
+      "loss": 0.753,
+      "step": 12532
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.18977643239654e-06,
+      "loss": 0.764,
+      "step": 12533
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1887400922182735e-06,
+      "loss": 0.7354,
+      "step": 12534
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1877041750597174e-06,
+      "loss": 0.7581,
+      "step": 12535
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1866686809706018e-06,
+      "loss": 0.6955,
+      "step": 12536
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.185633610000646e-06,
+      "loss": 0.727,
+      "step": 12537
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.184598962199539e-06,
+      "loss": 0.6904,
+      "step": 12538
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1835647376169535e-06,
+      "loss": 0.6891,
+      "step": 12539
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1825309363025427e-06,
+      "loss": 0.6911,
+      "step": 12540
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1814975583059352e-06,
+      "loss": 0.3846,
+      "step": 12541
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1804646036767465e-06,
+      "loss": 0.7152,
+      "step": 12542
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1794320724645658e-06,
+      "loss": 0.7125,
+      "step": 12543
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.178399964718967e-06,
+      "loss": 0.4021,
+      "step": 12544
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1773682804894926e-06,
+      "loss": 0.7083,
+      "step": 12545
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1763370198256797e-06,
+      "loss": 0.7269,
+      "step": 12546
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1753061827770362e-06,
+      "loss": 0.7183,
+      "step": 12547
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1742757693930517e-06,
+      "loss": 0.748,
+      "step": 12548
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1732457797231933e-06,
+      "loss": 0.7039,
+      "step": 12549
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1722162138169135e-06,
+      "loss": 0.6689,
+      "step": 12550
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1711870717236384e-06,
+      "loss": 0.3923,
+      "step": 12551
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1701583534927763e-06,
+      "loss": 0.6726,
+      "step": 12552
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1691300591737153e-06,
+      "loss": 0.7111,
+      "step": 12553
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1681021888158196e-06,
+      "loss": 0.4239,
+      "step": 12554
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1670747424684425e-06,
+      "loss": 0.6923,
+      "step": 12555
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1660477201809062e-06,
+      "loss": 0.7518,
+      "step": 12556
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1650211220025187e-06,
+      "loss": 0.737,
+      "step": 12557
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1639949479825652e-06,
+      "loss": 0.7042,
+      "step": 12558
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1629691981703116e-06,
+      "loss": 0.6959,
+      "step": 12559
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1619438726150022e-06,
+      "loss": 0.6686,
+      "step": 12560
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.160918971365864e-06,
+      "loss": 0.3965,
+      "step": 12561
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1598944944720991e-06,
+      "loss": 0.7454,
+      "step": 12562
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1588704419828912e-06,
+      "loss": 0.8329,
+      "step": 12563
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1578468139474064e-06,
+      "loss": 0.7479,
+      "step": 12564
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1568236104147878e-06,
+      "loss": 0.7064,
+      "step": 12565
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1558008314341573e-06,
+      "loss": 0.3935,
+      "step": 12566
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1547784770546179e-06,
+      "loss": 0.7495,
+      "step": 12567
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1537565473252487e-06,
+      "loss": 0.3919,
+      "step": 12568
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1527350422951178e-06,
+      "loss": 0.7594,
+      "step": 12569
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1517139620132634e-06,
+      "loss": 0.7286,
+      "step": 12570
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1506933065287062e-06,
+      "loss": 0.7333,
+      "step": 12571
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1496730758904472e-06,
+      "loss": 0.6907,
+      "step": 12572
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.148653270147464e-06,
+      "loss": 0.7183,
+      "step": 12573
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1476338893487237e-06,
+      "loss": 0.715,
+      "step": 12574
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1466149335431586e-06,
+      "loss": 0.71,
+      "step": 12575
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1455964027796906e-06,
+      "loss": 0.7261,
+      "step": 12576
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1445782971072161e-06,
+      "loss": 0.3971,
+      "step": 12577
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.143560616574616e-06,
+      "loss": 0.7237,
+      "step": 12578
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.142543361230749e-06,
+      "loss": 0.6875,
+      "step": 12579
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1415265311244495e-06,
+      "loss": 0.7315,
+      "step": 12580
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1405101263045338e-06,
+      "loss": 0.7134,
+      "step": 12581
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.139494146819803e-06,
+      "loss": 0.3735,
+      "step": 12582
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.138478592719031e-06,
+      "loss": 0.7193,
+      "step": 12583
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.137463464050973e-06,
+      "loss": 0.7276,
+      "step": 12584
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1364487608643638e-06,
+      "loss": 0.672,
+      "step": 12585
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1354344832079178e-06,
+      "loss": 0.7372,
+      "step": 12586
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1344206311303318e-06,
+      "loss": 0.7774,
+      "step": 12587
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1334072046802792e-06,
+      "loss": 0.405,
+      "step": 12588
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.132394203906415e-06,
+      "loss": 0.738,
+      "step": 12589
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1313816288573675e-06,
+      "loss": 0.7146,
+      "step": 12590
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1303694795817532e-06,
+      "loss": 0.4373,
+      "step": 12591
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1293577561281633e-06,
+      "loss": 0.7156,
+      "step": 12592
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1283464585451697e-06,
+      "loss": 0.6947,
+      "step": 12593
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1273355868813252e-06,
+      "loss": 0.706,
+      "step": 12594
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.126325141185156e-06,
+      "loss": 0.7204,
+      "step": 12595
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1253151215051782e-06,
+      "loss": 0.6934,
+      "step": 12596
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1243055278898808e-06,
+      "loss": 0.408,
+      "step": 12597
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.123296360387731e-06,
+      "loss": 0.77,
+      "step": 12598
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1222876190471799e-06,
+      "loss": 0.3911,
+      "step": 12599
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1212793039166524e-06,
+      "loss": 0.6933,
+      "step": 12600
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1202714150445627e-06,
+      "loss": 0.7358,
+      "step": 12601
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1192639524792948e-06,
+      "loss": 0.7556,
+      "step": 12602
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1182569162692158e-06,
+      "loss": 0.7162,
+      "step": 12603
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.117250306462675e-06,
+      "loss": 0.7041,
+      "step": 12604
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1162441231079968e-06,
+      "loss": 0.3744,
+      "step": 12605
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.115238366253486e-06,
+      "loss": 0.4354,
+      "step": 12606
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1142330359474306e-06,
+      "loss": 0.3802,
+      "step": 12607
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1132281322380922e-06,
+      "loss": 0.7908,
+      "step": 12608
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1122236551737187e-06,
+      "loss": 0.3732,
+      "step": 12609
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1112196048025326e-06,
+      "loss": 0.7465,
+      "step": 12610
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1102159811727376e-06,
+      "loss": 0.7301,
+      "step": 12611
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1092127843325152e-06,
+      "loss": 0.698,
+      "step": 12612
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.108210014330028e-06,
+      "loss": 0.6926,
+      "step": 12613
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1072076712134205e-06,
+      "loss": 0.6798,
+      "step": 12614
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1062057550308137e-06,
+      "loss": 0.7398,
+      "step": 12615
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1052042658303063e-06,
+      "loss": 0.6988,
+      "step": 12616
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1042032036599804e-06,
+      "loss": 0.6943,
+      "step": 12617
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.103202568567896e-06,
+      "loss": 0.3809,
+      "step": 12618
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1022023606020927e-06,
+      "loss": 0.7504,
+      "step": 12619
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1012025798105875e-06,
+      "loss": 0.6536,
+      "step": 12620
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1002032262413808e-06,
+      "loss": 0.692,
+      "step": 12621
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0992042999424491e-06,
+      "loss": 0.6843,
+      "step": 12622
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0982058009617525e-06,
+      "loss": 0.7222,
+      "step": 12623
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0972077293472262e-06,
+      "loss": 0.7336,
+      "step": 12624
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.096210085146786e-06,
+      "loss": 0.678,
+      "step": 12625
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0952128684083285e-06,
+      "loss": 0.7264,
+      "step": 12626
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0942160791797273e-06,
+      "loss": 0.3555,
+      "step": 12627
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0932197175088421e-06,
+      "loss": 0.3412,
+      "step": 12628
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0922237834435034e-06,
+      "loss": 0.411,
+      "step": 12629
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0912282770315253e-06,
+      "loss": 0.6486,
+      "step": 12630
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0902331983207016e-06,
+      "loss": 0.7059,
+      "step": 12631
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.089238547358803e-06,
+      "loss": 0.7014,
+      "step": 12632
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.088244324193588e-06,
+      "loss": 0.397,
+      "step": 12633
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.087250528872782e-06,
+      "loss": 0.7188,
+      "step": 12634
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0862571614440964e-06,
+      "loss": 0.7255,
+      "step": 12635
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0852642219552233e-06,
+      "loss": 0.6951,
+      "step": 12636
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0842717104538337e-06,
+      "loss": 0.424,
+      "step": 12637
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0832796269875757e-06,
+      "loss": 0.4777,
+      "step": 12638
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0822879716040803e-06,
+      "loss": 0.6976,
+      "step": 12639
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0812967443509503e-06,
+      "loss": 0.7571,
+      "step": 12640
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0803059452757803e-06,
+      "loss": 0.7066,
+      "step": 12641
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0793155744261352e-06,
+      "loss": 0.7144,
+      "step": 12642
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0783256318495605e-06,
+      "loss": 0.7236,
+      "step": 12643
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0773361175935836e-06,
+      "loss": 0.7493,
+      "step": 12644
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0763470317057078e-06,
+      "loss": 0.6663,
+      "step": 12645
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0753583742334206e-06,
+      "loss": 0.7332,
+      "step": 12646
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0743701452241872e-06,
+      "loss": 0.4355,
+      "step": 12647
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.073382344725451e-06,
+      "loss": 0.7117,
+      "step": 12648
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0723949727846306e-06,
+      "loss": 0.7241,
+      "step": 12649
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0714080294491346e-06,
+      "loss": 0.3696,
+      "step": 12650
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0704215147663432e-06,
+      "loss": 0.7091,
+      "step": 12651
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0694354287836173e-06,
+      "loss": 0.7667,
+      "step": 12652
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0684497715482989e-06,
+      "loss": 0.7482,
+      "step": 12653
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0674645431077047e-06,
+      "loss": 0.723,
+      "step": 12654
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.066479743509141e-06,
+      "loss": 0.7594,
+      "step": 12655
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0654953727998828e-06,
+      "loss": 0.7018,
+      "step": 12656
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0645114310271898e-06,
+      "loss": 0.6962,
+      "step": 12657
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0635279182383008e-06,
+      "loss": 0.669,
+      "step": 12658
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.06254483448043e-06,
+      "loss": 0.4115,
+      "step": 12659
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.061562179800778e-06,
+      "loss": 0.7328,
+      "step": 12660
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.06057995424652e-06,
+      "loss": 0.4118,
+      "step": 12661
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0595981578648117e-06,
+      "loss": 0.6545,
+      "step": 12662
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0586167907027877e-06,
+      "loss": 0.7277,
+      "step": 12663
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0576358528075636e-06,
+      "loss": 0.6735,
+      "step": 12664
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0566553442262306e-06,
+      "loss": 0.7255,
+      "step": 12665
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0556752650058643e-06,
+      "loss": 0.726,
+      "step": 12666
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.054695615193515e-06,
+      "loss": 0.6946,
+      "step": 12667
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0537163948362173e-06,
+      "loss": 0.6882,
+      "step": 12668
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0527376039809821e-06,
+      "loss": 0.7169,
+      "step": 12669
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0517592426747992e-06,
+      "loss": 0.7195,
+      "step": 12670
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0507813109646392e-06,
+      "loss": 0.7399,
+      "step": 12671
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0498038088974504e-06,
+      "loss": 0.7779,
+      "step": 12672
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0488267365201642e-06,
+      "loss": 0.7245,
+      "step": 12673
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0478500938796876e-06,
+      "loss": 0.6802,
+      "step": 12674
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0468738810229073e-06,
+      "loss": 0.3919,
+      "step": 12675
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0458980979966927e-06,
+      "loss": 0.7258,
+      "step": 12676
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.044922744847886e-06,
+      "loss": 0.3967,
+      "step": 12677
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.043947821623319e-06,
+      "loss": 0.7083,
+      "step": 12678
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0429733283697917e-06,
+      "loss": 0.7102,
+      "step": 12679
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0419992651340905e-06,
+      "loss": 0.7166,
+      "step": 12680
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0410256319629763e-06,
+      "loss": 0.7161,
+      "step": 12681
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0400524289031965e-06,
+      "loss": 0.3799,
+      "step": 12682
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0390796560014726e-06,
+      "loss": 0.3964,
+      "step": 12683
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.038107313304505e-06,
+      "loss": 0.3608,
+      "step": 12684
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0371354008589762e-06,
+      "loss": 0.7493,
+      "step": 12685
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0361639187115446e-06,
+      "loss": 0.7239,
+      "step": 12686
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0351928669088528e-06,
+      "loss": 0.7184,
+      "step": 12687
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0342222454975193e-06,
+      "loss": 0.7595,
+      "step": 12688
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.033252054524142e-06,
+      "loss": 0.6843,
+      "step": 12689
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0322822940352994e-06,
+      "loss": 0.4198,
+      "step": 12690
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0313129640775454e-06,
+      "loss": 0.7071,
+      "step": 12691
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0303440646974229e-06,
+      "loss": 0.4222,
+      "step": 12692
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0293755959414453e-06,
+      "loss": 0.7024,
+      "step": 12693
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0284075578561048e-06,
+      "loss": 0.4052,
+      "step": 12694
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0274399504878763e-06,
+      "loss": 0.6686,
+      "step": 12695
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.026472773883218e-06,
+      "loss": 0.7161,
+      "step": 12696
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0255060280885599e-06,
+      "loss": 0.7026,
+      "step": 12697
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0245397131503143e-06,
+      "loss": 0.7103,
+      "step": 12698
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0235738291148734e-06,
+      "loss": 0.7266,
+      "step": 12699
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0226083760286088e-06,
+      "loss": 0.3962,
+      "step": 12700
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0216433539378723e-06,
+      "loss": 0.7474,
+      "step": 12701
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0206787628889914e-06,
+      "loss": 0.6951,
+      "step": 12702
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0197146029282757e-06,
+      "loss": 0.6961,
+      "step": 12703
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0187508741020124e-06,
+      "loss": 0.7417,
+      "step": 12704
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0177875764564716e-06,
+      "loss": 0.7115,
+      "step": 12705
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0168247100379003e-06,
+      "loss": 0.7282,
+      "step": 12706
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0158622748925229e-06,
+      "loss": 0.719,
+      "step": 12707
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0149002710665444e-06,
+      "loss": 0.7554,
+      "step": 12708
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0139386986061528e-06,
+      "loss": 0.7817,
+      "step": 12709
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0129775575575096e-06,
+      "loss": 0.4022,
+      "step": 12710
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0120168479667581e-06,
+      "loss": 0.6937,
+      "step": 12711
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0110565698800224e-06,
+      "loss": 0.7698,
+      "step": 12712
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0100967233434023e-06,
+      "loss": 0.7217,
+      "step": 12713
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.009137308402982e-06,
+      "loss": 0.7353,
+      "step": 12714
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0081783251048205e-06,
+      "loss": 0.3848,
+      "step": 12715
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0072197734949585e-06,
+      "loss": 0.4167,
+      "step": 12716
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0062616536194148e-06,
+      "loss": 0.6745,
+      "step": 12717
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0053039655241836e-06,
+      "loss": 0.6943,
+      "step": 12718
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0043467092552496e-06,
+      "loss": 0.7118,
+      "step": 12719
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0033898848585665e-06,
+      "loss": 0.3803,
+      "step": 12720
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0024334923800695e-06,
+      "loss": 0.7276,
+      "step": 12721
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0014775318656755e-06,
+      "loss": 0.7372,
+      "step": 12722
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0005220033612783e-06,
+      "loss": 0.7151,
+      "step": 12723
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.995669069127533e-07,
+      "loss": 0.6953,
+      "step": 12724
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.986122425659516e-07,
+      "loss": 0.4141,
+      "step": 12725
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.976580103667065e-07,
+      "loss": 0.4203,
+      "step": 12726
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.967042103608281e-07,
+      "loss": 0.6722,
+      "step": 12727
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.957508425941121e-07,
+      "loss": 0.6765,
+      "step": 12728
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.947979071123247e-07,
+      "loss": 0.7363,
+      "step": 12729
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.938454039612166e-07,
+      "loss": 0.693,
+      "step": 12730
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.928933331865153e-07,
+      "loss": 0.6934,
+      "step": 12731
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.91941694833931e-07,
+      "loss": 0.718,
+      "step": 12732
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.909904889491495e-07,
+      "loss": 0.4548,
+      "step": 12733
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.900397155778385e-07,
+      "loss": 0.6781,
+      "step": 12734
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.890893747656415e-07,
+      "loss": 0.393,
+      "step": 12735
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.881394665581834e-07,
+      "loss": 0.7345,
+      "step": 12736
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.871899910010728e-07,
+      "loss": 0.7219,
+      "step": 12737
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.862409481398882e-07,
+      "loss": 0.7062,
+      "step": 12738
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.85292338020194e-07,
+      "loss": 0.3847,
+      "step": 12739
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.843441606875292e-07,
+      "loss": 0.7477,
+      "step": 12740
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.8339641618742e-07,
+      "loss": 0.4269,
+      "step": 12741
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.82449104565364e-07,
+      "loss": 0.7124,
+      "step": 12742
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.815022258668406e-07,
+      "loss": 0.7574,
+      "step": 12743
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.80555780137309e-07,
+      "loss": 0.7041,
+      "step": 12744
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.796097674222039e-07,
+      "loss": 0.7281,
+      "step": 12745
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.786641877669477e-07,
+      "loss": 0.4049,
+      "step": 12746
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.777190412169346e-07,
+      "loss": 0.6946,
+      "step": 12747
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.76774327817539e-07,
+      "loss": 0.7036,
+      "step": 12748
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.75830047614117e-07,
+      "loss": 0.757,
+      "step": 12749
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.74886200652e-07,
+      "loss": 0.3666,
+      "step": 12750
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.73942786976505e-07,
+      "loss": 0.7433,
+      "step": 12751
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.72999806632925e-07,
+      "loss": 0.6819,
+      "step": 12752
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.720572596665267e-07,
+      "loss": 0.6981,
+      "step": 12753
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.71115146122561e-07,
+      "loss": 0.7056,
+      "step": 12754
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.701734660462625e-07,
+      "loss": 0.7325,
+      "step": 12755
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.692322194828364e-07,
+      "loss": 0.7782,
+      "step": 12756
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.68291406477474e-07,
+      "loss": 0.7037,
+      "step": 12757
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.673510270753382e-07,
+      "loss": 0.7378,
+      "step": 12758
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.66411081321581e-07,
+      "loss": 0.6968,
+      "step": 12759
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.65471569261326e-07,
+      "loss": 0.7419,
+      "step": 12760
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.64532490939678e-07,
+      "loss": 0.7429,
+      "step": 12761
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.635938464017214e-07,
+      "loss": 0.6808,
+      "step": 12762
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.626556356925176e-07,
+      "loss": 0.3899,
+      "step": 12763
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.617178588571118e-07,
+      "loss": 0.7077,
+      "step": 12764
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.607805159405258e-07,
+      "loss": 0.7674,
+      "step": 12765
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.598436069877604e-07,
+      "loss": 0.4582,
+      "step": 12766
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.589071320437949e-07,
+      "loss": 0.7308,
+      "step": 12767
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.579710911535879e-07,
+      "loss": 0.6807,
+      "step": 12768
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.570354843620777e-07,
+      "loss": 0.3843,
+      "step": 12769
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.561003117141832e-07,
+      "loss": 0.6781,
+      "step": 12770
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.551655732548004e-07,
+      "loss": 0.7023,
+      "step": 12771
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.542312690288035e-07,
+      "loss": 0.6748,
+      "step": 12772
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.532973990810512e-07,
+      "loss": 0.7699,
+      "step": 12773
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.523639634563752e-07,
+      "loss": 0.4099,
+      "step": 12774
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.514309621995899e-07,
+      "loss": 0.4002,
+      "step": 12775
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.50498395355487e-07,
+      "loss": 0.7469,
+      "step": 12776
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.495662629688363e-07,
+      "loss": 0.7515,
+      "step": 12777
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.48634565084392e-07,
+      "loss": 0.7226,
+      "step": 12778
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.477033017468829e-07,
+      "loss": 0.7582,
+      "step": 12779
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.467724730010164e-07,
+      "loss": 0.3958,
+      "step": 12780
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.458420788914823e-07,
+      "loss": 0.7247,
+      "step": 12781
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.449121194629485e-07,
+      "loss": 0.6353,
+      "step": 12782
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.439825947600589e-07,
+      "loss": 0.7233,
+      "step": 12783
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.430535048274414e-07,
+      "loss": 0.6824,
+      "step": 12784
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.421248497096991e-07,
+      "loss": 0.7201,
+      "step": 12785
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.411966294514141e-07,
+      "loss": 0.7205,
+      "step": 12786
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.402688440971541e-07,
+      "loss": 0.722,
+      "step": 12787
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.393414936914592e-07,
+      "loss": 0.4403,
+      "step": 12788
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.38414578278849e-07,
+      "loss": 0.713,
+      "step": 12789
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.374880979038225e-07,
+      "loss": 0.716,
+      "step": 12790
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.365620526108643e-07,
+      "loss": 0.6753,
+      "step": 12791
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.356364424444298e-07,
+      "loss": 0.4447,
+      "step": 12792
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.347112674489567e-07,
+      "loss": 0.7268,
+      "step": 12793
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.337865276688618e-07,
+      "loss": 0.7954,
+      "step": 12794
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.328622231485396e-07,
+      "loss": 0.7198,
+      "step": 12795
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.319383539323678e-07,
+      "loss": 0.6882,
+      "step": 12796
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.310149200647001e-07,
+      "loss": 0.7345,
+      "step": 12797
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.300919215898674e-07,
+      "loss": 0.7469,
+      "step": 12798
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.29169358552181e-07,
+      "loss": 0.7051,
+      "step": 12799
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.282472309959368e-07,
+      "loss": 0.7249,
+      "step": 12800
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.273255389654023e-07,
+      "loss": 0.7172,
+      "step": 12801
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.26404282504827e-07,
+      "loss": 0.7081,
+      "step": 12802
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.254834616584396e-07,
+      "loss": 0.352,
+      "step": 12803
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.245630764704472e-07,
+      "loss": 0.4546,
+      "step": 12804
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.236431269850387e-07,
+      "loss": 0.397,
+      "step": 12805
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.227236132463791e-07,
+      "loss": 0.7031,
+      "step": 12806
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.218045352986127e-07,
+      "loss": 0.7633,
+      "step": 12807
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.208858931858633e-07,
+      "loss": 0.7078,
+      "step": 12808
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.199676869522333e-07,
+      "loss": 0.7373,
+      "step": 12809
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.190499166418065e-07,
+      "loss": 0.695,
+      "step": 12810
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.181325822986454e-07,
+      "loss": 0.4054,
+      "step": 12811
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.172156839667889e-07,
+      "loss": 0.7498,
+      "step": 12812
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.162992216902522e-07,
+      "loss": 0.4053,
+      "step": 12813
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.153831955130399e-07,
