diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,72355 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "global_step": 12055,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.524861878453039e-08,
+      "loss": 1.5828,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1049723756906078e-07,
+      "loss": 1.6206,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6574585635359117e-07,
+      "loss": 1.5747,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2099447513812156e-07,
+      "loss": 1.4883,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7624309392265196e-07,
+      "loss": 1.6629,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.3149171270718233e-07,
+      "loss": 1.555,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.867403314917127e-07,
+      "loss": 1.58,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.419889502762431e-07,
+      "loss": 1.5511,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.972375690607735e-07,
+      "loss": 1.5632,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 5.524861878453039e-07,
+      "loss": 1.6031,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.077348066298343e-07,
+      "loss": 1.547,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.629834254143647e-07,
+      "loss": 1.451,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.18232044198895e-07,
+      "loss": 1.5676,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.734806629834254e-07,
+      "loss": 0.2876,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.287292817679559e-07,
+      "loss": 1.4648,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.839779005524863e-07,
+      "loss": 1.5238,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.392265193370166e-07,
+      "loss": 1.5139,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.94475138121547e-07,
+      "loss": 1.4122,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0497237569060774e-06,
+      "loss": 1.5683,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.1049723756906078e-06,
+      "loss": 1.4046,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.160220994475138e-06,
+      "loss": 1.3583,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.2154696132596686e-06,
+      "loss": 1.3618,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.270718232044199e-06,
+      "loss": 1.3727,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3259668508287293e-06,
+      "loss": 1.4294,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.3812154696132598e-06,
+      "loss": 1.3549,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.43646408839779e-06,
+      "loss": 1.2923,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.4917127071823205e-06,
+      "loss": 1.2565,
+      "step": 27
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5469613259668508e-06,
+      "loss": 1.1961,
+      "step": 28
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6022099447513815e-06,
+      "loss": 1.2591,
+      "step": 29
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.6574585635359118e-06,
+      "loss": 1.2112,
+      "step": 30
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7127071823204422e-06,
+      "loss": 1.2889,
+      "step": 31
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7679558011049725e-06,
+      "loss": 0.2905,
+      "step": 32
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.823204419889503e-06,
+      "loss": 1.2455,
+      "step": 33
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8784530386740332e-06,
+      "loss": 1.2197,
+      "step": 34
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.933701657458564e-06,
+      "loss": 1.194,
+      "step": 35
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.988950276243094e-06,
+      "loss": 1.2464,
+      "step": 36
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0441988950276245e-06,
+      "loss": 1.2226,
+      "step": 37
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0994475138121547e-06,
+      "loss": 1.1573,
+      "step": 38
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1546961325966854e-06,
+      "loss": 1.1913,
+      "step": 39
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.2099447513812157e-06,
+      "loss": 1.2057,
+      "step": 40
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.265193370165746e-06,
+      "loss": 1.0902,
+      "step": 41
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.320441988950276e-06,
+      "loss": 1.1811,
+      "step": 42
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.375690607734807e-06,
+      "loss": 1.1382,
+      "step": 43
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.430939226519337e-06,
+      "loss": 1.1388,
+      "step": 44
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.486187845303868e-06,
+      "loss": 1.2158,
+      "step": 45
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.541436464088398e-06,
+      "loss": 1.1479,
+      "step": 46
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5966850828729284e-06,
+      "loss": 1.1453,
+      "step": 47
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6519337016574586e-06,
+      "loss": 1.1798,
+      "step": 48
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.707182320441989e-06,
+      "loss": 1.1155,
+      "step": 49
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.7624309392265196e-06,
+      "loss": 1.1505,
+      "step": 50
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.81767955801105e-06,
+      "loss": 1.0767,
+      "step": 51
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.87292817679558e-06,
+      "loss": 0.3117,
+      "step": 52
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.9281767955801104e-06,
+      "loss": 1.0973,
+      "step": 53
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.983425414364641e-06,
+      "loss": 1.1187,
+      "step": 54
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.0386740331491713e-06,
+      "loss": 1.0861,
+      "step": 55
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.0939226519337016e-06,
+      "loss": 1.0786,
+      "step": 56
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.149171270718232e-06,
+      "loss": 1.0475,
+      "step": 57
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.204419889502763e-06,
+      "loss": 1.1204,
+      "step": 58
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.2596685082872933e-06,
+      "loss": 1.1206,
+      "step": 59
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.3149171270718235e-06,
+      "loss": 1.1427,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.370165745856354e-06,
+      "loss": 1.0783,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4254143646408845e-06,
+      "loss": 1.0366,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4806629834254147e-06,
+      "loss": 1.0522,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.535911602209945e-06,
+      "loss": 1.039,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.5911602209944757e-06,
+      "loss": 1.0238,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.646408839779006e-06,
+      "loss": 1.0937,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7016574585635362e-06,
+      "loss": 1.1213,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.7569060773480665e-06,
+      "loss": 1.1134,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.812154696132597e-06,
+      "loss": 0.9342,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.867403314917128e-06,
+      "loss": 1.0584,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.922651933701658e-06,
+      "loss": 1.0772,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.977900552486188e-06,
+      "loss": 1.0218,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.033149171270719e-06,
+      "loss": 1.0351,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.088397790055249e-06,
+      "loss": 1.0506,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.143646408839779e-06,
+      "loss": 1.0443,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.1988950276243095e-06,
+      "loss": 1.0348,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.2541436464088406e-06,
+      "loss": 1.0423,
+      "step": 77
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.309392265193371e-06,
+      "loss": 1.0659,
+      "step": 78
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.364640883977901e-06,
+      "loss": 1.0262,
+      "step": 79
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.419889502762431e-06,
+      "loss": 1.0515,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.475138121546962e-06,
+      "loss": 1.0847,
+      "step": 81
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.530386740331492e-06,
+      "loss": 1.0443,
+      "step": 82
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.585635359116022e-06,
+      "loss": 0.9492,
+      "step": 83
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.640883977900552e-06,
+      "loss": 0.9776,
+      "step": 84
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.6961325966850835e-06,
+      "loss": 1.0058,
+      "step": 85
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.751381215469614e-06,
+      "loss": 1.0409,
+      "step": 86
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.806629834254144e-06,
+      "loss": 1.0207,
+      "step": 87
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.861878453038674e-06,
+      "loss": 1.0355,
+      "step": 88
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.9171270718232054e-06,
+      "loss": 1.0156,
+      "step": 89
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.972375690607736e-06,
+      "loss": 1.0172,
+      "step": 90
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.027624309392266e-06,
+      "loss": 1.0346,
+      "step": 91
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.082872928176796e-06,
+      "loss": 1.0042,
+      "step": 92
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.1381215469613265e-06,
+      "loss": 1.0716,
+      "step": 93
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.193370165745857e-06,
+      "loss": 0.9623,
+      "step": 94
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.248618784530387e-06,
+      "loss": 0.987,
+      "step": 95
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.303867403314917e-06,
+      "loss": 0.9707,
+      "step": 96
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.3591160220994476e-06,
+      "loss": 0.9991,
+      "step": 97
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.414364640883978e-06,
+      "loss": 1.0335,
+      "step": 98
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.469613259668509e-06,
+      "loss": 1.0262,
+      "step": 99
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.524861878453039e-06,
+      "loss": 0.9837,
+      "step": 100
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.5801104972375695e-06,
+      "loss": 1.0113,
+      "step": 101
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.6353591160221e-06,
+      "loss": 0.9878,
+      "step": 102
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.69060773480663e-06,
+      "loss": 1.0521,
+      "step": 103
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.74585635359116e-06,
+      "loss": 1.0333,
+      "step": 104
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.8011049723756905e-06,
+      "loss": 1.0627,
+      "step": 105
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.856353591160221e-06,
+      "loss": 0.9636,
+      "step": 106
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.911602209944752e-06,
+      "loss": 0.9918,
+      "step": 107
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.966850828729282e-06,
+      "loss": 0.9634,
+      "step": 108
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.0220994475138124e-06,
+      "loss": 1.0023,
+      "step": 109
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.077348066298343e-06,
+      "loss": 0.9517,
+      "step": 110
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.132596685082873e-06,
+      "loss": 1.0405,
+      "step": 111
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.187845303867403e-06,
+      "loss": 1.0125,
+      "step": 112
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.2430939226519335e-06,
+      "loss": 1.0006,
+      "step": 113
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.298342541436464e-06,
+      "loss": 0.9761,
+      "step": 114
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.353591160220996e-06,
+      "loss": 1.0202,
+      "step": 115
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.408839779005526e-06,
+      "loss": 1.0011,
+      "step": 116
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.464088397790056e-06,
+      "loss": 0.9935,
+      "step": 117
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.5193370165745865e-06,
+      "loss": 0.9581,
+      "step": 118
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.574585635359117e-06,
+      "loss": 1.0279,
+      "step": 119
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.629834254143647e-06,
+      "loss": 0.9945,
+      "step": 120
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.685082872928177e-06,
+      "loss": 0.9911,
+      "step": 121
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.740331491712708e-06,
+      "loss": 0.9702,
+      "step": 122
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.795580110497239e-06,
+      "loss": 0.9992,
+      "step": 123
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.850828729281769e-06,
+      "loss": 0.9941,
+      "step": 124
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.906077348066299e-06,
+      "loss": 0.987,
+      "step": 125
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.9613259668508295e-06,
+      "loss": 1.0136,
+      "step": 126
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.01657458563536e-06,
+      "loss": 0.9313,
+      "step": 127
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.07182320441989e-06,
+      "loss": 1.0039,
+      "step": 128
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.12707182320442e-06,
+      "loss": 1.0646,
+      "step": 129
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.182320441988951e-06,
+      "loss": 1.0156,
+      "step": 130
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.237569060773482e-06,
+      "loss": 0.9976,
+      "step": 131
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.292817679558012e-06,
+      "loss": 0.9707,
+      "step": 132
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.348066298342542e-06,
+      "loss": 1.002,
+      "step": 133
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.4033149171270724e-06,
+      "loss": 0.9855,
+      "step": 134
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.458563535911603e-06,
+      "loss": 0.9957,
+      "step": 135
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.513812154696133e-06,
+      "loss": 1.016,
+      "step": 136
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.569060773480663e-06,
+      "loss": 0.9857,
+      "step": 137
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.624309392265194e-06,
+      "loss": 1.0103,
+      "step": 138
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.679558011049725e-06,
+      "loss": 0.9716,
+      "step": 139
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.734806629834256e-06,
+      "loss": 0.9765,
+      "step": 140
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.790055248618785e-06,
+      "loss": 0.9737,
+      "step": 141
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.845303867403316e-06,
+      "loss": 1.0211,
+      "step": 142
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.900552486187846e-06,
+      "loss": 0.9633,
+      "step": 143
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.955801104972377e-06,
+      "loss": 1.0448,
+      "step": 144
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.011049723756906e-06,
+      "loss": 0.918,
+      "step": 145
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.066298342541437e-06,
+      "loss": 0.948,
+      "step": 146
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.121546961325968e-06,
+      "loss": 0.8965,
+      "step": 147
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.176795580110498e-06,
+      "loss": 1.0187,
+      "step": 148
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.232044198895029e-06,
+      "loss": 0.9974,
+      "step": 149
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.287292817679558e-06,
+      "loss": 0.9307,
+      "step": 150
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.34254143646409e-06,
+      "loss": 0.9563,
+      "step": 151
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.397790055248619e-06,
+      "loss": 1.0447,
+      "step": 152
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.45303867403315e-06,
+      "loss": 0.9679,
+      "step": 153
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.508287292817681e-06,
+      "loss": 0.9942,
+      "step": 154
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.56353591160221e-06,
+      "loss": 0.9579,
+      "step": 155
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.618784530386742e-06,
+      "loss": 0.9503,
+      "step": 156
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.674033149171271e-06,
+      "loss": 0.9256,
+      "step": 157
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.729281767955802e-06,
+      "loss": 0.94,
+      "step": 158
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.784530386740332e-06,
+      "loss": 0.9623,
+      "step": 159
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.839779005524863e-06,
+      "loss": 0.3009,
+      "step": 160
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.895027624309392e-06,
+      "loss": 0.9441,
+      "step": 161
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.950276243093923e-06,
+      "loss": 0.9552,
+      "step": 162
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.005524861878454e-06,
+      "loss": 0.8686,
+      "step": 163
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.060773480662984e-06,
+      "loss": 0.9753,
+      "step": 164
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.116022099447515e-06,
+      "loss": 1.014,
+      "step": 165
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.171270718232044e-06,
+      "loss": 0.9363,
+      "step": 166
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.226519337016575e-06,
+      "loss": 0.9306,
+      "step": 167
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.281767955801105e-06,
+      "loss": 0.9768,
+      "step": 168
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.337016574585636e-06,
+      "loss": 0.9355,
+      "step": 169
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.392265193370167e-06,
+      "loss": 0.9079,
+      "step": 170
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.447513812154696e-06,
+      "loss": 0.9727,
+      "step": 171
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.502762430939228e-06,
+      "loss": 0.9579,
+      "step": 172
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.558011049723757e-06,
+      "loss": 1.0006,
+      "step": 173
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.613259668508288e-06,
+      "loss": 1.0012,
+      "step": 174
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.668508287292818e-06,
+      "loss": 1.0376,
+      "step": 175
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.723756906077349e-06,
+      "loss": 0.9209,
+      "step": 176
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.779005524861878e-06,
+      "loss": 0.9641,
+      "step": 177
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.834254143646411e-06,
+      "loss": 0.9204,
+      "step": 178
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.88950276243094e-06,
+      "loss": 0.9438,
+      "step": 179
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.944751381215471e-06,
+      "loss": 0.9443,
+      "step": 180
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1e-05,
+      "loss": 0.9968,
+      "step": 181
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0055248618784532e-05,
+      "loss": 0.9539,
+      "step": 182
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0110497237569061e-05,
+      "loss": 0.895,
+      "step": 183
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0165745856353592e-05,
+      "loss": 0.9835,
+      "step": 184
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0220994475138122e-05,
+      "loss": 0.9411,
+      "step": 185
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0276243093922653e-05,
+      "loss": 0.9402,
+      "step": 186
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0331491712707182e-05,
+      "loss": 0.9158,
+      "step": 187
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0386740331491714e-05,
+      "loss": 0.9665,
+      "step": 188
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0441988950276245e-05,
+      "loss": 0.8975,
+      "step": 189
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0497237569060774e-05,
+      "loss": 0.3108,
+      "step": 190
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0552486187845305e-05,
+      "loss": 0.9062,
+      "step": 191
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0607734806629835e-05,
+      "loss": 0.9724,
+      "step": 192
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0662983425414366e-05,
+      "loss": 0.9749,
+      "step": 193
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0718232044198895e-05,
+      "loss": 0.9184,
+      "step": 194
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0773480662983426e-05,
+      "loss": 0.9573,
+      "step": 195
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0828729281767956e-05,
+      "loss": 0.9475,
+      "step": 196
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0883977900552487e-05,
+      "loss": 0.9405,
+      "step": 197
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0939226519337018e-05,
+      "loss": 0.9882,
+      "step": 198
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0994475138121547e-05,
+      "loss": 0.8959,
+      "step": 199
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1049723756906078e-05,
+      "loss": 0.9767,
+      "step": 200
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1104972375690608e-05,
+      "loss": 0.8999,
+      "step": 201
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1160220994475139e-05,
+      "loss": 0.9277,
+      "step": 202
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1215469613259668e-05,
+      "loss": 0.9845,
+      "step": 203
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.12707182320442e-05,
+      "loss": 0.9098,
+      "step": 204
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.132596685082873e-05,
+      "loss": 0.9887,
+      "step": 205
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.138121546961326e-05,
+      "loss": 0.8849,
+      "step": 206
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1436464088397791e-05,
+      "loss": 0.8736,
+      "step": 207
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.149171270718232e-05,
+      "loss": 0.9618,
+      "step": 208
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1546961325966852e-05,
+      "loss": 1.008,
+      "step": 209
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1602209944751381e-05,
+      "loss": 0.9057,
+      "step": 210
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1657458563535912e-05,
+      "loss": 0.9397,
+      "step": 211
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1712707182320442e-05,
+      "loss": 0.9846,
+      "step": 212
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1767955801104973e-05,
+      "loss": 0.9711,
+      "step": 213
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1823204419889504e-05,
+      "loss": 1.0127,
+      "step": 214
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1878453038674033e-05,
+      "loss": 0.9861,
+      "step": 215
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1933701657458564e-05,
+      "loss": 0.9825,
+      "step": 216
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1988950276243094e-05,
+      "loss": 0.9555,
+      "step": 217
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2044198895027625e-05,
+      "loss": 1.0494,
+      "step": 218
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2099447513812154e-05,
+      "loss": 0.2851,
+      "step": 219
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2154696132596685e-05,
+      "loss": 0.8949,
+      "step": 220
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2209944751381218e-05,
+      "loss": 0.9735,
+      "step": 221
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2265193370165746e-05,
+      "loss": 0.9,
+      "step": 222
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2320441988950279e-05,
+      "loss": 1.0231,
+      "step": 223
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2375690607734806e-05,
+      "loss": 1.0438,
+      "step": 224
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.243093922651934e-05,
+      "loss": 1.0022,
+      "step": 225
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2486187845303867e-05,
+      "loss": 0.9805,
+      "step": 226
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.25414364640884e-05,
+      "loss": 0.9007,
+      "step": 227
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2596685082872928e-05,
+      "loss": 0.9573,
+      "step": 228
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.265193370165746e-05,
+      "loss": 0.9231,
+      "step": 229
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2707182320441991e-05,
+      "loss": 0.972,
+      "step": 230
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2762430939226519e-05,
+      "loss": 0.9236,
+      "step": 231
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2817679558011052e-05,
+      "loss": 0.9714,
+      "step": 232
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.287292817679558e-05,
+      "loss": 0.9614,
+      "step": 233
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2928176795580112e-05,
+      "loss": 1.0339,
+      "step": 234
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.298342541436464e-05,
+      "loss": 0.9634,
+      "step": 235
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3038674033149173e-05,
+      "loss": 0.9378,
+      "step": 236
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3093922651933704e-05,
+      "loss": 0.9764,
+      "step": 237
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3149171270718234e-05,
+      "loss": 0.991,
+      "step": 238
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3204419889502765e-05,
+      "loss": 0.9265,
+      "step": 239
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3259668508287294e-05,
+      "loss": 0.8665,
+      "step": 240
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3314917127071825e-05,
+      "loss": 0.9138,
+      "step": 241
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3370165745856355e-05,
+      "loss": 0.9557,
+      "step": 242
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3425414364640886e-05,
+      "loss": 0.9126,
+      "step": 243
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3480662983425417e-05,
+      "loss": 0.9525,
+      "step": 244
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3535911602209946e-05,
+      "loss": 0.9815,
+      "step": 245
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3591160220994477e-05,
+      "loss": 1.029,
+      "step": 246
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3646408839779007e-05,
+      "loss": 0.9003,
+      "step": 247
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3701657458563538e-05,
+      "loss": 0.9227,
+      "step": 248
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3756906077348067e-05,
+      "loss": 0.9562,
+      "step": 249
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3812154696132598e-05,
+      "loss": 0.932,
+      "step": 250
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3867403314917128e-05,
+      "loss": 0.9694,
+      "step": 251
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3922651933701659e-05,
+      "loss": 0.9502,
+      "step": 252
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.397790055248619e-05,
+      "loss": 0.9338,
+      "step": 253
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.403314917127072e-05,
+      "loss": 0.9424,
+      "step": 254
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.408839779005525e-05,
+      "loss": 0.8951,
+      "step": 255
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.414364640883978e-05,
+      "loss": 0.9276,
+      "step": 256
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4198895027624311e-05,
+      "loss": 0.9,
+      "step": 257
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.425414364640884e-05,
+      "loss": 0.9147,
+      "step": 258
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4309392265193372e-05,
+      "loss": 1.0205,
+      "step": 259
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4364640883977903e-05,
+      "loss": 0.9692,
+      "step": 260
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4419889502762432e-05,
+      "loss": 0.9709,
+      "step": 261
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4475138121546963e-05,
+      "loss": 0.9116,
+      "step": 262
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4530386740331493e-05,
+      "loss": 0.9707,
+      "step": 263
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4585635359116024e-05,
+      "loss": 0.914,
+      "step": 264
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4640883977900553e-05,
+      "loss": 0.876,
+      "step": 265
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4696132596685084e-05,
+      "loss": 0.9648,
+      "step": 266
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4751381215469614e-05,
+      "loss": 0.9389,
+      "step": 267
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4806629834254145e-05,
+      "loss": 1.0046,
+      "step": 268
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4861878453038676e-05,
+      "loss": 0.9355,
+      "step": 269
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4917127071823205e-05,
+      "loss": 1.0023,
+      "step": 270
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4972375690607737e-05,
+      "loss": 0.922,
+      "step": 271
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5027624309392266e-05,
+      "loss": 0.9294,
+      "step": 272
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5082872928176797e-05,
+      "loss": 0.9394,
+      "step": 273
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5138121546961326e-05,
+      "loss": 0.9255,
+      "step": 274
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5193370165745858e-05,
+      "loss": 0.9549,
+      "step": 275
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5248618784530389e-05,
+      "loss": 0.9673,
+      "step": 276
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5303867403314918e-05,
+      "loss": 0.9342,
+      "step": 277
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.535911602209945e-05,
+      "loss": 0.9533,
+      "step": 278
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.541436464088398e-05,
+      "loss": 0.9084,
+      "step": 279
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.546961325966851e-05,
+      "loss": 0.991,
+      "step": 280
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.552486187845304e-05,
+      "loss": 0.9078,
+      "step": 281
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.558011049723757e-05,
+      "loss": 0.9806,
+      "step": 282
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5635359116022098e-05,
+      "loss": 0.8654,
+      "step": 283
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5690607734806633e-05,
+      "loss": 0.7945,
+      "step": 284
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5745856353591164e-05,
+      "loss": 0.9716,
+      "step": 285
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.580110497237569e-05,
+      "loss": 0.8892,
+      "step": 286
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5856353591160222e-05,
+      "loss": 0.9751,
+      "step": 287
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5911602209944754e-05,
+      "loss": 0.962,
+      "step": 288
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.5966850828729285e-05,
+      "loss": 0.93,
+      "step": 289
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6022099447513812e-05,
+      "loss": 0.9015,
+      "step": 290
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6077348066298344e-05,
+      "loss": 0.8991,
+      "step": 291
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6132596685082875e-05,
+      "loss": 0.9375,
+      "step": 292
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6187845303867406e-05,
+      "loss": 0.9313,
+      "step": 293
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6243093922651937e-05,
+      "loss": 0.984,
+      "step": 294
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6298342541436465e-05,
+      "loss": 1.0052,
+      "step": 295
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6353591160220996e-05,
+      "loss": 0.9125,
+      "step": 296
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6408839779005527e-05,
+      "loss": 0.921,
+      "step": 297
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6464088397790058e-05,
+      "loss": 0.9287,
+      "step": 298
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6519337016574586e-05,
+      "loss": 0.9748,
+      "step": 299
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6574585635359117e-05,
+      "loss": 0.9533,
+      "step": 300
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.6629834254143648e-05,
+      "loss": 0.957,
+      "step": 301
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.668508287292818e-05,
+      "loss": 0.9771,
+      "step": 302
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.674033149171271e-05,
+      "loss": 0.9178,
+      "step": 303
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6795580110497238e-05,
+      "loss": 0.9349,
+      "step": 304
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.685082872928177e-05,
+      "loss": 0.8795,
+      "step": 305
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.69060773480663e-05,
+      "loss": 0.9527,
+      "step": 306
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.696132596685083e-05,
+      "loss": 0.9212,
+      "step": 307
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7016574585635362e-05,
+      "loss": 0.9989,
+      "step": 308
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.707182320441989e-05,
+      "loss": 0.3213,
+      "step": 309
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.712707182320442e-05,
+      "loss": 0.871,
+      "step": 310
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7182320441988952e-05,
+      "loss": 0.9531,
+      "step": 311
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7237569060773483e-05,
+      "loss": 0.9578,
+      "step": 312