+      "loss": 0.7226,
+      "step": 12814
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.144676054791279e-07,
+      "loss": 0.6792,
+      "step": 12815
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.135524516324701e-07,
+      "loss": 0.7041,
+      "step": 12816
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.126377340170056e-07,
+      "loss": 0.7353,
+      "step": 12817
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.11723452676645e-07,
+      "loss": 0.7316,
+      "step": 12818
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.108096076552853e-07,
+      "loss": 0.7569,
+      "step": 12819
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.098961989967991e-07,
+      "loss": 0.3922,
+      "step": 12820
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.089832267450382e-07,
+      "loss": 0.7742,
+      "step": 12821
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.080706909438308e-07,
+      "loss": 0.7414,
+      "step": 12822
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.071585916369896e-07,
+      "loss": 0.7289,
+      "step": 12823
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.06246928868304e-07,
+      "loss": 0.6333,
+      "step": 12824
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.053357026815401e-07,
+      "loss": 0.7132,
+      "step": 12825
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.044249131204463e-07,
+      "loss": 0.7132,
+      "step": 12826
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.035145602287476e-07,
+      "loss": 0.7043,
+      "step": 12827
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.026046440501501e-07,
+      "loss": 0.708,
+      "step": 12828
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.016951646283367e-07,
+      "loss": 0.7223,
+      "step": 12829
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.007861220069714e-07,
+      "loss": 0.7354,
+      "step": 12830
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.998775162296946e-07,
+      "loss": 0.7216,
+      "step": 12831
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.989693473401318e-07,
+      "loss": 0.4127,
+      "step": 12832
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.9806161538188e-07,
+      "loss": 0.711,
+      "step": 12833
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.971543203985201e-07,
+      "loss": 0.7038,
+      "step": 12834
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.962474624336093e-07,
+      "loss": 0.6992,
+      "step": 12835
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.953410415306829e-07,
+      "loss": 0.7433,
+      "step": 12836
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.944350577332617e-07,
+      "loss": 0.7184,
+      "step": 12837
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.935295110848397e-07,
+      "loss": 0.6847,
+      "step": 12838
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.926244016288898e-07,
+      "loss": 0.7732,
+      "step": 12839
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.917197294088653e-07,
+      "loss": 0.6912,
+      "step": 12840
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.90815494468199e-07,
+      "loss": 0.7457,
+      "step": 12841
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.899116968503052e-07,
+      "loss": 0.7429,
+      "step": 12842
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.890083365985713e-07,
+      "loss": 0.7011,
+      "step": 12843
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.881054137563661e-07,
+      "loss": 0.3663,
+      "step": 12844
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.87202928367038e-07,
+      "loss": 0.6585,
+      "step": 12845
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.863008804739181e-07,
+      "loss": 0.7007,
+      "step": 12846
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.853992701203096e-07,
+      "loss": 0.691,
+      "step": 12847
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.844980973494988e-07,
+      "loss": 0.7673,
+      "step": 12848
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.835973622047478e-07,
+      "loss": 0.6955,
+      "step": 12849
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.826970647293032e-07,
+      "loss": 0.4045,
+      "step": 12850
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.81797204966387e-07,
+      "loss": 0.7242,
+      "step": 12851
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.808977829592002e-07,
+      "loss": 0.7038,
+      "step": 12852
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.799987987509229e-07,
+      "loss": 0.4157,
+      "step": 12853
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.791002523847114e-07,
+      "loss": 0.6414,
+      "step": 12854
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.782021439037091e-07,
+      "loss": 0.7094,
+      "step": 12855
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.773044733510338e-07,
+      "loss": 0.7206,
+      "step": 12856
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.764072407697765e-07,
+      "loss": 0.7697,
+      "step": 12857
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.755104462030139e-07,
+      "loss": 0.419,
+      "step": 12858
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.746140896938015e-07,
+      "loss": 0.7323,
+      "step": 12859
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.737181712851739e-07,
+      "loss": 0.713,
+      "step": 12860
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.728226910201409e-07,
+      "loss": 0.3887,
+      "step": 12861
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.71927648941694e-07,
+      "loss": 0.4494,
+      "step": 12862
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.710330450928029e-07,
+      "loss": 0.3865,
+      "step": 12863
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.701388795164178e-07,
+      "loss": 0.6557,
+      "step": 12864
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.692451522554668e-07,
+      "loss": 0.4157,
+      "step": 12865
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.683518633528565e-07,
+      "loss": 0.7314,
+      "step": 12866
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.674590128514726e-07,
+      "loss": 0.7458,
+      "step": 12867
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.665666007941775e-07,
+      "loss": 0.6888,
+      "step": 12868
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.656746272238192e-07,
+      "loss": 0.4022,
+      "step": 12869
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.64783092183219e-07,
+      "loss": 0.4111,
+      "step": 12870
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.638919957151792e-07,
+      "loss": 0.7065,
+      "step": 12871
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.630013378624768e-07,
+      "loss": 0.6819,
+      "step": 12872
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.621111186678755e-07,
+      "loss": 0.6947,
+      "step": 12873
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.61221338174113e-07,
+      "loss": 0.7146,
+      "step": 12874
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.603319964239054e-07,
+      "loss": 0.7207,
+      "step": 12875
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.594430934599507e-07,
+      "loss": 0.6885,
+      "step": 12876
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.585546293249214e-07,
+      "loss": 0.7076,
+      "step": 12877
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.576666040614745e-07,
+      "loss": 0.6904,
+      "step": 12878
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.567790177122436e-07,
+      "loss": 0.7396,
+      "step": 12879
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.558918703198393e-07,
+      "loss": 0.731,
+      "step": 12880
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.550051619268518e-07,
+      "loss": 0.7352,
+      "step": 12881
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.541188925758537e-07,
+      "loss": 0.7001,
+      "step": 12882
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.53233062309392e-07,
+      "loss": 0.3851,
+      "step": 12883
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.523476711699952e-07,
+      "loss": 0.7023,
+      "step": 12884
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.514627192001701e-07,
+      "loss": 0.6745,
+      "step": 12885
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.505782064424029e-07,
+      "loss": 0.7254,
+      "step": 12886
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.496941329391561e-07,
+      "loss": 0.6945,
+      "step": 12887
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.488104987328749e-07,
+      "loss": 0.6891,
+      "step": 12888
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.479273038659819e-07,
+      "loss": 0.6813,
+      "step": 12889
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.470445483808754e-07,
+      "loss": 0.6976,
+      "step": 12890
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.461622323199403e-07,
+      "loss": 0.6846,
+      "step": 12891
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.452803557255329e-07,
+      "loss": 0.6926,
+      "step": 12892
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.443989186399914e-07,
+      "loss": 0.4087,
+      "step": 12893
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.435179211056332e-07,
+      "loss": 0.7555,
+      "step": 12894
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.426373631647522e-07,
+      "loss": 0.6912,
+      "step": 12895
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.417572448596267e-07,
+      "loss": 0.7503,
+      "step": 12896
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.408775662325086e-07,
+      "loss": 0.3935,
+      "step": 12897
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.399983273256296e-07,
+      "loss": 0.7109,
+      "step": 12898
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.391195281812015e-07,
+      "loss": 0.3977,
+      "step": 12899
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.382411688414139e-07,
+      "loss": 0.7514,
+      "step": 12900
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.373632493484396e-07,
+      "loss": 0.7027,
+      "step": 12901
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.364857697444218e-07,
+      "loss": 0.7045,
+      "step": 12902
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.356087300714899e-07,
+      "loss": 0.7184,
+      "step": 12903
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.347321303717471e-07,
+      "loss": 0.7093,
+      "step": 12904
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.338559706872817e-07,
+      "loss": 0.4285,
+      "step": 12905
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.329802510601559e-07,
+      "loss": 0.4144,
+      "step": 12906
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.321049715324114e-07,
+      "loss": 0.7343,
+      "step": 12907
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.312301321460692e-07,
+      "loss": 0.704,
+      "step": 12908
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.303557329431322e-07,
+      "loss": 0.3905,
+      "step": 12909
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.294817739655759e-07,
+      "loss": 0.6808,
+      "step": 12910
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.286082552553609e-07,
+      "loss": 0.7125,
+      "step": 12911
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.277351768544228e-07,
+      "loss": 0.7018,
+      "step": 12912
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.268625388046758e-07,
+      "loss": 0.733,
+      "step": 12913
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.259903411480164e-07,
+      "loss": 0.3738,
+      "step": 12914
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.251185839263187e-07,
+      "loss": 0.6842,
+      "step": 12915
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.242472671814338e-07,
+      "loss": 0.677,
+      "step": 12916
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.233763909551906e-07,
+      "loss": 0.7152,
+      "step": 12917
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.22505955289401e-07,
+      "loss": 0.6866,
+      "step": 12918
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.21635960225855e-07,
+      "loss": 0.729,
+      "step": 12919
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.20766405806318e-07,
+      "loss": 0.3857,
+      "step": 12920
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.198972920725367e-07,
+      "loss": 0.4376,
+      "step": 12921
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.190286190662367e-07,
+      "loss": 0.4017,
+      "step": 12922
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.181603868291232e-07,
+      "loss": 0.3596,
+      "step": 12923
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.172925954028777e-07,
+      "loss": 0.3498,
+      "step": 12924
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.164252448291632e-07,
+      "loss": 0.4326,
+      "step": 12925
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.155583351496199e-07,
+      "loss": 0.7037,
+      "step": 12926
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.146918664058645e-07,
+      "loss": 0.3793,
+      "step": 12927
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.138258386395004e-07,
+      "loss": 0.7472,
+      "step": 12928
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.129602518921009e-07,
+      "loss": 0.71,
+      "step": 12929
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.120951062052241e-07,
+      "loss": 0.7176,
+      "step": 12930
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.112304016204042e-07,
+      "loss": 0.7553,
+      "step": 12931
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.103661381791539e-07,
+      "loss": 0.6772,
+      "step": 12932
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.095023159229665e-07,
+      "loss": 0.4115,
+      "step": 12933
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.086389348933121e-07,
+      "loss": 0.723,
+      "step": 12934
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.077759951316421e-07,
+      "loss": 0.7237,
+      "step": 12935
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.069134966793834e-07,
+      "loss": 0.6857,
+      "step": 12936
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.060514395779473e-07,
+      "loss": 0.4236,
+      "step": 12937
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.051898238687173e-07,
+      "loss": 0.7086,
+      "step": 12938
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.043286495930591e-07,
+      "loss": 0.7045,
+      "step": 12939
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.034679167923165e-07,
+      "loss": 0.714,
+      "step": 12940
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.026076255078141e-07,
+      "loss": 0.732,
+      "step": 12941
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.017477757808534e-07,
+      "loss": 0.7189,
+      "step": 12942
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.008883676527135e-07,
+      "loss": 0.3946,
+      "step": 12943
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.000294011646547e-07,
+      "loss": 0.6576,
+      "step": 12944
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.991708763579132e-07,
+      "loss": 0.4533,
+      "step": 12945
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.983127932737112e-07,
+      "loss": 0.7093,
+      "step": 12946
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.974551519532381e-07,
+      "loss": 0.7229,
+      "step": 12947
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.965979524376721e-07,
+      "loss": 0.7037,
+      "step": 12948
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.957411947681626e-07,
+      "loss": 0.6942,
+      "step": 12949
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.948848789858465e-07,
+      "loss": 0.7643,
+      "step": 12950
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.940290051318334e-07,
+      "loss": 0.7095,
+      "step": 12951
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.931735732472113e-07,
+      "loss": 0.7388,
+      "step": 12952
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.923185833730496e-07,
+      "loss": 0.7264,
+      "step": 12953
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.914640355503944e-07,
+      "loss": 0.4154,
+      "step": 12954
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.90609929820274e-07,
+      "loss": 0.7247,
+      "step": 12955
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.897562662236912e-07,
+      "loss": 0.4123,
+      "step": 12956
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.88903044801631e-07,
+      "loss": 0.7272,
+      "step": 12957
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.880502655950551e-07,
+      "loss": 0.7015,
+      "step": 12958
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.871979286449016e-07,
+      "loss": 0.7418,
+      "step": 12959
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.863460339920958e-07,
+      "loss": 0.7256,
+      "step": 12960
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.854945816775339e-07,
+      "loss": 0.684,
+      "step": 12961
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.84643571742093e-07,
+      "loss": 0.707,
+      "step": 12962
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.837930042266262e-07,
+      "loss": 0.683,
+      "step": 12963
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.82942879171974e-07,
+      "loss": 0.3897,
+      "step": 12964
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.82093196618946e-07,
+      "loss": 0.3507,
+      "step": 12965
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.812439566083374e-07,
+      "loss": 0.7497,
+      "step": 12966
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.803951591809178e-07,
+      "loss": 0.7972,
+      "step": 12967
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.795468043774346e-07,
+      "loss": 0.7267,
+      "step": 12968
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.786988922386218e-07,
+      "loss": 0.6987,
+      "step": 12969
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.778514228051837e-07,
+      "loss": 0.3994,
+      "step": 12970
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.770043961178076e-07,
+      "loss": 0.7115,
+      "step": 12971
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.761578122171554e-07,
+      "loss": 0.7245,
+      "step": 12972
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.753116711438746e-07,
+      "loss": 0.6427,
+      "step": 12973
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.744659729385862e-07,
+      "loss": 0.6842,
+      "step": 12974
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.736207176418931e-07,
+      "loss": 0.7398,
+      "step": 12975
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.727759052943696e-07,
+      "loss": 0.4059,
+      "step": 12976
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.719315359365786e-07,
+      "loss": 0.7154,
+      "step": 12977
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.71087609609058e-07,
+      "loss": 0.6852,
+      "step": 12978
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.702441263523219e-07,
+      "loss": 0.7524,
+      "step": 12979
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.694010862068668e-07,
+      "loss": 0.7043,
+      "step": 12980
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.685584892131625e-07,
+      "loss": 0.72,
+      "step": 12981
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.677163354116657e-07,
+      "loss": 0.7168,
+      "step": 12982
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.66874624842806e-07,
+      "loss": 0.7039,
+      "step": 12983
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.660333575469914e-07,
+      "loss": 0.6863,
+      "step": 12984
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.651925335646127e-07,
+      "loss": 0.6098,
+      "step": 12985
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.643521529360343e-07,
+      "loss": 0.7575,
+      "step": 12986
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.63512215701604e-07,
+      "loss": 0.7353,
+      "step": 12987
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.626727219016472e-07,
+      "loss": 0.4281,
+      "step": 12988
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.618336715764651e-07,
+      "loss": 0.6923,
+      "step": 12989
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.6099506476634e-07,
+      "loss": 0.703,
+      "step": 12990
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.60156901511534e-07,
+      "loss": 0.4393,
+      "step": 12991
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.59319181852285e-07,
+      "loss": 0.746,
+      "step": 12992
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.584819058288117e-07,
+      "loss": 0.6839,
+      "step": 12993
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.576450734813101e-07,
+      "loss": 0.6983,
+      "step": 12994
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.568086848499556e-07,
+      "loss": 0.7122,
+      "step": 12995
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.55972739974904e-07,
+      "loss": 0.4074,
+      "step": 12996
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.551372388962874e-07,
+      "loss": 0.7139,
+      "step": 12997
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.543021816542173e-07,
+      "loss": 0.4001,
+      "step": 12998
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.534675682887815e-07,
+      "loss": 0.7523,
+      "step": 12999
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.526333988400536e-07,
+      "loss": 0.7381,
+      "step": 13000
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.517996733480782e-07,
+      "loss": 0.7557,
+      "step": 13001
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.509663918528831e-07,
+      "loss": 0.7013,
+      "step": 13002
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.501335543944721e-07,
+      "loss": 0.7857,
+      "step": 13003
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.493011610128287e-07,
+      "loss": 0.7356,
+      "step": 13004
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.484692117479187e-07,
+      "loss": 0.6899,
+      "step": 13005
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.476377066396778e-07,
+      "loss": 0.7105,
+      "step": 13006
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.468066457280288e-07,
+      "loss": 0.7154,
+      "step": 13007
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.459760290528684e-07,
+      "loss": 0.4096,
+      "step": 13008
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.45145856654076e-07,
+      "loss": 0.6926,
+      "step": 13009
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.443161285715061e-07,
+      "loss": 0.3436,
+      "step": 13010
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.434868448449939e-07,
+      "loss": 0.4185,
+      "step": 13011
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.426580055143518e-07,
+      "loss": 0.3684,
+      "step": 13012
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.41829610619369e-07,
+      "loss": 0.6915,
+      "step": 13013
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.410016601998204e-07,
+      "loss": 0.6818,
+      "step": 13014
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.401741542954532e-07,
+      "loss": 0.7479,
+      "step": 13015
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.393470929459945e-07,
+      "loss": 0.7222,
+      "step": 13016
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.385204761911524e-07,
+      "loss": 0.7295,
+      "step": 13017
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.376943040706075e-07,
+      "loss": 0.7349,
+      "step": 13018
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.368685766240291e-07,
+      "loss": 0.738,
+      "step": 13019
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.360432938910589e-07,
+      "loss": 0.7309,
+      "step": 13020
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.352184559113151e-07,
+      "loss": 0.6979,
+      "step": 13021
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.343940627243962e-07,
+      "loss": 0.4308,
+      "step": 13022
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.335701143698837e-07,
+      "loss": 0.7254,
+      "step": 13023
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.327466108873338e-07,
+      "loss": 0.707,
+      "step": 13024
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.319235523162827e-07,
+      "loss": 0.3758,
+      "step": 13025
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.31100938696242e-07,
+      "loss": 0.7348,
+      "step": 13026
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.302787700667057e-07,
+      "loss": 0.7445,
+      "step": 13027
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.29457046467148e-07,
+      "loss": 0.6899,
+      "step": 13028
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.286357679370159e-07,
+      "loss": 0.694,
+      "step": 13029
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.278149345157392e-07,
+      "loss": 0.7222,
+      "step": 13030
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.269945462427242e-07,
+      "loss": 0.6974,
+      "step": 13031
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.261746031573592e-07,
+      "loss": 0.7107,
+      "step": 13032
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.253551052990071e-07,
+      "loss": 0.7016,
+      "step": 13033
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.245360527070122e-07,
+      "loss": 0.7496,
+      "step": 13034
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.237174454206952e-07,
+      "loss": 0.7347,
+      "step": 13035
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.22899283479358e-07,
+      "loss": 0.3868,
+      "step": 13036
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.22081566922278e-07,
+      "loss": 0.4212,
+      "step": 13037
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.21264295788715e-07,
+      "loss": 0.6859,
+      "step": 13038
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.204474701179031e-07,
+      "loss": 0.7458,
+      "step": 13039
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.196310899490577e-07,
+      "loss": 0.6875,
+      "step": 13040
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.188151553213741e-07,
+      "loss": 0.7158,
+      "step": 13041
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.179996662740241e-07,
+      "loss": 0.6987,
+      "step": 13042
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.171846228461576e-07,
+      "loss": 0.6906,
+      "step": 13043
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.163700250769046e-07,
+      "loss": 0.7069,
+      "step": 13044
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.155558730053713e-07,
+      "loss": 0.7483,
+      "step": 13045
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.147421666706467e-07,
+      "loss": 0.4564,
+      "step": 13046
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.13928906111796e-07,
+      "loss": 0.7468,
+      "step": 13047
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.131160913678625e-07,
+      "loss": 0.7028,
+      "step": 13048
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.123037224778673e-07,
+      "loss": 0.4156,
+      "step": 13049
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.114917994808135e-07,
+      "loss": 0.7052,
+      "step": 13050
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.106803224156789e-07,
+      "loss": 0.7286,
+      "step": 13051
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.098692913214233e-07,
+      "loss": 0.759,
+      "step": 13052
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.090587062369813e-07,
+      "loss": 0.7274,
+      "step": 13053
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.082485672012696e-07,
+      "loss": 0.66,
+      "step": 13054
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.074388742531824e-07,
+      "loss": 0.718,
+      "step": 13055
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.066296274315931e-07,
+      "loss": 0.7136,
+      "step": 13056
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.058208267753508e-07,
+      "loss": 0.3938,
+      "step": 13057
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.050124723232854e-07,
+      "loss": 0.7164,
+      "step": 13058
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.04204564114207e-07,
+      "loss": 0.7512,
+      "step": 13059
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.033971021869013e-07,
+      "loss": 0.7219,
+      "step": 13060
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.025900865801338e-07,
+      "loss": 0.8105,
+      "step": 13061
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.017835173326482e-07,
+      "loss": 0.3752,
+      "step": 13062
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.009773944831666e-07,
+      "loss": 0.7566,
+      "step": 13063
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.001717180703927e-07,
+      "loss": 0.7099,
+      "step": 13064
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.993664881330054e-07,
+      "loss": 0.6911,
+      "step": 13065
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.985617047096605e-07,
+      "loss": 0.7653,
+      "step": 13066
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.977573678389949e-07,
+      "loss": 0.7003,
+      "step": 13067
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.969534775596276e-07,
+      "loss": 0.7446,
+      "step": 13068
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.961500339101501e-07,
+      "loss": 0.7569,
+      "step": 13069
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.953470369291349e-07,
+      "loss": 0.7294,
+      "step": 13070
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.945444866551332e-07,
+      "loss": 0.7079,
+      "step": 13071
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.937423831266743e-07,
+      "loss": 0.7247,
+      "step": 13072
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.929407263822673e-07,
+      "loss": 0.7209,
+      "step": 13073
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.921395164603995e-07,
+      "loss": 0.7042,
+      "step": 13074
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.913387533995353e-07,
+      "loss": 0.7248,
+      "step": 13075
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.905384372381185e-07,
+      "loss": 0.6764,
+      "step": 13076
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.897385680145696e-07,
+      "loss": 0.6848,
+      "step": 13077
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.889391457672923e-07,
+      "loss": 0.7166,
+      "step": 13078
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.88140170534668e-07,
+      "loss": 0.7471,
+      "step": 13079
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.873416423550494e-07,
+      "loss": 0.7182,
+      "step": 13080
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.865435612667748e-07,
+      "loss": 0.7434,
+      "step": 13081
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.857459273081601e-07,
+      "loss": 0.7148,
+      "step": 13082
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.849487405175004e-07,
+      "loss": 0.7391,
+      "step": 13083
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.841520009330649e-07,
+      "loss": 0.7168,
+      "step": 13084
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.833557085931064e-07,
+      "loss": 0.4159,
+      "step": 13085
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.825598635358511e-07,
+      "loss": 0.6899,
+      "step": 13086
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.817644657995104e-07,
+      "loss": 0.744,
+      "step": 13087
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.809695154222685e-07,
+      "loss": 0.704,
+      "step": 13088
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.801750124422912e-07,
+      "loss": 0.6916,
+      "step": 13089
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.793809568977205e-07,
+      "loss": 0.695,
+      "step": 13090
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.785873488266792e-07,
+      "loss": 0.3776,
+      "step": 13091
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.777941882672679e-07,
+      "loss": 0.7069,
+      "step": 13092
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.770014752575649e-07,
+      "loss": 0.4156,
+      "step": 13093
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.762092098356277e-07,
+      "loss": 0.6969,
+      "step": 13094
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.754173920394924e-07,
+      "loss": 0.4091,
+      "step": 13095
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.746260219071732e-07,
+      "loss": 0.3977,
+      "step": 13096
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.738350994766618e-07,
+      "loss": 0.4152,
+      "step": 13097
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.730446247859312e-07,
+      "loss": 0.6888,
+      "step": 13098
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.722545978729289e-07,
+      "loss": 0.6822,
+      "step": 13099
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.714650187755867e-07,
+      "loss": 0.7368,
+      "step": 13100
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.706758875318098e-07,
+      "loss": 0.7386,
+      "step": 13101
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.698872041794835e-07,
+      "loss": 0.6683,
+      "step": 13102
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.690989687564719e-07,
+      "loss": 0.7142,
+      "step": 13103
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.68311181300616e-07,
+      "loss": 0.764,
+      "step": 13104
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.675238418497398e-07,
+      "loss": 0.4173,
+      "step": 13105
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.667369504416399e-07,
+      "loss": 0.7411,
+      "step": 13106
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.65950507114096e-07,
+      "loss": 0.4176,
+      "step": 13107
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.651645119048633e-07,
+      "loss": 0.7166,
+      "step": 13108
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.643789648516774e-07,
+      "loss": 0.7388,
+      "step": 13109
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.635938659922514e-07,
+      "loss": 0.385,
+      "step": 13110
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.628092153642763e-07,
+      "loss": 0.6704,
+      "step": 13111
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.620250130054229e-07,
+      "loss": 0.7311,
+      "step": 13112
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.61241258953339e-07,
+      "loss": 0.6933,
+      "step": 13113
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.604579532456545e-07,
+      "loss": 0.6973,
+      "step": 13114
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.596750959199727e-07,
+      "loss": 0.7336,
+      "step": 13115
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.58892687013879e-07,
+      "loss": 0.729,
+      "step": 13116
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.581107265649356e-07,
+      "loss": 0.3968,
+      "step": 13117
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.573292146106824e-07,
+      "loss": 0.3841,
+      "step": 13118
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.565481511886418e-07,
+      "loss": 0.4085,
+      "step": 13119
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.557675363363103e-07,
+      "loss": 0.6703,
+      "step": 13120
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.549873700911635e-07,
+      "loss": 0.7383,
+      "step": 13121
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.542076524906571e-07,
+      "loss": 0.672,
+      "step": 13122
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.534283835722255e-07,
+      "loss": 0.7111,
+      "step": 13123
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.526495633732821e-07,
+      "loss": 0.7486,
+      "step": 13124
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.518711919312137e-07,
+      "loss": 0.7442,
+      "step": 13125
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.510932692833893e-07,
+      "loss": 0.7263,
+      "step": 13126
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.50315795467158e-07,
+      "loss": 0.738,
+      "step": 13127
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.495387705198453e-07,
+      "loss": 0.7143,
+      "step": 13128
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.487621944787558e-07,
+      "loss": 0.4064,
+      "step": 13129
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.4798606738117e-07,
+      "loss": 0.7308,
+      "step": 13130
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.4721038926435e-07,
+      "loss": 0.6683,
+      "step": 13131
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.464351601655372e-07,
+      "loss": 0.7314,
+      "step": 13132
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.456603801219474e-07,
+      "loss": 0.4365,
+      "step": 13133
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.448860491707776e-07,
+      "loss": 0.7342,
+      "step": 13134
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.441121673492034e-07,
+      "loss": 0.6869,
+      "step": 13135
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.433387346943753e-07,
+      "loss": 0.73,
+      "step": 13136
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.425657512434292e-07,
+      "loss": 0.7024,
+      "step": 13137
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.417932170334728e-07,
+      "loss": 0.7258,
+      "step": 13138
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.410211321015969e-07,
+      "loss": 0.7241,
+      "step": 13139
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.402494964848638e-07,
+      "loss": 0.6917,
+      "step": 13140
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.394783102203228e-07,
+      "loss": 0.7012,
+      "step": 13141
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.387075733449976e-07,
+      "loss": 0.7038,
+      "step": 13142
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.379372858958888e-07,
+      "loss": 0.7485,
+      "step": 13143
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.371674479099787e-07,
+      "loss": 0.706,
+      "step": 13144
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.363980594242236e-07,
+      "loss": 0.7265,
+      "step": 13145
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.356291204755649e-07,
+      "loss": 0.6823,
+      "step": 13146
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.348606311009176e-07,
+      "loss": 0.709,
+      "step": 13147
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.340925913371753e-07,
+      "loss": 0.3993,
+      "step": 13148
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.333250012212089e-07,
+      "loss": 0.6892,
+      "step": 13149
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.325578607898741e-07,
+      "loss": 0.4108,
+      "step": 13150
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.317911700799973e-07,
+      "loss": 0.6728,
+      "step": 13151
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.310249291283876e-07,
+      "loss": 0.6871,
+      "step": 13152
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.302591379718314e-07,
+      "loss": 0.6911,
+      "step": 13153
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.294937966470926e-07,
+      "loss": 0.6763,
+      "step": 13154
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.287289051909162e-07,
+      "loss": 0.446,
+      "step": 13155
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.279644636400229e-07,
+      "loss": 0.7319,
+      "step": 13156
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.27200472031112e-07,
+      "loss": 0.721,
+      "step": 13157
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.264369304008621e-07,
+      "loss": 0.4354,
+      "step": 13158
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.256738387859318e-07,
+      "loss": 0.7152,
+      "step": 13159
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.249111972229549e-07,
+      "loss": 0.3777,
+      "step": 13160
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.241490057485466e-07,
+      "loss": 0.4054,
+      "step": 13161
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.233872643992966e-07,
+      "loss": 0.6812,
+      "step": 13162
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.226259732117757e-07,
+      "loss": 0.7107,
+      "step": 13163
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.218651322225344e-07,
+      "loss": 0.7483,
+      "step": 13164
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.211047414681004e-07,
+      "loss": 0.4152,
+      "step": 13165
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.203448009849766e-07,
+      "loss": 0.8079,
+      "step": 13166
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.195853108096483e-07,
+      "loss": 0.7152,
+      "step": 13167
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.188262709785775e-07,
+      "loss": 0.7107,
+      "step": 13168
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.180676815282071e-07,
+      "loss": 0.7358,
+      "step": 13169
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.173095424949537e-07,
+      "loss": 0.7418,
+      "step": 13170
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.165518539152149e-07,
+      "loss": 0.438,
+      "step": 13171
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.15794615825367e-07,
+      "loss": 0.404,
+      "step": 13172
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.150378282617641e-07,
+      "loss": 0.7421,
+      "step": 13173
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.142814912607409e-07,
+      "loss": 0.4461,
+      "step": 13174
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.135256048586058e-07,
+      "loss": 0.3772,
+      "step": 13175
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.127701690916487e-07,
+      "loss": 0.7221,
+      "step": 13176
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.120151839961363e-07,
+      "loss": 0.6985,
+      "step": 13177
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.112606496083173e-07,
+      "loss": 0.6678,
+      "step": 13178
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.105065659644149e-07,
+      "loss": 0.405,
+      "step": 13179
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.097529331006313e-07,
+      "loss": 0.6995,
+      "step": 13180
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.089997510531476e-07,
+      "loss": 0.6829,
+      "step": 13181
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.082470198581247e-07,
+      "loss": 0.3968,
+      "step": 13182
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.074947395516994e-07,
+      "loss": 0.719,
+      "step": 13183
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.067429101699907e-07,
+      "loss": 0.4146,
+      "step": 13184
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.059915317490861e-07,
+      "loss": 0.7225,
+      "step": 13185
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.052406043250658e-07,
+      "loss": 0.7037,
+      "step": 13186
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.044901279339777e-07,
+      "loss": 0.7168,
+      "step": 13187
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.037401026118516e-07,
+      "loss": 0.7103,
+      "step": 13188
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.029905283946969e-07,
+      "loss": 0.7357,
+      "step": 13189
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.022414053184967e-07,
+      "loss": 0.3956,
+      "step": 13190
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.014927334192189e-07,
+      "loss": 0.6673,
+      "step": 13191
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.00744512732806e-07,
+      "loss": 0.4044,
+      "step": 13192
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.999967432951792e-07,
+      "loss": 0.743,
+      "step": 13193
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.992494251422365e-07,
+      "loss": 0.3864,
+      "step": 13194
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.985025583098558e-07,
+      "loss": 0.4034,
+      "step": 13195
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.977561428338973e-07,
+      "loss": 0.751,
+      "step": 13196
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.970101787501914e-07,
+      "loss": 0.3945,
+      "step": 13197
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.962646660945559e-07,
+      "loss": 0.7006,
+      "step": 13198
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.955196049027745e-07,
+      "loss": 0.4127,
+      "step": 13199
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.947749952106229e-07,
+      "loss": 0.6981,
+      "step": 13200
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.940308370538483e-07,
+      "loss": 0.6912,
+      "step": 13201
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.932871304681753e-07,
+      "loss": 0.6962,
+      "step": 13202
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.925438754893098e-07,
+      "loss": 0.7039,
+      "step": 13203
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.918010721529332e-07,
+      "loss": 0.4111,
+      "step": 13204
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.910587204947082e-07,
+      "loss": 0.7377,
+      "step": 13205
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.903168205502752e-07,
+      "loss": 0.7322,
+      "step": 13206
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.8957537235525e-07,
+      "loss": 0.7355,
+      "step": 13207
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.888343759452275e-07,
+      "loss": 0.6562,
+      "step": 13208
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.880938313557871e-07,
+      "loss": 0.3596,
+      "step": 13209
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.873537386224781e-07,
+      "loss": 0.6956,
+      "step": 13210
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.866140977808322e-07,
+      "loss": 0.7466,
+      "step": 13211
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.858749088663584e-07,
+      "loss": 0.7176,
+      "step": 13212
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.851361719145465e-07,
+      "loss": 0.3764,
+      "step": 13213
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.843978869608591e-07,
+      "loss": 0.7315,
+      "step": 13214
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.836600540407433e-07,
+      "loss": 0.7132,
+      "step": 13215
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.82922673189621e-07,
+      "loss": 0.6887,
+      "step": 13216
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.821857444428914e-07,
+      "loss": 0.6993,
+      "step": 13217
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.814492678359362e-07,
+      "loss": 0.7008,
+      "step": 13218
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.807132434041119e-07,
+      "loss": 0.6843,
+      "step": 13219
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.799776711827543e-07,
+      "loss": 0.6667,
+      "step": 13220
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.792425512071776e-07,
+      "loss": 0.6792,
+      "step": 13221
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.785078835126723e-07,
+      "loss": 0.7232,
+      "step": 13222
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.777736681345125e-07,
+      "loss": 0.7336,
+      "step": 13223
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.770399051079445e-07,
+      "loss": 0.72,
+      "step": 13224
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.763065944681978e-07,
+      "loss": 0.4118,
+      "step": 13225
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.755737362504754e-07,
+      "loss": 0.7364,
+      "step": 13226
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.748413304899614e-07,
+      "loss": 0.6613,
+      "step": 13227
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.74109377221822e-07,
+      "loss": 0.7181,
+      "step": 13228
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.733778764811926e-07,
+      "loss": 0.706,
+      "step": 13229
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.726468283031928e-07,
+      "loss": 0.731,
+      "step": 13230
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.719162327229189e-07,
+      "loss": 0.7829,
+      "step": 13231
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.711860897754485e-07,
+      "loss": 0.3917,
+      "step": 13232
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.704563994958346e-07,
+      "loss": 0.7107,
+      "step": 13233
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.69727161919108e-07,
+      "loss": 0.7717,
+      "step": 13234
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.689983770802787e-07,
+      "loss": 0.6364,
+      "step": 13235
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.682700450143331e-07,
+      "loss": 0.749,
+      "step": 13236
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.675421657562408e-07,
+      "loss": 0.697,
+      "step": 13237
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.668147393409462e-07,
+      "loss": 0.7319,
+      "step": 13238
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.660877658033704e-07,
+      "loss": 0.7616,
+      "step": 13239
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.653612451784151e-07,
+      "loss": 0.6757,
+      "step": 13240
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.646351775009617e-07,
+      "loss": 0.7299,
+      "step": 13241
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.639095628058666e-07,
+      "loss": 0.403,
+      "step": 13242
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.631844011279674e-07,
+      "loss": 0.3966,
+      "step": 13243
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.624596925020742e-07,
+      "loss": 0.7423,
+      "step": 13244
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.617354369629835e-07,
+      "loss": 0.6771,
+      "step": 13245
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.610116345454653e-07,
+      "loss": 0.6831,
+      "step": 13246
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.602882852842684e-07,
+      "loss": 0.683,
+      "step": 13247
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.595653892141184e-07,
+      "loss": 0.4119,
+      "step": 13248
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.58842946369722e-07,
+      "loss": 0.6969,
+      "step": 13249
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.581209567857638e-07,
+      "loss": 0.702,
+      "step": 13250
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.573994204969058e-07,
+      "loss": 0.7481,
+      "step": 13251
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.566783375377882e-07,
+      "loss": 0.715,
+      "step": 13252
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.559577079430279e-07,
+      "loss": 0.7566,
+      "step": 13253
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.552375317472214e-07,
+      "loss": 0.6983,
+      "step": 13254
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.545178089849456e-07,
+      "loss": 0.4468,
+      "step": 13255
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.537985396907542e-07,
+      "loss": 0.714,
+      "step": 13256
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.530797238991758e-07,
+      "loss": 0.3732,
+      "step": 13257
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.523613616447232e-07,
+      "loss": 0.7187,
+      "step": 13258
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.516434529618808e-07,
+      "loss": 0.6884,
+      "step": 13259
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.509259978851178e-07,
+      "loss": 0.7083,
+      "step": 13260
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.502089964488766e-07,
+      "loss": 0.4239,
+      "step": 13261
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.494924486875808e-07,
+      "loss": 0.7389,
+      "step": 13262
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.487763546356284e-07,
+      "loss": 0.737,
+      "step": 13263
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.480607143274019e-07,
+      "loss": 0.6879,
+      "step": 13264
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.473455277972573e-07,
+      "loss": 0.6753,
+      "step": 13265
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.466307950795291e-07,
+      "loss": 0.7244,
+      "step": 13266
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.459165162085322e-07,
+      "loss": 0.7582,
+      "step": 13267
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.452026912185571e-07,
+      "loss": 0.4085,
+      "step": 13268
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.444893201438751e-07,
+      "loss": 0.7099,
+      "step": 13269
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.437764030187343e-07,
+      "loss": 0.6808,
+      "step": 13270
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.430639398773596e-07,
+      "loss": 0.6465,
+      "step": 13271
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.42351930753956e-07,
+      "loss": 0.7376,
+      "step": 13272
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.416403756827104e-07,
+      "loss": 0.3875,
+      "step": 13273
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.409292746977779e-07,
+      "loss": 0.7101,
+      "step": 13274
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.40218627833301e-07,
+      "loss": 0.7555,
+      "step": 13275
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.395084351233948e-07,
+      "loss": 0.7011,
+      "step": 13276
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.387986966021585e-07,
+      "loss": 0.6965,
+      "step": 13277
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.38089412303664e-07,
+      "loss": 0.7092,
+      "step": 13278
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.373805822619626e-07,
+      "loss": 0.7204,
+      "step": 13279
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.366722065110863e-07,
+      "loss": 0.382,
+      "step": 13280
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.3596428508504e-07,
+      "loss": 0.7251,
+      "step": 13281
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.352568180178153e-07,
+      "loss": 0.7552,
+      "step": 13282
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.34549805343374e-07,
+      "loss": 0.3926,
+      "step": 13283
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.33843247095659e-07,
+      "loss": 0.6981,
+      "step": 13284
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.331371433085919e-07,
+      "loss": 0.7419,
+      "step": 13285
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.324314940160713e-07,
+      "loss": 0.3921,
+      "step": 13286
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.317262992519768e-07,
+      "loss": 0.684,
+      "step": 13287
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.310215590501644e-07,
+      "loss": 0.7557,
+      "step": 13288
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.303172734444639e-07,
+      "loss": 0.7039,
+      "step": 13289
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.296134424686894e-07,
+      "loss": 0.6982,
+      "step": 13290
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.289100661566326e-07,
+      "loss": 0.7403,
+      "step": 13291
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.282071445420611e-07,
+      "loss": 0.4072,
+      "step": 13292
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.275046776587212e-07,
+      "loss": 0.4305,
+      "step": 13293
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.26802665540338e-07,
+      "loss": 0.6845,
+      "step": 13294
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.261011082206113e-07,
+      "loss": 0.7408,
+      "step": 13295
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.254000057332276e-07,
+      "loss": 0.7131,
+      "step": 13296
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.24699358111842e-07,
+      "loss": 0.6963,
+      "step": 13297
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.239991653900945e-07,
+      "loss": 0.7297,
+      "step": 13298
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.232994276015968e-07,
+      "loss": 0.6607,
+      "step": 13299
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.226001447799478e-07,
+      "loss": 0.7135,
+      "step": 13300
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.219013169587162e-07,
+      "loss": 0.7362,
+      "step": 13301
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.212029441714528e-07,
+      "loss": 0.7257,
+      "step": 13302
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.205050264516854e-07,
+      "loss": 0.6917,
+      "step": 13303
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.198075638329203e-07,
+      "loss": 0.6767,
+      "step": 13304
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.191105563486432e-07,
+      "loss": 0.3997,
+      "step": 13305
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.184140040323149e-07,
+      "loss": 0.7346,
+      "step": 13306
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.177179069173766e-07,
+      "loss": 0.7284,
+      "step": 13307
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.17022265037247e-07,
+      "loss": 0.6904,
+      "step": 13308
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.163270784253249e-07,
+      "loss": 0.4534,
+      "step": 13309
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.156323471149838e-07,
+      "loss": 0.7453,
+      "step": 13310
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.14938071139578e-07,
+      "loss": 0.6804,
+      "step": 13311
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.142442505324396e-07,
+      "loss": 0.6825,
+      "step": 13312
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.135508853268744e-07,
+      "loss": 0.7077,
+      "step": 13313
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.128579755561747e-07,
+      "loss": 0.7034,
+      "step": 13314
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.121655212536058e-07,
+      "loss": 0.7029,
+      "step": 13315
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.114735224524092e-07,
+      "loss": 0.3836,
+      "step": 13316
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.107819791858093e-07,
+      "loss": 0.6873,
+      "step": 13317
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.100908914870051e-07,
+      "loss": 0.7202,
+      "step": 13318
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.094002593891767e-07,
+      "loss": 0.7138,
+      "step": 13319
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.087100829254787e-07,
+      "loss": 0.6655,
+      "step": 13320
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.080203621290458e-07,
+      "loss": 0.7233,
+      "step": 13321
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.073310970329914e-07,
+      "loss": 0.39,
+      "step": 13322
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.066422876704069e-07,
+      "loss": 0.727,
+      "step": 13323
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.059539340743625e-07,
+      "loss": 0.7357,
+      "step": 13324
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.052660362779027e-07,
+      "loss": 0.7452,
+      "step": 13325
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.045785943140535e-07,
+      "loss": 0.7001,
+      "step": 13326
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.038916082158185e-07,
+      "loss": 0.7205,
+      "step": 13327
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.032050780161801e-07,
+      "loss": 0.7029,
+      "step": 13328
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.025190037480976e-07,
+      "loss": 0.6888,
+      "step": 13329
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.01833385444509e-07,
+      "loss": 0.7781,
+      "step": 13330
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.01148223138328e-07,
+      "loss": 0.4521,
+      "step": 13331
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.004635168624516e-07,
+      "loss": 0.6928,
+      "step": 13332
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.997792666497503e-07,
+      "loss": 0.4067,
+      "step": 13333
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.990954725330744e-07,
+      "loss": 0.3879,
+      "step": 13334
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.984121345452498e-07,
+      "loss": 0.7893,
+      "step": 13335
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.977292527190869e-07,
+      "loss": 0.4656,
+      "step": 13336
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.970468270873696e-07,
+      "loss": 0.7081,
+      "step": 13337
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.963648576828584e-07,
+      "loss": 0.6947,
+      "step": 13338
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.956833445382947e-07,
+      "loss": 0.714,
+      "step": 13339
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.950022876863958e-07,
+      "loss": 0.7134,
+      "step": 13340
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.943216871598622e-07,
+      "loss": 0.7091,
+      "step": 13341
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.936415429913666e-07,
+      "loss": 0.7039,
+      "step": 13342
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.929618552135618e-07,
+      "loss": 0.3727,
+      "step": 13343
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.922826238590805e-07,
+      "loss": 0.7633,
+      "step": 13344
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.916038489605279e-07,
+      "loss": 0.7045,
+      "step": 13345
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.909255305504968e-07,
+      "loss": 0.7367,
+      "step": 13346
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.902476686615509e-07,
+      "loss": 0.6841,
+      "step": 13347
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.895702633262323e-07,
+      "loss": 0.4078,
+      "step": 13348
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.888933145770602e-07,
+      "loss": 0.7662,
+      "step": 13349
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.8821682244654e-07,
+      "loss": 0.7367,
+      "step": 13350
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.875407869671455e-07,
+      "loss": 0.7359,
+      "step": 13351
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.868652081713332e-07,
+      "loss": 0.7507,
+      "step": 13352
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.86190086091537e-07,
+      "loss": 0.3788,
+      "step": 13353
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.855154207601686e-07,
+      "loss": 0.704,
+      "step": 13354
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.848412122096192e-07,
+      "loss": 0.6955,
+      "step": 13355
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.841674604722557e-07,
+      "loss": 0.7093,
+      "step": 13356
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.83494165580426e-07,
+      "loss": 0.3627,
+      "step": 13357
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.828213275664507e-07,
+      "loss": 0.7236,
+      "step": 13358
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.82148946462635e-07,
+      "loss": 0.6826,
+      "step": 13359
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.814770223012599e-07,
+      "loss": 0.6858,
+      "step": 13360
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.808055551145818e-07,
+      "loss": 0.7121,
+      "step": 13361
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.801345449348371e-07,
+      "loss": 0.7365,
+      "step": 13362
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.794639917942423e-07,
+      "loss": 0.3936,
+      "step": 13363
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.787938957249883e-07,
+      "loss": 0.4073,
+      "step": 13364
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.781242567592458e-07,
+      "loss": 0.3866,
+      "step": 13365
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.77455074929164e-07,
+      "loss": 0.7259,
+      "step": 13366
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.767863502668679e-07,
+      "loss": 0.7338,
+      "step": 13367
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7611808280446534e-07,
+      "loss": 0.4091,
+      "step": 13368
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7545027257403843e-07,
+      "loss": 0.7333,
+      "step": 13369
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.74782919607647e-07,
+      "loss": 0.7106,
+      "step": 13370
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7411602393732993e-07,
+      "loss": 0.4123,
+      "step": 13371
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.734495855951038e-07,
+      "loss": 0.3988,
+      "step": 13372
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7278360461296527e-07,
+      "loss": 0.7024,
+      "step": 13373
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7211808102288757e-07,
+      "loss": 0.6877,
+      "step": 13374
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.714530148568197e-07,
+      "loss": 0.6846,
+      "step": 13375
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7078840614669275e-07,
+      "loss": 0.6975,
+      "step": 13376
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.701242549244134e-07,
+      "loss": 0.7696,
+      "step": 13377
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6946056122186614e-07,