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.729281767955801e-05,
+      "loss": 0.9215,
+      "step": 313
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7348066298342542e-05,
+      "loss": 0.8852,
+      "step": 314
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7403314917127073e-05,
+      "loss": 0.9419,
+      "step": 315
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7458563535911604e-05,
+      "loss": 0.9578,
+      "step": 316
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7513812154696135e-05,
+      "loss": 0.996,
+      "step": 317
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7569060773480663e-05,
+      "loss": 0.3242,
+      "step": 318
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7624309392265194e-05,
+      "loss": 0.976,
+      "step": 319
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7679558011049725e-05,
+      "loss": 0.999,
+      "step": 320
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7734806629834257e-05,
+      "loss": 0.9782,
+      "step": 321
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7790055248618784e-05,
+      "loss": 0.9372,
+      "step": 322
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7845303867403315e-05,
+      "loss": 0.9391,
+      "step": 323
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7900552486187846e-05,
+      "loss": 0.9497,
+      "step": 324
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.7955801104972378e-05,
+      "loss": 0.9473,
+      "step": 325
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.801104972375691e-05,
+      "loss": 0.8961,
+      "step": 326
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8066298342541436e-05,
+      "loss": 0.8936,
+      "step": 327
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8121546961325968e-05,
+      "loss": 0.8793,
+      "step": 328
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.81767955801105e-05,
+      "loss": 0.9279,
+      "step": 329
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.823204419889503e-05,
+      "loss": 0.9673,
+      "step": 330
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.828729281767956e-05,
+      "loss": 0.9398,
+      "step": 331
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.834254143646409e-05,
+      "loss": 1.0061,
+      "step": 332
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.839779005524862e-05,
+      "loss": 0.9659,
+      "step": 333
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.845303867403315e-05,
+      "loss": 0.9304,
+      "step": 334
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8508287292817682e-05,
+      "loss": 0.948,
+      "step": 335
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.856353591160221e-05,
+      "loss": 0.9096,
+      "step": 336
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.861878453038674e-05,
+      "loss": 0.917,
+      "step": 337
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8674033149171272e-05,
+      "loss": 0.8971,
+      "step": 338
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8729281767955803e-05,
+      "loss": 0.97,
+      "step": 339
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8784530386740334e-05,
+      "loss": 0.9758,
+      "step": 340
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8839779005524862e-05,
+      "loss": 0.9623,
+      "step": 341
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8895027624309393e-05,
+      "loss": 0.9594,
+      "step": 342
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8950276243093924e-05,
+      "loss": 0.9677,
+      "step": 343
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9005524861878455e-05,
+      "loss": 0.9034,
+      "step": 344
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9060773480662983e-05,
+      "loss": 0.9121,
+      "step": 345
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9116022099447514e-05,
+      "loss": 0.9392,
+      "step": 346
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9171270718232045e-05,
+      "loss": 0.9786,
+      "step": 347
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9226519337016576e-05,
+      "loss": 0.922,
+      "step": 348
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9281767955801107e-05,
+      "loss": 0.9014,
+      "step": 349
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9337016574585635e-05,
+      "loss": 1.0125,
+      "step": 350
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9392265193370166e-05,
+      "loss": 1.0059,
+      "step": 351
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9447513812154697e-05,
+      "loss": 0.9726,
+      "step": 352
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.950276243093923e-05,
+      "loss": 0.9342,
+      "step": 353
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9558011049723756e-05,
+      "loss": 0.9663,
+      "step": 354
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9613259668508287e-05,
+      "loss": 1.0017,
+      "step": 355
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9668508287292822e-05,
+      "loss": 0.9544,
+      "step": 356
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.972375690607735e-05,
+      "loss": 0.9355,
+      "step": 357
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.977900552486188e-05,
+      "loss": 0.8943,
+      "step": 358
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9834254143646408e-05,
+      "loss": 0.9801,
+      "step": 359
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9889502762430943e-05,
+      "loss": 0.9943,
+      "step": 360
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.994475138121547e-05,
+      "loss": 0.9794,
+      "step": 361
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2e-05,
+      "loss": 0.8829,
+      "step": 362
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999999963907427e-05,
+      "loss": 0.9323,
+      "step": 363
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999998556297094e-05,
+      "loss": 0.9213,
+      "step": 364
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999996751668558e-05,
+      "loss": 0.9178,
+      "step": 365
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999994225188786e-05,
+      "loss": 0.9484,
+      "step": 366
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999999097685797e-05,
+      "loss": 0.8977,
+      "step": 367
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999998700667633e-05,
+      "loss": 0.9216,
+      "step": 368
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999982314644167e-05,
+      "loss": 0.9329,
+      "step": 369
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999976900761812e-05,
+      "loss": 0.3337,
+      "step": 370
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999970765029658e-05,
+      "loss": 0.8732,
+      "step": 371
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999963907448145e-05,
+      "loss": 1.0321,
+      "step": 372
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999956328017772e-05,
+      "loss": 0.9187,
+      "step": 373
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999948026739085e-05,
+      "loss": 0.9128,
+      "step": 374
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999993900361268e-05,
+      "loss": 0.3369,
+      "step": 375
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999929258639217e-05,
+      "loss": 0.9701,
+      "step": 376
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999991879181939e-05,
+      "loss": 0.9918,
+      "step": 377
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999907603153958e-05,
+      "loss": 0.9696,
+      "step": 378
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999895692643728e-05,
+      "loss": 0.9647,
+      "step": 379
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999883060289564e-05,
+      "loss": 0.9681,
+      "step": 380
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999986970609237e-05,
+      "loss": 0.9597,
+      "step": 381
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999855630053113e-05,
+      "loss": 0.9985,
+      "step": 382
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999840832172815e-05,
+      "loss": 0.926,
+      "step": 383
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999825312452537e-05,
+      "loss": 0.9875,
+      "step": 384
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.99998090708934e-05,
+      "loss": 0.9513,
+      "step": 385
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999979210749658e-05,
+      "loss": 0.9476,
+      "step": 386
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999774422263302e-05,
+      "loss": 0.9036,
+      "step": 387
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999975601519484e-05,
+      "loss": 0.963,
+      "step": 388
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999736886292524e-05,
+      "loss": 0.8811,
+      "step": 389
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999971703555773e-05,
+      "loss": 0.9156,
+      "step": 390
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999696462991895e-05,
+      "loss": 0.9796,
+      "step": 391
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999675168596507e-05,
+      "loss": 0.9455,
+      "step": 392
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999653152373098e-05,
+      "loss": 0.9369,
+      "step": 393
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999963041432326e-05,
+      "loss": 0.9639,
+      "step": 394
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999606954448635e-05,
+      "loss": 0.9995,
+      "step": 395
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999958277275091e-05,
+      "loss": 0.9203,
+      "step": 396
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999955786923184e-05,
+      "loss": 0.325,
+      "step": 397
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999532243893218e-05,
+      "loss": 0.8792,
+      "step": 398
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999505896736892e-05,
+      "loss": 0.8949,
+      "step": 399
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999478827764767e-05,
+      "loss": 0.9398,
+      "step": 400
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999451036978795e-05,
+      "loss": 0.8796,
+      "step": 401
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999422524380988e-05,
+      "loss": 0.8592,
+      "step": 402
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999393289973396e-05,
+      "loss": 0.9626,
+      "step": 403
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999936333375813e-05,
+      "loss": 0.9052,
+      "step": 404
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999332655737357e-05,
+      "loss": 0.9404,
+      "step": 405
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999301255913288e-05,
+      "loss": 0.8812,
+      "step": 406
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999269134288192e-05,
+      "loss": 0.3269,
+      "step": 407
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999236290864387e-05,
+      "loss": 0.9779,
+      "step": 408
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999202725644244e-05,
+      "loss": 0.9326,
+      "step": 409
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999168438630183e-05,
+      "loss": 0.9559,
+      "step": 410
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999913342982468e-05,
+      "loss": 0.9152,
+      "step": 411
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999909769923027e-05,
+      "loss": 0.9804,
+      "step": 412
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999906124684952e-05,
+      "loss": 0.9281,
+      "step": 413
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999902407268507e-05,
+      "loss": 0.9178,
+      "step": 414
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998986176739602e-05,
+      "loss": 0.918,
+      "step": 415
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998947559015844e-05,
+      "loss": 0.9938,
+      "step": 416
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998908219516596e-05,
+      "loss": 0.8846,
+      "step": 417
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998868158244686e-05,
+      "loss": 0.93,
+      "step": 418
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9998827375203015e-05,
+      "loss": 0.9281,
+      "step": 419
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.999878587039452e-05,
+      "loss": 0.9302,
+      "step": 420
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.99987436438222e-05,
+      "loss": 0.9599,
+      "step": 421
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999870069548911e-05,
+      "loss": 0.8903,
+      "step": 422
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999865702539834e-05,
+      "loss": 0.8966,
+      "step": 423
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999861263355304e-05,
+      "loss": 0.9327,
+      "step": 424
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998567519956428e-05,
+      "loss": 0.9193,
+      "step": 425
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998521684611746e-05,
+      "loss": 0.9387,
+      "step": 426
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998475127522315e-05,
+      "loss": 0.9446,
+      "step": 427
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998427848691487e-05,
+      "loss": 0.9019,
+      "step": 428
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999837984812268e-05,
+      "loss": 0.9787,
+      "step": 429
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998331125819358e-05,
+      "loss": 0.994,
+      "step": 430
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999828168178503e-05,
+      "loss": 0.9235,
+      "step": 431
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999823151602328e-05,
+      "loss": 0.9571,
+      "step": 432
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999818062853772e-05,
+      "loss": 0.9296,
+      "step": 433
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999812901933202e-05,
+      "loss": 0.9461,
+      "step": 434
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998076688409912e-05,
+      "loss": 0.9278,
+      "step": 435
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9998023635775175e-05,
+      "loss": 0.9527,
+      "step": 436
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997969861431632e-05,
+      "loss": 1.0266,
+      "step": 437
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997915365383167e-05,
+      "loss": 0.927,
+      "step": 438
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997860147633717e-05,
+      "loss": 1.01,
+      "step": 439
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997804208187266e-05,
+      "loss": 0.883,
+      "step": 440
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999774754704785e-05,
+      "loss": 0.9658,
+      "step": 441
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997690164219562e-05,
+      "loss": 0.9265,
+      "step": 442
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999763205970654e-05,
+      "loss": 0.986,
+      "step": 443
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999757323351298e-05,
+      "loss": 0.9686,
+      "step": 444
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997513685643133e-05,
+      "loss": 0.9192,
+      "step": 445
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997453416101294e-05,
+      "loss": 0.8924,
+      "step": 446
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997392424891812e-05,
+      "loss": 0.8765,
+      "step": 447
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997330712019095e-05,
+      "loss": 0.9661,
+      "step": 448
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999726827748759e-05,
+      "loss": 0.8799,
+      "step": 449
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999720512130181e-05,
+      "loss": 0.3092,
+      "step": 450
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999714124346631e-05,
+      "loss": 0.8919,
+      "step": 451
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997076643985702e-05,
+      "loss": 0.9428,
+      "step": 452
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999701132286465e-05,
+      "loss": 0.9175,
+      "step": 453
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999694528010787e-05,
+      "loss": 0.9363,
+      "step": 454
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999687851572013e-05,
+      "loss": 0.9307,
+      "step": 455
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996811029706247e-05,
+      "loss": 0.9533,
+      "step": 456
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996742822071092e-05,
+      "loss": 0.9074,
+      "step": 457
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999667389281959e-05,
+      "loss": 0.8989,
+      "step": 458
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999660424195672e-05,
+      "loss": 0.92,
+      "step": 459
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996533869487504e-05,
+      "loss": 0.8871,
+      "step": 460
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999646277541702e-05,
+      "loss": 0.9673,
+      "step": 461
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999639095975041e-05,
+      "loss": 0.9355,
+      "step": 462
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996318422492847e-05,
+      "loss": 0.9365,
+      "step": 463
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996245163649577e-05,
+      "loss": 0.9355,
+      "step": 464
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996171183225882e-05,
+      "loss": 0.9388,
+      "step": 465
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996096481227104e-05,
+      "loss": 0.9005,
+      "step": 466
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999602105765863e-05,
+      "loss": 0.9274,
+      "step": 467
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995944912525917e-05,
+      "loss": 0.8972,
+      "step": 468
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999586804583445e-05,
+      "loss": 0.8761,
+      "step": 469
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995790457589782e-05,
+      "loss": 0.9202,
+      "step": 470
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995712147797514e-05,
+      "loss": 0.9102,
+      "step": 471
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99956331164633e-05,
+      "loss": 0.9298,
+      "step": 472
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999555336359284e-05,
+      "loss": 0.9585,
+      "step": 473
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995472889191897e-05,
+      "loss": 0.9488,
+      "step": 474
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995391693266276e-05,
+      "loss": 0.8636,
+      "step": 475
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999530977582184e-05,
+      "loss": 0.9061,
+      "step": 476
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99952271368645e-05,
+      "loss": 0.9372,
+      "step": 477
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999514377640022e-05,
+      "loss": 0.8535,
+      "step": 478
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9995059694435028e-05,
+      "loss": 0.9425,
+      "step": 479
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999497489097498e-05,
+      "loss": 0.9217,
+      "step": 480
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994889366026203e-05,
+      "loss": 0.9478,
+      "step": 481
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999480311959487e-05,
+      "loss": 0.9309,
+      "step": 482
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994716151687213e-05,
+      "loss": 0.9267,
+      "step": 483
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99946284623095e-05,
+      "loss": 0.9288,
+      "step": 484
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999454005146807e-05,
+      "loss": 0.8845,
+      "step": 485
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994450919169295e-05,
+      "loss": 0.9085,
+      "step": 486
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994361065419613e-05,
+      "loss": 0.9926,
+      "step": 487
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994270490225516e-05,
+      "loss": 0.976,
+      "step": 488
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994179193593536e-05,
+      "loss": 0.9144,
+      "step": 489
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9994087175530265e-05,
+      "loss": 0.9126,
+      "step": 490
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993994436042345e-05,
+      "loss": 0.8724,
+      "step": 491
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999390097513647e-05,
+      "loss": 0.9446,
+      "step": 492
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993806792819387e-05,
+      "loss": 0.9164,
+      "step": 493
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993711889097892e-05,
+      "loss": 0.9333,
+      "step": 494
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993616263978843e-05,
+      "loss": 0.9408,
+      "step": 495
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993519917469134e-05,
+      "loss": 0.9063,
+      "step": 496
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993422849575728e-05,
+      "loss": 0.8669,
+      "step": 497
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993325060305624e-05,
+      "loss": 0.9392,
+      "step": 498
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993226549665888e-05,
+      "loss": 0.9624,
+      "step": 499
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993127317663624e-05,
+      "loss": 0.9648,
+      "step": 500
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993027364305998e-05,
+      "loss": 0.9335,
+      "step": 501
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999292668960023e-05,
+      "loss": 0.9498,
+      "step": 502
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999282529355358e-05,
+      "loss": 0.9245,
+      "step": 503
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999272317617337e-05,
+      "loss": 0.8742,
+      "step": 504
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992620337466976e-05,
+      "loss": 0.9187,
+      "step": 505
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999251677744181e-05,
+      "loss": 0.9623,
+      "step": 506
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999241249610536e-05,
+      "loss": 0.8531,
+      "step": 507
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9992307493465146e-05,
+      "loss": 0.3388,
+      "step": 508
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999220176952875e-05,
+      "loss": 0.9452,
+      "step": 509
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99920953243038e-05,
+      "loss": 0.8526,
+      "step": 510
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991988157797988e-05,
+      "loss": 0.9167,
+      "step": 511
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991880270019043e-05,
+      "loss": 0.8743,
+      "step": 512
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991771660974755e-05,
+      "loss": 0.8867,
+      "step": 513
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991662330672965e-05,
+      "loss": 0.9702,
+      "step": 514
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991552279121563e-05,
+      "loss": 0.9451,
+      "step": 515
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999144150632849e-05,
+      "loss": 0.9326,
+      "step": 516
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991330012301752e-05,
+      "loss": 0.8717,
+      "step": 517
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999121779704939e-05,
+      "loss": 0.9749,
+      "step": 518
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9991104860579508e-05,
+      "loss": 0.9329,
+      "step": 519
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990991202900252e-05,
+      "loss": 0.9976,
+      "step": 520
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999087682401983e-05,
+      "loss": 0.8995,
+      "step": 521
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990761723946502e-05,
+      "loss": 0.9447,
+      "step": 522
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999064590268857e-05,
+      "loss": 0.9297,
+      "step": 523
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99905293602544e-05,
+      "loss": 0.96,
+      "step": 524
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.99904120966524e-05,
+      "loss": 0.8901,
+      "step": 525
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999029411189104e-05,
+      "loss": 0.9408,
+      "step": 526
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9990175405978835e-05,
+      "loss": 0.9385,
+      "step": 527
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999005597892435e-05,
+      "loss": 0.9682,
+      "step": 528
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998993583073621e-05,
+      "loss": 0.9411,
+      "step": 529
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989814961423084e-05,
+      "loss": 0.9453,
+      "step": 530
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989693370993702e-05,
+      "loss": 0.9312,
+      "step": 531
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989571059456837e-05,
+      "loss": 0.8448,
+      "step": 532
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989448026821322e-05,
+      "loss": 0.9411,
+      "step": 533
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9989324273096034e-05,
+      "loss": 0.3621,
+      "step": 534
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998919979828991e-05,
+      "loss": 0.8904,
+      "step": 535
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998907460241193e-05,
+      "loss": 0.8795,
+      "step": 536
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988948685471135e-05,
+      "loss": 0.9108,
+      "step": 537
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988822047476618e-05,
+      "loss": 0.9047,
+      "step": 538
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998869468843751e-05,
+      "loss": 0.8602,
+      "step": 539
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9988566608363014e-05,
+      "loss": 0.9082,
+      "step": 540
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998843780726237e-05,
+      "loss": 0.8812,
+      "step": 541
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.998830828514488e-05,
+      "loss": 0.9474,
+      "step": 542
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998817804201989e-05,
+      "loss": 0.9254,
+      "step": 543
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9988047077896803e-05,
+      "loss": 0.9067,
+      "step": 544
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987915392785075e-05,
+      "loss": 0.9838,
+      "step": 545
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987782986694206e-05,
+      "loss": 0.9667,
+      "step": 546
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9987649859633754e-05,
+      "loss": 0.9278,
+      "step": 547
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998751601161334e-05,
+      "loss": 0.9467,
+      "step": 548
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998738144264261e-05,
+      "loss": 0.914,
+      "step": 549
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998724615273129e-05,
+      "loss": 0.9576,
+      "step": 550
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998711014188914e-05,
+      "loss": 0.9035,
+      "step": 551
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986973410125977e-05,
+      "loss": 0.915,
+      "step": 552
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986835957451673e-05,
+      "loss": 0.9373,
+      "step": 553
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986697783876153e-05,
+      "loss": 0.9297,
+      "step": 554
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998655888940939e-05,
+      "loss": 0.8597,
+      "step": 555
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986419274061407e-05,
+      "loss": 0.9578,
+      "step": 556
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998627893784228e-05,
+      "loss": 0.9883,
+      "step": 557
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986137880762148e-05,
+      "loss": 0.9094,
+      "step": 558
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985996102831186e-05,
+      "loss": 0.9424,
+      "step": 559
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998585360405963e-05,
+      "loss": 0.9124,
+      "step": 560
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985710384457767e-05,
+      "loss": 0.9294,
+      "step": 561
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985566444035936e-05,
+      "loss": 0.8944,
+      "step": 562
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9985421782804523e-05,
+      "loss": 0.8371,
+      "step": 563
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998527640077398e-05,
+      "loss": 0.8918,
+      "step": 564
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998513029795479e-05,
+      "loss": 0.9414,
+      "step": 565
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984983474357508e-05,
+      "loss": 0.9038,
+      "step": 566
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984835929992724e-05,
+      "loss": 0.9483,
+      "step": 567
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99846876648711e-05,
+      "loss": 0.8992,
+      "step": 568
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984538679003327e-05,
+      "loss": 0.89,
+      "step": 569
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984388972400168e-05,
+      "loss": 0.9172,
+      "step": 570
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984238545072422e-05,
+      "loss": 0.8912,
+      "step": 571
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9984087397030958e-05,
+      "loss": 0.9291,
+      "step": 572
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983935528286676e-05,
+      "loss": 0.901,
+      "step": 573
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983782938850544e-05,
+      "loss": 0.8622,
+      "step": 574
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998362962873358e-05,
+      "loss": 0.8418,
+      "step": 575
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983475597946844e-05,
+      "loss": 0.866,
+      "step": 576
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983320846501455e-05,
+      "loss": 0.9064,
+      "step": 577
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983165374408586e-05,
+      "loss": 0.9001,
+      "step": 578
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983009181679463e-05,
+      "loss": 0.8944,
+      "step": 579
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982852268325357e-05,
+      "loss": 0.9616,
+      "step": 580
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982694634357595e-05,
+      "loss": 1.0152,
+      "step": 581
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982536279787558e-05,
+      "loss": 0.9297,
+      "step": 582
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982377204626674e-05,
+      "loss": 0.8861,
+      "step": 583
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9982217408886425e-05,
+      "loss": 0.9422,
+      "step": 584
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998205689257835e-05,
+      "loss": 0.9505,
+      "step": 585
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981895655714034e-05,
+      "loss": 0.9331,
+      "step": 586
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981733698305115e-05,
+      "loss": 0.9784,
+      "step": 587
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981571020363285e-05,
+      "loss": 0.8903,
+      "step": 588
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981407621900288e-05,
+      "loss": 0.9486,
+      "step": 589
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981243502927915e-05,
+      "loss": 0.8964,
+      "step": 590
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981078663458015e-05,
+      "loss": 0.9297,
+      "step": 591
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998091310350249e-05,
+      "loss": 0.8764,
+      "step": 592
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980746823073288e-05,
+      "loss": 0.9623,
+      "step": 593
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980579822182412e-05,
+      "loss": 0.9774,
+      "step": 594
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980412100841917e-05,
+      "loss": 0.8992,
+      "step": 595
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980243659063912e-05,
+      "loss": 0.8927,
+      "step": 596
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9980074496860553e-05,
+      "loss": 0.9702,
+      "step": 597
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979904614244054e-05,
+      "loss": 0.9639,
+      "step": 598
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979734011226674e-05,
+      "loss": 0.9773,
+      "step": 599
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997956268782073e-05,
+      "loss": 0.9585,
+      "step": 600
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997939064403859e-05,
+      "loss": 0.9179,
+      "step": 601
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9979217879892676e-05,
+      "loss": 0.981,
+      "step": 602
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997904439539545e-05,
+      "loss": 0.9341,
+      "step": 603
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978870190559442e-05,
+      "loss": 0.9129,
+      "step": 604
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9978695265397227e-05,
+      "loss": 0.9599,
+      "step": 605
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997851961992143e-05,
+      "loss": 0.8757,
+      "step": 606
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997834325414473e-05,
+      "loss": 0.8714,
+      "step": 607
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997816616807986e-05,
+      "loss": 0.3194,
+      "step": 608
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99779883617396e-05,
+      "loss": 0.8508,
+      "step": 609
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977809835136785e-05,
+      "loss": 0.9002,
+      "step": 610
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977630588284303e-05,
+      "loss": 0.94,
+      "step": 611
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9977450621195095e-05,
+      "loss": 0.875,
+      "step": 612
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997726993388215e-05,
+      "loss": 0.9398,
+      "step": 613
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997708852635851e-05,
+      "loss": 0.9669,
+      "step": 614
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976906398637274e-05,
+      "loss": 0.9303,
+      "step": 615
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976723550731584e-05,
+      "loss": 0.851,
+      "step": 616
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976539982654643e-05,
+      "loss": 0.8815,
+      "step": 617
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976355694419695e-05,
+      "loss": 0.9495,
+      "step": 618
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9976170686040053e-05,
+      "loss": 0.9645,
+      "step": 619
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975984957529062e-05,
+      "loss": 0.9692,
+      "step": 620
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975798508900136e-05,
+      "loss": 0.8999,
+      "step": 621
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975611340166733e-05,
+      "loss": 0.9231,
+      "step": 622
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975423451342358e-05,
+      "loss": 0.9115,
+      "step": 623
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997523484244058e-05,
+      "loss": 0.8629,
+      "step": 624
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997504551347501e-05,
+      "loss": 1.0235,
+      "step": 625
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9974855464459317e-05,
+      "loss": 0.8767,
+      "step": 626
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9974664695407218e-05,
+      "loss": 0.3369,
+      "step": 627
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997447320633249e-05,
+      "loss": 0.9354,
+      "step": 628
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9974280997248946e-05,
+      "loss": 0.8935,
+      "step": 629
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997408806817046e-05,
+      "loss": 0.8918,
+      "step": 630
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9973894419110973e-05,
+      "loss": 0.9389,
+      "step": 631
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997370005008445e-05,
+      "loss": 0.933,
+      "step": 632
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9973504961104923e-05,
+      "loss": 0.8884,
+      "step": 633
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9973309152186483e-05,
+      "loss": 0.9104,
+      "step": 634
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9973112623343255e-05,
+      "loss": 0.9009,
+      "step": 635
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997291537458943e-05,
+      "loss": 0.8348,
+      "step": 636
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972717405939245e-05,
+      "loss": 0.902,
+      "step": 637
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997251871740699e-05,
+      "loss": 0.97,
+      "step": 638
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972319309007015e-05,
+      "loss": 0.9432,
+      "step": 639
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99721191807537e-05,
+      "loss": 0.971,
+      "step": 640
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99719183326615e-05,
+      "loss": 0.8948,
+      "step": 641
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9971716764744914e-05,
+      "loss": 0.9065,
+      "step": 642
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997151447701849e-05,
+      "loss": 0.9071,
+      "step": 643
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9971311469496834e-05,
+      "loss": 0.9333,
+      "step": 644
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9971107742194597e-05,
+      "loss": 0.8825,
+      "step": 645
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997090329512648e-05,
+      "loss": 0.9224,
+      "step": 646
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997069812830725e-05,
+      "loss": 0.8896,
+      "step": 647
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997049224175171e-05,
+      "loss": 0.9519,
+      "step": 648
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9970285635474725e-05,
+      "loss": 0.8854,
+      "step": 649
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997007830949121e-05,
+      "loss": 0.8984,
+      "step": 650
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9969870263816127e-05,
+      "loss": 0.9332,
+      "step": 651
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.99696614984645e-05,
+      "loss": 0.8813,
+      "step": 652
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9969452013451396e-05,
+      "loss": 0.9333,
+      "step": 653
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996924180879193e-05,
+      "loss": 0.9892,
+      "step": 654
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9969030884501287e-05,
+      "loss": 0.8748,
+      "step": 655
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9968819240594687e-05,
+      "loss": 0.9669,
+      "step": 656
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9968606877087408e-05,
+      "loss": 0.8992,
+      "step": 657
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9968393793994774e-05,
+      "loss": 0.9694,
+      "step": 658
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9968179991332175e-05,
+      "loss": 0.9258,
+      "step": 659
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.996796546911504e-05,
+      "loss": 0.8514,
+      "step": 660
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9967750227358858e-05,
+      "loss": 0.322,
+      "step": 661
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9967534266079162e-05,
+      "loss": 0.9137,
+      "step": 662
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9967317585291544e-05,
+      "loss": 0.9551,
+      "step": 663
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996710018501164e-05,
+      "loss": 0.8722,
+      "step": 664
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966882065255148e-05,
+      "loss": 0.9107,
+      "step": 665
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966663226037815e-05,
+      "loss": 0.9039,
+      "step": 666
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966443667375433e-05,
+      "loss": 0.9075,
+      "step": 667
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966223389283856e-05,
+      "loss": 0.9048,
+      "step": 668
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9966002391778976e-05,
+      "loss": 0.9731,
+      "step": 669
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9965780674876757e-05,
+      "loss": 0.8981,
+      "step": 670
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9965558238593195e-05,
+      "loss": 0.9535,
+      "step": 671
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996533508294435e-05,
+      "loss": 0.8845,
+      "step": 672
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9965111207946327e-05,
+      "loss": 0.9359,
+      "step": 673
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964886613615293e-05,
+      "loss": 0.914,
+      "step": 674
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964661299967457e-05,
+      "loss": 0.8788,
+      "step": 675
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964435267019084e-05,
+      "loss": 0.8855,
+      "step": 676
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9964208514786486e-05,
+      "loss": 0.8538,
+      "step": 677
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996398104328604e-05,
+      "loss": 0.8914,
+      "step": 678
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9963752852534154e-05,
+      "loss": 0.8931,
+      "step": 679
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9963523942547312e-05,
+      "loss": 0.8763,
+      "step": 680
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996329431334203e-05,
+      "loss": 0.9142,
+      "step": 681
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9963063964934886e-05,
+      "loss": 0.9508,
+      "step": 682
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996283289734251e-05,
+      "loss": 0.9452,
+      "step": 683
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9962601110581578e-05,
+      "loss": 0.9461,
+      "step": 684
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9962368604668825e-05,
+      "loss": 0.9548,
+      "step": 685
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9962135379621033e-05,
+      "loss": 0.8803,
+      "step": 686
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9961901435455036e-05,
+      "loss": 0.9236,
+      "step": 687
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9961666772187723e-05,
+      "loss": 0.8945,
+      "step": 688
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996143138983603e-05,
+      "loss": 0.8492,
+      "step": 689
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9961195288416956e-05,
+      "loss": 0.9751,
+      "step": 690
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9960958467947535e-05,
+      "loss": 0.904,
+      "step": 691
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9960720928444867e-05,
+      "loss": 0.9607,
+      "step": 692
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9960482669926097e-05,
+      "loss": 0.9337,
+      "step": 693
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9960243692408424e-05,
+      "loss": 0.9354,
+      "step": 694
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9960003995909098e-05,
+      "loss": 0.8791,
+      "step": 695
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9959763580445424e-05,
+      "loss": 0.8159,
+      "step": 696
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9959522446034758e-05,
+      "loss": 0.9159,
+      "step": 697
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99592805926945e-05,
+      "loss": 0.9026,
+      "step": 698
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995903802044211e-05,
+      "loss": 0.904,
+      "step": 699
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99587947292951e-05,
+      "loss": 0.8628,
+      "step": 700
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9958550719271033e-05,
+      "loss": 0.9466,
+      "step": 701
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9958305990387523e-05,
+      "loss": 0.8494,
+      "step": 702
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995806054266223e-05,
+      "loss": 0.9092,
+      "step": 703
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995781437611288e-05,
+      "loss": 0.8986,
+      "step": 704
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9957567490757235e-05,
+      "loss": 0.9409,
+      "step": 705
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995731988661312e-05,
+      "loss": 0.9284,
+      "step": 706
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995707156369841e-05,
+      "loss": 0.8702,
+      "step": 707
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995682252203103e-05,
+      "loss": 0.8689,
+      "step": 708
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995657276162895e-05,
+      "loss": 0.9272,
+      "step": 709
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995632228251021e-05,
+      "loss": 0.9514,
+      "step": 710
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9956071084692884e-05,
+      "loss": 0.8688,
+      "step": 711
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9955819168195104e-05,
+      "loss": 0.8797,
+      "step": 712
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995556653303506e-05,
+      "loss": 0.9301,
+      "step": 713
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9955313179230983e-05,
+      "loss": 0.8876,
+      "step": 714
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9955059106801164e-05,
+      "loss": 0.9012,
+      "step": 715
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9954804315763944e-05,
+      "loss": 0.8374,
+      "step": 716
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9954548806137714e-05,
+      "loss": 0.8753,
+      "step": 717
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9954292577940918e-05,
+      "loss": 0.3588,
+      "step": 718
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9954035631192054e-05,
+      "loss": 0.922,
+      "step": 719
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9953777965909663e-05,
+      "loss": 0.9001,
+      "step": 720
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9953519582112354e-05,
+      "loss": 0.8467,
+      "step": 721
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995326047981877e-05,
+      "loss": 0.8772,
+      "step": 722
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995300065904762e-05,
+      "loss": 0.8781,
+      "step": 723
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9952740119817658e-05,
+      "loss": 0.9425,
+      "step": 724
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9952478862147687e-05,
+      "loss": 0.8809,
+      "step": 725
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9952216886056574e-05,
+      "loss": 0.8556,
+      "step": 726
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9951954191563222e-05,
+      "loss": 0.3271,
+      "step": 727
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9951690778686597e-05,
+      "loss": 0.9194,
+      "step": 728
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9951426647445715e-05,
+      "loss": 0.8833,
+      "step": 729
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995116179785964e-05,
+      "loss": 0.9534,
+      "step": 730
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9950896229947488e-05,
+      "loss": 0.9328,
+      "step": 731
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9950629943728434e-05,
+      "loss": 0.9609,
+      "step": 732
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99503629392217e-05,
+      "loss": 0.9049,
+      "step": 733
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9950095216446554e-05,
+      "loss": 0.9543,
+      "step": 734
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9949826775422327e-05,
+      "loss": 0.3429,
+      "step": 735
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9949557616168393e-05,
+      "loss": 0.876,
+      "step": 736
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9949287738704182e-05,
+      "loss": 0.9361,
+      "step": 737
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9949017143049182e-05,
+      "loss": 0.9342,
+      "step": 738
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994874582922291e-05,
+      "loss": 0.327,
+      "step": 739
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994847379724497e-05,
+      "loss": 0.9168,
+      "step": 740
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9948201047134987e-05,
+      "loss": 0.9597,
+      "step": 741
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9947927578912653e-05,
+      "loss": 0.8121,
+      "step": 742
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9947653392597705e-05,
+      "loss": 0.917,
+      "step": 743
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9947378488209938e-05,
+      "loss": 0.8932,
+      "step": 744
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9947102865769197e-05,
+      "loss": 0.9153,
+      "step": 745
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9946826525295375e-05,
+      "loss": 0.8756,
+      "step": 746
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994654946680842e-05,
+      "loss": 0.9579,
+      "step": 747
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9946271690328335e-05,
+      "loss": 0.9236,
+      "step": 748
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9945993195875166e-05,
+      "loss": 0.9328,
+      "step": 749
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9945713983469023e-05,
+      "loss": 0.8736,
+      "step": 750
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9945434053130054e-05,
+      "loss": 0.9214,
+      "step": 751
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994515340487847e-05,
+      "loss": 0.9154,
+      "step": 752
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9944872038734528e-05,
+      "loss": 0.8853,
+      "step": 753
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9944589954718538e-05,
+      "loss": 0.9325,
+      "step": 754
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9944307152850867e-05,
+      "loss": 0.867,
+      "step": 755
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9944023633151924e-05,
+      "loss": 0.8995,
+      "step": 756
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994373939564217e-05,
+      "loss": 0.8878,
+      "step": 757
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9943454440342137e-05,
+      "loss": 0.9091,
+      "step": 758
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9943168767272386e-05,
+      "loss": 0.9167,
+      "step": 759
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9942882376453533e-05,
+      "loss": 0.9056,
+      "step": 760
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9942595267906263e-05,
+      "loss": 0.8946,
+      "step": 761
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9942307441651293e-05,
+      "loss": 0.8564,
+      "step": 762
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.99420188977094e-05,
+      "loss": 0.3658,
+      "step": 763
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9941729636101416e-05,
+      "loss": 0.866,
+      "step": 764
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994143965684822e-05,
+      "loss": 0.9633,
+      "step": 765
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9941148959970743e-05,
+      "loss": 0.9038,
+      "step": 766
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994085754548997e-05,
+      "loss": 0.9406,
+      "step": 767
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994056541342694e-05,
+      "loss": 0.9289,
+      "step": 768
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994027256380273e-05,
+      "loss": 0.9268,
+      "step": 769
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9939978996638497e-05,
+      "loss": 0.8976,
+      "step": 770
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9939684711955415e-05,
+      "loss": 0.9145,
+      "step": 771
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9939389709774735e-05,
+      "loss": 0.803,
+      "step": 772
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9939093990117752e-05,
+      "loss": 0.8857,
+      "step": 773
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993879755300581e-05,
+      "loss": 0.9394,
+      "step": 774
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9938500398460306e-05,
+      "loss": 0.912,
+      "step": 775
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9938202526502697e-05,
+      "loss": 0.9336,
+      "step": 776
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993790393715448e-05,
+      "loss": 0.8575,
+      "step": 777
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993760463043721e-05,
+      "loss": 0.8786,
+      "step": 778
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9937304606372487e-05,
+      "loss": 0.8254,
+      "step": 779
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.993700386498198e-05,
+      "loss": 0.9017,
+      "step": 780
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9936702406287387e-05,
+      "loss": 0.9337,
+      "step": 781
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9936400230310472e-05,
+      "loss": 0.348,
+      "step": 782
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9936097337073054e-05,
+      "loss": 0.9303,
+      "step": 783
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993579372659699e-05,
+      "loss": 0.9157,
+      "step": 784
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9935489398904196e-05,
+      "loss": 0.9209,
+      "step": 785
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9935184354016642e-05,
+      "loss": 0.9174,
+      "step": 786
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993487859195635e-05,
+      "loss": 0.9505,
+      "step": 787
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993457211274539e-05,
+      "loss": 0.8794,
+      "step": 788
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9934264916405882e-05,
+      "loss": 0.9166,
+      "step": 789
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9933957002960006e-05,
+      "loss": 0.9134,
+      "step": 790
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9933648372429985e-05,
+      "loss": 0.8613,
+      "step": 791
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9933339024838097e-05,
+      "loss": 0.8723,
+      "step": 792
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993302896020668e-05,
+      "loss": 0.9695,
+      "step": 793
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9932718178558105e-05,
+      "loss": 0.3273,
+      "step": 794
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9932406679914816e-05,
+      "loss": 0.9233,
+      "step": 795
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993209446429929e-05,
+      "loss": 0.9005,
+      "step": 796
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993178153173407e-05,
+      "loss": 0.8988,
+      "step": 797
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993146788224174e-05,
+      "loss": 0.8548,
+      "step": 798
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9931153515844947e-05,
+      "loss": 0.9123,
+      "step": 799
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.993083843256638e-05,
+      "loss": 0.8906,
+      "step": 800
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9930522632428783e-05,
+      "loss": 0.8858,
+      "step": 801
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9930206115454952e-05,
+      "loss": 0.9614,
+      "step": 802
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992988888166774e-05,
+      "loss": 0.951,
+      "step": 803
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9929570931090038e-05,
+      "loss": 0.9315,
+      "step": 804
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9929252263744804e-05,
+      "loss": 0.8733,
+      "step": 805
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9928932879655038e-05,
+      "loss": 0.931,
+      "step": 806
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9928612778843797e-05,
+      "loss": 0.8571,
+      "step": 807
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9928291961334184e-05,
+      "loss": 0.3231,
+      "step": 808
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992797042714936e-05,
+      "loss": 0.8785,
+      "step": 809
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9927648176312534e-05,
+      "loss": 0.9286,
+      "step": 810
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992732520884697e-05,
+      "loss": 0.9307,
+      "step": 811
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9927001524775976e-05,
+      "loss": 0.9066,
+      "step": 812
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9926677124122925e-05,
+      "loss": 0.8857,
+      "step": 813
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9926352006911225e-05,
+      "loss": 0.9594,
+      "step": 814
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9926026173164353e-05,
+      "loss": 0.871,
+      "step": 815
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9925699622905822e-05,
+      "loss": 0.9162,
+      "step": 816
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992537235615921e-05,
+      "loss": 0.9034,
+      "step": 817
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9925044372948136e-05,
+      "loss": 0.9417,
+      "step": 818
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992471567329628e-05,
+      "loss": 0.9391,
+      "step": 819
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992438625722737e-05,
+      "loss": 0.9324,
+      "step": 820
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992405612476518e-05,
+      "loss": 0.9372,
+      "step": 821
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992372527593354e-05,
+      "loss": 0.9004,
+      "step": 822
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9923393710756338e-05,
+      "loss": 0.9214,
+      "step": 823
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9923061429257503e-05,
+      "loss": 0.8962,
+      "step": 824
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9922728431461024e-05,
+      "loss": 0.9083,
+      "step": 825
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9922394717390937e-05,
+      "loss": 0.8767,
+      "step": 826
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9922060287071332e-05,
+      "loss": 0.8824,
+      "step": 827
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9921725140526352e-05,
+      "loss": 0.8805,
+      "step": 828
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9921389277780185e-05,
+      "loss": 0.9325,
+      "step": 829
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9921052698857076e-05,
+      "loss": 0.9001,
+      "step": 830
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9920715403781326e-05,
+      "loss": 0.9062,
+      "step": 831
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9920377392577278e-05,
+      "loss": 0.9012,
+      "step": 832
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.992003866526933e-05,
+      "loss": 0.9144,
+      "step": 833
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991969922188194e-05,
+      "loss": 0.8982,
+      "step": 834
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9919359062439602e-05,
+      "loss": 0.9222,
+      "step": 835
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991901818696688e-05,
+      "loss": 0.8682,
+      "step": 836
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991867659548837e-05,
+      "loss": 0.955,
+      "step": 837
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991833428802874e-05,
+      "loss": 0.8964,
+      "step": 838
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991799126461269e-05,
+      "loss": 0.8704,
+      "step": 839
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9917647525264988e-05,
+      "loss": 0.9592,
+      "step": 840
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9917303070010442e-05,
+      "loss": 0.8903,
+      "step": 841
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991695789887392e-05,
+      "loss": 0.8885,
+      "step": 842
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991661201188034e-05,
+      "loss": 0.8905,
+      "step": 843
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9916265409054665e-05,
+      "loss": 0.893,
+      "step": 844
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991591809042192e-05,
+      "loss": 0.3359,
+      "step": 845
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9915570056007168e-05,
+      "loss": 0.9368,
+      "step": 846
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9915221305835536e-05,
+      "loss": 0.9651,
+      "step": 847
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9914871839932204e-05,
+      "loss": 0.8805,
+      "step": 848
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991452165832239e-05,
+      "loss": 1.0026,
+      "step": 849
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991417076103138e-05,
+      "loss": 0.8114,
+      "step": 850
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9913819148084498e-05,
+      "loss": 0.9365,
+      "step": 851
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9913466819507123e-05,
+      "loss": 0.9058,
+      "step": 852
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9913113775324696e-05,
+      "loss": 0.9035,
+      "step": 853
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991276001556269e-05,
+      "loss": 0.933,
+      "step": 854
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9912405540246656e-05,
+      "loss": 0.8891,
+      "step": 855
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991205034940217e-05,
+      "loss": 0.3328,
+      "step": 856
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9911694443054876e-05,
+      "loss": 0.8979,
+      "step": 857
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9911337821230465e-05,
+      "loss": 0.9186,
+      "step": 858
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9910980483954678e-05,
+      "loss": 0.8553,
+      "step": 859
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991062243125331e-05,
+      "loss": 0.8284,
+      "step": 860
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.991026366315221e-05,
+      "loss": 0.9223,
+      "step": 861
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9909904179677272e-05,
+      "loss": 0.9434,
+      "step": 862
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990954398085445e-05,
+      "loss": 0.3135,
+      "step": 863
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990918306670974e-05,
+      "loss": 0.8801,
+      "step": 864
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9908821437269198e-05,
+      "loss": 0.3158,
+      "step": 865
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9908459092558928e-05,
+      "loss": 0.8452,
+      "step": 866
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990809603260508e-05,
+      "loss": 0.838,
+      "step": 867
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990773225743387e-05,
+      "loss": 0.932,
+      "step": 868
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990736776707155e-05,
+      "loss": 0.9186,
+      "step": 869
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990700256154444e-05,
+      "loss": 0.8892,
+      "step": 870
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9906636640878892e-05,
+      "loss": 0.909,
+      "step": 871
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9906270005101328e-05,
+      "loss": 0.8775,
+      "step": 872
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990590265423821e-05,
+      "loss": 0.8619,
+      "step": 873
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9905534588316055e-05,
+      "loss": 0.8748,
+      "step": 874
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9905165807361432e-05,
+      "loss": 0.9155,
+      "step": 875
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990479631140096e-05,
+      "loss": 0.3419,
+      "step": 876
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990442610046132e-05,
+      "loss": 0.8899,
+      "step": 877
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990405517456922e-05,
+      "loss": 0.9093,
+      "step": 878
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990368353375145e-05,
+      "loss": 0.8922,
+      "step": 879
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990331117803483e-05,
+      "loss": 0.8795,
+      "step": 880
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990293810744624e-05,
+      "loss": 0.8268,
+      "step": 881
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990256432201261e-05,
+      "loss": 0.8196,
+      "step": 882
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9902189821760924e-05,
+      "loss": 0.9898,
+      "step": 883
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990181460671821e-05,
+      "loss": 0.9537,
+      "step": 884
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9901438676911556e-05,
+      "loss": 0.9108,
+      "step": 885
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9901062032368098e-05,
+      "loss": 0.8578,
+      "step": 886
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9900684673115023e-05,
+      "loss": 0.8775,
+      "step": 887
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9900306599179577e-05,
+      "loss": 0.8785,
+      "step": 888
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989992781058905e-05,
+      "loss": 0.9234,
+      "step": 889
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9899548307370777e-05,
+      "loss": 0.8901,
+      "step": 890
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9899168089552155e-05,
+      "loss": 0.9032,
+      "step": 891
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9898787157160635e-05,
+      "loss": 0.8799,
+      "step": 892
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9898405510223712e-05,
+      "loss": 0.911,
+      "step": 893
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989802314876893e-05,
+      "loss": 0.9646,
+      "step": 894
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9897640072823902e-05,
+      "loss": 0.8843,
+      "step": 895
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989725628241627e-05,
+      "loss": 0.8529,
+      "step": 896
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989687177757374e-05,
+      "loss": 0.9457,
+      "step": 897
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9896486558324075e-05,
+      "loss": 0.9029,
+      "step": 898
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989610062469507e-05,
+      "loss": 0.9178,
+      "step": 899
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.989571397671459e-05,
+      "loss": 0.9027,
+      "step": 900
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9895326614410548e-05,
+      "loss": 0.9102,
+      "step": 901
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.98949385378109e-05,
+      "loss": 0.9304,
+      "step": 902
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9894549746943666e-05,
+      "loss": 0.8774,
+      "step": 903
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9894160241836903e-05,
+      "loss": 0.9282,
+      "step": 904
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9893770022518734e-05,
+      "loss": 0.8616,
+      "step": 905
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9893379089017322e-05,
+      "loss": 0.8888,
+      "step": 906
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989298744136089e-05,
+      "loss": 0.8534,
+      "step": 907
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9892595079577708e-05,
+      "loss": 0.9231,
+      "step": 908
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98922020036961e-05,
+      "loss": 0.877,
+      "step": 909
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9891808213744438e-05,
+      "loss": 0.9011,
+      "step": 910
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.989141370975115e-05,
+      "loss": 0.8806,
+      "step": 911
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9891018491744712e-05,
+      "loss": 0.8459,
+      "step": 912
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9890622559753655e-05,
+      "loss": 0.8797,
+      "step": 913
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9890225913806557e-05,
+      "loss": 0.8994,
+      "step": 914
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9889828553932047e-05,
+      "loss": 0.972,
+      "step": 915
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9889430480158815e-05,
+      "loss": 0.8573,
+      "step": 916
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988903169251559e-05,
+      "loss": 0.8931,
+      "step": 917
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9888632191031167e-05,
+      "loss": 0.899,
+      "step": 918
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9888231975734374e-05,
+      "loss": 0.923,
+      "step": 919
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9887831046654108e-05,
+      "loss": 0.9246,
+      "step": 920
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9887429403819307e-05,
+      "loss": 0.8756,
+      "step": 921
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988702704725896e-05,
+      "loss": 0.8834,
+      "step": 922
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9886623977002116e-05,
+      "loss": 0.889,
+      "step": 923
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9886220193077875e-05,
+      "loss": 0.8536,
+      "step": 924
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9885815695515375e-05,
+      "loss": 0.914,
+      "step": 925
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9885410484343822e-05,
+      "loss": 0.8889,
+      "step": 926
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988500455959246e-05,