+      "loss": 0.6533,
+      "step": 13378
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6879732507091434e-07,
+      "loss": 0.7076,
+      "step": 13379
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.681345465034004e-07,
+      "loss": 0.3834,
+      "step": 13380
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.67472225551141e-07,
+      "loss": 0.3881,
+      "step": 13381
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6681036224593747e-07,
+      "loss": 0.7352,
+      "step": 13382
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6614895661956204e-07,
+      "loss": 0.397,
+      "step": 13383
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6548800870377055e-07,
+      "loss": 0.7458,
+      "step": 13384
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.64827518530292e-07,
+      "loss": 0.4205,
+      "step": 13385
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.641674861308365e-07,
+      "loss": 0.7351,
+      "step": 13386
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.635079115370933e-07,
+      "loss": 0.6789,
+      "step": 13387
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6284879478072584e-07,
+      "loss": 0.3952,
+      "step": 13388
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6219013589338005e-07,
+      "loss": 0.6986,
+      "step": 13389
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.615319349066738e-07,
+      "loss": 0.7202,
+      "step": 13390
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6087419185220973e-07,
+      "loss": 0.7481,
+      "step": 13391
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.602169067615658e-07,
+      "loss": 0.7467,
+      "step": 13392
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.595600796662947e-07,
+      "loss": 0.7222,
+      "step": 13393
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5890371059793106e-07,
+      "loss": 0.7435,
+      "step": 13394
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.582477995879875e-07,
+      "loss": 0.3896,
+      "step": 13395
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.575923466679544e-07,
+      "loss": 0.6951,
+      "step": 13396
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5693735186929656e-07,
+      "loss": 0.4156,
+      "step": 13397
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5628281522346217e-07,
+      "loss": 0.3936,
+      "step": 13398
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.556287367618728e-07,
+      "loss": 0.683,
+      "step": 13399
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.549751165159322e-07,
+      "loss": 0.7248,
+      "step": 13400
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5432195451701965e-07,
+      "loss": 0.645,
+      "step": 13401
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.536692507964913e-07,
+      "loss": 0.7312,
+      "step": 13402
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.530170053856842e-07,
+      "loss": 0.674,
+      "step": 13403
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5236521831591e-07,
+      "loss": 0.7564,
+      "step": 13404
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5171388961846363e-07,
+      "loss": 0.721,
+      "step": 13405
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5106301932461236e-07,
+      "loss": 0.7209,
+      "step": 13406
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5041260746560565e-07,
+      "loss": 0.7238,
+      "step": 13407
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.497626540726652e-07,
+      "loss": 0.383,
+      "step": 13408
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.491131591769993e-07,
+      "loss": 0.6764,
+      "step": 13409
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4846412280978637e-07,
+      "loss": 0.7263,
+      "step": 13410
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4781554500218813e-07,
+      "loss": 0.7392,
+      "step": 13411
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4716742578534087e-07,
+      "loss": 0.7363,
+      "step": 13412
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4651976519035853e-07,
+      "loss": 0.7335,
+      "step": 13413
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.458725632483385e-07,
+      "loss": 0.7623,
+      "step": 13414
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.452258199903503e-07,
+      "loss": 0.3727,
+      "step": 13415
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.445795354474436e-07,
+      "loss": 0.4289,
+      "step": 13416
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4393370965064575e-07,
+      "loss": 0.3972,
+      "step": 13417
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.432883426309598e-07,
+      "loss": 0.6724,
+      "step": 13418
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.426434344193742e-07,
+      "loss": 0.6634,
+      "step": 13419
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.419989850468476e-07,
+      "loss": 0.692,
+      "step": 13420
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4135499454431854e-07,
+      "loss": 0.7274,
+      "step": 13421
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.407114629427056e-07,
+      "loss": 0.4105,
+      "step": 13422
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4006839027290415e-07,
+      "loss": 0.7323,
+      "step": 13423
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3942577656578613e-07,
+      "loss": 0.7263,
+      "step": 13424
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.387836218522046e-07,
+      "loss": 0.7577,
+      "step": 13425
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.38141926162986e-07,
+      "loss": 0.411,
+      "step": 13426
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.375006895289413e-07,
+      "loss": 0.6829,
+      "step": 13427
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.368599119808536e-07,
+      "loss": 0.6969,
+      "step": 13428
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3621959354948507e-07,
+      "loss": 0.7264,
+      "step": 13429
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3557973426557874e-07,
+      "loss": 0.7335,
+      "step": 13430
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3494033415985126e-07,
+      "loss": 0.7558,
+      "step": 13431
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3430139326300245e-07,
+      "loss": 0.6568,
+      "step": 13432
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3366291160570564e-07,
+      "loss": 0.4205,
+      "step": 13433
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3302488921861285e-07,
+      "loss": 0.732,
+      "step": 13434
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.323873261323563e-07,
+      "loss": 0.7123,
+      "step": 13435
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.317502223775449e-07,
+      "loss": 0.3621,
+      "step": 13436
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3111357798476416e-07,
+      "loss": 0.7045,
+      "step": 13437
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.304773929845785e-07,
+      "loss": 0.6592,
+      "step": 13438
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2984166740753233e-07,
+      "loss": 0.7399,
+      "step": 13439
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.292064012841435e-07,
+      "loss": 0.6704,
+      "step": 13440
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2857159464491425e-07,
+      "loss": 0.4283,
+      "step": 13441
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2793724752031807e-07,
+      "loss": 0.6746,
+      "step": 13442
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2730335994081054e-07,
+      "loss": 0.4049,
+      "step": 13443
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2666993193682395e-07,
+      "loss": 0.756,
+      "step": 13444
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2603696353876623e-07,
+      "loss": 0.6934,
+      "step": 13445
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2540445477702863e-07,
+      "loss": 0.7417,
+      "step": 13446
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.247724056819769e-07,
+      "loss": 0.66,
+      "step": 13447
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2414081628395444e-07,
+      "loss": 0.7018,
+      "step": 13448
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2350968661328154e-07,
+      "loss": 0.416,
+      "step": 13449
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.228790167002605e-07,
+      "loss": 0.6921,
+      "step": 13450
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2224880657516954e-07,
+      "loss": 0.7068,
+      "step": 13451
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2161905626826317e-07,
+      "loss": 0.6915,
+      "step": 13452
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.209897658097728e-07,
+      "loss": 0.6837,
+      "step": 13453
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2036093522991316e-07,
+      "loss": 0.3523,
+      "step": 13454
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1973256455887456e-07,
+      "loss": 0.7312,
+      "step": 13455
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1910465382682177e-07,
+      "loss": 0.7043,
+      "step": 13456
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.184772030639006e-07,
+      "loss": 0.3839,
+      "step": 13457
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.178502123002348e-07,
+      "loss": 0.7156,
+      "step": 13458
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1722368156592583e-07,
+      "loss": 0.6558,
+      "step": 13459
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.165976108910541e-07,
+      "loss": 0.7206,
+      "step": 13460
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1597200030567443e-07,
+      "loss": 0.719,
+      "step": 13461
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1534684983982277e-07,
+      "loss": 0.67,
+      "step": 13462
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.147221595235096e-07,
+      "loss": 0.7324,
+      "step": 13463
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1409792938672977e-07,
+      "loss": 0.7182,
+      "step": 13464
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1347415945944937e-07,
+      "loss": 0.6896,
+      "step": 13465
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.128508497716177e-07,
+      "loss": 0.7358,
+      "step": 13466
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1222800035315427e-07,
+      "loss": 0.4089,
+      "step": 13467
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1160561123396504e-07,
+      "loss": 0.4296,
+      "step": 13468
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.109836824439306e-07,
+      "loss": 0.7496,
+      "step": 13469
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.103622140129071e-07,
+      "loss": 0.3457,
+      "step": 13470
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.097412059707317e-07,
+      "loss": 0.3853,
+      "step": 13471
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.091206583472174e-07,
+      "loss": 0.7539,
+      "step": 13472
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.085005711721579e-07,
+      "loss": 0.4002,
+      "step": 13473
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0788094447532176e-07,
+      "loss": 0.7274,
+      "step": 13474
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.072617782864563e-07,
+      "loss": 0.6976,
+      "step": 13475
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.066430726352888e-07,
+      "loss": 0.3978,
+      "step": 13476
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0602482755151884e-07,
+      "loss": 0.4269,
+      "step": 13477
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.054070430648327e-07,
+      "loss": 0.7317,
+      "step": 13478
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0478971920488663e-07,
+      "loss": 0.7346,
+      "step": 13479
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.041728560013192e-07,
+      "loss": 0.7215,
+      "step": 13480
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.035564534837444e-07,
+      "loss": 0.7208,
+      "step": 13481
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.029405116817542e-07,
+      "loss": 0.7008,
+      "step": 13482
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.023250306249216e-07,
+      "loss": 0.7392,
+      "step": 13483
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.01710010342794e-07,
+      "loss": 0.7361,
+      "step": 13484
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.010954508648957e-07,
+      "loss": 0.6915,
+      "step": 13485
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.004813522207351e-07,
+      "loss": 0.7013,
+      "step": 13486
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.998677144397922e-07,
+      "loss": 0.7103,
+      "step": 13487
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9925453755152885e-07,
+      "loss": 0.7614,
+      "step": 13488
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9864182158538046e-07,
+      "loss": 0.6538,
+      "step": 13489
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.980295665707634e-07,
+      "loss": 0.7457,
+      "step": 13490
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9741777253707425e-07,
+      "loss": 0.383,
+      "step": 13491
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.968064395136817e-07,
+      "loss": 0.4006,
+      "step": 13492
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.961955675299367e-07,
+      "loss": 0.7622,
+      "step": 13493
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9558515661516585e-07,
+      "loss": 0.6958,
+      "step": 13494
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.949752067986734e-07,
+      "loss": 0.7651,
+      "step": 13495
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.94365718109746e-07,
+      "loss": 0.7366,
+      "step": 13496
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9375669057764133e-07,
+      "loss": 0.4309,
+      "step": 13497
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.931481242315993e-07,
+      "loss": 0.7265,
+      "step": 13498
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.925400191008344e-07,
+      "loss": 0.4177,
+      "step": 13499
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.919323752145443e-07,
+      "loss": 0.7213,
+      "step": 13500
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9132519260190016e-07,
+      "loss": 0.6893,
+      "step": 13501
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.907184712920531e-07,
+      "loss": 0.4074,
+      "step": 13502
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.901122113141287e-07,
+      "loss": 0.7368,
+      "step": 13503
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8950641269723475e-07,
+      "loss": 0.6905,
+      "step": 13504
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8890107547045473e-07,
+      "loss": 0.7449,
+      "step": 13505
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.882961996628509e-07,
+      "loss": 0.4071,
+      "step": 13506
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.876917853034623e-07,
+      "loss": 0.6747,
+      "step": 13507
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8708783242130453e-07,
+      "loss": 0.725,
+      "step": 13508
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.864843410453767e-07,
+      "loss": 0.7748,
+      "step": 13509
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8588131120464777e-07,
+      "loss": 0.7054,
+      "step": 13510
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.852787429280724e-07,
+      "loss": 0.4232,
+      "step": 13511
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.846766362445742e-07,
+      "loss": 0.6912,
+      "step": 13512
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8407499118306433e-07,
+      "loss": 0.424,
+      "step": 13513
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8347380777242536e-07,
+      "loss": 0.7117,
+      "step": 13514
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.828730860415197e-07,
+      "loss": 0.7213,
+      "step": 13515
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.822728260191877e-07,
+      "loss": 0.644,
+      "step": 13516
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.816730277342451e-07,
+      "loss": 0.4482,
+      "step": 13517
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8107369121549e-07,
+      "loss": 0.7086,
+      "step": 13518
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8047481649169603e-07,
+      "loss": 0.4105,
+      "step": 13519
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7987640359161473e-07,
+      "loss": 0.7436,
+      "step": 13520
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.79278452543973e-07,
+      "loss": 0.6876,
+      "step": 13521
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7868096337747907e-07,
+      "loss": 0.7061,
+      "step": 13522
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7808393612081887e-07,
+      "loss": 0.7288,
+      "step": 13523
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7748737080265384e-07,
+      "loss": 0.7357,
+      "step": 13524
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7689126745162455e-07,
+      "loss": 0.7083,
+      "step": 13525
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7629562609635017e-07,
+      "loss": 0.4095,
+      "step": 13526
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.757004467654268e-07,
+      "loss": 0.3721,
+      "step": 13527
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7510572948742717e-07,
+      "loss": 0.3736,
+      "step": 13528
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7451147429090397e-07,
+      "loss": 0.7286,
+      "step": 13529
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.739176812043865e-07,
+      "loss": 0.7295,
+      "step": 13530
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.733243502563821e-07,
+      "loss": 0.7425,
+      "step": 13531
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7273148147537575e-07,
+      "loss": 0.7021,
+      "step": 13532
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.721390748898324e-07,
+      "loss": 0.7127,
+      "step": 13533
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7154713052819035e-07,
+      "loss": 0.4174,
+      "step": 13534
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.709556484188703e-07,
+      "loss": 0.3825,
+      "step": 13535
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.703646285902651e-07,
+      "loss": 0.7164,
+      "step": 13536
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6977407107075426e-07,
+      "loss": 0.4115,
+      "step": 13537
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6918397588868616e-07,
+      "loss": 0.7079,
+      "step": 13538
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.685943430723926e-07,
+      "loss": 0.7167,
+      "step": 13539
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6800517265017877e-07,
+      "loss": 0.7285,
+      "step": 13540
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.674164646503331e-07,
+      "loss": 0.7749,
+      "step": 13541
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.668282191011174e-07,
+      "loss": 0.6839,
+      "step": 13542
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6624043603077363e-07,
+      "loss": 0.7087,
+      "step": 13543
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.65653115467518e-07,
+      "loss": 0.6964,
+      "step": 13544
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.650662574395503e-07,
+      "loss": 0.6938,
+      "step": 13545
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6447986197504336e-07,
+      "loss": 0.7409,
+      "step": 13546
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.638939291021504e-07,
+      "loss": 0.6791,
+      "step": 13547
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.633084588489999e-07,
+      "loss": 0.444,
+      "step": 13548
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6272345124370057e-07,
+      "loss": 0.7054,
+      "step": 13549
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.621389063143388e-07,
+      "loss": 0.7383,
+      "step": 13550
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.615548240889788e-07,
+      "loss": 0.4193,
+      "step": 13551
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6097120459565925e-07,
+      "loss": 0.7649,
+      "step": 13552
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.603880478624e-07,
+      "loss": 0.6852,
+      "step": 13553
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.598053539171975e-07,
+      "loss": 0.7309,
+      "step": 13554
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5922312278802827e-07,
+      "loss": 0.7083,
+      "step": 13555
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.586413545028433e-07,
+      "loss": 0.7301,
+      "step": 13556
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.580600490895714e-07,
+      "loss": 0.7165,
+      "step": 13557
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5747920657612124e-07,
+      "loss": 0.4297,
+      "step": 13558
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5689882699038057e-07,
+      "loss": 0.6718,
+      "step": 13559
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5631891036021047e-07,
+      "loss": 0.7252,
+      "step": 13560
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.557394567134531e-07,
+      "loss": 0.7295,
+      "step": 13561
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.551604660779273e-07,
+      "loss": 0.4497,
+      "step": 13562
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.545819384814286e-07,
+      "loss": 0.6732,
+      "step": 13563
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.540038739517337e-07,
+      "loss": 0.4332,
+      "step": 13564
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.534262725165938e-07,
+      "loss": 0.7221,
+      "step": 13565
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5284913420373876e-07,
+      "loss": 0.6852,
+      "step": 13566
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.522724590408777e-07,
+      "loss": 0.3749,
+      "step": 13567
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5169624705569284e-07,
+      "loss": 0.6891,
+      "step": 13568
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5112049827585205e-07,
+      "loss": 0.7117,
+      "step": 13569
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.505452127289943e-07,
+      "loss": 0.7553,
+      "step": 13570
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.499703904427376e-07,
+      "loss": 0.7404,
+      "step": 13571
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.493960314446798e-07,
+      "loss": 0.7119,
+      "step": 13572
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4882213576239444e-07,
+      "loss": 0.6659,
+      "step": 13573
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4824870342343496e-07,
+      "loss": 0.4151,
+      "step": 13574
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4767573445533054e-07,
+      "loss": 0.6925,
+      "step": 13575
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.471032288855869e-07,
+      "loss": 0.6827,
+      "step": 13576
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4653118674169316e-07,
+      "loss": 0.7467,
+      "step": 13577
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.459596080511107e-07,
+      "loss": 0.6766,
+      "step": 13578
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4538849284127983e-07,
+      "loss": 0.6819,
+      "step": 13579
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.448178411396197e-07,
+      "loss": 0.6554,
+      "step": 13580
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4424765297352613e-07,
+      "loss": 0.715,
+      "step": 13581
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.436779283703762e-07,
+      "loss": 0.6908,
+      "step": 13582
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4310866735751793e-07,
+      "loss": 0.7173,
+      "step": 13583
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4253986996228284e-07,
+      "loss": 0.7136,
+      "step": 13584
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.419715362119791e-07,
+      "loss": 0.3923,
+      "step": 13585
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4140366613389043e-07,
+      "loss": 0.7742,
+      "step": 13586
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4083625975527944e-07,
+      "loss": 0.7241,
+      "step": 13587
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4026931710338776e-07,
+      "loss": 0.3863,
+      "step": 13588
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3970283820543347e-07,
+      "loss": 0.4288,
+      "step": 13589
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3913682308861163e-07,
+      "loss": 0.7197,
+      "step": 13590
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.385712717800982e-07,
+      "loss": 0.6786,
+      "step": 13591
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.380061843070426e-07,
+      "loss": 0.4249,
+      "step": 13592
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.374415606965764e-07,
+      "loss": 0.391,
+      "step": 13593
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.368774009758047e-07,
+      "loss": 0.6925,
+      "step": 13594
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.363137051718113e-07,
+      "loss": 0.6975,
+      "step": 13595
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.357504733116612e-07,
+      "loss": 0.7061,
+      "step": 13596
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.35187705422394e-07,
+      "loss": 0.6876,
+      "step": 13597
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3462540153102684e-07,
+      "loss": 0.744,
+      "step": 13598
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.340635616645549e-07,
+      "loss": 0.4076,
+      "step": 13599
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.335021858499543e-07,
+      "loss": 0.711,
+      "step": 13600
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3294127411417356e-07,
+      "loss": 0.7498,
+      "step": 13601
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3238082648414105e-07,
+      "loss": 0.7554,
+      "step": 13602
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.318208429867642e-07,
+      "loss": 0.7631,
+      "step": 13603
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3126132364892814e-07,
+      "loss": 0.7179,
+      "step": 13604
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3070226849749367e-07,
+      "loss": 0.4026,
+      "step": 13605
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3014367755930143e-07,
+      "loss": 0.7196,
+      "step": 13606
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2958555086116784e-07,
+      "loss": 0.69,
+      "step": 13607
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.290278884298881e-07,
+      "loss": 0.4159,
+      "step": 13608
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.284706902922363e-07,
+      "loss": 0.4034,
+      "step": 13609
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2791395647496116e-07,
+      "loss": 0.7223,
+      "step": 13610
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.273576870047923e-07,
+      "loss": 0.6857,
+      "step": 13611
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.268018819084351e-07,
+      "loss": 0.746,
+      "step": 13612
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.262465412125715e-07,
+      "loss": 0.637,
+      "step": 13613
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.256916649438668e-07,
+      "loss": 0.6895,
+      "step": 13614
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.251372531289576e-07,
+      "loss": 0.331,
+      "step": 13615
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.245833057944603e-07,
+      "loss": 0.7189,
+      "step": 13616
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2402982296696917e-07,
+      "loss": 0.728,
+      "step": 13617
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2347680467305744e-07,
+      "loss": 0.7518,
+      "step": 13618
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2292425093927605e-07,
+      "loss": 0.6975,
+      "step": 13619
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2237216179214934e-07,
+      "loss": 0.3828,
+      "step": 13620
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2182053725818506e-07,
+      "loss": 0.7444,
+      "step": 13621
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.212693773638653e-07,
+      "loss": 0.6814,
+      "step": 13622
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.207186821356512e-07,
+      "loss": 0.7198,
+      "step": 13623
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.201684515999803e-07,
+      "loss": 0.7008,
+      "step": 13624
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1961868578326835e-07,
+      "loss": 0.662,
+      "step": 13625
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1906938471191086e-07,
+      "loss": 0.7263,
+      "step": 13626
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1852054841227666e-07,
+      "loss": 0.679,
+      "step": 13627
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.17972176910718e-07,
+      "loss": 0.423,
+      "step": 13628
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1742427023355946e-07,
+      "loss": 0.6975,
+      "step": 13629
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.168768284071055e-07,
+      "loss": 0.7131,
+      "step": 13630
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.163298514576385e-07,
+      "loss": 0.6622,
+      "step": 13631
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.157833394114185e-07,
+      "loss": 0.7199,
+      "step": 13632
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.152372922946834e-07,
+      "loss": 0.7127,
+      "step": 13633
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1469171013364775e-07,
+      "loss": 0.733,
+      "step": 13634
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1414659295450403e-07,
+      "loss": 0.6813,
+      "step": 13635
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.136019407834234e-07,
+      "loss": 0.7277,
+      "step": 13636
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.13057753646554e-07,
+      "loss": 0.6996,
+      "step": 13637
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1251403157002257e-07,
+      "loss": 0.7239,
+      "step": 13638
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1197077457993054e-07,
+      "loss": 0.685,
+      "step": 13639
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.114279827023603e-07,
+      "loss": 0.6811,
+      "step": 13640
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.108856559633711e-07,
+      "loss": 0.4157,
+      "step": 13641
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1034379438899973e-07,
+      "loss": 0.6593,
+      "step": 13642
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0980239800526e-07,
+      "loss": 0.7326,
+      "step": 13643
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.092614668381433e-07,
+      "loss": 0.6875,
+      "step": 13644
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.087210009136188e-07,
+      "loss": 0.6838,
+      "step": 13645
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0818100025763574e-07,
+      "loss": 0.3909,
+      "step": 13646
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.076414648961168e-07,
+      "loss": 0.783,
+      "step": 13647
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0710239485496673e-07,
+      "loss": 0.3693,
+      "step": 13648
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0656379016006154e-07,
+      "loss": 0.7672,
+      "step": 13649
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.06025650837265e-07,
+      "loss": 0.3732,
+      "step": 13650
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.054879769124086e-07,
+      "loss": 0.3956,
+      "step": 13651
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.049507684113073e-07,
+      "loss": 0.6684,
+      "step": 13652
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.044140253597505e-07,
+      "loss": 0.7377,
+      "step": 13653
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.038777477835064e-07,
+      "loss": 0.6831,
+      "step": 13654
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0334193570832335e-07,
+      "loss": 0.6515,
+      "step": 13655
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.028065891599241e-07,
+      "loss": 0.7097,
+      "step": 13656
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.022717081640103e-07,
+      "loss": 0.6602,
+      "step": 13657
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.017372927462603e-07,
+      "loss": 0.701,
+      "step": 13658
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0120334293233025e-07,
+      "loss": 0.7639,
+      "step": 13659
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.006698587478585e-07,