+      "loss": 0.8239,
+      "step": 927
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9884597921290596e-05,
+      "loss": 0.9342,
+      "step": 928
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988419056946758e-05,
+      "loss": 0.7987,
+      "step": 929
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9883782504152816e-05,
+      "loss": 0.8688,
+      "step": 930
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9883373725375763e-05,
+      "loss": 0.8413,
+      "step": 931
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9882964233165928e-05,
+      "loss": 0.9489,
+      "step": 932
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9882554027552873e-05,
+      "loss": 0.9474,
+      "step": 933
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98821431085662e-05,
+      "loss": 0.8911,
+      "step": 934
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9881731476235582e-05,
+      "loss": 0.9081,
+      "step": 935
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9881319130590723e-05,
+      "loss": 0.9677,
+      "step": 936
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9880906071661397e-05,
+      "loss": 0.8583,
+      "step": 937
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9880492299477415e-05,
+      "loss": 0.8508,
+      "step": 938
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9880077814068648e-05,
+      "loss": 0.8937,
+      "step": 939
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9879662615465013e-05,
+      "loss": 0.8711,
+      "step": 940
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9879246703696484e-05,
+      "loss": 0.9476,
+      "step": 941
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987883007879308e-05,
+      "loss": 0.9357,
+      "step": 942
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987841274078488e-05,
+      "loss": 0.8531,
+      "step": 943
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9877994689702007e-05,
+      "loss": 0.9594,
+      "step": 944
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987757592557464e-05,
+      "loss": 0.9072,
+      "step": 945
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9877156448433005e-05,
+      "loss": 0.9295,
+      "step": 946
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9876736258307377e-05,
+      "loss": 0.89,
+      "step": 947
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98763153552281e-05,
+      "loss": 0.92,
+      "step": 948
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9875893739225546e-05,
+      "loss": 0.9131,
+      "step": 949
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987547141033016e-05,
+      "loss": 0.9493,
+      "step": 950
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9875048368572414e-05,
+      "loss": 0.9194,
+      "step": 951
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9874624613982857e-05,
+      "loss": 1.0044,
+      "step": 952
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9874200146592072e-05,
+      "loss": 1.0006,
+      "step": 953
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98737749664307e-05,
+      "loss": 0.8801,
+      "step": 954
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987334907352944e-05,
+      "loss": 0.9056,
+      "step": 955
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.987292246791902e-05,
+      "loss": 0.884,
+      "step": 956
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9872495149630247e-05,
+      "loss": 0.9131,
+      "step": 957
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9872067118693967e-05,
+      "loss": 0.8881,
+      "step": 958
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9871638375141065e-05,
+      "loss": 0.9469,
+      "step": 959
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9871208919002507e-05,
+      "loss": 0.8854,
+      "step": 960
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9870778750309282e-05,
+      "loss": 0.9455,
+      "step": 961
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9870347869092447e-05,
+      "loss": 0.8505,
+      "step": 962
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9869916275383098e-05,
+      "loss": 0.9024,
+      "step": 963
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98694839692124e-05,
+      "loss": 0.9505,
+      "step": 964
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9869050950611547e-05,
+      "loss": 0.9454,
+      "step": 965
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986861721961181e-05,
+      "loss": 0.9341,
+      "step": 966
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9868182776244488e-05,
+      "loss": 0.9116,
+      "step": 967
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9867747620540946e-05,
+      "loss": 0.9333,
+      "step": 968
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9867311752532597e-05,
+      "loss": 0.93,
+      "step": 969
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98668751722509e-05,
+      "loss": 0.8346,
+      "step": 970
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986643787972737e-05,
+      "loss": 0.412,
+      "step": 971
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9865999874993572e-05,
+      "loss": 0.9044,
+      "step": 972
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9865561158081127e-05,
+      "loss": 1.0112,
+      "step": 973
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9865121729021706e-05,
+      "loss": 0.8611,
+      "step": 974
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9864681587847023e-05,
+      "loss": 0.9052,
+      "step": 975
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9864240734588854e-05,
+      "loss": 0.8551,
+      "step": 976
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9863799169279018e-05,
+      "loss": 0.9141,
+      "step": 977
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9863356891949393e-05,
+      "loss": 0.8651,
+      "step": 978
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9862913902631903e-05,
+      "loss": 0.8513,
+      "step": 979
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986247020135853e-05,
+      "loss": 0.9776,
+      "step": 980
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9862025788161293e-05,
+      "loss": 0.9573,
+      "step": 981
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.986158066307228e-05,
+      "loss": 0.8424,
+      "step": 982
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9861134826123622e-05,
+      "loss": 0.7636,
+      "step": 983
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98606882773475e-05,
+      "loss": 0.926,
+      "step": 984
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9860241016776144e-05,
+      "loss": 0.8707,
+      "step": 985
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9859793044441847e-05,
+      "loss": 0.8705,
+      "step": 986
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9859344360376945e-05,
+      "loss": 0.952,
+      "step": 987
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9858894964613823e-05,
+      "loss": 0.9116,
+      "step": 988
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9858444857184918e-05,
+      "loss": 0.953,
+      "step": 989
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9857994038122726e-05,
+      "loss": 0.8884,
+      "step": 990
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985754250745979e-05,
+      "loss": 0.834,
+      "step": 991
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98570902652287e-05,
+      "loss": 0.8422,
+      "step": 992
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9856637311462102e-05,
+      "loss": 0.8912,
+      "step": 993
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9856183646192696e-05,
+      "loss": 0.8961,
+      "step": 994
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985572926945323e-05,
+      "loss": 0.9214,
+      "step": 995
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9855274181276498e-05,
+      "loss": 0.9155,
+      "step": 996
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9854818381695354e-05,
+      "loss": 0.8591,
+      "step": 997
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9854361870742698e-05,
+      "loss": 0.8797,
+      "step": 998
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9853904648451484e-05,
+      "loss": 0.9051,
+      "step": 999
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985344671485472e-05,
+      "loss": 0.9579,
+      "step": 1000
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985298806998546e-05,
+      "loss": 0.9853,
+      "step": 1001
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.985252871387681e-05,
+      "loss": 0.8501,
+      "step": 1002
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9852068646561928e-05,
+      "loss": 0.8809,
+      "step": 1003
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9851607868074026e-05,
+      "loss": 0.8924,
+      "step": 1004
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9851146378446365e-05,
+      "loss": 0.889,
+      "step": 1005
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9850684177712258e-05,
+      "loss": 0.9084,
+      "step": 1006
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9850221265905068e-05,
+      "loss": 0.8713,
+      "step": 1007
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9849757643058208e-05,
+      "loss": 0.9646,
+      "step": 1008
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9849293309205153e-05,
+      "loss": 0.9425,
+      "step": 1009
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9848828264379412e-05,
+      "loss": 0.8378,
+      "step": 1010
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984836250861456e-05,
+      "loss": 0.9374,
+      "step": 1011
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9847896041944214e-05,
+      "loss": 0.9443,
+      "step": 1012
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9847428864402048e-05,
+      "loss": 0.3965,
+      "step": 1013
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9846960976021786e-05,
+      "loss": 0.9204,
+      "step": 1014
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.98464923768372e-05,
+      "loss": 0.9297,
+      "step": 1015
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9846023066882117e-05,
+      "loss": 0.8819,
+      "step": 1016
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9845553046190414e-05,
+      "loss": 0.9214,
+      "step": 1017
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9845082314796022e-05,
+      "loss": 0.9271,
+      "step": 1018
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984461087273292e-05,
+      "loss": 0.9172,
+      "step": 1019
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9844138720035135e-05,
+      "loss": 0.866,
+      "step": 1020
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9843665856736752e-05,
+      "loss": 0.9238,
+      "step": 1021
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.984319228287191e-05,
+      "loss": 0.8895,
+      "step": 1022
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9842717998474783e-05,
+      "loss": 0.8381,
+      "step": 1023
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9842243003579614e-05,
+      "loss": 0.8966,
+      "step": 1024
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9841767298220694e-05,
+      "loss": 0.8538,
+      "step": 1025
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9841290882432357e-05,
+      "loss": 0.8817,
+      "step": 1026
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.984081375624899e-05,
+      "loss": 0.9103,
+      "step": 1027
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9840335919705043e-05,
+      "loss": 0.8545,
+      "step": 1028
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9839857372835005e-05,
+      "loss": 0.933,
+      "step": 1029
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983937811567341e-05,
+      "loss": 0.9158,
+      "step": 1030
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983889814825487e-05,
+      "loss": 0.975,
+      "step": 1031
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9838417470614022e-05,
+      "loss": 0.891,
+      "step": 1032
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983793608278557e-05,
+      "loss": 0.8215,
+      "step": 1033
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983745398480425e-05,
+      "loss": 0.9036,
+      "step": 1034
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9836971176704877e-05,
+      "loss": 0.9126,
+      "step": 1035
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9836487658522295e-05,
+      "loss": 0.8891,
+      "step": 1036
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983600343029141e-05,
+      "loss": 0.9308,
+      "step": 1037
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983551849204717e-05,
+      "loss": 0.8876,
+      "step": 1038
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983503284382459e-05,
+      "loss": 0.9195,
+      "step": 1039
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9834546485658718e-05,
+      "loss": 0.8685,
+      "step": 1040
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9834059417584663e-05,
+      "loss": 0.7918,
+      "step": 1041
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983357163963759e-05,
+      "loss": 0.8712,
+      "step": 1042
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9833083151852704e-05,
+      "loss": 0.904,
+      "step": 1043
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983259395426527e-05,
+      "loss": 0.8922,
+      "step": 1044
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.98321040469106e-05,
+      "loss": 0.9602,
+      "step": 1045
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9831613429824054e-05,
+      "loss": 0.9258,
+      "step": 1046
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983112210304105e-05,
+      "loss": 0.8864,
+      "step": 1047
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983063006659706e-05,
+      "loss": 0.9248,
+      "step": 1048
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983013732052759e-05,
+      "loss": 0.9649,
+      "step": 1049
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9829643864868217e-05,
+      "loss": 0.8663,
+      "step": 1050
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982914969965456e-05,
+      "loss": 0.8977,
+      "step": 1051
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9828654824922294e-05,
+      "loss": 0.8302,
+      "step": 1052
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9828159240707135e-05,
+      "loss": 0.892,
+      "step": 1053
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9827662947044857e-05,
+      "loss": 0.838,
+      "step": 1054
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.982716594397129e-05,
+      "loss": 0.9494,
+      "step": 1055
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9826668231522304e-05,
+      "loss": 0.8841,
+      "step": 1056
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9826169809733836e-05,
+      "loss": 0.9081,
+      "step": 1057
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9825670678641856e-05,
+      "loss": 0.9525,
+      "step": 1058
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9825170838282395e-05,
+      "loss": 0.8779,
+      "step": 1059
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9824670288691537e-05,
+      "loss": 0.9427,
+      "step": 1060
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9824169029905414e-05,
+      "loss": 0.8982,
+      "step": 1061
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9823667061960207e-05,
+      "loss": 0.9256,
+      "step": 1062
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9823164384892152e-05,
+      "loss": 0.8647,
+      "step": 1063
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9822660998737534e-05,
+      "loss": 0.9251,
+      "step": 1064
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9822156903532695e-05,
+      "loss": 0.8756,
+      "step": 1065
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9821652099314016e-05,
+      "loss": 0.9026,
+      "step": 1066
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9821146586117938e-05,
+      "loss": 0.879,
+      "step": 1067
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9820640363980958e-05,
+      "loss": 0.9537,
+      "step": 1068
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9820133432939607e-05,
+      "loss": 0.8791,
+      "step": 1069
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9819625793030485e-05,
+      "loss": 0.9256,
+      "step": 1070
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9819117444290236e-05,
+      "loss": 0.9178,
+      "step": 1071
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981860838675555e-05,
+      "loss": 0.9376,
+      "step": 1072
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981809862046318e-05,
+      "loss": 0.8409,
+      "step": 1073
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9817588145449925e-05,
+      "loss": 0.9266,
+      "step": 1074
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9817076961752624e-05,
+      "loss": 0.9116,
+      "step": 1075
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9816565069408184e-05,
+      "loss": 0.9541,
+      "step": 1076
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9816052468453554e-05,
+      "loss": 0.8969,
+      "step": 1077
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981553915892574e-05,
+      "loss": 0.9824,
+      "step": 1078
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9815025140861787e-05,
+      "loss": 0.8153,
+      "step": 1079
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9814510414298806e-05,
+      "loss": 0.9653,
+      "step": 1080
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981399497927395e-05,
+      "loss": 0.8662,
+      "step": 1081
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981347883582443e-05,
+      "loss": 0.9121,
+      "step": 1082
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9812961983987497e-05,
+      "loss": 0.9015,
+      "step": 1083
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9812444423800464e-05,
+      "loss": 0.8972,
+      "step": 1084
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9811926155300693e-05,
+      "loss": 0.8151,
+      "step": 1085
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981140717852559e-05,
+      "loss": 0.8819,
+      "step": 1086
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.981088749351262e-05,
+      "loss": 0.896,
+      "step": 1087
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.98103671002993e-05,
+      "loss": 0.8732,
+      "step": 1088
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9809845998923195e-05,
+      "loss": 0.8659,
+      "step": 1089
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9809324189421913e-05,
+      "loss": 0.8646,
+      "step": 1090
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9808801671833125e-05,
+      "loss": 0.9036,
+      "step": 1091
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9808278446194552e-05,
+      "loss": 0.938,
+      "step": 1092
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9807754512543957e-05,
+      "loss": 0.9104,
+      "step": 1093
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9807229870919166e-05,
+      "loss": 0.8894,
+      "step": 1094
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9806704521358048e-05,
+      "loss": 0.9684,
+      "step": 1095
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9806178463898522e-05,
+      "loss": 0.8866,
+      "step": 1096
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980565169857857e-05,
+      "loss": 0.7954,
+      "step": 1097
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980512422543621e-05,
+      "loss": 0.9166,
+      "step": 1098
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9804596044509517e-05,
+      "loss": 0.9142,
+      "step": 1099
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9804067155836622e-05,
+      "loss": 0.8828,
+      "step": 1100
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9803537559455703e-05,
+      "loss": 0.8795,
+      "step": 1101
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9803007255404986e-05,
+      "loss": 0.8996,
+      "step": 1102
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9802476243722753e-05,
+      "loss": 0.8813,
+      "step": 1103
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980194452444733e-05,
+      "loss": 0.8753,
+      "step": 1104
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9801412097617108e-05,
+      "loss": 0.8662,
+      "step": 1105
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9800878963270515e-05,
+      "loss": 0.9363,
+      "step": 1106
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9800345121446036e-05,
+      "loss": 0.8572,
+      "step": 1107
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9799810572182205e-05,
+      "loss": 0.9362,
+      "step": 1108
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979927531551761e-05,
+      "loss": 0.8734,
+      "step": 1109
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979873935149089e-05,
+      "loss": 0.8833,
+      "step": 1110
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9798202680140736e-05,
+      "loss": 0.8581,
+      "step": 1111
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979766530150588e-05,
+      "loss": 0.8648,
+      "step": 1112
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979712721562512e-05,
+      "loss": 0.8966,
+      "step": 1113
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979658842253729e-05,
+      "loss": 0.8931,
+      "step": 1114
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979604892228129e-05,
+      "loss": 0.9198,
+      "step": 1115
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979550871489606e-05,
+      "loss": 0.9096,
+      "step": 1116
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.97949678004206e-05,
+      "loss": 0.8437,
+      "step": 1117
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979442617889395e-05,
+      "loss": 0.843,
+      "step": 1118
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979388385035521e-05,
+      "loss": 0.8879,
+      "step": 1119
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979334081484353e-05,
+      "loss": 0.943,
+      "step": 1120
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.97927970723981e-05,
+      "loss": 0.9188,
+      "step": 1121
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979225262305818e-05,
+      "loss": 0.8706,
+      "step": 1122
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979170746686307e-05,
+      "loss": 0.8892,
+      "step": 1123
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.979116160385212e-05,
+      "loss": 0.8955,
+      "step": 1124
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9790615034064732e-05,
+      "loss": 0.8488,
+      "step": 1125
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9790067757540364e-05,
+      "loss": 0.9121,
+      "step": 1126
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9789519774318516e-05,
+      "loss": 0.883,
+      "step": 1127
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9788971084438743e-05,
+      "loss": 0.9254,
+      "step": 1128
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9788421687940664e-05,
+      "loss": 0.8766,
+      "step": 1129
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9787871584863925e-05,
+      "loss": 0.864,
+      "step": 1130
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978732077524824e-05,
+      "loss": 0.8474,
+      "step": 1131
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978676925913337e-05,
+      "loss": 0.8962,
+      "step": 1132
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9786217036559125e-05,
+      "loss": 0.8883,
+      "step": 1133
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978566410756537e-05,
+      "loss": 0.9389,
+      "step": 1134
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978511047219201e-05,
+      "loss": 0.8627,
+      "step": 1135
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978455613047902e-05,
+      "loss": 0.9088,
+      "step": 1136
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9784001082466408e-05,
+      "loss": 0.8936,
+      "step": 1137
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9783445328194243e-05,
+      "loss": 0.892,
+      "step": 1138
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9782888867702642e-05,
+      "loss": 0.8908,
+      "step": 1139
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9782331701031776e-05,
+      "loss": 0.974,
+      "step": 1140
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9781773828221857e-05,
+      "loss": 0.8744,
+      "step": 1141
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.978121524931316e-05,
+      "loss": 0.8868,
+      "step": 1142
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9780655964346006e-05,
+      "loss": 0.9127,
+      "step": 1143
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9780095973360766e-05,
+      "loss": 0.8154,
+      "step": 1144
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9779535276397864e-05,
+      "loss": 0.9199,
+      "step": 1145
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9778973873497774e-05,
+      "loss": 0.846,
+      "step": 1146
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9778411764701016e-05,
+      "loss": 0.9031,
+      "step": 1147
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9777848950048176e-05,
+      "loss": 0.9318,
+      "step": 1148
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977728542957987e-05,
+      "loss": 0.8951,
+      "step": 1149
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9776721203336785e-05,
+      "loss": 0.8284,
+      "step": 1150
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9776156271359644e-05,
+      "loss": 0.8825,
+      "step": 1151
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977559063368923e-05,
+      "loss": 0.9466,
+      "step": 1152
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9775024290366372e-05,
+      "loss": 0.8793,
+      "step": 1153
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9774457241431947e-05,
+      "loss": 0.9155,
+      "step": 1154
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97738894869269e-05,
+      "loss": 0.9309,
+      "step": 1155
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9773321026892202e-05,
+      "loss": 0.957,
+      "step": 1156
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.977275186136889e-05,
+      "loss": 0.8269,
+      "step": 1157
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9772181990398052e-05,
+      "loss": 0.8871,
+      "step": 1158
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9771611414020827e-05,
+      "loss": 0.9119,
+      "step": 1159
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9771040132278395e-05,
+      "loss": 0.9267,
+      "step": 1160
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9770468145212e-05,
+      "loss": 0.8614,
+      "step": 1161
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9769895452862928e-05,
+      "loss": 0.9191,
+      "step": 1162
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976932205527252e-05,
+      "loss": 0.843,
+      "step": 1163
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9768747952482164e-05,
+      "loss": 0.8646,
+      "step": 1164
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9768173144533305e-05,
+      "loss": 0.9012,
+      "step": 1165
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976759763146744e-05,
+      "loss": 0.8371,
+      "step": 1166
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9767021413326098e-05,
+      "loss": 0.8998,
+      "step": 1167
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9766444490150887e-05,
+      "loss": 0.8833,
+      "step": 1168
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976586686198345e-05,
+      "loss": 0.9173,
+      "step": 1169
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9765288528865474e-05,
+      "loss": 0.8411,
+      "step": 1170
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976470949083872e-05,
+      "loss": 0.9057,
+      "step": 1171
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9764129747944975e-05,
+      "loss": 0.8576,
+      "step": 1172
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9763549300226094e-05,
+      "loss": 0.9301,
+      "step": 1173
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9762968147723972e-05,
+      "loss": 0.874,
+      "step": 1174
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9762386290480565e-05,
+      "loss": 0.9045,
+      "step": 1175
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976180372853787e-05,
+      "loss": 0.8372,
+      "step": 1176
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9761220461937937e-05,
+      "loss": 0.8981,
+      "step": 1177
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9760636490722877e-05,
+      "loss": 0.8875,
+      "step": 1178
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.976005181493484e-05,
+      "loss": 0.9065,
+      "step": 1179
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9759466434616034e-05,
+      "loss": 0.8651,
+      "step": 1180
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9758880349808707e-05,
+      "loss": 0.9874,
+      "step": 1181
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975829356055517e-05,
+      "loss": 0.3472,
+      "step": 1182
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9757706066897783e-05,
+      "loss": 0.9164,
+      "step": 1183
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975711786887895e-05,
+      "loss": 0.9216,
+      "step": 1184
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9756528966541134e-05,
+      "loss": 0.9091,
+      "step": 1185
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9755939359926842e-05,
+      "loss": 0.896,
+      "step": 1186
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9755349049078642e-05,
+      "loss": 0.9318,
+      "step": 1187
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975475803403913e-05,
+      "loss": 0.9087,
+      "step": 1188
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9754166314850988e-05,
+      "loss": 0.347,
+      "step": 1189
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9753573891556912e-05,
+      "loss": 0.8975,
+      "step": 1190
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975298076419968e-05,
+      "loss": 0.9437,
+      "step": 1191
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9752386932822097e-05,
+      "loss": 0.9,
+      "step": 1192
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.975179239746703e-05,
+      "loss": 0.9048,
+      "step": 1193
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9751197158177403e-05,
+      "loss": 0.8341,
+      "step": 1194
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9750601214996176e-05,
+      "loss": 0.9378,
+      "step": 1195
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9750004567966372e-05,
+      "loss": 0.9426,
+      "step": 1196
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9749407217131055e-05,
+      "loss": 0.8639,
+      "step": 1197
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974880916253335e-05,
+      "loss": 0.8603,
+      "step": 1198
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9748210404216427e-05,
+      "loss": 0.9355,
+      "step": 1199
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9747610942223504e-05,
+      "loss": 0.8942,
+      "step": 1200
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9747010776597855e-05,
+      "loss": 0.3477,
+      "step": 1201
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97464099073828e-05,
+      "loss": 0.8584,
+      "step": 1202
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.974580833462172e-05,
+      "loss": 0.3468,
+      "step": 1203
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9745206058358034e-05,
+      "loss": 0.8495,
+      "step": 1204
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9744603078635224e-05,
+      "loss": 0.9433,
+      "step": 1205
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9743999395496803e-05,
+      "loss": 0.8899,
+      "step": 1206
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9743395008986363e-05,
+      "loss": 0.9069,
+      "step": 1207
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9742789919147523e-05,
+      "loss": 0.3505,
+      "step": 1208
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9742184126023962e-05,
+      "loss": 0.9032,
+      "step": 1209
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9741577629659412e-05,
+      "loss": 0.9034,
+      "step": 1210
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9740970430097653e-05,
+      "loss": 0.8685,
+      "step": 1211
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9740362527382514e-05,
+      "loss": 0.952,
+      "step": 1212
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9739753921557872e-05,
+      "loss": 0.8802,
+      "step": 1213
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9739144612667668e-05,
+      "loss": 0.8794,
+      "step": 1214
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9738534600755882e-05,
+      "loss": 0.8836,
+      "step": 1215
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9737923885866547e-05,
+      "loss": 0.9148,
+      "step": 1216
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9737312468043747e-05,
+      "loss": 0.8721,
+      "step": 1217
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973670034733162e-05,
+      "loss": 0.8552,
+      "step": 1218
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9736087523774346e-05,
+      "loss": 0.8828,
+      "step": 1219
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9735473997416173e-05,
+      "loss": 0.9056,
+      "step": 1220
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9734859768301375e-05,
+      "loss": 0.8602,
+      "step": 1221
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9734244836474297e-05,
+      "loss": 0.9289,
+      "step": 1222
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973362920197933e-05,
+      "loss": 0.918,
+      "step": 1223
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973301286486091e-05,
+      "loss": 0.822,
+      "step": 1224
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973239582516353e-05,
+      "loss": 0.8767,
+      "step": 1225
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9731778082931728e-05,
+      "loss": 0.9573,
+      "step": 1226
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97311596382101e-05,
+      "loss": 0.8428,
+      "step": 1227
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9730540491043278e-05,
+      "loss": 0.8977,
+      "step": 1228
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972992064147597e-05,
+      "loss": 0.8379,
+      "step": 1229
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9729300089552912e-05,
+      "loss": 0.8544,
+      "step": 1230
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97286788353189e-05,
+      "loss": 0.8754,
+      "step": 1231
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9728056878818778e-05,
+      "loss": 0.8503,
+      "step": 1232
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9727434220097445e-05,
+      "loss": 0.8926,
+      "step": 1233
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9726810859199843e-05,
+      "loss": 0.9203,
+      "step": 1234
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9726186796170973e-05,
+      "loss": 0.9404,
+      "step": 1235
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9725562031055886e-05,
+      "loss": 0.8481,
+      "step": 1236
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9724936563899672e-05,
+      "loss": 0.9467,
+      "step": 1237
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972431039474749e-05,
+      "loss": 0.8669,
+      "step": 1238
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9723683523644532e-05,
+      "loss": 0.8277,
+      "step": 1239
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9723055950636052e-05,
+      "loss": 0.8715,
+      "step": 1240
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9722427675767354e-05,
+      "loss": 0.8132,
+      "step": 1241
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.972179869908379e-05,
+      "loss": 0.3872,
+      "step": 1242
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9721169020630757e-05,
+      "loss": 0.9211,
+      "step": 1243
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9720538640453713e-05,
+      "loss": 0.9114,
+      "step": 1244
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971990755859816e-05,
+      "loss": 0.8778,
+      "step": 1245
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9719275775109656e-05,
+      "loss": 0.8424,
+      "step": 1246
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9718643290033804e-05,
+      "loss": 0.8342,
+      "step": 1247
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9718010103416262e-05,
+      "loss": 0.9239,
+      "step": 1248
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9717376215302734e-05,
+      "loss": 0.9138,
+      "step": 1249
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9716741625738975e-05,
+      "loss": 0.8108,
+      "step": 1250
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9716106334770798e-05,
+      "loss": 0.8129,
+      "step": 1251
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9715470342444063e-05,
+      "loss": 0.8513,
+      "step": 1252
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9714833648804676e-05,
+      "loss": 0.9111,
+      "step": 1253
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971419625389859e-05,
+      "loss": 0.9266,
+      "step": 1254
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.971355815777183e-05,
+      "loss": 0.9066,
+      "step": 1255
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9712919360470447e-05,
+      "loss": 0.8802,
+      "step": 1256
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9712279862040552e-05,
+      "loss": 0.9049,
+      "step": 1257
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9711639662528313e-05,
+      "loss": 0.8953,
+      "step": 1258
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9710998761979942e-05,
+      "loss": 0.9782,
+      "step": 1259
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97103571604417e-05,
+      "loss": 0.9326,
+      "step": 1260
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.97097148579599e-05,
+      "loss": 0.956,
+      "step": 1261
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.970907185458091e-05,
+      "loss": 0.8697,
+      "step": 1262
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9708428150351146e-05,
+      "loss": 0.8563,
+      "step": 1263
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9707783745317065e-05,
+      "loss": 0.8416,
+      "step": 1264
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9707138639525193e-05,
+      "loss": 0.9549,
+      "step": 1265
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9706492833022097e-05,
+      "loss": 0.8342,
+      "step": 1266
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970584632585439e-05,
+      "loss": 0.8686,
+      "step": 1267
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9705199118068745e-05,
+      "loss": 0.8729,
+      "step": 1268
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9704551209711877e-05,
+      "loss": 0.8799,
+      "step": 1269
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9703902600830554e-05,
+      "loss": 0.8883,
+      "step": 1270
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.97032532914716e-05,
+      "loss": 0.9275,
+      "step": 1271
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970260328168188e-05,
+      "loss": 0.9631,
+      "step": 1272
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9701952571508323e-05,
+      "loss": 0.9533,
+      "step": 1273
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9701301160997893e-05,
+      "loss": 0.8657,
+      "step": 1274
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.970064905019762e-05,
+      "loss": 0.8547,
+      "step": 1275
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969999623915457e-05,
+      "loss": 0.9589,
+      "step": 1276
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9699342727915867e-05,
+      "loss": 0.8553,
+      "step": 1277
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969868851652869e-05,
+      "loss": 0.862,
+      "step": 1278
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9698033605040253e-05,
+      "loss": 0.8378,
+      "step": 1279
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969737799349784e-05,
+      "loss": 0.8404,
+      "step": 1280
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969672168194878e-05,
+      "loss": 0.8601,
+      "step": 1281
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9696064670440436e-05,
+      "loss": 0.8894,
+      "step": 1282
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9695406959020245e-05,
+      "loss": 0.9311,
+      "step": 1283
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969474854773568e-05,
+      "loss": 0.8145,
+      "step": 1284
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9694089436634268e-05,
+      "loss": 0.8666,
+      "step": 1285
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9693429625763584e-05,
+      "loss": 0.9082,
+      "step": 1286
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9692769115171267e-05,
+      "loss": 0.9093,
+      "step": 1287
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9692107904904986e-05,
+      "loss": 0.9478,
+      "step": 1288
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9691445995012475e-05,
+      "loss": 0.9017,
+      "step": 1289
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.969078338554151e-05,
+      "loss": 0.9079,
+      "step": 1290
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9690120076539928e-05,
+      "loss": 0.8662,
+      "step": 1291
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968945606805561e-05,
+      "loss": 0.8314,
+      "step": 1292
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968879136013648e-05,
+      "loss": 0.9056,
+      "step": 1293
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9688125952830527e-05,
+      "loss": 0.8742,
+      "step": 1294
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9687459846185777e-05,
+      "loss": 0.8607,
+      "step": 1295
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9686793040250318e-05,
+      "loss": 0.8815,
+      "step": 1296
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9686125535072284e-05,
+      "loss": 0.9202,
+      "step": 1297
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9685457330699856e-05,
+      "loss": 0.8264,
+      "step": 1298
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968478842718127e-05,
+      "loss": 0.909,
+      "step": 1299
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.968411882456481e-05,
+      "loss": 0.9735,
+      "step": 1300
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9683448522898812e-05,
+      "loss": 0.8838,
+      "step": 1301
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9682777522231663e-05,
+      "loss": 0.8861,
+      "step": 1302
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9682105822611798e-05,
+      "loss": 0.3414,
+      "step": 1303
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9681433424087704e-05,
+      "loss": 0.9329,
+      "step": 1304
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9680760326707916e-05,
+      "loss": 0.8657,
+      "step": 1305
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9680086530521024e-05,
+      "loss": 0.9266,
+      "step": 1306
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9679412035575668e-05,
+      "loss": 0.8958,
+      "step": 1307
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9678736841920535e-05,
+      "loss": 0.8825,
+      "step": 1308
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967806094960436e-05,
+      "loss": 0.898,
+      "step": 1309
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967738435867594e-05,
+      "loss": 0.8771,
+      "step": 1310
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9676707069184106e-05,
+      "loss": 0.8199,
+      "step": 1311
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9676029081177752e-05,
+      "loss": 0.8938,
+      "step": 1312
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9675350394705825e-05,
+      "loss": 0.9029,
+      "step": 1313
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9674671009817302e-05,
+      "loss": 0.8492,
+      "step": 1314
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967399092656124e-05,
+      "loss": 0.9106,
+      "step": 1315
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9673310144986722e-05,
+      "loss": 0.8759,
+      "step": 1316
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9672628665142893e-05,
+      "loss": 0.8318,
+      "step": 1317
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967194648707894e-05,
+      "loss": 0.8572,
+      "step": 1318
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9671263610844117e-05,
+      "loss": 0.8558,
+      "step": 1319
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.967058003648771e-05,
+      "loss": 0.8544,
+      "step": 1320
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9669895764059063e-05,
+      "loss": 0.8525,
+      "step": 1321
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966921079360757e-05,
+      "loss": 0.8485,
+      "step": 1322
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966852512518268e-05,
+      "loss": 0.8629,
+      "step": 1323
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9667838758833882e-05,
+      "loss": 0.9468,
+      "step": 1324
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966715169461073e-05,
+      "loss": 0.9403,
+      "step": 1325
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966646393256281e-05,
+      "loss": 0.9259,
+      "step": 1326
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9665775472739776e-05,
+      "loss": 0.9392,
+      "step": 1327
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966508631519132e-05,
+      "loss": 0.9138,
+      "step": 1328
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966439645996719e-05,
+      "loss": 0.9135,
+      "step": 1329
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966370590711719e-05,
+      "loss": 0.9362,
+      "step": 1330
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9663014656691155e-05,
+      "loss": 0.8816,
+      "step": 1331
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9662322708738993e-05,
+      "loss": 0.8817,
+      "step": 1332
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9661630063310644e-05,
+      "loss": 0.8043,
+      "step": 1333
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9660936720456117e-05,
+      "loss": 0.9132,
+      "step": 1334
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966024268022545e-05,
+      "loss": 0.9511,
+      "step": 1335
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965954794266875e-05,
+      "loss": 0.9238,
+      "step": 1336
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965885250783617e-05,
+      "loss": 0.8961,
+      "step": 1337
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96581563757779e-05,
+      "loss": 0.8593,
+      "step": 1338
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965745954654419e-05,
+      "loss": 0.8239,
+      "step": 1339
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9656762020185355e-05,
+      "loss": 0.8915,
+      "step": 1340
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9656063796751735e-05,
+      "loss": 0.8882,
+      "step": 1341
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965536487629373e-05,
+      "loss": 0.9112,
+      "step": 1342
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9654665258861798e-05,
+      "loss": 0.824,
+      "step": 1343
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9653964944506436e-05,
+      "loss": 0.8442,
+      "step": 1344
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96532639332782e-05,
+      "loss": 0.874,
+      "step": 1345
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.965256222522769e-05,
+      "loss": 0.8418,
+      "step": 1346
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9651859820405557e-05,
+      "loss": 0.859,
+      "step": 1347
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9651156718862512e-05,
+      "loss": 0.9149,
+      "step": 1348
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9650452920649302e-05,
+      "loss": 0.8867,
+      "step": 1349
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9649748425816728e-05,
+      "loss": 0.8913,
+      "step": 1350
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9649043234415652e-05,
+      "loss": 0.9101,
+      "step": 1351
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964833734649697e-05,
+      "loss": 0.8963,
+      "step": 1352
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964763076211165e-05,
+      "loss": 0.8734,
+      "step": 1353
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9646923481310683e-05,
+      "loss": 0.9341,
+      "step": 1354
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964621550414513e-05,
+      "loss": 0.8567,
+      "step": 1355
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9645506830666095e-05,
+      "loss": 0.907,
+      "step": 1356
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9644797460924732e-05,
+      "loss": 0.8928,
+      "step": 1357
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9644087394972254e-05,
+      "loss": 0.8947,
+      "step": 1358
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964337663285991e-05,
+      "loss": 0.9108,
+      "step": 1359
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9642665174639012e-05,
+      "loss": 0.8712,
+      "step": 1360
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964195302036091e-05,
+      "loss": 0.8343,
+      "step": 1361
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9641240170077016e-05,
+      "loss": 0.876,
+      "step": 1362
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9640526623838787e-05,
+      "loss": 0.8778,
+      "step": 1363
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963981238169773e-05,
+      "loss": 0.8344,
+      "step": 1364
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.96390974437054e-05,
+      "loss": 0.8946,
+      "step": 1365
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.963838180991341e-05,
+      "loss": 0.8648,
+      "step": 1366
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9637665480373412e-05,
+      "loss": 0.9256,
+      "step": 1367
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9636948455137118e-05,
+      "loss": 0.9543,
+      "step": 1368
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9636230734256286e-05,
+      "loss": 0.8606,
+      "step": 1369
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9635512317782724e-05,
+      "loss": 0.9223,
+      "step": 1370
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9634793205768296e-05,
+      "loss": 0.8661,
+      "step": 1371
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9634073398264902e-05,
+      "loss": 0.9138,
+      "step": 1372
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9633352895324507e-05,
+      "loss": 0.8945,
+      "step": 1373
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9632631696999123e-05,
+      "loss": 0.9006,
+      "step": 1374
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9631909803340808e-05,
+      "loss": 0.8878,
+      "step": 1375
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9631187214401664e-05,
+      "loss": 0.9178,
+      "step": 1376
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9630463930233866e-05,
+      "loss": 0.8715,
+      "step": 1377
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.962973995088961e-05,
+      "loss": 0.8465,
+      "step": 1378
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9629015276421168e-05,
+      "loss": 0.8318,
+      "step": 1379
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9628289906880842e-05,
+      "loss": 0.8874,
+      "step": 1380
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9627563842320998e-05,
+      "loss": 0.9161,
+      "step": 1381
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.962683708279405e-05,
+      "loss": 0.7978,
+      "step": 1382
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.962610962835245e-05,
+      "loss": 0.8412,
+      "step": 1383
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9625381479048714e-05,
+      "loss": 0.8556,
+      "step": 1384
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9624652634935403e-05,
+      "loss": 0.3464,
+      "step": 1385
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.962392309606513e-05,
+      "loss": 0.9932,
+      "step": 1386
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.962319286249056e-05,
+      "loss": 0.8436,
+      "step": 1387
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.96224619342644e-05,
+      "loss": 0.7986,
+      "step": 1388
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9621730311439413e-05,
+      "loss": 0.8602,
+      "step": 1389
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.962099799406841e-05,
+      "loss": 0.8699,
+      "step": 1390
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9620264982204258e-05,
+      "loss": 0.922,
+      "step": 1391
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9619531275899867e-05,
+      "loss": 0.9022,
+      "step": 1392
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9618796875208202e-05,
+      "loss": 0.8662,
+      "step": 1393
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961806178018227e-05,
+      "loss": 0.8898,
+      "step": 1394
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9617325990875135e-05,
+      "loss": 0.8175,
+      "step": 1395
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9616589507339916e-05,
+      "loss": 0.9168,
+      "step": 1396
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9615852329629775e-05,
+      "loss": 0.8635,
+      "step": 1397
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961511445779792e-05,
+      "loss": 0.8752,
+      "step": 1398
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9614375891897617e-05,
+      "loss": 0.8919,
+      "step": 1399
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961363663198218e-05,
+      "loss": 0.8881,
+      "step": 1400
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961289667810497e-05,
+      "loss": 0.8898,
+      "step": 1401
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9612156030319407e-05,
+      "loss": 0.8465,
+      "step": 1402
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.961141468867895e-05,
+      "loss": 0.9292,
+      "step": 1403
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9610672653237117e-05,
+      "loss": 0.8608,
+      "step": 1404
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9609929924047462e-05,
+      "loss": 0.9454,
+      "step": 1405
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9609186501163608e-05,
+      "loss": 0.825,
+      "step": 1406
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9608442384639218e-05,
+      "loss": 0.9054,
+      "step": 1407
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9607697574528005e-05,
+      "loss": 0.9803,
+      "step": 1408
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960695207088373e-05,
+      "loss": 0.8196,
+      "step": 1409
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9606205873760217e-05,
+      "loss": 0.9342,
+      "step": 1410
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9605458983211317e-05,
+      "loss": 0.891,
+      "step": 1411
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9604711399290958e-05,
+      "loss": 0.858,
+      "step": 1412
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960396312205309e-05,
+      "loss": 0.8834,
+      "step": 1413
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960321415155174e-05,
+      "loss": 0.9223,
+      "step": 1414
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9602464487840968e-05,
+      "loss": 0.9275,
+      "step": 1415
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9601714130974886e-05,
+      "loss": 0.9269,
+      "step": 1416
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.960096308100766e-05,
+      "loss": 0.861,
+      "step": 1417
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9600211337993508e-05,
+      "loss": 0.8805,
+      "step": 1418
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959945890198669e-05,
+      "loss": 0.8552,
+      "step": 1419
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9598705773041526e-05,
+      "loss": 0.9132,
+      "step": 1420
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9597951951212376e-05,
+      "loss": 0.9533,
+      "step": 1421
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9597197436553654e-05,
+      "loss": 0.9092,
+      "step": 1422
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9596442229119833e-05,
+      "loss": 0.9375,
+      "step": 1423
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9595686328965416e-05,
+      "loss": 0.4068,
+      "step": 1424
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9594929736144978e-05,
+      "loss": 0.8279,
+      "step": 1425
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9594172450713126e-05,
+      "loss": 0.9207,
+      "step": 1426
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9593414472724527e-05,
+      "loss": 0.7812,
+      "step": 1427
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9592655802233903e-05,
+      "loss": 0.3554,
+      "step": 1428
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9591896439296007e-05,
+      "loss": 0.9217,
+      "step": 1429
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9591136383965664e-05,
+      "loss": 0.8769,
+      "step": 1430
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.959037563629773e-05,
+      "loss": 0.8693,
+      "step": 1431
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9589614196347124e-05,
+      "loss": 0.8637,
+      "step": 1432
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9588852064168814e-05,
+      "loss": 0.8194,
+      "step": 1433
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958808923981781e-05,
+      "loss": 0.8318,
+      "step": 1434
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9587325723349172e-05,
+      "loss": 0.8985,
+      "step": 1435
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9586561514818026e-05,
+      "loss": 0.931,
+      "step": 1436
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9585796614279532e-05,
+      "loss": 0.8792,
+      "step": 1437
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9585031021788897e-05,
+      "loss": 0.9345,
+      "step": 1438
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9584264737401398e-05,
+      "loss": 0.8886,
+      "step": 1439
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958349776117234e-05,
+      "loss": 0.8325,
+      "step": 1440
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.958273009315709e-05,
+      "loss": 0.8835,
+      "step": 1441
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9581961733411063e-05,
+      "loss": 0.9051,
+      "step": 1442
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9581192681989726e-05,
+      "loss": 0.9428,
+      "step": 1443
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9580422938948582e-05,
+      "loss": 0.9269,
+      "step": 1444
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9579652504343207e-05,
+      "loss": 0.8073,
+      "step": 1445
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9578881378229215e-05,
+      "loss": 0.8601,
+      "step": 1446
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9578109560662258e-05,
+      "loss": 0.844,
+      "step": 1447
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957733705169806e-05,
+      "loss": 0.9241,
+      "step": 1448
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9576563851392382e-05,
+      "loss": 0.8694,
+      "step": 1449
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957578995980104e-05,
+      "loss": 0.9179,
+      "step": 1450
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957501537697989e-05,
+      "loss": 0.8212,
+      "step": 1451
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9574240102984853e-05,
+      "loss": 0.8713,
+      "step": 1452
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.957346413787189e-05,
+      "loss": 0.8648,
+      "step": 1453
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9572687481697016e-05,
+      "loss": 0.8431,
+      "step": 1454
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9571910134516288e-05,
+      "loss": 0.9205,
+      "step": 1455
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9571132096385824e-05,
+      "loss": 0.8572,
+      "step": 1456
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9570353367361784e-05,
+      "loss": 0.837,
+      "step": 1457
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9569573947500385e-05,
+      "loss": 0.8974,
+      "step": 1458
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956879383685789e-05,
+      "loss": 0.8326,
+      "step": 1459
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.95680130354906e-05,
+      "loss": 0.9283,
+      "step": 1460
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9567231543454888e-05,
+      "loss": 0.3477,
+      "step": 1461
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9566449360807166e-05,
+      "loss": 0.8796,
+      "step": 1462
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9565666487603893e-05,
+      "loss": 0.8223,
+      "step": 1463
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956488292390158e-05,
+      "loss": 0.8866,
+      "step": 1464
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956409866975679e-05,
+      "loss": 0.8863,
+      "step": 1465
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9563313725226135e-05,
+      "loss": 0.896,
+      "step": 1466
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9562528090366277e-05,
+      "loss": 0.8954,
+      "step": 1467
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9561741765233923e-05,
+      "loss": 0.8748,
+      "step": 1468
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.956095474988584e-05,
+      "loss": 0.8736,
+      "step": 1469
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9560167044378834e-05,
+      "loss": 0.9019,
+      "step": 1470
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955937864876977e-05,
+      "loss": 0.8798,
+      "step": 1471
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955858956311555e-05,
+      "loss": 0.874,
+      "step": 1472
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9557799787473146e-05,
+      "loss": 0.8531,
+      "step": 1473
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955700932189956e-05,
+      "loss": 0.8266,
+      "step": 1474
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9556218166451854e-05,
+      "loss": 0.8871,
+      "step": 1475
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955542632118714e-05,
+      "loss": 0.8618,
+      "step": 1476
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9554633786162572e-05,
+      "loss": 0.8891,
+      "step": 1477
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955384056143536e-05,
+      "loss": 0.9163,
+      "step": 1478
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9553046647062773e-05,
+      "loss": 0.902,
+      "step": 1479
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.955225204310211e-05,
+      "loss": 0.9441,
+      "step": 1480
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9551456749610728e-05,
+      "loss": 0.8773,
+      "step": 1481
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9550660766646044e-05,
+      "loss": 0.8425,
+      "step": 1482
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954986409426551e-05,
+      "loss": 0.8627,
+      "step": 1483
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9549066732526635e-05,
+      "loss": 0.8358,
+      "step": 1484
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9548268681486978e-05,
+      "loss": 0.831,
+      "step": 1485
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9547469941204143e-05,
+      "loss": 0.945,
+      "step": 1486
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954667051173579e-05,
+      "loss": 0.9102,
+      "step": 1487
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9545870393139626e-05,
+      "loss": 0.8614,
+      "step": 1488
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9545069585473407e-05,
+      "loss": 0.8986,
+      "step": 1489
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9544268088794943e-05,
+      "loss": 0.9096,
+      "step": 1490
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9543465903162085e-05,
+      "loss": 0.858,
+      "step": 1491
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954266302863274e-05,
+      "loss": 0.8416,
+      "step": 1492
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9541859465264865e-05,
+      "loss": 0.839,
+      "step": 1493
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9541055213116463e-05,
+      "loss": 0.8313,
+      "step": 1494
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9540250272245595e-05,
+      "loss": 0.8878,
+      "step": 1495
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9539444642710357e-05,
+      "loss": 0.8686,
+      "step": 1496
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953863832456891e-05,
+      "loss": 0.8537,
+      "step": 1497
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9537831317879457e-05,
+      "loss": 0.9008,
+      "step": 1498
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953702362270025e-05,
+      "loss": 0.9068,
+      "step": 1499
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9536215239089593e-05,
+      "loss": 0.884,
+      "step": 1500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.953540616710584e-05,
+      "loss": 0.8063,
+      "step": 1501
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9534596406807397e-05,
+      "loss": 0.3978,
+      "step": 1502
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9533785958252708e-05,
+      "loss": 0.8469,
+      "step": 1503
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9532974821500287e-05,
+      "loss": 0.85,
+      "step": 1504
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9532162996608678e-05,
+      "loss": 0.9345,
+      "step": 1505
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9531350483636484e-05,
+      "loss": 0.851,
+      "step": 1506
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.953053728264236e-05,
+      "loss": 0.8666,
+      "step": 1507
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9529723393685e-05,
+      "loss": 0.358,
+      "step": 1508
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952890881682316e-05,
+      "loss": 0.9202,
+      "step": 1509
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952809355211564e-05,
+      "loss": 0.8346,
+      "step": 1510
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9527277599621287e-05,
+      "loss": 0.8385,
+      "step": 1511
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9526460959399004e-05,
+      "loss": 0.8983,
+      "step": 1512
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952564363150774e-05,
+      "loss": 0.8759,
+      "step": 1513
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9524825616006495e-05,
+      "loss": 0.8709,
+      "step": 1514
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952400691295431e-05,
+      "loss": 0.3851,
+      "step": 1515
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9523187522410295e-05,
+      "loss": 0.9181,
+      "step": 1516
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9522367444433587e-05,
+      "loss": 0.8753,
+      "step": 1517
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9521546679083392e-05,
+      "loss": 0.9141,
+      "step": 1518
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.952072522641895e-05,
+      "loss": 0.8898,
+      "step": 1519
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9519903086499564e-05,
+      "loss": 0.8321,
+      "step": 1520
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9519080259384575e-05,
+      "loss": 0.8925,
+      "step": 1521
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9518256745133378e-05,
+      "loss": 0.9373,
+      "step": 1522
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9517432543805426e-05,
+      "loss": 0.8789,
+      "step": 1523
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9516607655460208e-05,
+      "loss": 0.9346,
+      "step": 1524
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.951578208015727e-05,
+      "loss": 0.8568,
+      "step": 1525
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9514955817956206e-05,
+      "loss": 0.9103,
+      "step": 1526
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9514128868916663e-05,
+      "loss": 0.8918,
+      "step": 1527
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9513301233098332e-05,
+      "loss": 0.9238,
+      "step": 1528
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9512472910560954e-05,
+      "loss": 0.8613,
+      "step": 1529
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.951164390136432e-05,
+      "loss": 0.9462,
+      "step": 1530
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9510814205568284e-05,
+      "loss": 0.8875,
+      "step": 1531
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9509983823232722e-05,
+      "loss": 0.876,
+      "step": 1532
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9509152754417587e-05,
+      "loss": 0.8808,
+      "step": 1533
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9508320999182862e-05,
+      "loss": 0.8253,
+      "step": 1534