+      "loss": 0.7655,
+      "step": 13660
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0013684021845235e-07,
+      "loss": 0.3763,
+      "step": 13661
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.996042873697025e-07,
+      "loss": 0.6984,
+      "step": 13662
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9907220022717734e-07,
+      "loss": 0.4247,
+      "step": 13663
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.98540578816422e-07,
+      "loss": 0.6877,
+      "step": 13664
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.980094231629593e-07,
+      "loss": 0.3832,
+      "step": 13665
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9747873329228793e-07,
+      "loss": 0.6994,
+      "step": 13666
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.969485092298863e-07,
+      "loss": 0.4631,
+      "step": 13667
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9641875100121065e-07,
+      "loss": 0.6631,
+      "step": 13668
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.958894586316952e-07,
+      "loss": 0.736,
+      "step": 13669
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.953606321467473e-07,
+      "loss": 0.7659,
+      "step": 13670
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9483227157175887e-07,
+      "loss": 0.6695,
+      "step": 13671
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9430437693209303e-07,
+      "loss": 0.4023,
+      "step": 13672
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.937769482530961e-07,
+      "loss": 0.6735,
+      "step": 13673
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.932499855600901e-07,
+      "loss": 0.3916,
+      "step": 13674
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.927234888783692e-07,
+      "loss": 0.6817,
+      "step": 13675
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9219745823321324e-07,
+      "loss": 0.7169,
+      "step": 13676
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9167189364987745e-07,
+      "loss": 0.7487,
+      "step": 13677
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9114679515359177e-07,
+      "loss": 0.7724,
+      "step": 13678
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.90622162769566e-07,
+      "loss": 0.6963,
+      "step": 13679
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9009799652298775e-07,
+      "loss": 0.7909,
+      "step": 13680
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.895742964390191e-07,
+      "loss": 0.706,
+      "step": 13681
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8905106254280666e-07,
+      "loss": 0.7223,
+      "step": 13682
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.885282948594692e-07,
+      "loss": 0.7525,
+      "step": 13683
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.880059934141022e-07,
+      "loss": 0.7031,
+      "step": 13684
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8748415823178223e-07,
+      "loss": 0.7462,
+      "step": 13685
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.869627893375615e-07,
+      "loss": 0.7393,
+      "step": 13686
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.864418867564711e-07,
+      "loss": 0.7535,
+      "step": 13687
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8592145051351883e-07,
+      "loss": 0.7149,
+      "step": 13688
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.854014806336902e-07,
+      "loss": 0.4334,
+      "step": 13689
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8488197714194866e-07,
+      "loss": 0.7168,
+      "step": 13690
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.843629400632342e-07,
+      "loss": 0.7464,
+      "step": 13691
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8384436942246573e-07,
+      "loss": 0.7383,
+      "step": 13692
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.833262652445401e-07,
+      "loss": 0.3749,
+      "step": 13693
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8280862755432957e-07,
+      "loss": 0.689,
+      "step": 13694
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.822914563766865e-07,
+      "loss": 0.7113,
+      "step": 13695
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8177475173644e-07,
+      "loss": 0.7316,
+      "step": 13696
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.812585136583956e-07,
+      "loss": 0.7319,
+      "step": 13697
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.807427421673381e-07,
+      "loss": 0.7192,
+      "step": 13698
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.802274372880265e-07,
+      "loss": 0.7225,
+      "step": 13699
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.797125990452043e-07,
+      "loss": 0.6822,
+      "step": 13700
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7919822746358736e-07,
+      "loss": 0.7212,
+      "step": 13701
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7868432256786816e-07,
+      "loss": 0.7528,
+      "step": 13702
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7817088438272023e-07,
+      "loss": 0.426,
+      "step": 13703
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7765791293279165e-07,
+      "loss": 0.6879,
+      "step": 13704
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.771454082427116e-07,
+      "loss": 0.6942,
+      "step": 13705
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.766333703370849e-07,
+      "loss": 0.6877,
+      "step": 13706
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.761217992404919e-07,
+      "loss": 0.6658,
+      "step": 13707
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.756106949774939e-07,
+      "loss": 0.3637,
+      "step": 13708
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7510005757262923e-07,
+      "loss": 0.3636,
+      "step": 13709
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.745898870504116e-07,
+      "loss": 0.358,
+      "step": 13710
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.740801834353357e-07,
+      "loss": 0.6971,
+      "step": 13711
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.735709467518699e-07,
+      "loss": 0.7751,
+      "step": 13712
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7306217702446236e-07,
+      "loss": 0.7,
+      "step": 13713
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.725538742775402e-07,
+      "loss": 0.7338,
+      "step": 13714
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7204603853550506e-07,
+      "loss": 0.713,
+      "step": 13715
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7153866982273845e-07,
+      "loss": 0.6692,
+      "step": 13716
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7103176816359877e-07,
+      "loss": 0.3853,
+      "step": 13717
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7052533358241984e-07,
+      "loss": 0.7131,
+      "step": 13718
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7001936610351887e-07,
+      "loss": 0.7608,
+      "step": 13719
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.695138657511831e-07,
+      "loss": 0.7485,
+      "step": 13720
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6900883254968203e-07,
+      "loss": 0.3823,
+      "step": 13721
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.685042665232629e-07,
+      "loss": 0.7121,
+      "step": 13722
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.680001676961497e-07,
+      "loss": 0.6886,
+      "step": 13723
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.674965360925419e-07,
+      "loss": 0.7273,
+      "step": 13724
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.669933717366191e-07,
+      "loss": 0.728,
+      "step": 13725
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.664906746525375e-07,
+      "loss": 0.3872,
+      "step": 13726
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.659884448644323e-07,
+      "loss": 0.7475,
+      "step": 13727
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.654866823964142e-07,
+      "loss": 0.7246,
+      "step": 13728
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.649853872725727e-07,
+      "loss": 0.6972,
+      "step": 13729
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.644845595169743e-07,
+      "loss": 0.7333,
+      "step": 13730
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.639841991536618e-07,
+      "loss": 0.7486,
+      "step": 13731
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6348430620665945e-07,
+      "loss": 0.7145,
+      "step": 13732
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6298488069996464e-07,
+      "loss": 0.778,
+      "step": 13733
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.624859226575571e-07,
+      "loss": 0.6991,
+      "step": 13734
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.619874321033866e-07,
+      "loss": 0.75,
+      "step": 13735
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.614894090613884e-07,
+      "loss": 0.7289,
+      "step": 13736
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.609918535554723e-07,
+      "loss": 0.6985,
+      "step": 13737
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.604947656095247e-07,
+      "loss": 0.7045,
+      "step": 13738
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5999814524740985e-07,
+      "loss": 0.6951,
+      "step": 13739
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5950199249296983e-07,
+      "loss": 0.7244,
+      "step": 13740
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5900630737002553e-07,
+      "loss": 0.7248,
+      "step": 13741
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.585110899023735e-07,
+      "loss": 0.7146,
+      "step": 13742
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.580163401137892e-07,
+      "loss": 0.673,
+      "step": 13743
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5752205802802466e-07,
+      "loss": 0.4284,
+      "step": 13744
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5702824366880986e-07,
+      "loss": 0.6514,
+      "step": 13745
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5653489705985247e-07,
+      "loss": 0.721,
+      "step": 13746
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.56042018224838e-07,
+      "loss": 0.7076,
+      "step": 13747
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5554960718742863e-07,
+      "loss": 0.6186,
+      "step": 13748
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.550576639712632e-07,
+      "loss": 0.7223,
+      "step": 13749
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.545661885999617e-07,
+      "loss": 0.6853,
+      "step": 13750
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5407518109711873e-07,
+      "loss": 0.757,
+      "step": 13751
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5358464148630636e-07,
+      "loss": 0.6906,
+      "step": 13752
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.530945697910747e-07,
+      "loss": 0.7172,
+      "step": 13753
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5260496603495275e-07,
+      "loss": 0.7562,
+      "step": 13754
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.52115830241445e-07,
+      "loss": 0.3891,
+      "step": 13755
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5162716243403494e-07,
+      "loss": 0.7302,
+      "step": 13756
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5113896263618263e-07,
+      "loss": 0.7082,
+      "step": 13757
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.506512308713249e-07,
+      "loss": 0.7351,
+      "step": 13758
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.501639671628797e-07,
+      "loss": 0.7746,
+      "step": 13759
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.496771715342383e-07,
+      "loss": 0.6854,
+      "step": 13760
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4919084400877316e-07,
+      "loss": 0.3752,
+      "step": 13761
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4870498460982997e-07,
+      "loss": 0.7469,
+      "step": 13762
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4821959336073454e-07,
+      "loss": 0.7377,
+      "step": 13763
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.477346702847938e-07,
+      "loss": 0.4289,
+      "step": 13764
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.472502154052836e-07,
+      "loss": 0.6709,
+      "step": 13765
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.467662287454642e-07,
+      "loss": 0.6909,
+      "step": 13766
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.462827103285703e-07,
+      "loss": 0.7494,
+      "step": 13767
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4579966017781675e-07,
+      "loss": 0.6793,
+      "step": 13768
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.453170783163938e-07,
+      "loss": 0.7822,
+      "step": 13769
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.448349647674697e-07,
+      "loss": 0.7043,
+      "step": 13770
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4435331955418916e-07,
+      "loss": 0.7147,
+      "step": 13771
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4387214269967596e-07,
+      "loss": 0.7423,
+      "step": 13772
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.433914342270327e-07,
+      "loss": 0.6421,
+      "step": 13773
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.429111941593365e-07,
+      "loss": 0.7008,
+      "step": 13774
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.424314225196422e-07,
+      "loss": 0.6896,
+      "step": 13775
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.419521193309837e-07,
+      "loss": 0.6945,
+      "step": 13776
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4147328461637143e-07,
+      "loss": 0.6878,
+      "step": 13777
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4099491839879586e-07,
+      "loss": 0.4119,
+      "step": 13778
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4051702070122197e-07,
+      "loss": 0.7318,
+      "step": 13779
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.400395915465903e-07,
+      "loss": 0.4095,
+      "step": 13780
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3956263095782583e-07,
+      "loss": 0.6504,
+      "step": 13781
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.390861389578247e-07,
+      "loss": 0.7922,
+      "step": 13782
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.386101155694642e-07,
+      "loss": 0.4125,
+      "step": 13783
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3813456081559606e-07,
+      "loss": 0.7178,
+      "step": 13784
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3765947471905082e-07,
+      "loss": 0.7219,
+      "step": 13785
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3718485730263918e-07,
+      "loss": 0.3947,
+      "step": 13786
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3671070858914623e-07,
+      "loss": 0.7407,
+      "step": 13787
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3623702860133492e-07,
+      "loss": 0.6811,
+      "step": 13788
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3576381736194588e-07,
+      "loss": 0.7305,
+      "step": 13789
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3529107489369762e-07,
+      "loss": 0.7479,
+      "step": 13790
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3481880121928645e-07,
+      "loss": 0.7178,
+      "step": 13791
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3434699636138646e-07,
+      "loss": 0.6709,
+      "step": 13792
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3387566034264842e-07,
+      "loss": 0.7243,
+      "step": 13793
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3340479318569753e-07,
+      "loss": 0.7103,
+      "step": 13794
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.329343949131435e-07,
+      "loss": 0.3433,
+      "step": 13795
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.324644655475683e-07,
+      "loss": 0.6937,
+      "step": 13796
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3199500511153272e-07,
+      "loss": 0.7073,
+      "step": 13797
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3152601362757433e-07,
+      "loss": 0.7063,
+      "step": 13798
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3105749111820952e-07,
+      "loss": 0.4162,
+      "step": 13799
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3058943760593256e-07,
+      "loss": 0.7238,
+      "step": 13800
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3012185311321323e-07,
+      "loss": 0.657,
+      "step": 13801
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2965473766250023e-07,
+      "loss": 0.7161,
+      "step": 13802
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2918809127621788e-07,
+      "loss": 0.7059,
+      "step": 13803
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2872191397677156e-07,
+      "loss": 0.7191,
+      "step": 13804
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2825620578654006e-07,
+      "loss": 0.7155,
+      "step": 13805
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2779096672788325e-07,
+      "loss": 0.7377,
+      "step": 13806
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2732619682313662e-07,
+      "loss": 0.7202,
+      "step": 13807
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2686189609461229e-07,
+      "loss": 0.6947,
+      "step": 13808
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2639806456460133e-07,
+      "loss": 0.7451,
+      "step": 13809
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.259347022553715e-07,
+      "loss": 0.4217,
+      "step": 13810
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2547180918916834e-07,
+      "loss": 0.7109,
+      "step": 13811
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2500938538821515e-07,
+      "loss": 0.7115,
+      "step": 13812
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.245474308747131e-07,
+      "loss": 0.6733,
+      "step": 13813
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2408594567083887e-07,
+      "loss": 0.3863,
+      "step": 13814
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2362492979874916e-07,
+      "loss": 0.7035,
+      "step": 13815
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2316438328057631e-07,
+      "loss": 0.7518,
+      "step": 13816
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.227043061384293e-07,
+      "loss": 0.7248,
+      "step": 13817
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2224469839439932e-07,
+      "loss": 0.7278,
+      "step": 13818
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2178556007054876e-07,
+      "loss": 0.7328,
+      "step": 13819
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2132689118892105e-07,
+      "loss": 0.6693,
+      "step": 13820
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2086869177153746e-07,
+      "loss": 0.7189,
+      "step": 13821
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2041096184039268e-07,
+      "loss": 0.6568,
+      "step": 13822
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1995370141746687e-07,
+      "loss": 0.7265,
+      "step": 13823
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.194969105247091e-07,
+      "loss": 0.4095,
+      "step": 13824
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.190405891840497e-07,
+      "loss": 0.7028,
+      "step": 13825
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.185847374173966e-07,
+      "loss": 0.4253,
+      "step": 13826
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1812935524663458e-07,
+      "loss": 0.7463,
+      "step": 13827
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1767444269362724e-07,
+      "loss": 0.4124,
+      "step": 13828
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.172199997802138e-07,
+      "loss": 0.7027,
+      "step": 13829
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1676602652821121e-07,
+      "loss": 0.4176,
+      "step": 13830
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1631252295941318e-07,
+      "loss": 0.6996,
+      "step": 13831
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.158594890955945e-07,
+      "loss": 0.7296,
+      "step": 13832
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1540692495850334e-07,
+      "loss": 0.7187,
+      "step": 13833
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1495483056986677e-07,
+      "loss": 0.4128,
+      "step": 13834
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1450320595139074e-07,
+      "loss": 0.7475,
+      "step": 13835
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1405205112475458e-07,
+      "loss": 0.7277,
+      "step": 13836
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1360136611161987e-07,
+      "loss": 0.7172,
+      "step": 13837
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1315115093362593e-07,
+      "loss": 0.4023,
+      "step": 13838
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1270140561238106e-07,
+      "loss": 0.3897,
+      "step": 13839
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1225213016948132e-07,
+      "loss": 0.7213,
+      "step": 13840
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.118033246264961e-07,
+      "loss": 0.7433,
+      "step": 13841
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1135498900497043e-07,
+      "loss": 0.6761,
+      "step": 13842
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1090712332642926e-07,
+      "loss": 0.7054,
+      "step": 13843
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.104597276123721e-07,
+      "loss": 0.7355,
+      "step": 13844
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1001280188428176e-07,
+      "loss": 0.4538,
+      "step": 13845
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0956634616361217e-07,
+      "loss": 0.6879,
+      "step": 13846
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.091203604717984e-07,
+      "loss": 0.7074,
+      "step": 13847
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0867484483025113e-07,
+      "loss": 0.7347,
+      "step": 13848
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.082297992603588e-07,
+      "loss": 0.4059,
+      "step": 13849
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0778522378348765e-07,
+      "loss": 0.647,
+      "step": 13850
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0734111842098282e-07,
+      "loss": 0.7099,
+      "step": 13851
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.068974831941628e-07,
+      "loss": 0.7282,
+      "step": 13852
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0645431812432836e-07,
+      "loss": 0.6788,
+      "step": 13853
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.060116232327547e-07,
+      "loss": 0.7157,
+      "step": 13854
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.055693985406948e-07,
+      "loss": 0.6808,
+      "step": 13855
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0512764406937945e-07,
+      "loss": 0.7744,
+      "step": 13856
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0468635984001728e-07,
+      "loss": 0.6679,
+      "step": 13857
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0424554587379242e-07,
+      "loss": 0.7054,
+      "step": 13858
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0380520219186907e-07,
+      "loss": 0.6738,
+      "step": 13859
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0336532881538917e-07,
+      "loss": 0.42,
+      "step": 13860
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0292592576546811e-07,
+      "loss": 0.4475,
+      "step": 13861
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.024869930632023e-07,
+      "loss": 0.4032,
+      "step": 13862
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.020485307296638e-07,
+      "loss": 0.7144,
+      "step": 13863
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0161053878590465e-07,
+      "loss": 0.7246,
+      "step": 13864
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0117301725295025e-07,
+      "loss": 0.7094,
+      "step": 13865
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0073596615180714e-07,
+      "loss": 0.388,
+      "step": 13866
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.002993855034563e-07,
+      "loss": 0.3963,
+      "step": 13867
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9986327532885764e-07,
+      "loss": 0.7086,
+      "step": 13868
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9942763564894996e-07,
+      "loss": 0.6832,
+      "step": 13869
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9899246648464653e-07,
+      "loss": 0.7163,
+      "step": 13870
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9855776785683844e-07,
+      "loss": 0.7524,
+      "step": 13871
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9812353978639786e-07,
+      "loss": 0.7262,
+      "step": 13872
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9768978229416924e-07,
+      "loss": 0.3806,
+      "step": 13873
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9725649540097814e-07,
+      "loss": 0.6895,
+      "step": 13874
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9682367912762567e-07,
+      "loss": 0.4124,
+      "step": 13875
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9639133349489082e-07,
+      "loss": 0.7441,
+      "step": 13876
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9595945852353027e-07,
+      "loss": 0.7108,
+      "step": 13877
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9552805423427855e-07,
+      "loss": 0.6604,
+      "step": 13878
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.950971206478469e-07,
+      "loss": 0.6812,
+      "step": 13879
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9466665778492212e-07,
+      "loss": 0.7413,
+      "step": 13880
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9423666566617206e-07,
+      "loss": 0.4114,
+      "step": 13881
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9380714431223914e-07,
+      "loss": 0.7292,
+      "step": 13882
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9337809374374683e-07,
+      "loss": 0.7166,
+      "step": 13883
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.929495139812898e-07,
+      "loss": 0.723,
+      "step": 13884
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9252140504544604e-07,
+      "loss": 0.6879,
+      "step": 13885
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9209376695676686e-07,
+      "loss": 0.3725,
+      "step": 13886
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9166659973578472e-07,
+      "loss": 0.7061,
+      "step": 13887
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.912399034030077e-07,
+      "loss": 0.7121,
+      "step": 13888
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9081367797891825e-07,
+      "loss": 0.6902,
+      "step": 13889
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9038792348398116e-07,
+      "loss": 0.7651,
+      "step": 13890
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8996263993863673e-07,
+      "loss": 0.7388,
+      "step": 13891
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8953782736330084e-07,
+      "loss": 0.7331,
+      "step": 13892
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.891134857783694e-07,
+      "loss": 0.6572,
+      "step": 13893
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8868961520421502e-07,
+      "loss": 0.709,
+      "step": 13894
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8826621566118586e-07,
+      "loss": 0.6984,
+      "step": 13895
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.878432871696101e-07,
+      "loss": 0.7421,
+      "step": 13896
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.874208297497926e-07,
+      "loss": 0.6526,
+      "step": 13897
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.869988434220138e-07,
+      "loss": 0.7497,
+      "step": 13898
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8657732820653306e-07,
+      "loss": 0.3786,
+      "step": 13899
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8615628412358755e-07,
+      "loss": 0.3924,
+      "step": 13900
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8573571119339107e-07,
+      "loss": 0.7193,
+      "step": 13901
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8531560943613414e-07,
+      "loss": 0.6974,
+      "step": 13902
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8489597887198507e-07,
+      "loss": 0.6876,
+      "step": 13903
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.844768195210911e-07,
+      "loss": 0.7356,
+      "step": 13904
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.840581314035761e-07,
+      "loss": 0.7402,
+      "step": 13905
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8363991453953955e-07,
+      "loss": 0.3736,
+      "step": 13906
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8322216894905986e-07,
+      "loss": 0.7343,
+      "step": 13907
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.828048946521932e-07,
+      "loss": 0.402,
+      "step": 13908
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8238809166897131e-07,
+      "loss": 0.6934,
+      "step": 13909
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8197176001940596e-07,
+      "loss": 0.435,
+      "step": 13910
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8155589972348453e-07,
+      "loss": 0.7041,
+      "step": 13911
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.811405108011721e-07,
+      "loss": 0.7056,
+      "step": 13912
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8072559327240945e-07,
+      "loss": 0.4015,
+      "step": 13913
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.803111471571184e-07,
+      "loss": 0.6826,
+      "step": 13914
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7989717247519411e-07,
+      "loss": 0.7386,
+      "step": 13915
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7948366924651294e-07,
+      "loss": 0.6675,
+      "step": 13916
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7907063749092567e-07,
+      "loss": 0.7091,
+      "step": 13917
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.78658077228262e-07,
+      "loss": 0.7239,
+      "step": 13918
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.782459884783294e-07,
+      "loss": 0.7247,
+      "step": 13919
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7783437126090986e-07,
+      "loss": 0.7624,
+      "step": 13920
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.774232255957664e-07,
+      "loss": 0.6889,
+      "step": 13921
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7701255150263553e-07,
+      "loss": 0.7274,
+      "step": 13922
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7660234900123585e-07,
+      "loss": 0.7045,
+      "step": 13923
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7619261811125943e-07,
+      "loss": 0.7457,
+      "step": 13924
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.757833588523783e-07,
+      "loss": 0.6739,
+      "step": 13925
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7537457124423896e-07,
+      "loss": 0.6918,
+      "step": 13926
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7496625530646793e-07,
+      "loss": 0.3869,
+      "step": 13927
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7455841105866733e-07,
+      "loss": 0.6911,
+      "step": 13928
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7415103852041703e-07,
+      "loss": 0.7001,
+      "step": 13929
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7374413771127473e-07,
+      "loss": 0.7084,
+      "step": 13930
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7333770865077704e-07,
+      "loss": 0.6923,
+      "step": 13931
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.72931751358435e-07,
+      "loss": 0.7096,
+      "step": 13932
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7252626585373854e-07,
+      "loss": 0.7889,
+      "step": 13933
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7212125215615327e-07,
+      "loss": 0.721,
+      "step": 13934
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7171671028512472e-07,
+      "loss": 0.7176,
+      "step": 13935
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7131264026007399e-07,
+      "loss": 0.421,
+      "step": 13936
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7090904210040115e-07,
+      "loss": 0.7353,
+      "step": 13937
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7050591582548182e-07,
+      "loss": 0.7468,
+      "step": 13938
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7010326145466938e-07,
+      "loss": 0.6906,
+      "step": 13939
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6970107900729395e-07,
+      "loss": 0.7027,
+      "step": 13940
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.692993685026667e-07,
+      "loss": 0.4234,
+      "step": 13941