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9507488557588595e-05,
+      "loss": 0.8788,
+      "step": 1535
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950665542969487e-05,
+      "loss": 0.9154,
+      "step": 1536
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9505821615561827e-05,
+      "loss": 0.9198,
+      "step": 1537
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9504987115249652e-05,
+      "loss": 0.8983,
+      "step": 1538
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9504151928818594e-05,
+      "loss": 0.8829,
+      "step": 1539
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9503316056328935e-05,
+      "loss": 0.8532,
+      "step": 1540
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950247949784101e-05,
+      "loss": 0.9146,
+      "step": 1541
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950164225341521e-05,
+      "loss": 0.3623,
+      "step": 1542
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.950080432311197e-05,
+      "loss": 0.9156,
+      "step": 1543
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9499965706991774e-05,
+      "loss": 0.8459,
+      "step": 1544
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949912640511516e-05,
+      "loss": 0.9088,
+      "step": 1545
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9498286417542717e-05,
+      "loss": 0.9189,
+      "step": 1546
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949744574433507e-05,
+      "loss": 0.8811,
+      "step": 1547
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949660438555291e-05,
+      "loss": 0.9417,
+      "step": 1548
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949576234125697e-05,
+      "loss": 0.8213,
+      "step": 1549
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9494919611508037e-05,
+      "loss": 0.9581,
+      "step": 1550
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9494076196366935e-05,
+      "loss": 0.882,
+      "step": 1551
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9493232095894547e-05,
+      "loss": 0.9277,
+      "step": 1552
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.949238731015181e-05,
+      "loss": 0.9353,
+      "step": 1553
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9491541839199702e-05,
+      "loss": 0.8974,
+      "step": 1554
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9490695683099257e-05,
+      "loss": 0.822,
+      "step": 1555
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9489848841911548e-05,
+      "loss": 0.853,
+      "step": 1556
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.948900131569771e-05,
+      "loss": 0.8786,
+      "step": 1557
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.948815310451892e-05,
+      "loss": 0.8671,
+      "step": 1558
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9487304208436407e-05,
+      "loss": 0.8456,
+      "step": 1559
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9486454627511445e-05,
+      "loss": 0.9137,
+      "step": 1560
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.948560436180537e-05,
+      "loss": 0.9021,
+      "step": 1561
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9484753411379547e-05,
+      "loss": 0.3596,
+      "step": 1562
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9483901776295407e-05,
+      "loss": 0.8583,
+      "step": 1563
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.948304945661443e-05,
+      "loss": 0.8043,
+      "step": 1564
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9482196452398136e-05,
+      "loss": 0.8196,
+      "step": 1565
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9481342763708098e-05,
+      "loss": 0.879,
+      "step": 1566
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.948048839060594e-05,
+      "loss": 0.87,
+      "step": 1567
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.947963333315334e-05,
+      "loss": 0.9005,
+      "step": 1568
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9478777591412018e-05,
+      "loss": 0.8928,
+      "step": 1569
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9477921165443745e-05,
+      "loss": 0.8662,
+      "step": 1570
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.947706405531034e-05,
+      "loss": 0.8137,
+      "step": 1571
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9476206261073675e-05,
+      "loss": 0.9577,
+      "step": 1572
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9475347782795674e-05,
+      "loss": 0.8747,
+      "step": 1573
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9474488620538304e-05,
+      "loss": 0.8398,
+      "step": 1574
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.947362877436358e-05,
+      "loss": 0.8798,
+      "step": 1575
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9472768244333572e-05,
+      "loss": 0.8999,
+      "step": 1576
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9471907030510403e-05,
+      "loss": 0.9388,
+      "step": 1577
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9471045132956234e-05,
+      "loss": 0.8808,
+      "step": 1578
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.947018255173328e-05,
+      "loss": 0.848,
+      "step": 1579
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9469319286903814e-05,
+      "loss": 0.8521,
+      "step": 1580
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946845533853014e-05,
+      "loss": 0.8172,
+      "step": 1581
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9467590706674633e-05,
+      "loss": 0.8483,
+      "step": 1582
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.94667253913997e-05,
+      "loss": 0.8329,
+      "step": 1583
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9465859392767806e-05,
+      "loss": 0.8448,
+      "step": 1584
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9464992710841466e-05,
+      "loss": 0.791,
+      "step": 1585
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9464125345683237e-05,
+      "loss": 0.8829,
+      "step": 1586
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946325729735573e-05,
+      "loss": 0.8287,
+      "step": 1587
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9462388565921607e-05,
+      "loss": 0.9433,
+      "step": 1588
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9461519151443577e-05,
+      "loss": 0.8904,
+      "step": 1589
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.94606490539844e-05,
+      "loss": 0.8045,
+      "step": 1590
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9459778273606885e-05,
+      "loss": 0.8672,
+      "step": 1591
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9458906810373884e-05,
+      "loss": 0.9221,
+      "step": 1592
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9458034664348312e-05,
+      "loss": 0.8301,
+      "step": 1593
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9457161835593114e-05,
+      "loss": 0.8831,
+      "step": 1594
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9456288324171306e-05,
+      "loss": 0.8504,
+      "step": 1595
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9455414130145936e-05,
+      "loss": 0.8283,
+      "step": 1596
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945453925358011e-05,
+      "loss": 0.8439,
+      "step": 1597
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9453663694536983e-05,
+      "loss": 0.8578,
+      "step": 1598
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9452787453079752e-05,
+      "loss": 0.8773,
+      "step": 1599
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9451910529271674e-05,
+      "loss": 0.8543,
+      "step": 1600
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.945103292317605e-05,
+      "loss": 0.8982,
+      "step": 1601
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9450154634856225e-05,
+      "loss": 0.8449,
+      "step": 1602
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9449275664375603e-05,
+      "loss": 0.8614,
+      "step": 1603
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.944839601179763e-05,
+      "loss": 0.8891,
+      "step": 1604
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9447515677185803e-05,
+      "loss": 0.8219,
+      "step": 1605
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9446634660603676e-05,
+      "loss": 0.801,
+      "step": 1606
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9445752962114833e-05,
+      "loss": 0.874,
+      "step": 1607
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9444870581782935e-05,
+      "loss": 0.8859,
+      "step": 1608
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9443987519671664e-05,
+      "loss": 0.8793,
+      "step": 1609
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.944310377584477e-05,
+      "loss": 0.9026,
+      "step": 1610
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.944221935036604e-05,
+      "loss": 0.9052,
+      "step": 1611
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9441334243299327e-05,
+      "loss": 0.8414,
+      "step": 1612
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9440448454708512e-05,
+      "loss": 0.9197,
+      "step": 1613
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9439561984657542e-05,
+      "loss": 0.8924,
+      "step": 1614
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9438674833210406e-05,
+      "loss": 0.3466,
+      "step": 1615
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9437787000431143e-05,
+      "loss": 0.8754,
+      "step": 1616
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9436898486383843e-05,
+      "loss": 0.9278,
+      "step": 1617
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9436009291132637e-05,
+      "loss": 0.9145,
+      "step": 1618
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9435119414741715e-05,
+      "loss": 0.8567,
+      "step": 1619
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9434228857275318e-05,
+      "loss": 0.8638,
+      "step": 1620
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9433337618797728e-05,
+      "loss": 0.9375,
+      "step": 1621
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.943244569937327e-05,
+      "loss": 0.8762,
+      "step": 1622
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9431553099066345e-05,
+      "loss": 0.8286,
+      "step": 1623
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.943065981794137e-05,
+      "loss": 0.8423,
+      "step": 1624
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9429765856062836e-05,
+      "loss": 0.8383,
+      "step": 1625
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9428871213495268e-05,
+      "loss": 0.8529,
+      "step": 1626
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9427975890303252e-05,
+      "loss": 0.3879,
+      "step": 1627
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.942707988655141e-05,
+      "loss": 0.905,
+      "step": 1628
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.942618320230442e-05,
+      "loss": 0.9107,
+      "step": 1629
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.942528583762702e-05,
+      "loss": 0.8991,
+      "step": 1630
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9424387792583976e-05,
+      "loss": 0.8824,
+      "step": 1631
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9423489067240116e-05,
+      "loss": 0.8687,
+      "step": 1632
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9422589661660315e-05,
+      "loss": 0.8642,
+      "step": 1633
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.94216895759095e-05,
+      "loss": 0.9494,
+      "step": 1634
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9420788810052637e-05,
+      "loss": 0.8518,
+      "step": 1635
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9419887364154753e-05,
+      "loss": 0.8783,
+      "step": 1636
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.941898523828092e-05,
+      "loss": 0.8774,
+      "step": 1637
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9418082432496253e-05,
+      "loss": 0.9313,
+      "step": 1638
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.941717894686592e-05,
+      "loss": 0.8599,
+      "step": 1639
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9416274781455153e-05,
+      "loss": 0.8576,
+      "step": 1640
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9415369936329203e-05,
+      "loss": 0.941,
+      "step": 1641
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9414464411553394e-05,
+      "loss": 0.8763,
+      "step": 1642
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9413558207193093e-05,
+      "loss": 0.8881,
+      "step": 1643
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.941265132331371e-05,
+      "loss": 0.8451,
+      "step": 1644
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.941174375998071e-05,
+      "loss": 0.8586,
+      "step": 1645
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9410835517259605e-05,
+      "loss": 0.8932,
+      "step": 1646
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9409926595215957e-05,
+      "loss": 0.9599,
+      "step": 1647
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.940901699391538e-05,
+      "loss": 0.8948,
+      "step": 1648
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.940810671342353e-05,
+      "loss": 0.8693,
+      "step": 1649
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9407195753806115e-05,
+      "loss": 0.8989,
+      "step": 1650
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9406284115128898e-05,
+      "loss": 0.8384,
+      "step": 1651
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.940537179745768e-05,
+      "loss": 0.9159,
+      "step": 1652
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.940445880085832e-05,
+      "loss": 0.7976,
+      "step": 1653
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.940354512539672e-05,
+      "loss": 0.9197,
+      "step": 1654
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9402630771138837e-05,
+      "loss": 0.861,
+      "step": 1655
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9401715738150672e-05,
+      "loss": 0.7968,
+      "step": 1656
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9400800026498276e-05,
+      "loss": 0.3843,
+      "step": 1657
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9399883636247755e-05,
+      "loss": 0.8892,
+      "step": 1658
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9398966567465253e-05,
+      "loss": 0.851,
+      "step": 1659
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.939804882021697e-05,
+      "loss": 0.8738,
+      "step": 1660
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9397130394569153e-05,
+      "loss": 0.8871,
+      "step": 1661
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.93962112905881e-05,
+      "loss": 0.8455,
+      "step": 1662
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9395291508340158e-05,
+      "loss": 0.8931,
+      "step": 1663
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.939437104789172e-05,
+      "loss": 0.8857,
+      "step": 1664
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.939344990930923e-05,
+      "loss": 0.8272,
+      "step": 1665
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9392528092659178e-05,
+      "loss": 0.8812,
+      "step": 1666
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.939160559800811e-05,
+      "loss": 0.8993,
+      "step": 1667
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9390682425422614e-05,
+      "loss": 0.8863,
+      "step": 1668
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.938975857496933e-05,
+      "loss": 0.4144,
+      "step": 1669
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9388834046714944e-05,
+      "loss": 0.8798,
+      "step": 1670
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9387908840726197e-05,
+      "loss": 0.8602,
+      "step": 1671
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9386982957069873e-05,
+      "loss": 0.894,
+      "step": 1672
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9386056395812804e-05,
+      "loss": 0.8372,
+      "step": 1673
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9385129157021878e-05,
+      "loss": 0.8912,
+      "step": 1674
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.938420124076403e-05,
+      "loss": 0.8861,
+      "step": 1675
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9383272647106237e-05,
+      "loss": 0.8663,
+      "step": 1676
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.938234337611553e-05,
+      "loss": 0.9158,
+      "step": 1677
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9381413427858994e-05,
+      "loss": 0.8668,
+      "step": 1678
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.938048280240375e-05,
+      "loss": 0.798,
+      "step": 1679
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.937955149981698e-05,
+      "loss": 0.8586,
+      "step": 1680
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9378619520165907e-05,
+      "loss": 0.8426,
+      "step": 1681
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9377686863517812e-05,
+      "loss": 0.9253,
+      "step": 1682
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9376753529940013e-05,
+      "loss": 0.9456,
+      "step": 1683
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9375819519499886e-05,
+      "loss": 0.8655,
+      "step": 1684
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.937488483226485e-05,
+      "loss": 0.8843,
+      "step": 1685
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9373949468302377e-05,
+      "loss": 0.957,
+      "step": 1686
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9373013427679987e-05,
+      "loss": 0.9431,
+      "step": 1687
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9372076710465247e-05,
+      "loss": 0.8611,
+      "step": 1688
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9371139316725777e-05,
+      "loss": 0.8465,
+      "step": 1689
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.937020124652924e-05,
+      "loss": 0.8384,
+      "step": 1690
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.936926249994335e-05,
+      "loss": 0.9194,
+      "step": 1691
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.936832307703587e-05,
+      "loss": 0.8638,
+      "step": 1692
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.936738297787462e-05,
+      "loss": 0.8757,
+      "step": 1693
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9366442202527454e-05,
+      "loss": 0.8616,
+      "step": 1694
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.936550075106228e-05,
+      "loss": 0.8358,
+      "step": 1695
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9364558623547064e-05,
+      "loss": 0.8606,
+      "step": 1696
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9363615820049804e-05,
+      "loss": 0.8385,
+      "step": 1697
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.936267234063857e-05,
+      "loss": 0.7909,
+      "step": 1698
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9361728185381456e-05,
+      "loss": 0.8758,
+      "step": 1699
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.936078335434662e-05,
+      "loss": 0.4046,
+      "step": 1700
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9359837847602264e-05,
+      "loss": 0.8684,
+      "step": 1701
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.935889166521664e-05,
+      "loss": 0.8683,
+      "step": 1702
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9357944807258052e-05,
+      "loss": 0.902,
+      "step": 1703
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.935699727379484e-05,
+      "loss": 0.7772,
+      "step": 1704
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.935604906489541e-05,
+      "loss": 0.8644,
+      "step": 1705
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.93551001806282e-05,
+      "loss": 0.8945,
+      "step": 1706
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.935415062106172e-05,
+      "loss": 0.791,
+      "step": 1707
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.93532003862645e-05,
+      "loss": 0.9437,
+      "step": 1708
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9352249476305137e-05,
+      "loss": 0.8784,
+      "step": 1709
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9351297891252276e-05,
+      "loss": 0.9733,
+      "step": 1710
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9350345631174607e-05,
+      "loss": 0.8406,
+      "step": 1711
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9349392696140864e-05,
+      "loss": 0.904,
+      "step": 1712
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9348439086219835e-05,
+      "loss": 0.8718,
+      "step": 1713
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9347484801480363e-05,
+      "loss": 0.8729,
+      "step": 1714
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9346529841991327e-05,
+      "loss": 0.8605,
+      "step": 1715
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9345574207821664e-05,
+      "loss": 0.8157,
+      "step": 1716
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9344617899040358e-05,
+      "loss": 0.8979,
+      "step": 1717
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9343660915716436e-05,
+      "loss": 0.8652,
+      "step": 1718
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9342703257918983e-05,
+      "loss": 0.8867,
+      "step": 1719
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.934174492571712e-05,
+      "loss": 0.8098,
+      "step": 1720
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.934078591918003e-05,
+      "loss": 0.3566,
+      "step": 1721
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9339826238376944e-05,
+      "loss": 0.8554,
+      "step": 1722
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9338865883377125e-05,
+      "loss": 0.9155,
+      "step": 1723
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9337904854249903e-05,
+      "loss": 0.9106,
+      "step": 1724
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9336943151064652e-05,
+      "loss": 0.873,
+      "step": 1725
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.933598077389079e-05,
+      "loss": 0.8709,
+      "step": 1726
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.933501772279778e-05,
+      "loss": 0.8974,
+      "step": 1727
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9334053997855153e-05,
+      "loss": 0.8161,
+      "step": 1728
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9333089599132465e-05,
+      "loss": 0.9043,
+      "step": 1729
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9332124526699333e-05,
+      "loss": 0.8619,
+      "step": 1730
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9331158780625426e-05,
+      "loss": 0.8679,
+      "step": 1731
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9330192360980456e-05,
+      "loss": 0.8783,
+      "step": 1732
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.932922526783418e-05,
+      "loss": 0.876,
+      "step": 1733
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.932825750125641e-05,
+      "loss": 0.8869,
+      "step": 1734
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9327289061317003e-05,
+      "loss": 0.8718,
+      "step": 1735
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9326319948085865e-05,
+      "loss": 0.8802,
+      "step": 1736
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9325350161632958e-05,
+      "loss": 0.8638,
+      "step": 1737
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9324379702028276e-05,
+      "loss": 0.8743,
+      "step": 1738
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9323408569341883e-05,
+      "loss": 0.9138,
+      "step": 1739
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.932243676364387e-05,
+      "loss": 0.919,
+      "step": 1740
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.932146428500439e-05,
+      "loss": 0.8679,
+      "step": 1741
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9320491133493647e-05,
+      "loss": 0.8878,
+      "step": 1742
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9319517309181886e-05,
+      "loss": 0.8907,
+      "step": 1743
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9318542812139398e-05,
+      "loss": 0.9075,
+      "step": 1744
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.931756764243653e-05,
+      "loss": 0.8204,
+      "step": 1745
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9316591800143676e-05,
+      "loss": 0.8882,
+      "step": 1746
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9315615285331274e-05,
+      "loss": 0.8847,
+      "step": 1747
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.931463809806982e-05,
+      "loss": 0.8825,
+      "step": 1748
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9313660238429846e-05,
+      "loss": 0.8624,
+      "step": 1749
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.931268170648194e-05,
+      "loss": 0.8759,
+      "step": 1750
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.931170250229674e-05,
+      "loss": 0.8681,
+      "step": 1751
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.931072262594493e-05,
+      "loss": 0.8865,
+      "step": 1752
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.930974207749724e-05,
+      "loss": 0.8497,
+      "step": 1753
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9308760857024454e-05,
+      "loss": 0.8671,
+      "step": 1754
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.93077789645974e-05,
+      "loss": 0.8122,
+      "step": 1755
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9306796400286953e-05,
+      "loss": 0.8701,
+      "step": 1756
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9305813164164045e-05,
+      "loss": 0.8568,
+      "step": 1757
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.930482925629965e-05,
+      "loss": 0.8268,
+      "step": 1758
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9303844676764783e-05,
+      "loss": 0.8441,
+      "step": 1759
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.930285942563053e-05,
+      "loss": 0.9339,
+      "step": 1760
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9301873502968003e-05,
+      "loss": 0.8744,
+      "step": 1761
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9300886908848373e-05,
+      "loss": 0.884,
+      "step": 1762
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9299899643342857e-05,
+      "loss": 0.8133,
+      "step": 1763
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.929891170652272e-05,
+      "loss": 0.8672,
+      "step": 1764
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9297923098459276e-05,
+      "loss": 0.9323,
+      "step": 1765
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9296933819223892e-05,
+      "loss": 0.9372,
+      "step": 1766
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9295943868887976e-05,
+      "loss": 0.8401,
+      "step": 1767
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.929495324752299e-05,
+      "loss": 0.894,
+      "step": 1768
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9293961955200436e-05,
+      "loss": 0.8582,
+      "step": 1769
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9292969991991874e-05,
+      "loss": 0.9276,
+      "step": 1770
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9291977357968915e-05,
+      "loss": 0.8798,
+      "step": 1771
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9290984053203205e-05,
+      "loss": 0.9355,
+      "step": 1772
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9289990077766447e-05,
+      "loss": 0.8406,
+      "step": 1773
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9288995431730393e-05,
+      "loss": 0.8609,
+      "step": 1774
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.928800011516684e-05,
+      "loss": 0.8985,
+      "step": 1775
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.928700412814764e-05,
+      "loss": 0.9118,
+      "step": 1776
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.928600747074468e-05,
+      "loss": 0.8203,
+      "step": 1777
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9285010143029912e-05,
+      "loss": 0.8209,
+      "step": 1778
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9284012145075324e-05,
+      "loss": 0.9106,
+      "step": 1779
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9283013476952957e-05,
+      "loss": 0.8911,
+      "step": 1780
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.92820141387349e-05,
+      "loss": 0.8928,
+      "step": 1781
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9281014130493288e-05,
+      "loss": 0.4271,
+      "step": 1782
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9280013452300314e-05,
+      "loss": 0.8253,
+      "step": 1783
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9279012104228206e-05,
+      "loss": 0.8488,
+      "step": 1784
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9278010086349247e-05,
+      "loss": 0.9158,
+      "step": 1785
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.927700739873577e-05,
+      "loss": 0.8705,
+      "step": 1786
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.927600404146015e-05,
+      "loss": 0.8365,
+      "step": 1787
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9275000014594822e-05,
+      "loss": 0.8977,
+      "step": 1788
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9273995318212255e-05,
+      "loss": 0.8472,
+      "step": 1789
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9272989952384974e-05,
+      "loss": 0.8711,
+      "step": 1790
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9271983917185556e-05,
+      "loss": 0.8573,
+      "step": 1791
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9270977212686618e-05,
+      "loss": 0.8293,
+      "step": 1792
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.926996983896083e-05,
+      "loss": 0.96,
+      "step": 1793
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9268961796080908e-05,
+      "loss": 0.8763,
+      "step": 1794
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9267953084119622e-05,
+      "loss": 0.8602,
+      "step": 1795
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.926694370314978e-05,
+      "loss": 0.8301,
+      "step": 1796
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9265933653244245e-05,
+      "loss": 0.9095,
+      "step": 1797
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9264922934475935e-05,
+      "loss": 0.9436,
+      "step": 1798
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.92639115469178e-05,
+      "loss": 0.9002,
+      "step": 1799
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9262899490642854e-05,
+      "loss": 0.8679,
+      "step": 1800
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9261886765724148e-05,
+      "loss": 0.8644,
+      "step": 1801
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9260873372234788e-05,
+      "loss": 0.8477,
+      "step": 1802
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9259859310247924e-05,
+      "loss": 0.8851,
+      "step": 1803
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.925884457983676e-05,
+      "loss": 0.8226,
+      "step": 1804
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9257829181074536e-05,
+      "loss": 0.7946,
+      "step": 1805
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.925681311403456e-05,
+      "loss": 0.8553,
+      "step": 1806
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9255796378790168e-05,
+      "loss": 0.8669,
+      "step": 1807
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9254778975414757e-05,
+      "loss": 0.9554,
+      "step": 1808
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.925376090398177e-05,
+      "loss": 0.9106,
+      "step": 1809
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9252742164564693e-05,
+      "loss": 0.8462,
+      "step": 1810
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9251722757237064e-05,
+      "loss": 0.8371,
+      "step": 1811
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9250702682072475e-05,
+      "loss": 0.8755,
+      "step": 1812
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9249681939144552e-05,
+      "loss": 0.8462,
+      "step": 1813
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9248660528526985e-05,
+      "loss": 0.868,
+      "step": 1814
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9247638450293498e-05,
+      "loss": 0.8941,
+      "step": 1815
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9246615704517872e-05,
+      "loss": 0.8568,
+      "step": 1816
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9245592291273935e-05,
+      "loss": 0.937,
+      "step": 1817
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9244568210635567e-05,
+      "loss": 0.8165,
+      "step": 1818
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.924354346267668e-05,
+      "loss": 0.9433,
+      "step": 1819
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9242518047471253e-05,
+      "loss": 0.8868,
+      "step": 1820
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.924149196509331e-05,
+      "loss": 0.8344,
+      "step": 1821
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.924046521561691e-05,
+      "loss": 0.8419,
+      "step": 1822
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9239437799116174e-05,
+      "loss": 0.8291,
+      "step": 1823
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9238409715665266e-05,
+      "loss": 0.8783,
+      "step": 1824
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9237380965338393e-05,
+      "loss": 0.9317,
+      "step": 1825
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9236351548209824e-05,
+      "loss": 0.8806,
+      "step": 1826
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9235321464353863e-05,
+      "loss": 0.9132,
+      "step": 1827
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9234290713844866e-05,
+      "loss": 0.8907,
+      "step": 1828
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.923325929675724e-05,
+      "loss": 0.7906,
+      "step": 1829
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.923222721316544e-05,
+      "loss": 0.9218,
+      "step": 1830
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9231194463143958e-05,
+      "loss": 0.8273,
+      "step": 1831
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9230161046767353e-05,
+      "loss": 0.8124,
+      "step": 1832
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9229126964110217e-05,
+      "loss": 0.8151,
+      "step": 1833
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.92280922152472e-05,
+      "loss": 0.9281,
+      "step": 1834
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9227056800252992e-05,
+      "loss": 0.8073,
+      "step": 1835
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9226020719202333e-05,
+      "loss": 0.8761,
+      "step": 1836
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9224983972170017e-05,
+      "loss": 0.9228,
+      "step": 1837
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.922394655923088e-05,
+      "loss": 0.9175,
+      "step": 1838
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.922290848045981e-05,
+      "loss": 0.9139,
+      "step": 1839
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9221869735931733e-05,
+      "loss": 0.8734,
+      "step": 1840
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9220830325721637e-05,
+      "loss": 0.8733,
+      "step": 1841
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9219790249904553e-05,
+      "loss": 0.9188,
+      "step": 1842
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9218749508555556e-05,
+      "loss": 0.8858,
+      "step": 1843
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9217708101749774e-05,
+      "loss": 0.8002,
+      "step": 1844
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.921666602956238e-05,
+      "loss": 0.8892,
+      "step": 1845
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.92156232920686e-05,
+      "loss": 0.8876,
+      "step": 1846
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.92145798893437e-05,
+      "loss": 0.8175,
+      "step": 1847
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9213535821462995e-05,
+      "loss": 0.9135,
+      "step": 1848
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9212491088501856e-05,
+      "loss": 0.8818,
+      "step": 1849
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9211445690535696e-05,
+      "loss": 0.8548,
+      "step": 1850
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.921039962763998e-05,
+      "loss": 0.879,
+      "step": 1851
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9209352899890214e-05,
+      "loss": 0.901,
+      "step": 1852
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.920830550736196e-05,
+      "loss": 0.8549,
+      "step": 1853
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9207257450130822e-05,
+      "loss": 0.7522,
+      "step": 1854
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.920620872827245e-05,
+      "loss": 0.8382,
+      "step": 1855
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9205159341862553e-05,
+      "loss": 0.9208,
+      "step": 1856
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9204109290976877e-05,
+      "loss": 0.9461,
+      "step": 1857
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9203058575691226e-05,
+      "loss": 0.8545,
+      "step": 1858
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.920200719608144e-05,
+      "loss": 0.8425,
+      "step": 1859
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9200955152223413e-05,
+      "loss": 0.915,
+      "step": 1860
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9199902444193088e-05,
+      "loss": 0.9835,
+      "step": 1861
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9198849072066458e-05,
+      "loss": 0.901,
+      "step": 1862
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9197795035919554e-05,
+      "loss": 0.8346,
+      "step": 1863
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.919674033582847e-05,
+      "loss": 0.9068,
+      "step": 1864
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9195684971869337e-05,
+      "loss": 0.9262,
+      "step": 1865
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.919462894411833e-05,
+      "loss": 0.8663,
+      "step": 1866
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9193572252651685e-05,
+      "loss": 0.9211,
+      "step": 1867
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.919251489754568e-05,
+      "loss": 0.8794,
+      "step": 1868
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9191456878876643e-05,
+      "loss": 0.8639,
+      "step": 1869
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9190398196720935e-05,
+      "loss": 0.3606,
+      "step": 1870
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.918933885115499e-05,
+      "loss": 0.3733,
+      "step": 1871
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.918827884225527e-05,
+      "loss": 0.8837,
+      "step": 1872
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9187218170098294e-05,
+      "loss": 0.8478,
+      "step": 1873
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9186156834760625e-05,
+      "loss": 0.9513,
+      "step": 1874
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.918509483631888e-05,
+      "loss": 0.7966,
+      "step": 1875
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9184032174849717e-05,
+      "loss": 0.8817,
+      "step": 1876
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9182968850429842e-05,
+      "loss": 0.938,
+      "step": 1877
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9181904863136017e-05,
+      "loss": 0.8356,
+      "step": 1878
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9180840213045036e-05,
+      "loss": 0.8237,
+      "step": 1879
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9179774900233763e-05,
+      "loss": 0.8801,
+      "step": 1880
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.917870892477909e-05,
+      "loss": 0.8569,
+      "step": 1881
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.917764228675797e-05,
+      "loss": 0.8956,
+      "step": 1882
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.917657498624739e-05,
+      "loss": 0.9319,
+      "step": 1883
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9175507023324403e-05,
+      "loss": 0.8997,
+      "step": 1884
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.917443839806609e-05,
+      "loss": 0.8547,
+      "step": 1885
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.91733691105496e-05,
+      "loss": 0.8122,
+      "step": 1886
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9172299160852112e-05,
+      "loss": 0.8536,
+      "step": 1887
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9171228549050866e-05,
+      "loss": 0.8053,
+      "step": 1888
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9170157275223143e-05,
+      "loss": 0.8146,
+      "step": 1889
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9169085339446266e-05,
+      "loss": 0.8547,
+      "step": 1890
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9168012741797624e-05,
+      "loss": 0.9037,
+      "step": 1891
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9166939482354636e-05,
+      "loss": 0.8555,
+      "step": 1892
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9165865561194777e-05,
+      "loss": 0.8408,
+      "step": 1893
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9164790978395566e-05,
+      "loss": 0.8917,
+      "step": 1894
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9163715734034576e-05,
+      "loss": 0.8858,
+      "step": 1895
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.916263982818942e-05,
+      "loss": 0.8698,
+      "step": 1896
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9161563260937764e-05,
+      "loss": 0.898,
+      "step": 1897
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.916048603235732e-05,
+      "loss": 0.907,
+      "step": 1898
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.915940814252585e-05,
+      "loss": 0.8696,
+      "step": 1899
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9158329591521158e-05,
+      "loss": 0.9112,
+      "step": 1900
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.91572503794211e-05,
+      "loss": 0.8899,
+      "step": 1901
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.915617050630358e-05,
+      "loss": 0.9621,
+      "step": 1902
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.915508997224655e-05,
+      "loss": 0.8705,
+      "step": 1903
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9154008777328006e-05,
+      "loss": 0.8544,
+      "step": 1904
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9152926921625998e-05,
+      "loss": 0.9071,
+      "step": 1905
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9151844405218613e-05,
+      "loss": 0.8684,
+      "step": 1906
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9150761228184e-05,
+      "loss": 0.3736,
+      "step": 1907
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9149677390600344e-05,
+      "loss": 0.8795,
+      "step": 1908
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9148592892545887e-05,
+      "loss": 0.9087,
+      "step": 1909
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9147507734098907e-05,
+      "loss": 0.9374,
+      "step": 1910
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.914642191533774e-05,
+      "loss": 0.7434,
+      "step": 1911
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9145335436340763e-05,
+      "loss": 0.8387,
+      "step": 1912
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.914424829718641e-05,
+      "loss": 0.8512,
+      "step": 1913
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.914316049795315e-05,
+      "loss": 0.91,
+      "step": 1914
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9142072038719507e-05,
+      "loss": 0.9345,
+      "step": 1915
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9140982919564056e-05,
+      "loss": 0.8416,
+      "step": 1916
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.913989314056541e-05,
+      "loss": 0.8236,
+      "step": 1917
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9138802701802236e-05,
+      "loss": 0.8666,
+      "step": 1918
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.913771160335325e-05,
+      "loss": 0.8484,
+      "step": 1919
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.913661984529721e-05,
+      "loss": 0.8693,
+      "step": 1920
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9135527427712923e-05,
+      "loss": 0.3692,
+      "step": 1921
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9134434350679254e-05,
+      "loss": 0.9068,
+      "step": 1922
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9133340614275098e-05,
+      "loss": 0.9026,
+      "step": 1923
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9132246218579412e-05,
+      "loss": 0.8536,
+      "step": 1924
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.913115116367119e-05,
+      "loss": 0.7653,
+      "step": 1925
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9130055449629488e-05,
+      "loss": 0.8898,
+      "step": 1926
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9128959076533388e-05,
+      "loss": 0.8613,
+      "step": 1927
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9127862044462043e-05,
+      "loss": 0.9207,
+      "step": 1928
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.912676435349463e-05,
+      "loss": 0.8413,
+      "step": 1929
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.91256660037104e-05,
+      "loss": 0.8771,
+      "step": 1930
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9124566995188625e-05,
+      "loss": 0.9098,
+      "step": 1931
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.912346732800865e-05,
+      "loss": 0.8661,
+      "step": 1932
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9122367002249843e-05,
+      "loss": 0.8515,
+      "step": 1933
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9121266017991637e-05,
+      "loss": 0.8683,
+      "step": 1934
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9120164375313504e-05,
+      "loss": 0.8486,
+      "step": 1935
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9119062074294968e-05,
+      "loss": 0.8287,
+      "step": 1936
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.91179591150156e-05,
+      "loss": 0.8563,
+      "step": 1937
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.911685549755501e-05,
+      "loss": 0.9249,
+      "step": 1938
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9115751221992875e-05,
+      "loss": 0.921,
+      "step": 1939
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.91146462884089e-05,
+      "loss": 0.8408,
+      "step": 1940
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9113540696882844e-05,
+      "loss": 0.9251,
+      "step": 1941
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9112434447494516e-05,
+      "loss": 0.8341,
+      "step": 1942
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.911132754032377e-05,
+      "loss": 0.9144,
+      "step": 1943
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.911021997545051e-05,
+      "loss": 0.8669,
+      "step": 1944
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9109111752954685e-05,
+      "loss": 0.8523,
+      "step": 1945
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9108002872916288e-05,
+      "loss": 0.8257,
+      "step": 1946
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9106893335415373e-05,
+      "loss": 0.8102,
+      "step": 1947
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9105783140532025e-05,
+      "loss": 0.8634,
+      "step": 1948
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9104672288346384e-05,
+      "loss": 0.922,
+      "step": 1949
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9103560778938635e-05,
+      "loss": 0.9104,
+      "step": 1950
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9102448612389022e-05,
+      "loss": 0.861,
+      "step": 1951
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9101335788777818e-05,
+      "loss": 0.8265,
+      "step": 1952
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9100222308185356e-05,
+      "loss": 0.7899,
+      "step": 1953
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.909910817069201e-05,
+      "loss": 0.878,
+      "step": 1954
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9097993376378206e-05,
+      "loss": 0.8671,
+      "step": 1955
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9096877925324417e-05,
+      "loss": 0.8212,
+      "step": 1956
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9095761817611157e-05,
+      "loss": 0.8796,
+      "step": 1957
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9094645053318998e-05,
+      "loss": 0.8673,
+      "step": 1958
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.909352763252855e-05,
+      "loss": 0.9207,
+      "step": 1959
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.909240955532048e-05,
+      "loss": 0.8299,
+      "step": 1960
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.909129082177549e-05,
+      "loss": 0.8977,
+      "step": 1961
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9090171431974338e-05,
+      "loss": 0.8823,
+      "step": 1962
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9089051385997826e-05,
+      "loss": 0.3538,
+      "step": 1963
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.908793068392681e-05,
+      "loss": 0.8564,
+      "step": 1964
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9086809325842183e-05,
+      "loss": 0.8428,
+      "step": 1965
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9085687311824894e-05,
+      "loss": 0.8121,
+      "step": 1966
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.908456464195593e-05,
+      "loss": 0.7602,
+      "step": 1967
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9083441316316333e-05,
+      "loss": 0.8429,
+      "step": 1968
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9082317334987194e-05,
+      "loss": 0.8393,
+      "step": 1969
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9081192698049648e-05,
+      "loss": 0.9087,
+      "step": 1970
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9080067405584874e-05,
+      "loss": 0.8796,
+      "step": 1971
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.90789414576741e-05,
+      "loss": 0.8792,
+      "step": 1972
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9077814854398607e-05,
+      "loss": 0.8709,
+      "step": 1973
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9076687595839717e-05,
+      "loss": 0.8361,
+      "step": 1974
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9075559682078802e-05,
+      "loss": 0.8795,
+      "step": 1975
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.907443111319728e-05,
+      "loss": 0.8672,
+      "step": 1976
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9073301889276613e-05,
+      "loss": 0.8366,
+      "step": 1977
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9072172010398322e-05,
+      "loss": 0.8155,
+      "step": 1978
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9071041476643962e-05,
+      "loss": 0.8722,
+      "step": 1979
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9069910288095142e-05,
+      "loss": 0.7788,
+      "step": 1980
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.906877844483352e-05,
+      "loss": 0.8941,
+      "step": 1981
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9067645946940795e-05,
+      "loss": 0.9018,
+      "step": 1982
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9066512794498714e-05,
+      "loss": 0.8925,
+      "step": 1983
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9065378987589078e-05,
+      "loss": 0.9129,
+      "step": 1984
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.906424452629373e-05,
+      "loss": 0.9061,
+      "step": 1985
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.906310941069456e-05,
+      "loss": 0.8135,
+      "step": 1986
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.906197364087351e-05,
+      "loss": 0.854,
+      "step": 1987
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9060837216912565e-05,
+      "loss": 0.8851,
+      "step": 1988
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9059700138893752e-05,
+      "loss": 0.9158,
+      "step": 1989
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9058562406899158e-05,
+      "loss": 0.7961,
+      "step": 1990
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9057424021010907e-05,
+      "loss": 0.9382,
+      "step": 1991
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9056284981311175e-05,
+      "loss": 0.9014,
+      "step": 1992
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9055145287882182e-05,
+      "loss": 0.8596,
+      "step": 1993
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9054004940806198e-05,
+      "loss": 0.8356,
+      "step": 1994
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9052863940165542e-05,
+      "loss": 0.8741,
+      "step": 1995
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9051722286042568e-05,
+      "loss": 0.8463,
+      "step": 1996
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9050579978519697e-05,
+      "loss": 0.7976,
+      "step": 1997
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9049437017679386e-05,
+      "loss": 0.8289,
+      "step": 1998
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.904829340360413e-05,
+      "loss": 0.8531,
+      "step": 1999
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.904714913637649e-05,
+      "loss": 0.8008,
+      "step": 2000
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9046004216079065e-05,
+      "loss": 0.8059,
+      "step": 2001
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9044858642794496e-05,
+      "loss": 0.8627,
+      "step": 2002
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.904371241660548e-05,
+      "loss": 0.8721,
+      "step": 2003
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9042565537594756e-05,
+      "loss": 0.8425,
+      "step": 2004
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.904141800584511e-05,
+      "loss": 0.7978,
+      "step": 2005
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.904026982143938e-05,
+      "loss": 0.9082,
+      "step": 2006
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9039120984460448e-05,
+      "loss": 0.8661,
+      "step": 2007
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9037971494991242e-05,
+      "loss": 0.8835,
+      "step": 2008
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9036821353114735e-05,
+      "loss": 0.7567,
+      "step": 2009
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9035670558913954e-05,
+      "loss": 0.8595,
+      "step": 2010
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.903451911247197e-05,
+      "loss": 0.8705,
+      "step": 2011
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.90333670138719e-05,
+      "loss": 0.8313,
+      "step": 2012
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9032214263196903e-05,
+      "loss": 0.9296,
+      "step": 2013
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9031060860530196e-05,
+      "loss": 0.8494,
+      "step": 2014
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9029906805955033e-05,
+      "loss": 0.8667,
+      "step": 2015
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9028752099554725e-05,
+      "loss": 0.891,
+      "step": 2016
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9027596741412624e-05,
+      "loss": 0.8904,
+      "step": 2017
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.902644073161213e-05,
+      "loss": 0.8369,
+      "step": 2018
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9025284070236682e-05,
+      "loss": 0.9325,
+      "step": 2019
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.902412675736978e-05,
+      "loss": 0.8562,
+      "step": 2020
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9022968793094966e-05,
+      "loss": 0.8608,
+      "step": 2021
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9021810177495824e-05,
+      "loss": 0.8736,
+      "step": 2022
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9020650910655993e-05,
+      "loss": 0.8761,
+      "step": 2023
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9019490992659154e-05,
+      "loss": 0.8808,
+      "step": 2024
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.901833042358903e-05,
+      "loss": 0.8496,
+      "step": 2025
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9017169203529403e-05,
+      "loss": 0.8548,
+      "step": 2026
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9016007332564096e-05,
+      "loss": 0.8407,
+      "step": 2027
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.901484481077698e-05,
+      "loss": 0.8958,
+      "step": 2028
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9013681638251962e-05,
+      "loss": 0.8667,
+      "step": 2029
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9012517815073016e-05,
+      "loss": 0.8617,
+      "step": 2030
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9011353341324147e-05,
+      "loss": 0.7844,
+      "step": 2031
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9010188217089422e-05,
+      "loss": 0.8297,
+      "step": 2032
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9009022442452933e-05,
+      "loss": 0.7902,
+      "step": 2033
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9007856017498834e-05,
+      "loss": 0.8733,
+      "step": 2034
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9006688942311332e-05,
+      "loss": 0.8685,
+      "step": 2035
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9005521216974666e-05,
+      "loss": 0.8769,
+      "step": 2036
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.900435284157313e-05,
+      "loss": 0.8585,
+      "step": 2037
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9003183816191066e-05,
+      "loss": 0.9432,
+      "step": 2038
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9002014140912855e-05,
+      "loss": 0.8476,
+      "step": 2039
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9000843815822934e-05,
+      "loss": 0.8512,
+      "step": 2040
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.899967284100578e-05,
+      "loss": 0.7983,
+      "step": 2041
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.899850121654592e-05,
+      "loss": 0.884,
+      "step": 2042
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8997328942527935e-05,
+      "loss": 0.9024,
+      "step": 2043
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8996156019036438e-05,
+      "loss": 0.8751,
+      "step": 2044
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8994982446156097e-05,
+      "loss": 0.8463,
+      "step": 2045
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.899380822397163e-05,
+      "loss": 0.8116,
+      "step": 2046
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8992633352567797e-05,
+      "loss": 0.8163,
+      "step": 2047
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8991457832029407e-05,
+      "loss": 0.8649,
+      "step": 2048
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.899028166244131e-05,
+      "loss": 0.8001,
+      "step": 2049
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.898910484388842e-05,
+      "loss": 0.8772,
+      "step": 2050
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8987927376455674e-05,
+      "loss": 0.9328,
+      "step": 2051
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.898674926022807e-05,
+      "loss": 0.8918,
+      "step": 2052
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8985570495290652e-05,
+      "loss": 0.8669,
+      "step": 2053
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8984391081728514e-05,
+      "loss": 0.8391,
+      "step": 2054
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8983211019626785e-05,
+      "loss": 0.9047,
+      "step": 2055
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.898203030907065e-05,
+      "loss": 0.842,
+      "step": 2056
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8980848950145342e-05,
+      "loss": 0.8138,
+      "step": 2057
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8979666942936133e-05,
+      "loss": 0.7366,
+      "step": 2058
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.897848428752835e-05,
+      "loss": 0.8231,
+      "step": 2059
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8977300984007362e-05,
+      "loss": 0.8012,
+      "step": 2060
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8976117032458583e-05,
+      "loss": 0.8297,
+      "step": 2061
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8974932432967482e-05,
+      "loss": 0.913,
+      "step": 2062
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8973747185619565e-05,
+      "loss": 0.8259,
+      "step": 2063
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.897256129050039e-05,
+      "loss": 0.8513,
+      "step": 2064
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.897137474769556e-05,
+      "loss": 0.7949,
+      "step": 2065
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8970187557290733e-05,
+      "loss": 0.8967,
+      "step": 2066
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.89689997193716e-05,
+      "loss": 0.8582,
+      "step": 2067
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8967811234023906e-05,
+      "loss": 0.884,
+      "step": 2068
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.896662210133344e-05,
+      "loss": 0.8565,
+      "step": 2069
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8965432321386042e-05,
+      "loss": 0.8415,
+      "step": 2070
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8964241894267603e-05,
+      "loss": 0.9107,
+      "step": 2071
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8963050820064043e-05,
+      "loss": 0.8379,
+      "step": 2072
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8961859098861345e-05,
+      "loss": 0.8292,
+      "step": 2073
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8960666730745535e-05,
+      "loss": 0.8532,
+      "step": 2074
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.895947371580268e-05,
+      "loss": 0.9296,
+      "step": 2075
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.89582800541189e-05,
+      "loss": 0.844,
+      "step": 2076
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8957085745780364e-05,
+      "loss": 0.8308,
+      "step": 2077
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8955890790873276e-05,
+      "loss": 0.8398,
+      "step": 2078
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.89546951894839e-05,
+      "loss": 0.434,
+      "step": 2079
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8953498941698537e-05,
+      "loss": 0.8569,
+      "step": 2080
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.895230204760354e-05,
+      "loss": 0.8812,
+      "step": 2081
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8951104507285304e-05,
+      "loss": 0.8159,
+      "step": 2082
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.894990632083028e-05,
+      "loss": 0.8924,
+      "step": 2083
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.894870748832495e-05,
+      "loss": 0.9369,
+      "step": 2084
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8947508009855858e-05,
+      "loss": 0.9063,
+      "step": 2085
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8946307885509593e-05,
+      "loss": 0.9163,
+      "step": 2086
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8945107115372777e-05,
+      "loss": 0.8503,
+      "step": 2087
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8943905699532094e-05,
+      "loss": 0.8852,
+      "step": 2088
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.894270363807426e-05,
+      "loss": 0.891,
+      "step": 2089
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8941500931086058e-05,
+      "loss": 0.807,
+      "step": 2090
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.89402975786543e-05,
+      "loss": 0.8736,
+      "step": 2091
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8939093580865844e-05,
+      "loss": 0.8818,
+      "step": 2092
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8937888937807612e-05,
+      "loss": 0.8324,
+      "step": 2093
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8936683649566553e-05,
+      "loss": 0.8389,
+      "step": 2094
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8935477716229677e-05,
+      "loss": 0.8685,
+      "step": 2095
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8934271137884028e-05,
+      "loss": 0.8599,
+      "step": 2096
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.893306391461671e-05,
+      "loss": 0.8038,
+      "step": 2097
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.893185604651486e-05,
+      "loss": 0.8455,
+      "step": 2098
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8930647533665668e-05,
+      "loss": 0.9259,
+      "step": 2099
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8929438376156377e-05,
+      "loss": 0.9375,
+      "step": 2100
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8928228574074265e-05,
+      "loss": 0.8573,
+      "step": 2101
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8927018127506668e-05,
+      "loss": 0.7874,
+      "step": 2102
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8925807036540954e-05,
+      "loss": 0.8414,
+      "step": 2103
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.892459530126455e-05,
+      "loss": 0.8594,
+      "step": 2104
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8923382921764926e-05,
+      "loss": 0.8762,
+      "step": 2105
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8922169898129595e-05,
+      "loss": 0.8954,
+      "step": 2106
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8920956230446125e-05,
+      "loss": 0.824,
+      "step": 2107
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8919741918802117e-05,
+      "loss": 0.793,
+      "step": 2108
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.891852696328523e-05,
+      "loss": 0.8486,
+      "step": 2109
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.891731136398317e-05,
+      "loss": 0.8553,
+      "step": 2110
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8916095120983677e-05,
+      "loss": 0.9087,
+      "step": 2111
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.891487823437455e-05,
+      "loss": 0.8791,
+      "step": 2112
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8913660704243632e-05,
+      "loss": 0.8912,
+      "step": 2113
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8912442530678807e-05,
+      "loss": 0.8464,
+      "step": 2114
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.891122371376801e-05,
+      "loss": 0.8208,
+      "step": 2115
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8910004253599223e-05,
+      "loss": 0.8587,
+      "step": 2116
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.890878415026047e-05,
+      "loss": 0.8625,
+      "step": 2117
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8907563403839828e-05,
+      "loss": 0.9027,
+      "step": 2118
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8906342014425415e-05,
+      "loss": 0.8952,
+      "step": 2119
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8905119982105397e-05,
+      "loss": 0.8958,
+      "step": 2120
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8903897306967988e-05,
+      "loss": 0.8341,
+      "step": 2121
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8902673989101444e-05,
+      "loss": 0.8485,
+      "step": 2122
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.890145002859407e-05,
+      "loss": 0.8237,
+      "step": 2123
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8900225425534222e-05,
+      "loss": 0.9141,
+      "step": 2124
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8899000180010297e-05,
+      "loss": 0.8946,
+      "step": 2125
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8897774292110736e-05,
+      "loss": 0.8682,
+      "step": 2126
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8896547761924035e-05,
+      "loss": 0.8538,
+      "step": 2127
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.889532058953873e-05,
+      "loss": 0.8596,
+      "step": 2128
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8894092775043398e-05,
+      "loss": 0.7965,
+      "step": 2129
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8892864318526677e-05,
+      "loss": 0.8758,
+      "step": 2130
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8891635220077242e-05,
+      "loss": 0.8259,
+      "step": 2131
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8890405479783813e-05,
+      "loss": 0.8577,
+      "step": 2132
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.888917509773516e-05,
+      "loss": 0.8675,
+      "step": 2133
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8887944074020098e-05,
+      "loss": 0.8856,
+      "step": 2134
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.888671240872749e-05,
+      "loss": 0.8701,
+      "step": 2135
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8885480101946245e-05,
+      "loss": 0.8311,
+      "step": 2136
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8884247153765313e-05,
+      "loss": 0.8331,
+      "step": 2137
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.88830135642737e-05,
+      "loss": 0.8623,
+      "step": 2138
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8881779333560445e-05,
+      "loss": 0.8452,
+      "step": 2139
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8880544461714648e-05,
+      "loss": 0.8244,
+      "step": 2140