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6889812996007227e-07,
+      "loss": 0.7478,
+      "step": 13942
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6849736339877188e-07,
+      "loss": 0.3524,
+      "step": 13943
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6809706883800681e-07,
+      "loss": 0.7167,
+      "step": 13944
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6769724629699613e-07,
+      "loss": 0.7119,
+      "step": 13945
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.672978957949334e-07,
+      "loss": 0.6922,
+      "step": 13946
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.668990173509921e-07,
+      "loss": 0.7232,
+      "step": 13947
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6650061098432145e-07,
+      "loss": 0.4201,
+      "step": 13948
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6610267671404833e-07,
+      "loss": 0.7518,
+      "step": 13949
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6570521455927745e-07,
+      "loss": 0.7021,
+      "step": 13950
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6530822453909135e-07,
+      "loss": 0.6736,
+      "step": 13951
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6491170667254697e-07,
+      "loss": 0.3849,
+      "step": 13952
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6451566097868355e-07,
+      "loss": 0.7011,
+      "step": 13953
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6412008747651142e-07,
+      "loss": 0.7636,
+      "step": 13954
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6372498618502542e-07,
+      "loss": 0.7105,
+      "step": 13955
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6333035712319146e-07,
+      "loss": 0.3889,
+      "step": 13956
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.629362003099566e-07,
+      "loss": 0.7004,
+      "step": 13957
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.625425157642435e-07,
+      "loss": 0.6809,
+      "step": 13958
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.621493035049515e-07,
+      "loss": 0.3739,
+      "step": 13959
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6175656355095993e-07,
+      "loss": 0.66,
+      "step": 13960
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6136429592112257e-07,
+      "loss": 0.3467,
+      "step": 13961
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6097250063427106e-07,
+      "loss": 0.7101,
+      "step": 13962
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.605811777092181e-07,
+      "loss": 0.3912,
+      "step": 13963
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.601903271647487e-07,
+      "loss": 0.741,
+      "step": 13964
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.597999490196278e-07,
+      "loss": 0.7658,
+      "step": 13965
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.59410043292596e-07,
+      "loss": 0.4274,
+      "step": 13966
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.590206100023728e-07,
+      "loss": 0.694,
+      "step": 13967
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5863164916765539e-07,
+      "loss": 0.7489,
+      "step": 13968
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5824316080711665e-07,
+      "loss": 0.6898,
+      "step": 13969
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5785514493940724e-07,
+      "loss": 0.4324,
+      "step": 13970
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5746760158315667e-07,
+      "loss": 0.7112,
+      "step": 13971
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5708053075696893e-07,
+      "loss": 0.3972,
+      "step": 13972
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.566939324794281e-07,
+      "loss": 0.6774,
+      "step": 13973
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5630780676909264e-07,
+      "loss": 0.708,
+      "step": 13974
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5592215364450214e-07,
+      "loss": 0.7432,
+      "step": 13975
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.555369731241696e-07,
+      "loss": 0.7696,
+      "step": 13976
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5515226522658912e-07,
+      "loss": 0.4302,
+      "step": 13977
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5476802997022812e-07,
+      "loss": 0.7461,
+      "step": 13978
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5438426737353518e-07,
+      "loss": 0.7156,
+      "step": 13979
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5400097745493338e-07,
+      "loss": 0.7008,
+      "step": 13980
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.536181602328235e-07,
+      "loss": 0.7319,
+      "step": 13981
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5323581572558534e-07,
+      "loss": 0.7052,
+      "step": 13982
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5285394395157527e-07,
+      "loss": 0.3649,
+      "step": 13983
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.524725449291242e-07,
+      "loss": 0.682,
+      "step": 13984
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.520916186765453e-07,
+      "loss": 0.7687,
+      "step": 13985
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5171116521212505e-07,
+      "loss": 0.7273,
+      "step": 13986
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5133118455412988e-07,
+      "loss": 0.7206,
+      "step": 13987
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.509516767208008e-07,
+      "loss": 0.7262,
+      "step": 13988
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5057264173035768e-07,
+      "loss": 0.3916,
+      "step": 13989
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.501940796009982e-07,
+      "loss": 0.7643,
+      "step": 13990
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4981599035089667e-07,
+      "loss": 0.7449,
+      "step": 13991
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4943837399820527e-07,
+      "loss": 0.704,
+      "step": 13992
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4906123056105281e-07,
+      "loss": 0.6967,
+      "step": 13993
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4868456005754372e-07,
+      "loss": 0.6806,
+      "step": 13994
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4830836250576464e-07,
+      "loss": 0.7568,
+      "step": 13995
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4793263792377444e-07,
+      "loss": 0.7709,
+      "step": 13996
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.475573863296109e-07,
+      "loss": 0.7271,
+      "step": 13997
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.471826077412908e-07,
+      "loss": 0.6987,
+      "step": 13998
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.468083021768063e-07,
+      "loss": 0.6924,
+      "step": 13999
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4643446965412755e-07,
+      "loss": 0.4393,
+      "step": 14000
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.460611101912024e-07,
+      "loss": 0.7412,
+      "step": 14001
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4568822380595539e-07,
+      "loss": 0.7531,
+      "step": 14002
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4531581051628662e-07,
+      "loss": 0.3717,
+      "step": 14003
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4494387034007738e-07,
+      "loss": 0.7267,
+      "step": 14004
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4457240329518229e-07,
+      "loss": 0.3845,
+      "step": 14005
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4420140939943706e-07,
+      "loss": 0.417,
+      "step": 14006
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4383088867065187e-07,
+      "loss": 0.7124,
+      "step": 14007
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4346084112661475e-07,
+      "loss": 0.3978,
+      "step": 14008
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4309126678509145e-07,
+      "loss": 0.6829,
+      "step": 14009
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4272216566382559e-07,
+      "loss": 0.7008,
+      "step": 14010
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.423535377805363e-07,
+      "loss": 0.7204,
+      "step": 14011
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4198538315292166e-07,
+      "loss": 0.728,
+      "step": 14012
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4161770179865531e-07,
+      "loss": 0.7201,
+      "step": 14013
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.412504937353909e-07,
+      "loss": 0.6999,
+      "step": 14014
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4088375898075658e-07,
+      "loss": 0.7293,
+      "step": 14015
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4051749755236045e-07,
+      "loss": 0.7171,
+      "step": 14016
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.40151709467784e-07,
+      "loss": 0.704,
+      "step": 14017
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3978639474458988e-07,
+      "loss": 0.6975,
+      "step": 14018
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.394215534003174e-07,
+      "loss": 0.7319,
+      "step": 14019
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3905718545247916e-07,
+      "loss": 0.7052,
+      "step": 14020
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.386932909185701e-07,
+      "loss": 0.6963,
+      "step": 14021
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3832986981606068e-07,
+      "loss": 0.7103,
+      "step": 14022
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3796692216239915e-07,
+      "loss": 0.7406,
+      "step": 14023
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3760444797500827e-07,
+      "loss": 0.728,
+      "step": 14024
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3724244727129077e-07,
+      "loss": 0.7263,
+      "step": 14025
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3688092006862496e-07,
+      "loss": 0.3905,
+      "step": 14026
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.365198663843692e-07,
+      "loss": 0.697,
+      "step": 14027
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3615928623585738e-07,
+      "loss": 0.7824,
+      "step": 14028
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3579917964039902e-07,
+      "loss": 0.771,
+      "step": 14029
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.354395466152836e-07,
+      "loss": 0.6763,
+      "step": 14030
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3508038717777617e-07,
+      "loss": 0.3923,
+      "step": 14031
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3472170134512075e-07,
+      "loss": 0.4124,
+      "step": 14032
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3436348913453578e-07,
+      "loss": 0.7164,
+      "step": 14033
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3400575056321974e-07,
+      "loss": 0.7038,
+      "step": 14034
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3364848564834553e-07,
+      "loss": 0.7189,
+      "step": 14035
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3329169440706834e-07,
+      "loss": 0.407,
+      "step": 14036
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3293537685651448e-07,
+      "loss": 0.3999,
+      "step": 14037
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3257953301379246e-07,
+      "loss": 0.7449,
+      "step": 14038
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3222416289598416e-07,
+      "loss": 0.7398,
+      "step": 14039
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3186926652015154e-07,
+      "loss": 0.6892,
+      "step": 14040
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3151484390333315e-07,
+      "loss": 0.673,
+      "step": 14041
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3116089506254315e-07,
+      "loss": 0.6918,
+      "step": 14042
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3080742001477686e-07,
+      "loss": 0.3898,
+      "step": 14043
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.304544187770007e-07,
+      "loss": 0.6561,
+      "step": 14044
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3010189136616447e-07,
+      "loss": 0.6888,
+      "step": 14045
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2974983779919237e-07,
+      "loss": 0.7265,
+      "step": 14046
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2939825809298533e-07,
+      "loss": 0.7346,
+      "step": 14047
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.290471522644221e-07,
+      "loss": 0.7375,
+      "step": 14048
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2869652033035916e-07,
+      "loss": 0.7001,
+      "step": 14049
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2834636230763198e-07,
+      "loss": 0.721,
+      "step": 14050
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2799667821304816e-07,
+      "loss": 0.7122,
+      "step": 14051
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2764746806339767e-07,
+      "loss": 0.7244,
+      "step": 14052
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2729873187544483e-07,
+      "loss": 0.3971,
+      "step": 14053
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2695046966593295e-07,
+      "loss": 0.723,
+      "step": 14054
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2660268145158195e-07,
+      "loss": 0.6826,
+      "step": 14055
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2625536724908737e-07,
+      "loss": 0.7025,
+      "step": 14056
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.259085270751248e-07,
+      "loss": 0.7116,
+      "step": 14057
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2556216094634422e-07,
+      "loss": 0.6874,
+      "step": 14058
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2521626887937565e-07,
+      "loss": 0.7432,
+      "step": 14059
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2487085089082362e-07,
+      "loss": 0.6843,
+      "step": 14060
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2452590699727485e-07,
+      "loss": 0.4277,
+      "step": 14061
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2418143721528387e-07,
+      "loss": 0.735,
+      "step": 14062
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.238374415613941e-07,
+      "loss": 0.6951,
+      "step": 14063
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.234939200521168e-07,
+      "loss": 0.3641,
+      "step": 14064
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2315087270394545e-07,
+      "loss": 0.7059,
+      "step": 14065
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2280829953334906e-07,
+      "loss": 0.7095,
+      "step": 14066
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2246620055677338e-07,
+      "loss": 0.3952,
+      "step": 14067
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2212457579064417e-07,
+      "loss": 0.7093,
+      "step": 14068
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2178342525136167e-07,
+      "loss": 0.4095,
+      "step": 14069
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2144274895530385e-07,
+      "loss": 0.6973,
+      "step": 14070
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.211025469188265e-07,
+      "loss": 0.4289,
+      "step": 14071
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2076281915826216e-07,
+      "loss": 0.3938,
+      "step": 14072
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2042356568991999e-07,
+      "loss": 0.7026,
+      "step": 14073
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2008478653008915e-07,
+      "loss": 0.3772,
+      "step": 14074
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1974648169503333e-07,
+      "loss": 0.753,
+      "step": 14075
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.19408651200994e-07,
+      "loss": 0.4179,
+      "step": 14076
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1907129506418923e-07,
+      "loss": 0.7402,
+      "step": 14077
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1873441330081614e-07,
+      "loss": 0.6753,
+      "step": 14078
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1839800592704842e-07,
+      "loss": 0.4136,
+      "step": 14079
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1806207295903538e-07,
+      "loss": 0.7399,
+      "step": 14080
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1772661441290634e-07,
+      "loss": 0.75,
+      "step": 14081
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1739163030476508e-07,
+      "loss": 0.7678,
+      "step": 14082
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.170571206506943e-07,
+      "loss": 0.7151,
+      "step": 14083
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1672308546675337e-07,
+      "loss": 0.4325,
+      "step": 14084
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1638952476897835e-07,
+      "loss": 0.7374,
+      "step": 14085
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1605643857338534e-07,
+      "loss": 0.698,
+      "step": 14086
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1572382689596373e-07,
+      "loss": 0.7266,
+      "step": 14087
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.153916897526819e-07,
+      "loss": 0.6535,
+      "step": 14088
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1506002715948595e-07,
+      "loss": 0.3842,
+      "step": 14089
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1472883913229759e-07,
+      "loss": 0.7279,
+      "step": 14090
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1439812568701858e-07,
+      "loss": 0.6704,
+      "step": 14091
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1406788683952508e-07,
+      "loss": 0.7069,
+      "step": 14092
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1373812260567107e-07,
+      "loss": 0.7323,
+      "step": 14093
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1340883300128836e-07,
+      "loss": 0.739,
+      "step": 14094
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.130800180421876e-07,
+      "loss": 0.4153,
+      "step": 14095
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1275167774415285e-07,
+      "loss": 0.7792,
+      "step": 14096
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1242381212294818e-07,
+      "loss": 0.3787,
+      "step": 14097
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.120964211943143e-07,
+      "loss": 0.7242,
+      "step": 14098
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1176950497396643e-07,
+      "loss": 0.7263,
+      "step": 14099
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1144306347760426e-07,
+      "loss": 0.7369,
+      "step": 14100
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1111709672089633e-07,
+      "loss": 0.3834,
+      "step": 14101
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1079160471949235e-07,
+      "loss": 0.6969,
+      "step": 14102
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1046658748901984e-07,
+      "loss": 0.3749,
+      "step": 14103
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1014204504508186e-07,
+      "loss": 0.7083,
+      "step": 14104
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0981797740326039e-07,
+      "loss": 0.6927,
+      "step": 14105
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0949438457911299e-07,
+      "loss": 0.448,
+      "step": 14106
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.091712665881739e-07,
+      "loss": 0.6893,
+      "step": 14107
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0884862344595737e-07,
+      "loss": 0.3947,
+      "step": 14108
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0852645516795213e-07,
+      "loss": 0.7065,
+      "step": 14109
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.082047617696258e-07,
+      "loss": 0.7242,
+      "step": 14110
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.078835432664227e-07,
+      "loss": 0.7411,
+      "step": 14111
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.075627996737627e-07,
+      "loss": 0.3702,
+      "step": 14112
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0724253100704573e-07,
+      "loss": 0.7073,
+      "step": 14113
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0692273728164727e-07,
+      "loss": 0.7455,
+      "step": 14114
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0660341851292055e-07,
+      "loss": 0.7248,
+      "step": 14115
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0628457471619557e-07,
+      "loss": 0.7213,
+      "step": 14116
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0596620590677897e-07,
+      "loss": 0.6639,
+      "step": 14117
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.056483120999563e-07,
+      "loss": 0.6951,
+      "step": 14118
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0533089331098978e-07,
+      "loss": 0.7238,
+      "step": 14119
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0501394955511723e-07,
+      "loss": 0.4053,
+      "step": 14120
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0469748084755426e-07,
+      "loss": 0.4081,
+      "step": 14121
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.043814872034965e-07,
+      "loss": 0.7348,
+      "step": 14122
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.040659686381118e-07,
+      "loss": 0.4004,
+      "step": 14123
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0375092516655028e-07,
+      "loss": 0.6912,
+      "step": 14124
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0343635680393538e-07,
+      "loss": 0.7355,
+      "step": 14125
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0312226356536837e-07,
+      "loss": 0.4015,
+      "step": 14126
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0280864546593161e-07,
+      "loss": 0.3776,
+      "step": 14127
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0249550252067863e-07,
+      "loss": 0.4333,
+      "step": 14128
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0218283474464518e-07,
+      "loss": 0.4061,
+      "step": 14129
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0187064215284037e-07,
+      "loss": 0.7002,
+      "step": 14130
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.015589247602522e-07,
+      "loss": 0.7182,
+      "step": 14131
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.012476825818487e-07,
+      "loss": 0.7089,
+      "step": 14132
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0093691563256902e-07,
+      "loss": 0.3771,
+      "step": 14133
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0062662392733458e-07,
+      "loss": 0.6998,
+      "step": 14134
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0031680748104122e-07,
+      "loss": 0.7239,
+      "step": 14135
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0000746630856484e-07,
+      "loss": 0.389,
+      "step": 14136
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.969860042475354e-08,
+      "loss": 0.7091,
+      "step": 14137
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.93902098444377e-08,
+      "loss": 0.7037,
+      "step": 14138
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.908229458242214e-08,
+      "loss": 0.7776,
+      "step": 14139
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.877485465349057e-08,
+      "loss": 0.3781,
+      "step": 14140
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.84678900724012e-08,
+      "loss": 0.7046,
+      "step": 14141
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.816140085389337e-08,
+      "loss": 0.7202,
+      "step": 14142
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.78553870126786e-08,
+      "loss": 0.7363,
+      "step": 14143
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.754984856345074e-08,
+      "loss": 0.7498,
+      "step": 14144
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.724478552087691e-08,
+      "loss": 0.7014,
+      "step": 14145
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.694019789960318e-08,
+      "loss": 0.7029,
+      "step": 14146
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.663608571425231e-08,
+      "loss": 0.7649,
+      "step": 14147
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.633244897942485e-08,
+      "loss": 0.7236,
+      "step": 14148
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.602928770969799e-08,
+      "loss": 0.7273,
+      "step": 14149
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.57266019196268e-08,
+      "loss": 0.6598,
+      "step": 14150
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.542439162374184e-08,
+      "loss": 0.7429,
+      "step": 14151
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.512265683655264e-08,
+      "loss": 0.7472,
+      "step": 14152
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.482139757254427e-08,
+      "loss": 0.7463,
+      "step": 14153
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.452061384618072e-08,
+      "loss": 0.3515,
+      "step": 14154
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.422030567190265e-08,
+      "loss": 0.7304,
+      "step": 14155
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.392047306412633e-08,
+      "loss": 0.705,
+      "step": 14156
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.362111603724688e-08,
+      "loss": 0.7102,
+      "step": 14157
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.332223460563727e-08,
+      "loss": 0.7064,
+      "step": 14158
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.30238287836449e-08,
+      "loss": 0.3921,
+      "step": 14159
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.272589858559721e-08,
+      "loss": 0.4306,
+      "step": 14160
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.242844402579609e-08,
+      "loss": 0.7056,
+      "step": 14161
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.213146511852344e-08,
+      "loss": 0.7178,
+      "step": 14162
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.183496187803564e-08,
+      "loss": 0.7058,
+      "step": 14163
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.153893431856909e-08,
+      "loss": 0.73,
+      "step": 14164
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.124338245433573e-08,
+      "loss": 0.69,
+      "step": 14165
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.094830629952312e-08,
+      "loss": 0.7163,
+      "step": 14166
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.06537058682977e-08,
+      "loss": 0.4366,
+      "step": 14167
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.03595811748037e-08,
+      "loss": 0.3382,
+      "step": 14168
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.006593223316207e-08,
+      "loss": 0.3686,
+      "step": 14169
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.977275905747041e-08,
+      "loss": 0.6986,
+      "step": 14170
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.94800616618019e-08,
+      "loss": 0.3803,
+      "step": 14171
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.918784006021198e-08,
+      "loss": 0.7593,
+      "step": 14172
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.889609426672718e-08,
+      "loss": 0.7261,
+      "step": 14173
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.860482429535522e-08,
+      "loss": 0.7249,
+      "step": 14174
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.831403016007934e-08,
+      "loss": 0.687,
+      "step": 14175
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.802371187485948e-08,
+      "loss": 0.4161,
+      "step": 14176
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.773386945363449e-08,
+      "loss": 0.7293,
+      "step": 14177
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.744450291031992e-08,
+      "loss": 0.7061,
+      "step": 14178
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.715561225880687e-08,
+      "loss": 0.6799,
+      "step": 14179
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.686719751296536e-08,
+      "loss": 0.7219,
+      "step": 14180
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.657925868664208e-08,
+      "loss": 0.7443,
+      "step": 14181
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.629179579365931e-08,
+      "loss": 0.77,
+      "step": 14182
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.600480884781825e-08,
+      "loss": 0.7109,
+      "step": 14183
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.571829786289898e-08,
+      "loss": 0.6975,
+      "step": 14184
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.543226285265271e-08,
+      "loss": 0.7022,
+      "step": 14185
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.514670383081513e-08,
+      "loss": 0.713,
+      "step": 14186
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.486162081109528e-08,
+      "loss": 0.3949,
+      "step": 14187
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.457701380717776e-08,
+      "loss": 0.7187,
+      "step": 14188
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.429288283272829e-08,
+      "loss": 0.3564,
+      "step": 14189
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.400922790138599e-08,
+      "loss": 0.6711,
+      "step": 14190
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.372604902676996e-08,
+      "loss": 0.6581,
+      "step": 14191
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.344334622247486e-08,
+      "loss": 0.737,
+      "step": 14192
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.316111950207429e-08,
+      "loss": 0.6958,
+      "step": 14193
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.287936887911519e-08,
+      "loss": 0.6743,
+      "step": 14194
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.259809436712673e-08,
+      "loss": 0.4167,
+      "step": 14195
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.231729597961035e-08,
+      "loss": 0.4182,
+      "step": 14196
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.20369737300486e-08,
+      "loss": 0.7031,
+      "step": 14197
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.175712763189736e-08,
+      "loss": 0.7524,
+      "step": 14198
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.147775769859478e-08,
+      "loss": 0.432,
+      "step": 14199
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.119886394355126e-08,
+      "loss": 0.4059,
+      "step": 14200
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.092044638015607e-08,
+      "loss": 0.6936,
+      "step": 14201
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.064250502177628e-08,
+      "loss": 0.7003,
+      "step": 14202
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.036503988175459e-08,
+      "loss": 0.4054,
+      "step": 14203
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.008805097341366e-08,
+      "loss": 0.7399,
+      "step": 14204
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.981153831005062e-08,
+      "loss": 0.7234,
+      "step": 14205
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.953550190494042e-08,
+      "loss": 0.6945,
+      "step": 14206
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.925994177133467e-08,
+      "loss": 0.7043,
+      "step": 14207
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.898485792246279e-08,
+      "loss": 0.4258,
+      "step": 14208
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.871025037153312e-08,
+      "loss": 0.422,
+      "step": 14209
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.843611913172844e-08,
+      "loss": 0.7186,
+      "step": 14210
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.816246421620821e-08,
+      "loss": 0.7209,
+      "step": 14211
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.788928563811082e-08,
+      "loss": 0.7579,
+      "step": 14212
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.761658341055244e-08,
+      "loss": 0.7111,
+      "step": 14213
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.734435754662373e-08,
+      "loss": 0.6807,
+      "step": 14214
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.707260805939531e-08,
+      "loss": 0.6492,
+      "step": 14215
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.680133496191344e-08,
+      "loss": 0.7282,
+      "step": 14216
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.653053826719992e-08,
+      "loss": 0.7001,
+      "step": 14217
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.626021798825767e-08,
+      "loss": 0.7263,
+      "step": 14218
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.599037413806298e-08,
+      "loss": 0.4031,
+      "step": 14219
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.572100672957216e-08,
+      "loss": 0.7018,
+      "step": 14220
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.545211577571599e-08,
+      "loss": 0.6994,
+      "step": 14221
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.518370128940411e-08,
+      "loss": 0.6846,
+      "step": 14222
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.49157632835229e-08,
+      "loss": 0.3869,
+      "step": 14223
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.464830177093651e-08,
+      "loss": 0.3779,
+      "step": 14224
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.438131676448468e-08,
+      "loss": 0.7133,
+      "step": 14225
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.41148082769838e-08,
+      "loss": 0.4153,
+      "step": 14226
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.384877632123033e-08,
+      "loss": 0.7573,
+      "step": 14227
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.358322090999736e-08,
+      "loss": 0.7408,
+      "step": 14228
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.331814205603138e-08,
+      "loss": 0.3952,
+      "step": 14229
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.305353977205887e-08,
+      "loss": 0.7256,
+      "step": 14230
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.278941407078522e-08,
+      "loss": 0.7162,
+      "step": 14231
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.252576496488917e-08,
+      "loss": 0.7582,
+      "step": 14232
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.226259246702838e-08,
+      "loss": 0.6884,
+      "step": 14233
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.19998965898383e-08,
+      "loss": 0.7768,
+      "step": 14234
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.173767734592885e-08,
+      "loss": 0.7252,
+      "step": 14235
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.147593474789105e-08,
+      "loss": 0.706,
+      "step": 14236
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.121466880829042e-08,
+      "loss": 0.3802,
+      "step": 14237
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.095387953966915e-08,
+      "loss": 0.6985,
+      "step": 14238
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.069356695454832e-08,
+      "loss": 0.6868,
+      "step": 14239
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.043373106542461e-08,
+      "loss": 0.7229,
+      "step": 14240
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.01743718847725e-08,
+      "loss": 0.68,
+      "step": 14241
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.991548942504423e-08,
+      "loss": 0.6893,
+      "step": 14242
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.965708369866874e-08,
+      "loss": 0.7388,
+      "step": 14243
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.939915471804947e-08,
+      "loss": 0.7143,
+      "step": 14244
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.914170249557317e-08,
+      "loss": 0.7522,
+      "step": 14245
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.888472704359661e-08,
+      "loss": 0.688,
+      "step": 14246
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.862822837445882e-08,
+      "loss": 0.7287,
+      "step": 14247
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.837220650047216e-08,
+      "loss": 0.3745,
+      "step": 14248
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.811666143393014e-08,
+      "loss": 0.4002,
+      "step": 14249
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.786159318709962e-08,
+      "loss": 0.4048,
+      "step": 14250
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.760700177222745e-08,
+      "loss": 0.7153,
+      "step": 14251
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.7352887201535e-08,
+      "loss": 0.4137,
+      "step": 14252
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.709924948722246e-08,
+      "loss": 0.6923,
+      "step": 14253
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.684608864146791e-08,
+      "loss": 0.4128,
+      "step": 14254
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.659340467642384e-08,
+      "loss": 0.437,
+      "step": 14255
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.634119760422165e-08,
+      "loss": 0.6931,
+      "step": 14256
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.608946743696943e-08,
+      "loss": 0.374,
+      "step": 14257
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.583821418675417e-08,
+      "loss": 0.7325,
+      "step": 14258
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.558743786563627e-08,
+      "loss": 0.7693,
+      "step": 14259
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.533713848565604e-08,
+      "loss": 0.7405,
+      "step": 14260
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.508731605882946e-08,
+      "loss": 0.7245,
+      "step": 14261
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.483797059715136e-08,
+      "loss": 0.3897,
+      "step": 14262
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.458910211259217e-08,
+      "loss": 0.7399,
+      "step": 14263
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.434071061709901e-08,
+      "loss": 0.7234,
+      "step": 14264
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.409279612259789e-08,
+      "loss": 0.7219,
+      "step": 14265
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.38453586409904e-08,
+      "loss": 0.7054,
+      "step": 14266
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.359839818415592e-08,
+      "loss": 0.6921,
+      "step": 14267
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.335191476395053e-08,
+      "loss": 0.7692,
+      "step": 14268
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.31059083922092e-08,
+      "loss": 0.6685,
+      "step": 14269
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.286037908074027e-08,
+      "loss": 0.764,
+      "step": 14270
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.26153268413321e-08,
+      "loss": 0.722,
+      "step": 14271
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.237075168574969e-08,
+      "loss": 0.4431,
+      "step": 14272
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.212665362573477e-08,
+      "loss": 0.7199,
+      "step": 14273
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.188303267300689e-08,
+      "loss": 0.4362,
+      "step": 14274
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.163988883925997e-08,
+      "loss": 0.3966,
+      "step": 14275
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.139722213616916e-08,
+      "loss": 0.7532,
+      "step": 14276
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.115503257538402e-08,
+      "loss": 0.7203,
+      "step": 14277
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.09133201685319e-08,
+      "loss": 0.7059,
+      "step": 14278
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.067208492721687e-08,
+      "loss": 0.6711,
+      "step": 14279
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.043132686302077e-08,
+      "loss": 0.3886,
+      "step": 14280
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.019104598750103e-08,
+      "loss": 0.6834,
+      "step": 14281
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.995124231219507e-08,
+      "loss": 0.6818,
+      "step": 14282
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.97119158486148e-08,
+      "loss": 0.3579,
+      "step": 14283
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.947306660824992e-08,
+      "loss": 0.7223,
+      "step": 14284
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.9234694602566814e-08,
+      "loss": 0.7512,
+      "step": 14285
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.8996799843010766e-08,
+      "loss": 0.3904,
+      "step": 14286
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.8759382341001534e-08,
+      "loss": 0.6864,
+      "step": 14287
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.852244210793889e-08,
+      "loss": 0.3766,
+      "step": 14288
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.828597915519596e-08,
+      "loss": 0.7168,
+      "step": 14289
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.804999349412699e-08,
+      "loss": 0.7355,
+      "step": 14290
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.781448513606069e-08,
+      "loss": 0.4224,
+      "step": 14291
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.757945409230359e-08,
+      "loss": 0.7411,
+      "step": 14292
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.734490037413887e-08,
+      "loss": 0.7224,
+      "step": 14293
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.711082399282752e-08,
+      "loss": 0.7118,
+      "step": 14294
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.687722495960834e-08,
+      "loss": 0.751,
+      "step": 14295
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.664410328569458e-08,
+      "loss": 0.7351,
+      "step": 14296
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.641145898227951e-08,
+      "loss": 0.7192,
+      "step": 14297
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.617929206053086e-08,
+      "loss": 0.383,
+      "step": 14298
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.594760253159637e-08,
+      "loss": 0.6608,
+      "step": 14299
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.5716390406597155e-08,
+      "loss": 0.74,
+      "step": 14300
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.548565569663433e-08,
+      "loss": 0.3806,
+      "step": 14301
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.525539841278682e-08,
+      "loss": 0.7588,
+      "step": 14302
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.502561856610578e-08,
+      "loss": 0.4408,
+      "step": 14303
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.479631616762682e-08,
+      "loss": 0.6994,
+      "step": 14304
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.4567491228354474e-08,
+      "loss": 0.3916,
+      "step": 14305
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.433914375927774e-08,
+      "loss": 0.7212,
+      "step": 14306
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.411127377135672e-08,
+      "loss": 0.3717,
+      "step": 14307
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.388388127553268e-08,
+      "loss": 0.7021,
+      "step": 14308
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.3656966282722437e-08,
+      "loss": 0.7137,
+      "step": 14309
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.3430528803819496e-08,
+      "loss": 0.6979,
+      "step": 14310
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.320456884969405e-08,
+      "loss": 0.667,
+      "step": 14311
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.297908643119631e-08,
+      "loss": 0.6982,
+      "step": 14312
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.275408155914985e-08,
+      "loss": 0.7339,
+      "step": 14313
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.252955424435935e-08,
+      "loss": 0.7088,
+      "step": 14314
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.2305504497600635e-08,
+      "loss": 0.6858,
+      "step": 14315
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.2081932329631766e-08,
+      "loss": 0.736,
+      "step": 14316
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.1858837751186384e-08,
+      "loss": 0.7088,
+      "step": 14317
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.163622077297481e-08,
+      "loss": 0.3936,
+      "step": 14318
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.1414081405684044e-08,
+      "loss": 0.7143,
+      "step": 14319
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.1192419659980005e-08,
+      "loss": 0.6723,
+      "step": 14320
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.097123554650418e-08,
+      "loss": 0.3872,
+      "step": 14321
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.0750529075874744e-08,
+      "loss": 0.7343,
+      "step": 14322
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.053030025868766e-08,
+      "loss": 0.3945,
+      "step": 14323
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.031054910551669e-08,
+      "loss": 0.7236,
+      "step": 14324
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.0091275626911187e-08,
+      "loss": 0.7168,
+      "step": 14325
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.9872479833398266e-08,
+      "loss": 0.6506,
+      "step": 14326
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.965416173548176e-08,
+      "loss": 0.7143,
+      "step": 14327
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.94363213436444e-08,
+      "loss": 0.74,
+      "step": 14328
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.9218958668343365e-08,
+      "loss": 0.4122,
+      "step": 14329
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.900207372001253e-08,
+      "loss": 0.7195,
+      "step": 14330
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.878566650906691e-08,
+      "loss": 0.3894,
+      "step": 14331
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.856973704589596e-08,
+      "loss": 0.6927,
+      "step": 14332
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.835428534086361e-08,
+      "loss": 0.7099,
+      "step": 14333
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.8139311404316045e-08,
+      "loss": 0.6468,
+      "step": 14334
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.792481524657278e-08,
+      "loss": 0.6977,
+      "step": 14335
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.771079687793112e-08,
+      "loss": 0.6732,
+      "step": 14336
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.7497256308668415e-08,
+      "loss": 0.703,
+      "step": 14337
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.728419354903313e-08,
+      "loss": 0.6838,
+      "step": 14338
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.707160860925597e-08,
+      "loss": 0.6965,
+      "step": 14339
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.68595014995421e-08,
+      "loss": 0.3998,
+      "step": 14340
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.664787223007672e-08,
+      "loss": 0.7758,
+      "step": 14341
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.643672081101724e-08,
+      "loss": 0.411,
+      "step": 14342
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.622604725250224e-08,
+      "loss": 0.74,
+      "step": 14343
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.601585156464583e-08,
+      "loss": 0.706,
+      "step": 14344
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.580613375753884e-08,
+      "loss": 0.749,
+      "step": 14345
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.559689384124988e-08,
+      "loss": 0.7453,
+      "step": 14346
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.5388131825825354e-08,
+      "loss": 0.7541,
+      "step": 14347
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.517984772128503e-08,
+      "loss": 0.3608,
+      "step": 14348
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.497204153763091e-08,
+      "loss": 0.6761,
+      "step": 14349
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.476471328483945e-08,
+      "loss": 0.6822,
+      "step": 14350
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.4557862972862686e-08,
+      "loss": 0.7017,
+      "step": 14351
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.4351490611632686e-08,
+      "loss": 0.7452,
+      "step": 14352
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.414559621105596e-08,
+      "loss": 0.6898,
+      "step": 14353
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.394017978101905e-08,
+      "loss": 0.4107,
+      "step": 14354
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.373524133138185e-08,
+      "loss": 0.7119,
+      "step": 14355
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.353078087198537e-08,
+      "loss": 0.4107,
+      "step": 14356
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.332679841264287e-08,
+      "loss": 0.744,
+      "step": 14357
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.312329396314985e-08,
+      "loss": 0.6988,
+      "step": 14358
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.292026753327516e-08,
+      "loss": 0.7106,
+      "step": 14359
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.271771913276546e-08,
+      "loss": 0.7065,
+      "step": 14360
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.251564877134629e-08,
+      "loss": 0.7352,
+      "step": 14361
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.231405645871767e-08,
+      "loss": 0.6766,
+      "step": 14362
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.2112942204557414e-08,
+      "loss": 0.7052,
+      "step": 14363
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.191230601852336e-08,
+      "loss": 0.6966,
+      "step": 14364
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.171214791024447e-08,
+      "loss": 0.7343,
+      "step": 14365
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.151246788933305e-08,
+      "loss": 0.6958,
+      "step": 14366
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.131326596537255e-08,
+      "loss": 0.7182,
+      "step": 14367
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.111454214792976e-08,
+      "loss": 0.6982,
+      "step": 14368
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.0916296446542603e-08,
+      "loss": 0.6907,
+      "step": 14369
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.0718528870730134e-08,
+      "loss": 0.7474,
+      "step": 14370
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.052123942998698e-08,
+      "loss": 0.6822,
+      "step": 14371
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.0324428133783343e-08,
+      "loss": 0.3998,
+      "step": 14372
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.0128094991570554e-08,
+      "loss": 0.3732,
+      "step": 14373
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.9932240012771073e-08,
+      "loss": 0.7915,
+      "step": 14374
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.97368632067896e-08,
+      "loss": 0.7226,
+      "step": 14375
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.954196458300641e-08,
+      "loss": 0.7431,
+      "step": 14376
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.9347544150777353e-08,
+      "loss": 0.6985,
+      "step": 14377
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.915360191943607e-08,
+      "loss": 0.6684,
+      "step": 14378
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.8960137898295117e-08,
+      "loss": 0.6893,
+      "step": 14379
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.8767152096641504e-08,
+      "loss": 0.7577,
+      "step": 14380
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.8574644523740044e-08,
+      "loss": 0.4252,
+      "step": 14381
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.838261518883335e-08,
+      "loss": 0.4336,
+      "step": 14382
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.8191064101139593e-08,
+      "loss": 0.7018,
+      "step": 14383
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.7999991269856984e-08,
+      "loss": 0.653,
+      "step": 14384
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.780939670415595e-08,
+      "loss": 0.7083,
+      "step": 14385
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.761928041318919e-08,
+      "loss": 0.6949,
+      "step": 14386
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.7429642406083824e-08,
+      "loss": 0.7536,
+      "step": 14387
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.724048269194258e-08,
+      "loss": 0.3726,
+      "step": 14388
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.7051801279847085e-08,
+      "loss": 0.7726,
+      "step": 14389
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.686359817885787e-08,
+      "loss": 0.4076,
+      "step": 14390
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.667587339800771e-08,
+      "loss": 0.3584,
+      "step": 14391
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.64886269463105e-08,
+      "loss": 0.7526,
+      "step": 14392
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.630185883275572e-08,
+      "loss": 0.6521,
+      "step": 14393
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.611556906630842e-08,
+      "loss": 0.7156,
+      "step": 14394
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.5929757655914776e-08,
+      "loss": 0.6988,
+      "step": 14395
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.574442461049321e-08,
+      "loss": 0.381,
+      "step": 14396
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.5559569938942163e-08,
+      "loss": 0.7337,
+      "step": 14397
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.537519365013675e-08,
+      "loss": 0.718,
+      "step": 14398
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.519129575292879e-08,
+      "loss": 0.7865,
+      "step": 14399
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.5007876256145653e-08,
+      "loss": 0.7827,
+      "step": 14400
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.482493516859475e-08,
+      "loss": 0.7348,
+      "step": 14401
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.464247249905795e-08,
+      "loss": 0.7076,
+      "step": 14402
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.446048825629489e-08,
+      "loss": 0.7264,
+      "step": 14403
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.427898244904304e-08,
+      "loss": 0.7218,
+      "step": 14404
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.409795508601654e-08,
+      "loss": 0.3763,
+      "step": 14405
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.391740617590622e-08,
+      "loss": 0.6774,
+      "step": 14406
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.373733572737958e-08,
+      "loss": 0.7415,
+      "step": 14407
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.355774374908083e-08,
+      "loss": 0.7163,
+      "step": 14408
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.337863024963417e-08,
+      "loss": 0.6994,
+      "step": 14409
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.319999523763717e-08,
+      "loss": 0.6966,
+      "step": 14410
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.302183872166631e-08,
+      "loss": 0.6789,
+      "step": 14411
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.284416071027363e-08,
+      "loss": 0.3669,
+      "step": 14412
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.26669612119912e-08,
+      "loss": 0.6965,
+      "step": 14413
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.2490240235325544e-08,
+      "loss": 0.4317,
+      "step": 14414
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.2313997788759876e-08,
+      "loss": 0.7289,
+      "step": 14415
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.213823388075632e-08,
+      "loss": 0.7214,
+      "step": 14416
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.196294851975257e-08,
+      "loss": 0.388,
+      "step": 14417
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.1788141714165223e-08,
+      "loss": 0.4518,
+      "step": 14418
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.161381347238535e-08,
+      "loss": 0.3782,
+      "step": 14419
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.14399638027818e-08,
+      "loss": 0.7247,
+      "step": 14420
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.126659271370236e-08,
+      "loss": 0.7021,
+      "step": 14421
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.109370021346925e-08,
+      "loss": 0.6979,
+      "step": 14422
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.0921286310383605e-08,
+      "loss": 0.7815,
+      "step": 14423
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.074935101272214e-08,
+      "loss": 0.7839,
+      "step": 14424
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.057789432874048e-08,
+      "loss": 0.4025,
+      "step": 14425
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.04069162666687e-08,
+      "loss": 0.7358,
+      "step": 14426
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.0236416834715785e-08,
+      "loss": 0.7022,
+      "step": 14427
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.0066396041067426e-08,
+      "loss": 0.7304,
+      "step": 14428
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.989685389388597e-08,
+      "loss": 0.6431,
+      "step": 14429
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9727790401311574e-08,
+      "loss": 0.745,
+      "step": 14430
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.955920557145997e-08,
+      "loss": 0.6934,
+      "step": 14431
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9391099412424683e-08,
+      "loss": 0.7796,
+      "step": 14432
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9223471932277037e-08,
+      "loss": 0.4292,
+      "step": 14433
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9056323139065035e-08,
+      "loss": 0.4218,
+      "step": 14434
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.888965304081115e-08,
+      "loss": 0.6724,
+      "step": 14435
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.872346164552009e-08,
+      "loss": 0.7415,
+      "step": 14436
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8557748961167698e-08,
+      "loss": 0.7057,
+      "step": 14437
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.839251499571205e-08,
+      "loss": 0.709,
+      "step": 14438
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8227759757083474e-08,
+      "loss": 0.7232,
+      "step": 14439
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8063483253193414e-08,
+      "loss": 0.7355,
+      "step": 14440
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7899685491928897e-08,
+      "loss": 0.4004,
+      "step": 14441
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.773636648115252e-08,
+      "loss": 0.6972,
+      "step": 14442
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7573526228704683e-08,
+      "loss": 0.6964,
+      "step": 14443
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7411164742404684e-08,
+      "loss": 0.7356,
+      "step": 14444
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7249282030047397e-08,
+      "loss": 0.734,
+      "step": 14445
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7087878099403275e-08,
+      "loss": 0.6761,
+      "step": 14446
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6926952958221673e-08,
+      "loss": 0.726,
+      "step": 14447
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6766506614228637e-08,
+      "loss": 0.3594,
+      "step": 14448
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6606539075126892e-08,
+      "loss": 0.7069,
+      "step": 14449
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.644705034859696e-08,
+      "loss": 0.7358,
+      "step": 14450
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6288040442293827e-08,
+      "loss": 0.6766,
+      "step": 14451
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6129509363853612e-08,
+      "loss": 0.6889,
+      "step": 14452
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.597145712088689e-08,
+      "loss": 0.7091,
+      "step": 14453
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5813883720979815e-08,
+      "loss": 0.4516,
+      "step": 14454
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.565678917169967e-08,
+      "loss": 0.7519,
+      "step": 14455
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5500173480587088e-08,
+      "loss": 0.7232,
+      "step": 14456
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5344036655161608e-08,
+      "loss": 0.7304,
+      "step": 14457
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5188378702917238e-08,
+      "loss": 0.6742,
+      "step": 14458
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5033199631330217e-08,
+      "loss": 0.6963,
+      "step": 14459
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4878499447847925e-08,
+      "loss": 0.7027,
+      "step": 14460
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4724278159898863e-08,
+      "loss": 0.7135,
+      "step": 14461
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4570535774887104e-08,
+      "loss": 0.6931,
+      "step": 14462
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4417272300192308e-08,
+      "loss": 0.6736,
+      "step": 14463
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4264487743173027e-08,
+      "loss": 0.7083,
+      "step": 14464
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4112182111164508e-08,
+      "loss": 0.7297,
+      "step": 14465
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3960355411478674e-08,
+      "loss": 0.6722,
+      "step": 14466
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.380900765140526e-08,
+      "loss": 0.7082,
+      "step": 14467
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3658138838209554e-08,
+      "loss": 0.6704,
+      "step": 14468
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.350774897913466e-08,
+      "loss": 0.7404,
+      "step": 14469
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3357838081401462e-08,
+      "loss": 0.7564,
+      "step": 14470
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.320840615220532e-08,
+      "loss": 0.7448,
+      "step": 14471
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3059453198721603e-08,
+      "loss": 0.7414,
+      "step": 14472
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.291097922810237e-08,
+      "loss": 0.7631,
+      "step": 14473
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.276298424747303e-08,
+      "loss": 0.3994,
+      "step": 14474
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.261546826394123e-08,
+      "loss": 0.6676,
+      "step": 14475
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2468431284586866e-08,
+      "loss": 0.3677,
+      "step": 14476
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2321873316470953e-08,
+      "loss": 0.7191,
+      "step": 14477
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2175794366628977e-08,
+      "loss": 0.4165,
+      "step": 14478
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2030194442073104e-08,
+      "loss": 0.7394,
+      "step": 14479
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1885073549794412e-08,
+      "loss": 0.705,
+      "step": 14480
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1740431696759545e-08,
+      "loss": 0.7504,
+      "step": 14481
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.159626888991295e-08,
+      "loss": 0.3916,
+      "step": 14482
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1452585136175763e-08,
+      "loss": 0.6983,
+      "step": 14483
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1309380442445792e-08,
+      "loss": 0.7211,
+      "step": 14484
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1166654815598654e-08,
+      "loss": 0.6606,
+      "step": 14485
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.102440826248553e-08,
+      "loss": 0.7126,
+      "step": 14486
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0882640789936514e-08,
+      "loss": 0.7406,
+      "step": 14487
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0741352404758386e-08,
+      "loss": 0.4234,
+      "step": 14488
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0600543113731276e-08,
+      "loss": 0.6715,
+      "step": 14489
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.046021292361866e-08,
+      "loss": 0.4352,
+      "step": 14490
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0320361841155156e-08,
+      "loss": 0.7042,
+      "step": 14491
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.01809898730565e-08,
+      "loss": 0.6983,
+      "step": 14492
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0042097026014006e-08,
+      "loss": 0.7442,
+      "step": 14493
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9903683306693457e-08,
+      "loss": 0.7097,
+      "step": 14494
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.976574872174286e-08,
+      "loss": 0.7026,
+      "step": 14495
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9628293277782483e-08,
+      "loss": 0.6886,
+      "step": 14496
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.949131698141149e-08,
+      "loss": 0.4215,
+      "step": 14497
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.935481983920684e-08,
+      "loss": 0.6754,
+      "step": 14498
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.921880185771996e-08,
+      "loss": 0.4203,
+      "step": 14499
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9083263043483403e-08,
+      "loss": 0.7153,
+      "step": 14500
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8948203403003073e-08,
+      "loss": 0.3533,
+      "step": 14501
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8813622942762676e-08,
+      "loss": 0.6721,
+      "step": 14502
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.867952166922371e-08,
+      "loss": 0.3522,
+      "step": 14503
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8545899588823245e-08,
+      "loss": 0.7103,
+      "step": 14504
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8412756707978373e-08,
+      "loss": 0.7124,
+      "step": 14505
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8280093033079537e-08,
+      "loss": 0.6998,
+      "step": 14506
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8147908570496088e-08,
+      "loss": 0.4125,
+      "step": 14507
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8016203326575167e-08,
+      "loss": 0.6729,
+      "step": 14508
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7884977307637275e-08,
+      "loss": 0.7083,
+      "step": 14509
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.775423051998515e-08,
+      "loss": 0.6653,
+      "step": 14510
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7623962969894882e-08,
+      "loss": 0.7162,
+      "step": 14511
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7494174663619245e-08,
+      "loss": 0.7252,
+      "step": 14512
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7364865607391035e-08,
+      "loss": 0.7229,
+      "step": 14513
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7236035807416397e-08,
+      "loss": 0.7034,
+      "step": 14514
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7107685269882603e-08,
+      "loss": 0.7217,
+      "step": 14515
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.697981400095028e-08,
+      "loss": 0.7291,
+      "step": 14516
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.685242200675785e-08,
+      "loss": 0.7388,
+      "step": 14517
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6725509293422648e-08,
+      "loss": 0.7137,
+      "step": 14518
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6599075867037572e-08,
+      "loss": 0.7322,
+      "step": 14519
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6473121733671104e-08,
+      "loss": 0.7472,
+      "step": 14520
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.634764689937174e-08,
+      "loss": 0.7323,
+      "step": 14521
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6222651370161323e-08,
+      "loss": 0.7036,
+      "step": 14522
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.609813515204395e-08,
+      "loss": 0.6578,
+      "step": 14523
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.597409825099483e-08,
+      "loss": 0.3812,
+      "step": 14524
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5850540672969205e-08,
+      "loss": 0.7035,
+      "step": 14525
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5727462423900108e-08,
+      "loss": 0.7454,
+      "step": 14526
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5604863509695034e-08,
+      "loss": 0.7102,
+      "step": 14527
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5482743936241496e-08,
+      "loss": 0.7287,
+      "step": 14528
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.536110370940147e-08,
+      "loss": 0.7153,
+      "step": 14529
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.523994283501362e-08,
+      "loss": 0.7278,
+      "step": 14530
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5119261318896627e-08,
+      "loss": 0.7148,
+      "step": 14531
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4999059166843633e-08,
+      "loss": 0.3907,
+      "step": 14532
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4879336384624466e-08,
+      "loss": 0.6697,
+      "step": 14533
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4760092977988971e-08,
+      "loss": 0.731,
+      "step": 14534
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.464132895265924e-08,
+      "loss": 0.7749,
+      "step": 14535
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4523044314339596e-08,
+      "loss": 0.7212,
+      "step": 14536
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4405239068707722e-08,
+      "loss": 0.7124,
+      "step": 14537
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4287913221417982e-08,
+      "loss": 0.4141,
+      "step": 14538
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4171066778104759e-08,
+      "loss": 0.715,
+      "step": 14539
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.405469974437801e-08,
+      "loss": 0.6928,
+      "step": 14540
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3938812125823265e-08,
+      "loss": 0.3889,
+      "step": 14541
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3823403928004964e-08,
+      "loss": 0.7486,
+      "step": 14542
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3708475156462009e-08,
+      "loss": 0.706,
+      "step": 14543
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3594025816714429e-08,
+      "loss": 0.3718,
+      "step": 14544
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3480055914255608e-08,
+      "loss": 0.7552,
+      "step": 14545
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3366565454557834e-08,
+      "loss": 0.687,
+      "step": 14546
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3253554443067863e-08,
+      "loss": 0.6937,
+      "step": 14547
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3141022885213573e-08,
+      "loss": 0.7373,
+      "step": 14548
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.302897078639509e-08,
+      "loss": 0.7226,
+      "step": 14549
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2917398151994775e-08,
+      "loss": 0.4095,
+      "step": 14550
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2806304987366125e-08,
+      "loss": 0.7612,
+      "step": 14551
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.269569129784487e-08,
+      "loss": 0.6738,
+      "step": 14552
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2585557088741207e-08,
+      "loss": 0.7382,
+      "step": 14553
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2475902365340908e-08,
+      "loss": 0.7458,
+      "step": 14554
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2366727132909761e-08,
+      "loss": 0.7121,
+      "step": 14555
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.225803139668913e-08,
+      "loss": 0.7153,
+      "step": 14556
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.214981516189706e-08,
+      "loss": 0.4014,
+      "step": 14557
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2042078433729398e-08,
+      "loss": 0.7155,
+      "step": 14558
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1934821217357561e-08,
+      "loss": 0.3931,
+      "step": 14559
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1828043517931875e-08,
+      "loss": 0.7315,
+      "step": 14560
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1721745340577129e-08,
+      "loss": 0.6954,
+      "step": 14561
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1615926690398127e-08,
+      "loss": 0.6951,
+      "step": 14562
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1510587572474141e-08,
+      "loss": 0.6852,
+      "step": 14563
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1405727991863348e-08,
+      "loss": 0.4121,
+      "step": 14564
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1301347953599495e-08,
+      "loss": 0.3837,
+      "step": 14565
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1197447462694133e-08,
+      "loss": 0.3774,
+      "step": 14566
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1094026524134383e-08,
+      "loss": 0.7819,
+      "step": 14567
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.099108514288627e-08,
+      "loss": 0.7205,
+      "step": 14568
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0888623323891401e-08,
+      "loss": 0.7516,
+      "step": 14569
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.078664107206806e-08,
+      "loss": 0.7213,
+      "step": 14570
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0685138392314553e-08,
+      "loss": 0.7177,
+      "step": 14571
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.058411528950254e-08,
+      "loss": 0.7604,
+      "step": 14572
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0483571768481471e-08,
+      "loss": 0.4502,
+      "step": 14573
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.038350783407971e-08,
+      "loss": 0.4622,
+      "step": 14574
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.028392349110119e-08,
+      "loss": 0.6512,
+      "step": 14575
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.018481874432542e-08,
+      "loss": 0.7518,
+      "step": 14576
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0086193598510819e-08,
+      "loss": 0.7109,
+      "step": 14577
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.988048058393596e-09,
+      "loss": 0.6767,
+      "step": 14578
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.890382128684429e-09,
+      "loss": 0.7298,
+      "step": 14579
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.793195814071788e-09,
+      "loss": 0.7397,
+      "step": 14580
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.696489119221942e-09,
+      "loss": 0.7221,
+      "step": 14581
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.600262048778952e-09,
+      "loss": 0.7342,
+      "step": 14582
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.504514607360237e-09,
+      "loss": 0.3591,
+      "step": 14583
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.409246799564342e-09,
+      "loss": 0.6989,
+      "step": 14584
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.314458629964274e-09,
+      "loss": 0.745,
+      "step": 14585
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.220150103110837e-09,
+      "loss": 0.7238,
+      "step": 14586
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.12632122353152e-09,
+      "loss": 0.6861,
+      "step": 14587
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.032971995730499e-09,
+      "loss": 0.7323,
+      "step": 14588
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.940102424190855e-09,
+      "loss": 0.3997,
+      "step": 14589
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.847712513370132e-09,
+      "loss": 0.6836,
+      "step": 14590
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.755802267703673e-09,
+      "loss": 0.406,
+      "step": 14591
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.664371691604611e-09,
+      "loss": 0.4004,
+      "step": 14592
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.573420789462771e-09,
+      "loss": 0.6903,
+      "step": 14593
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.48294956564355e-09,
+      "loss": 0.4126,
+      "step": 14594
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.392958024490139e-09,
+      "loss": 0.3849,
+      "step": 14595
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.303446170324636e-09,
+      "loss": 0.7187,
+      "step": 14596
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.214414007442494e-09,
+      "loss": 0.6711,
+      "step": 14597
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.125861540120295e-09,
+      "loss": 0.7204,
+      "step": 14598
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.037788772606858e-09,
+      "loss": 0.7398,
+      "step": 14599
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.950195709132136e-09,
+      "loss": 0.6943,
+      "step": 14600
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.863082353900542e-09,
+      "loss": 0.4312,
+      "step": 14601
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.776448711095397e-09,
+      "loss": 0.6853,
+      "step": 14602
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.690294784874486e-09,
+      "loss": 0.7064,
+      "step": 14603
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.6046205793745e-09,
+      "loss": 0.7141,
+      "step": 14604
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.519426098709925e-09,
+      "loss": 0.3819,
+      "step": 14605
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.434711346968604e-09,
+      "loss": 0.7463,
+      "step": 14606
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.350476328218392e-09,
+      "loss": 0.4157,
+      "step": 14607
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.266721046504943e-09,
+      "loss": 0.7013,
+      "step": 14608
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.183445505846154e-09,
+      "loss": 0.7037,
+      "step": 14609
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.100649710243268e-09,
+      "loss": 0.7044,
+      "step": 14610
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.018333663669774e-09,
+      "loss": 0.3807,
+      "step": 14611
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.936497370075845e-09,
+      "loss": 0.7386,
+      "step": 14612
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.855140833393892e-09,
+      "loss": 0.6939,
+      "step": 14613
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.7742640575263474e-09,
+      "loss": 0.3916,
+      "step": 14614
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.6938670463589925e-09,
+      "loss": 0.7291,
+      "step": 14615
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.613949803749853e-09,
+      "loss": 0.6795,
+      "step": 14616
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.534512333535859e-09,
+      "loss": 0.7134,
+      "step": 14617
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.455554639530626e-09,
+      "loss": 0.6802,
+      "step": 14618
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.3770767255244555e-09,
+      "loss": 0.7618,
+      "step": 14619
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.299078595286556e-09,
+      "loss": 0.352,
+      "step": 14620
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.221560252560599e-09,
+      "loss": 0.7522,
+      "step": 14621
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.144521701068051e-09,
+      "loss": 0.7428,
+      "step": 14622
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.067962944508177e-09,
+      "loss": 0.3995,
+      "step": 14623
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.991883986554703e-09,
+      "loss": 0.3718,
+      "step": 14624
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.916284830862485e-09,
+      "loss": 0.6879,
+      "step": 14625
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.841165481059729e-09,
+      "loss": 0.7658,
+      "step": 14626
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.766525940751333e-09,
+      "loss": 0.3835,
+      "step": 14627
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.692366213523315e-09,
+      "loss": 0.7417,
+      "step": 14628
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.61868630293505e-09,
+      "loss": 0.7297,
+      "step": 14629
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.545486212523709e-09,
+      "loss": 0.6585,
+      "step": 14630
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.472765945803149e-09,
+      "loss": 0.6337,
+      "step": 14631
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.40052550626502e-09,
+      "loss": 0.6525,
+      "step": 14632
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.328764897377658e-09,
+      "loss": 0.6714,
+      "step": 14633
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.257484122586087e-09,
+      "loss": 0.724,
+      "step": 14634
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.186683185313124e-09,
+      "loss": 0.7131,
+      "step": 14635
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.116362088956051e-09,
+      "loss": 0.3817,
+      "step": 14636
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.046520836892166e-09,
+      "loss": 0.7477,
+      "step": 14637
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.977159432474343e-09,
+      "loss": 0.7074,
+      "step": 14638
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.908277879032142e-09,
+      "loss": 0.3998,
+      "step": 14639
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.839876179872916e-09,
+      "loss": 0.6571,
+      "step": 14640
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.7719543382818145e-09,
+      "loss": 0.6997,
+      "step": 14641
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.704512357517344e-09,
+      "loss": 0.6928,
+      "step": 14642
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.637550240818023e-09,
+      "loss": 0.7063,
+      "step": 14643
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.571067991399059e-09,
+      "loss": 0.7157,
+      "step": 14644
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.505065612452342e-09,
+      "loss": 0.3951,
+      "step": 14645
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.439543107146449e-09,
+      "loss": 0.7459,
+      "step": 14646
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.374500478626642e-09,
+      "loss": 0.3909,
+      "step": 14647
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.309937730015978e-09,
+      "loss": 0.7063,
+      "step": 14648
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.245854864414201e-09,
+      "loss": 0.7431,
+      "step": 14649
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.182251884897737e-09,
+      "loss": 0.6937,
+      "step": 14650
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.11912879451859e-09,
+      "loss": 0.7411,
+      "step": 14651
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.0564855963098896e-09,
+      "loss": 0.4091,
+      "step": 14652
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.994322293278119e-09,
+      "loss": 0.7085,
+      "step": 14653
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.9326388884064485e-09,
+      "loss": 0.6784,
+      "step": 14654
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.871435384656952e-09,
+      "loss": 0.6794,
+      "step": 14655
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.8107117849683905e-09,
+      "loss": 0.7431,
+      "step": 14656
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.75046809225621e-09,
+      "loss": 0.6881,
+      "step": 14657
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.6907043094125405e-09,
+      "loss": 0.3913,
+      "step": 14658
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.631420439305089e-09,
+      "loss": 0.7354,
+      "step": 14659
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.572616484781577e-09,
+      "loss": 0.6587,
+      "step": 14660
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.5142924486653017e-09,
+      "loss": 0.7229,
+      "step": 14661
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.456448333755136e-09,
+      "loss": 0.7022,
+      "step": 14662
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.399084142828857e-09,
+      "loss": 0.6982,
+      "step": 14663
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3421998786409284e-09,
+      "loss": 0.3719,
+      "step": 14664
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.2857955439213886e-09,
+      "loss": 0.7171,
+      "step": 14665
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.2298711413780716e-09,
+      "loss": 0.7237,
+      "step": 14666
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.174426673696607e-09,
+      "loss": 0.4244,
+      "step": 14667
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.119462143538199e-09,
+      "loss": 0.7232,
+      "step": 14668
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.0649775535429584e-09,
+      "loss": 0.7258,
+      "step": 14669
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.0109729063243497e-09,
+      "loss": 0.742,
+      "step": 14670
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.9574482044769647e-09,
+      "loss": 0.3774,
+      "step": 14671
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.904403450569859e-09,
+      "loss": 0.6718,
+      "step": 14672
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.851838647149885e-09,
+      "loss": 0.6647,
+      "step": 14673
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.7997537967394682e-09,
+      "loss": 0.6789,
+      "step": 14674
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.7481489018410525e-09,
+      "loss": 0.7097,
+      "step": 14675
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.697023964930434e-09,
+      "loss": 0.6664,
+      "step": 14676
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.646378988463427e-09,
+      "loss": 0.7146,
+      "step": 14677
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5962139748703098e-09,
+      "loss": 0.6627,
+      "step": 14678
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5465289265602657e-09,
+      "loss": 0.7846,
+      "step": 14679
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.4973238459180537e-09,
+      "loss": 0.7015,
+      "step": 14680
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.448598735306229e-09,
+      "loss": 0.4097,
+      "step": 14681
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.4003535970640314e-09,
+      "loss": 0.6892,
+      "step": 14682
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.3525884335073855e-09,
+      "loss": 0.3974,
+      "step": 14683
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.3053032469311232e-09,
+      "loss": 0.766,
+      "step": 14684
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.2584980396023192e-09,
+      "loss": 0.6919,
+      "step": 14685
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.2121728137702856e-09,
+      "loss": 0.6596,
+      "step": 14686
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1663275716587995e-09,
+      "loss": 0.4198,
+      "step": 14687
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.120962315468322e-09,
+      "loss": 0.7381,
+      "step": 14688
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0760770473760016e-09,
+      "loss": 0.7088,
+      "step": 14689
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0316717695390008e-09,
+      "loss": 0.7269,
+      "step": 14690
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.987746484087838e-09,
+      "loss": 0.7173,
+      "step": 14691
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9443011931308266e-09,
+      "loss": 0.751,
+      "step": 14692
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9013358987540752e-09,
+      "loss": 0.6523,
+      "step": 14693
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8588506030203791e-09,
+      "loss": 0.7077,
+      "step": 14694
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8168453079703275e-09,
+      "loss": 0.6979,
+      "step": 14695
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7753200156189753e-09,
+      "loss": 0.3854,
+      "step": 14696
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.7342747279602835e-09,
+      "loss": 0.3804,
+      "step": 14697
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6937094469660076e-09,
+      "loss": 0.6985,
+      "step": 14698
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.653624174582369e-09,
+      "loss": 0.7068,
+      "step": 14699
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.6140189127356042e-09,
+      "loss": 0.3803,
+      "step": 14700
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5748936633241952e-09,
+      "loss": 0.4099,
+      "step": 14701
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5362484282288593e-09,
+      "loss": 0.7374,
+      "step": 14702
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4980832093036691e-09,
+      "loss": 0.4153,
+      "step": 14703
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4603980083816028e-09,
+      "loss": 0.6924,
+      "step": 14704
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4231928272723239e-09,
+      "loss": 0.7107,
+      "step": 14705
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.386467667761071e-09,
+      "loss": 0.7045,
+      "step": 14706
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3502225316108786e-09,
+      "loss": 0.3744,
+      "step": 14707
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.314457420562576e-09,
+      "loss": 0.7343,
+      "step": 14708
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2791723363325681e-09,
+      "loss": 0.7253,
+      "step": 14709
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2443672806150553e-09,
+      "loss": 0.7393,
+      "step": 14710
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2100422550809232e-09,
+      "loss": 0.6846,
+      "step": 14711
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1761972613788531e-09,
+      "loss": 0.7705,
+      "step": 14712
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1428323011331011e-09,
+      "loss": 0.4164,
+      "step": 14713
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1099473759457192e-09,
+      "loss": 0.6644,
+      "step": 14714
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.077542487394334e-09,
+      "loss": 0.7224,
+      "step": 14715
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0456176370365888e-09,
+      "loss": 0.7205,
+      "step": 14716
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0141728264034812e-09,
+      "loss": 0.7555,
+      "step": 14717
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.832080570060243e-10,
+      "loss": 0.3759,
+      "step": 14718
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.527233303296968e-10,
+      "loss": 0.6924,
+      "step": 14719
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.227186478377725e-10,
+      "loss": 0.7438,
+      "step": 14720
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.931940109724313e-10,
+      "loss": 0.7238,
+      "step": 14721
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.641494211503176e-10,
+      "loss": 0.3576,
+      "step": 14722
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.355848797647615e-10,
+      "loss": 0.7131,
+      "step": 14723
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.075003881879983e-10,
+      "loss": 0.7089,
+      "step": 14724
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.798959477689494e-10,
+      "loss": 0.7037,
+      "step": 14725
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.527715598321106e-10,
+      "loss": 0.7079,
+      "step": 14726
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.261272256797736e-10,
+      "loss": 0.6649,
+      "step": 14727
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.999629465909153e-10,
+      "loss": 0.6934,
+      "step": 14728
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.742787238223081e-10,
+      "loss": 0.7151,
+      "step": 14729
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.490745586062997e-10,
+      "loss": 0.4044,
+      "step": 14730
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.243504521541433e-10,
+      "loss": 0.4069,
+      "step": 14731
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.001064056504469e-10,
+      "loss": 0.7262,
+      "step": 14732
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.763424202620549e-10,
+      "loss": 0.7225,
+      "step": 14733
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.530584971280562e-10,
+      "loss": 0.7335,
+      "step": 14734
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.302546373664452e-10,
+      "loss": 0.6876,
+      "step": 14735
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.07930842071902e-10,
+      "loss": 0.6979,
+      "step": 14736
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.86087112316902e-10,
+      "loss": 0.6783,
+      "step": 14737
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.647234491494956e-10,
+      "loss": 0.397,
+      "step": 14738
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.4383985359552907e-10,
+      "loss": 0.7461,
+      "step": 14739
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.234363266575336e-10,
+      "loss": 0.651,
+      "step": 14740
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.0351286931583633e-10,
+      "loss": 0.6906,
+      "step": 14741
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.840694825252289e-10,
+      "loss": 0.7638,
+      "step": 14742
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.651061672216294e-10,
+      "loss": 0.6752,
+      "step": 14743
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.466229243132002e-10,
+      "loss": 0.4228,
+      "step": 14744
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.286197546881198e-10,
+      "loss": 0.7031,
+      "step": 14745
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.110966592101417e-10,
+      "loss": 0.7508,
+      "step": 14746
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.9405363872192506e-10,
+      "loss": 0.4367,
+      "step": 14747
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.7749069404059414e-10,
+      "loss": 0.7332,
+      "step": 14748
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.614078259621788e-10,
+      "loss": 0.6987,
+      "step": 14749
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.4580503525717393e-10,
+      "loss": 0.6581,
+      "step": 14750
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.3068232267609014e-10,
+      "loss": 0.7046,
+      "step": 14751
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.1603968894501338e-10,
+      "loss": 0.3905,
+      "step": 14752
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0187713476560456e-10,
+      "loss": 0.6814,
+      "step": 14753
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.8819466081954064e-10,
+      "loss": 0.3963,
+      "step": 14754
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.749922677629634e-10,
+      "loss": 0.6424,
+      "step": 14755
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.622699562287e-10,
+      "loss": 0.7586,
+      "step": 14756
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5002772682848334e-10,
+      "loss": 0.6946,
+      "step": 14757
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3826558015073155e-10,
+      "loss": 0.399,
+      "step": 14758
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2698351675832775e-10,
+      "loss": 0.4833,
+      "step": 14759
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1618153719528125e-10,
+      "loss": 0.4062,
+      "step": 14760
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0585964197784571e-10,
+      "loss": 0.3943,
+      "step": 14761
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.601783160229084e-11,
+      "loss": 0.725,
+      "step": 14762
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.665610654268186e-11,
+      "loss": 0.7249,
+      "step": 14763
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.777446724643867e-11,
+      "loss": 0.7083,
+      "step": 14764
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.93729141398869e-11,
+      "loss": 0.3972,
+      "step": 14765
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.145144762825794e-11,
+      "loss": 0.7606,
+      "step": 14766
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.401006809013787e-11,
+      "loss": 0.7234,
+      "step": 14767
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.704877588412871e-11,
+      "loss": 0.7533,
+      "step": 14768
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.0567571343297366e-11,
+      "loss": 0.6634,
+      "step": 14769
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.456645477961651e-11,
+      "loss": 0.6443,
+      "step": 14770
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.904542648063391e-11,
+      "loss": 0.7508,
+      "step": 14771
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.4004486711692863e-11,
+      "loss": 0.7808,
+      "step": 14772
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9443635713711774e-11,
+      "loss": 0.7698,
+      "step": 14773
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5362873707625016e-11,
+      "loss": 0.6957,
+      "step": 14774
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1762200888831843e-11,
+      "loss": 0.4224,
+      "step": 14775
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.641617428306604e-12,
+      "loss": 0.3902,
+      "step": 14776
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.001123478149851e-12,
+      "loss": 0.4174,
+      "step": 14777
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.8407191649270094e-12,
+      "loss": 0.7541,
+      "step": 14778
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.1604045907785976e-12,
+      "loss": 0.7325,
+      "step": 14779
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.601798367508962e-13,
+      "loss": 0.4265,
+      "step": 14780
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.400449616857259e-13,
+      "loss": 0.7144,
+      "step": 14781
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 0.4449,
+      "step": 14782
+    },
+    {
+      "epoch": 1.0,
+      "step": 14782,
+      "total_flos": 2.0987229186428817e+20,
+      "train_loss": 0.007514059227078391,
+      "train_runtime": 1786.5188,
+      "train_samples_per_second": 2118.135,
+      "train_steps_per_second": 8.274
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 14782,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 500,
+  "total_flos": 2.0987229186428817e+20,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}