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8879308948825446e-05,
+      "loss": 0.8662,
+      "step": 2141
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8878072794982025e-05,
+      "loss": 0.9042,
+      "step": 2142
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8876836000273616e-05,
+      "loss": 0.8858,
+      "step": 2143
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.88755985647895e-05,
+      "loss": 0.7825,
+      "step": 2144
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8874360488619e-05,
+      "loss": 0.8759,
+      "step": 2145
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8873121771851483e-05,
+      "loss": 0.8542,
+      "step": 2146
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.887188241457637e-05,
+      "loss": 0.875,
+      "step": 2147
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.887064241688313e-05,
+      "loss": 0.8106,
+      "step": 2148
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8869401778861258e-05,
+      "loss": 0.8572,
+      "step": 2149
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8868160500600318e-05,
+      "loss": 0.8629,
+      "step": 2150
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8866918582189915e-05,
+      "loss": 0.8171,
+      "step": 2151
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.886567602371969e-05,
+      "loss": 0.8334,
+      "step": 2152
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.886443282527934e-05,
+      "loss": 0.8126,
+      "step": 2153
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8863188986958612e-05,
+      "loss": 0.8872,
+      "step": 2154
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.886194450884728e-05,
+      "loss": 0.8389,
+      "step": 2155
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8860699391035187e-05,
+      "loss": 0.922,
+      "step": 2156
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8859453633612208e-05,
+      "loss": 0.7816,
+      "step": 2157
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.885820723666827e-05,
+      "loss": 0.8493,
+      "step": 2158
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.885696020029334e-05,
+      "loss": 0.849,
+      "step": 2159
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8855712524577444e-05,
+      "loss": 0.8956,
+      "step": 2160
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8854464209610638e-05,
+      "loss": 0.408,
+      "step": 2161
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.885321525548303e-05,
+      "loss": 0.8027,
+      "step": 2162
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8851965662284785e-05,
+      "loss": 0.9026,
+      "step": 2163
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.88507154301061e-05,
+      "loss": 0.8972,
+      "step": 2164
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.884946455903722e-05,
+      "loss": 0.877,
+      "step": 2165
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8848213049168446e-05,
+      "loss": 0.8688,
+      "step": 2166
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.884696090059011e-05,
+      "loss": 0.8831,
+      "step": 2167
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.884570811339261e-05,
+      "loss": 0.9403,
+      "step": 2168
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8844454687666368e-05,
+      "loss": 0.8819,
+      "step": 2169
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8843200623501867e-05,
+      "loss": 0.7824,
+      "step": 2170
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8841945920989633e-05,
+      "loss": 0.8314,
+      "step": 2171
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8840690580220235e-05,
+      "loss": 0.8655,
+      "step": 2172
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.883943460128429e-05,
+      "loss": 0.8186,
+      "step": 2173
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.883817798427246e-05,
+      "loss": 0.8609,
+      "step": 2174
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.883692072927546e-05,
+      "loss": 0.86,
+      "step": 2175
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8835662836384038e-05,
+      "loss": 0.8155,
+      "step": 2176
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8834404305689e-05,
+      "loss": 0.7865,
+      "step": 2177
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.883314513728119e-05,
+      "loss": 0.8728,
+      "step": 2178
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.88318853312515e-05,
+      "loss": 0.8421,
+      "step": 2179
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8830624887690872e-05,
+      "loss": 0.8491,
+      "step": 2180
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8829363806690294e-05,
+      "loss": 0.8692,
+      "step": 2181
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8828102088340795e-05,
+      "loss": 0.8524,
+      "step": 2182
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8826839732733448e-05,
+      "loss": 0.9305,
+      "step": 2183
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8825576739959383e-05,
+      "loss": 0.9054,
+      "step": 2184
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8824313110109768e-05,
+      "loss": 0.8503,
+      "step": 2185
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.882304884327581e-05,
+      "loss": 0.8574,
+      "step": 2186
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8821783939548783e-05,
+      "loss": 0.9121,
+      "step": 2187
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8820518399019986e-05,
+      "loss": 0.9287,
+      "step": 2188
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8819252221780774e-05,
+      "loss": 0.8861,
+      "step": 2189
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8817985407922547e-05,
+      "loss": 0.8572,
+      "step": 2190
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.881671795753675e-05,
+      "loss": 0.8878,
+      "step": 2191
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8815449870714874e-05,
+      "loss": 0.8314,
+      "step": 2192
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8814181147548455e-05,
+      "loss": 0.8578,
+      "step": 2193
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8812911788129078e-05,
+      "loss": 0.8374,
+      "step": 2194
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.881164179254837e-05,
+      "loss": 0.8146,
+      "step": 2195
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8810371160898007e-05,
+      "loss": 0.7693,
+      "step": 2196
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8809099893269708e-05,
+      "loss": 0.8988,
+      "step": 2197
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.880782798975524e-05,
+      "loss": 0.8946,
+      "step": 2198
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.880655545044642e-05,
+      "loss": 0.8769,
+      "step": 2199
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.88052822754351e-05,
+      "loss": 0.8458,
+      "step": 2200
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.880400846481319e-05,
+      "loss": 0.8017,
+      "step": 2201
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8802734018672636e-05,
+      "loss": 0.8449,
+      "step": 2202
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8801458937105433e-05,
+      "loss": 0.8353,
+      "step": 2203
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8800183220203626e-05,
+      "loss": 0.3897,
+      "step": 2204
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8798906868059307e-05,
+      "loss": 0.7787,
+      "step": 2205
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.87976298807646e-05,
+      "loss": 0.8702,
+      "step": 2206
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8796352258411694e-05,
+      "loss": 0.8363,
+      "step": 2207
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8795074001092806e-05,
+      "loss": 0.9016,
+      "step": 2208
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8793795108900213e-05,
+      "loss": 0.8538,
+      "step": 2209
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.879251558192623e-05,
+      "loss": 0.916,
+      "step": 2210
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8791235420263216e-05,
+      "loss": 0.8754,
+      "step": 2211
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8789954624003587e-05,
+      "loss": 0.8247,
+      "step": 2212
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8788673193239794e-05,
+      "loss": 0.9099,
+      "step": 2213
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8787391128064335e-05,
+      "loss": 0.9254,
+      "step": 2214
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8786108428569762e-05,
+      "loss": 0.8788,
+      "step": 2215
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.878482509484866e-05,
+      "loss": 0.8972,
+      "step": 2216
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.878354112699367e-05,
+      "loss": 0.8807,
+      "step": 2217
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8782256525097477e-05,
+      "loss": 0.8722,
+      "step": 2218
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.878097128925281e-05,
+      "loss": 0.8683,
+      "step": 2219
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.877968541955244e-05,
+      "loss": 0.878,
+      "step": 2220
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.877839891608919e-05,
+      "loss": 0.9007,
+      "step": 2221
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8777111778955927e-05,
+      "loss": 0.8516,
+      "step": 2222
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8775824008245564e-05,
+      "loss": 0.8902,
+      "step": 2223
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.877453560405105e-05,
+      "loss": 0.8794,
+      "step": 2224
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8773246566465407e-05,
+      "loss": 0.8669,
+      "step": 2225
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8771956895581666e-05,
+      "loss": 0.8627,
+      "step": 2226
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8770666591492933e-05,
+      "loss": 0.9046,
+      "step": 2227
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8769375654292343e-05,
+      "loss": 0.8859,
+      "step": 2228
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8768084084073086e-05,
+      "loss": 0.8426,
+      "step": 2229
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8766791880928394e-05,
+      "loss": 0.8779,
+      "step": 2230
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8765499044951543e-05,
+      "loss": 0.9163,
+      "step": 2231
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.876420557623586e-05,
+      "loss": 0.9604,
+      "step": 2232
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.876291147487471e-05,
+      "loss": 0.8968,
+      "step": 2233
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.876161674096151e-05,
+      "loss": 0.8316,
+      "step": 2234
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8760321374589717e-05,
+      "loss": 0.8882,
+      "step": 2235
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8759025375852845e-05,
+      "loss": 0.8316,
+      "step": 2236
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.875772874484444e-05,
+      "loss": 0.8576,
+      "step": 2237
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.87564314816581e-05,
+      "loss": 0.7496,
+      "step": 2238
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.875513358638747e-05,
+      "loss": 0.8187,
+      "step": 2239
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8753835059126236e-05,
+      "loss": 0.7956,
+      "step": 2240
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8752535899968135e-05,
+      "loss": 0.8561,
+      "step": 2241
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8751236109006946e-05,
+      "loss": 0.8673,
+      "step": 2242
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8749935686336496e-05,
+      "loss": 0.8713,
+      "step": 2243
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8748634632050654e-05,
+      "loss": 0.8374,
+      "step": 2244
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8747332946243335e-05,
+      "loss": 0.3967,
+      "step": 2245
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8746030629008508e-05,
+      "loss": 0.8598,
+      "step": 2246
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8744727680440175e-05,
+      "loss": 0.8724,
+      "step": 2247
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.874342410063239e-05,
+      "loss": 0.8485,
+      "step": 2248
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8742119889679257e-05,
+      "loss": 0.864,
+      "step": 2249
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8740815047674917e-05,
+      "loss": 0.8696,
+      "step": 2250
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.873950957471356e-05,
+      "loss": 0.8334,
+      "step": 2251
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.873820347088942e-05,
+      "loss": 0.8172,
+      "step": 2252
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8736896736296782e-05,
+      "loss": 0.8358,
+      "step": 2253
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8735589371029973e-05,
+      "loss": 0.7829,
+      "step": 2254
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.873428137518336e-05,
+      "loss": 0.9233,
+      "step": 2255
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8732972748851368e-05,
+      "loss": 0.8884,
+      "step": 2256
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8731663492128454e-05,
+      "loss": 0.8059,
+      "step": 2257
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8730353605109134e-05,
+      "loss": 0.8088,
+      "step": 2258
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8729043087887956e-05,
+      "loss": 0.9032,
+      "step": 2259
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8727731940559525e-05,
+      "loss": 0.8266,
+      "step": 2260
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.872642016321848e-05,
+      "loss": 0.906,
+      "step": 2261
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.872510775595951e-05,
+      "loss": 0.8824,
+      "step": 2262
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8723794718877366e-05,
+      "loss": 0.8157,
+      "step": 2263
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8722481052066818e-05,
+      "loss": 0.8811,
+      "step": 2264
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8721166755622696e-05,
+      "loss": 0.8478,
+      "step": 2265
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8719851829639874e-05,
+      "loss": 0.8803,
+      "step": 2266
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8718536274213265e-05,
+      "loss": 0.8299,
+      "step": 2267
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8717220089437832e-05,
+      "loss": 0.8499,
+      "step": 2268
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8715903275408595e-05,
+      "loss": 0.8192,
+      "step": 2269
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.87145858322206e-05,
+      "loss": 0.7865,
+      "step": 2270
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8713267759968947e-05,
+      "loss": 0.8026,
+      "step": 2271
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8711949058748782e-05,
+      "loss": 0.8891,
+      "step": 2272
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8710629728655298e-05,
+      "loss": 0.8807,
+      "step": 2273
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.870930976978373e-05,
+      "loss": 0.8106,
+      "step": 2274
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8707989182229355e-05,
+      "loss": 0.8746,
+      "step": 2275
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8706667966087505e-05,
+      "loss": 0.887,
+      "step": 2276
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.870534612145355e-05,
+      "loss": 0.8633,
+      "step": 2277
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.870402364842291e-05,
+      "loss": 0.854,
+      "step": 2278
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8702700547091045e-05,
+      "loss": 0.8154,
+      "step": 2279
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8701376817553468e-05,
+      "loss": 0.3947,
+      "step": 2280
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8700052459905723e-05,
+      "loss": 0.8491,
+      "step": 2281
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.869872747424342e-05,
+      "loss": 0.8292,
+      "step": 2282
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8697401860662193e-05,
+      "loss": 0.8323,
+      "step": 2283
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.869607561925774e-05,
+      "loss": 0.847,
+      "step": 2284
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8694748750125792e-05,
+      "loss": 0.8677,
+      "step": 2285
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.869342125336213e-05,
+      "loss": 0.8469,
+      "step": 2286
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.869209312906258e-05,
+      "loss": 0.8873,
+      "step": 2287
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8690764377323016e-05,
+      "loss": 0.8371,
+      "step": 2288
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.868943499823934e-05,
+      "loss": 0.9003,
+      "step": 2289
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.868810499190753e-05,
+      "loss": 0.8215,
+      "step": 2290
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8686774358423592e-05,
+      "loss": 0.848,
+      "step": 2291
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8685443097883568e-05,
+      "loss": 0.8108,
+      "step": 2292
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8684111210383556e-05,
+      "loss": 0.8305,
+      "step": 2293
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.868277869601971e-05,
+      "loss": 0.8729,
+      "step": 2294
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.86814455548882e-05,
+      "loss": 0.8822,
+      "step": 2295
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.868011178708528e-05,
+      "loss": 0.8749,
+      "step": 2296
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.867877739270721e-05,
+      "loss": 0.8393,
+      "step": 2297
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8677442371850323e-05,
+      "loss": 0.8493,
+      "step": 2298
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8676106724610986e-05,
+      "loss": 0.8347,
+      "step": 2299
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.867477045108561e-05,
+      "loss": 0.8312,
+      "step": 2300
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.867343355137066e-05,
+      "loss": 0.8647,
+      "step": 2301
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8672096025562634e-05,
+      "loss": 0.8983,
+      "step": 2302
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8670757873758087e-05,
+      "loss": 0.81,
+      "step": 2303
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8669419096053607e-05,
+      "loss": 0.8522,
+      "step": 2304
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.866807969254584e-05,
+      "loss": 0.8123,
+      "step": 2305
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8666739663331466e-05,
+      "loss": 0.9236,
+      "step": 2306
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8665399008507223e-05,
+      "loss": 0.8489,
+      "step": 2307
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8664057728169876e-05,
+      "loss": 0.9447,
+      "step": 2308
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8662715822416257e-05,
+      "loss": 0.8432,
+      "step": 2309
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.866137329134322e-05,
+      "loss": 0.831,
+      "step": 2310
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8660030135047684e-05,
+      "loss": 0.767,
+      "step": 2311
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.86586863536266e-05,
+      "loss": 0.8657,
+      "step": 2312
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8657341947176976e-05,
+      "loss": 0.8815,
+      "step": 2313
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.865599691579585e-05,
+      "loss": 0.9088,
+      "step": 2314
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8654651259580316e-05,
+      "loss": 0.8443,
+      "step": 2315
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8653304978627512e-05,
+      "loss": 0.8785,
+      "step": 2316
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.865195807303462e-05,
+      "loss": 0.838,
+      "step": 2317
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8650610542898864e-05,
+      "loss": 0.8403,
+      "step": 2318
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.864926238831752e-05,
+      "loss": 0.9215,
+      "step": 2319
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8647913609387897e-05,
+      "loss": 0.8975,
+      "step": 2320
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8646564206207362e-05,
+      "loss": 0.8224,
+      "step": 2321
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8645214178873325e-05,
+      "loss": 0.8535,
+      "step": 2322
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8643863527483234e-05,
+      "loss": 0.8695,
+      "step": 2323
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8642512252134582e-05,
+      "loss": 0.8577,
+      "step": 2324
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8641160352924918e-05,
+      "loss": 0.9226,
+      "step": 2325
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8639807829951826e-05,
+      "loss": 0.8954,
+      "step": 2326
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.863845468331294e-05,
+      "loss": 0.8074,
+      "step": 2327
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8637100913105933e-05,
+      "loss": 0.9055,
+      "step": 2328
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.863574651942853e-05,
+      "loss": 0.8731,
+      "step": 2329
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8634391502378498e-05,
+      "loss": 0.8092,
+      "step": 2330
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.863303586205365e-05,
+      "loss": 0.853,
+      "step": 2331
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8631679598551837e-05,
+      "loss": 0.8222,
+      "step": 2332
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.863032271197097e-05,
+      "loss": 0.9032,
+      "step": 2333
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.862896520240899e-05,
+      "loss": 0.8156,
+      "step": 2334
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8627607069963895e-05,
+      "loss": 0.8875,
+      "step": 2335
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8626248314733715e-05,
+      "loss": 0.858,
+      "step": 2336
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8624888936816535e-05,
+      "loss": 0.8592,
+      "step": 2337
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8623528936310483e-05,
+      "loss": 0.8688,
+      "step": 2338
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8622168313313728e-05,
+      "loss": 0.825,
+      "step": 2339
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.862080706792449e-05,
+      "loss": 0.8495,
+      "step": 2340
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8619445200241028e-05,
+      "loss": 0.9057,
+      "step": 2341
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8618082710361648e-05,
+      "loss": 0.7459,
+      "step": 2342
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8616719598384704e-05,
+      "loss": 0.8437,
+      "step": 2343
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8615355864408597e-05,
+      "loss": 0.8423,
+      "step": 2344
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8613991508531758e-05,
+      "loss": 0.9249,
+      "step": 2345
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8612626530852678e-05,
+      "loss": 0.7919,
+      "step": 2346
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.861126093146989e-05,
+      "loss": 0.8144,
+      "step": 2347
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.860989471048197e-05,
+      "loss": 0.7731,
+      "step": 2348
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8608527867987533e-05,
+      "loss": 0.83,
+      "step": 2349
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.860716040408525e-05,
+      "loss": 0.8752,
+      "step": 2350
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8605792318873836e-05,
+      "loss": 0.9044,
+      "step": 2351
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8604423612452035e-05,
+      "loss": 0.8428,
+      "step": 2352
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8603054284918657e-05,
+      "loss": 0.8502,
+      "step": 2353
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.860168433637254e-05,
+      "loss": 0.8266,
+      "step": 2354
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.860031376691258e-05,
+      "loss": 0.8404,
+      "step": 2355
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.859894257663771e-05,
+      "loss": 0.9439,
+      "step": 2356
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8597570765646906e-05,
+      "loss": 0.8255,
+      "step": 2357
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8596198334039194e-05,
+      "loss": 0.9387,
+      "step": 2358
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8594825281913648e-05,
+      "loss": 0.8632,
+      "step": 2359
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8593451609369377e-05,
+      "loss": 0.9145,
+      "step": 2360
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8592077316505538e-05,
+      "loss": 0.8242,
+      "step": 2361
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.859070240342134e-05,
+      "loss": 0.8732,
+      "step": 2362
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.858932687021603e-05,
+      "loss": 0.8446,
+      "step": 2363
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8587950716988892e-05,
+      "loss": 0.8826,
+      "step": 2364
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.858657394383928e-05,
+      "loss": 0.8375,
+      "step": 2365
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8585196550866568e-05,
+      "loss": 0.8615,
+      "step": 2366
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.858381853817018e-05,
+      "loss": 0.8379,
+      "step": 2367
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8582439905849594e-05,
+      "loss": 0.8576,
+      "step": 2368
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.858106065400432e-05,
+      "loss": 0.821,
+      "step": 2369
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8579680782733927e-05,
+      "loss": 0.8728,
+      "step": 2370
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.857830029213802e-05,
+      "loss": 0.8861,
+      "step": 2371
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8576919182316246e-05,
+      "loss": 0.8887,
+      "step": 2372
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8575537453368303e-05,
+      "loss": 0.8254,
+      "step": 2373
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.857415510539393e-05,
+      "loss": 0.898,
+      "step": 2374
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8572772138492915e-05,
+      "loss": 0.8913,
+      "step": 2375
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8571388552765082e-05,
+      "loss": 0.837,
+      "step": 2376
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8570004348310317e-05,
+      "loss": 0.8481,
+      "step": 2377
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.856861952522852e-05,
+      "loss": 0.8752,
+      "step": 2378
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8567234083619672e-05,
+      "loss": 0.8767,
+      "step": 2379
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8565848023583775e-05,
+      "loss": 0.8542,
+      "step": 2380
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.856446134522088e-05,
+      "loss": 0.8326,
+      "step": 2381
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8563074048631086e-05,
+      "loss": 0.894,
+      "step": 2382
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8561686133914537e-05,
+      "loss": 0.3956,
+      "step": 2383
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8560297601171417e-05,
+      "loss": 0.8728,
+      "step": 2384
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.855890845050196e-05,
+      "loss": 0.8588,
+      "step": 2385
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8557518682006442e-05,
+      "loss": 0.9227,
+      "step": 2386
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.855612829578518e-05,
+      "loss": 0.8615,
+      "step": 2387
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8554737291938543e-05,
+      "loss": 0.7127,
+      "step": 2388
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.855334567056694e-05,
+      "loss": 0.8818,
+      "step": 2389
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8551953431770823e-05,
+      "loss": 0.8495,
+      "step": 2390
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8550560575650695e-05,
+      "loss": 0.8711,
+      "step": 2391
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8549167102307097e-05,
+      "loss": 0.8057,
+      "step": 2392
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8547773011840618e-05,
+      "loss": 0.8994,
+      "step": 2393
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.854637830435189e-05,
+      "loss": 0.809,
+      "step": 2394
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.854498297994159e-05,
+      "loss": 0.8344,
+      "step": 2395
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.854358703871044e-05,
+      "loss": 0.802,
+      "step": 2396
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8542190480759207e-05,
+      "loss": 0.8621,
+      "step": 2397
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8540793306188697e-05,
+      "loss": 0.8147,
+      "step": 2398
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8539395515099774e-05,
+      "loss": 0.7951,
+      "step": 2399
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8537997107593333e-05,
+      "loss": 0.3881,
+      "step": 2400
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8536598083770318e-05,
+      "loss": 0.851,
+      "step": 2401
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8535198443731714e-05,
+      "loss": 0.8915,
+      "step": 2402
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8533798187578563e-05,
+      "loss": 0.8676,
+      "step": 2403
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8532397315411933e-05,
+      "loss": 0.8753,
+      "step": 2404
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8530995827332955e-05,
+      "loss": 0.8667,
+      "step": 2405
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8529593723442794e-05,
+      "loss": 0.9005,
+      "step": 2406
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.852819100384265e-05,
+      "loss": 0.8507,
+      "step": 2407
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8526787668633797e-05,
+      "loss": 0.8652,
+      "step": 2408
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8525383717917524e-05,
+      "loss": 0.9019,
+      "step": 2409
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8523979151795172e-05,
+      "loss": 0.8405,
+      "step": 2410
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8522573970368135e-05,
+      "loss": 0.8291,
+      "step": 2411
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.852116817373785e-05,
+      "loss": 0.8506,
+      "step": 2412
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8519761762005783e-05,
+      "loss": 0.9439,
+      "step": 2413
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8518354735273468e-05,
+      "loss": 0.8728,
+      "step": 2414
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8516947093642468e-05,
+      "loss": 0.9488,
+      "step": 2415
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8515538837214392e-05,
+      "loss": 0.7754,
+      "step": 2416
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8514129966090894e-05,
+      "loss": 0.797,
+      "step": 2417
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8512720480373677e-05,
+      "loss": 0.8413,
+      "step": 2418
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.851131038016448e-05,
+      "loss": 0.8127,
+      "step": 2419
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8509899665565098e-05,
+      "loss": 0.8764,
+      "step": 2420
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.850848833667736e-05,
+      "loss": 0.881,
+      "step": 2421
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8507076393603145e-05,
+      "loss": 0.8616,
+      "step": 2422
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8505663836444368e-05,
+      "loss": 0.8029,
+      "step": 2423
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8504250665303002e-05,
+      "loss": 0.876,
+      "step": 2424
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8502836880281054e-05,
+      "loss": 0.8987,
+      "step": 2425
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8501422481480577e-05,
+      "loss": 0.8012,
+      "step": 2426
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8500007469003673e-05,
+      "loss": 0.7944,
+      "step": 2427
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8498591842952483e-05,
+      "loss": 0.8113,
+      "step": 2428
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.849717560342919e-05,
+      "loss": 0.8748,
+      "step": 2429
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8495758750536035e-05,
+      "loss": 0.8681,
+      "step": 2430
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8494341284375285e-05,
+      "loss": 0.3861,
+      "step": 2431
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8492923205049265e-05,
+      "loss": 0.8608,
+      "step": 2432
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8491504512660337e-05,
+      "loss": 0.8134,
+      "step": 2433
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8490085207310907e-05,
+      "loss": 0.8715,
+      "step": 2434
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8488665289103437e-05,
+      "loss": 0.8964,
+      "step": 2435
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8487244758140412e-05,
+      "loss": 0.8101,
+      "step": 2436
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8485823614524383e-05,
+      "loss": 0.8527,
+      "step": 2437
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.848440185835793e-05,
+      "loss": 0.9081,
+      "step": 2438
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.848297948974368e-05,
+      "loss": 0.3389,
+      "step": 2439
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8481556508784318e-05,
+      "loss": 0.8998,
+      "step": 2440
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8480132915582555e-05,
+      "loss": 0.7964,
+      "step": 2441
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.847870871024115e-05,
+      "loss": 0.8381,
+      "step": 2442
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8477283892862915e-05,
+      "loss": 0.8406,
+      "step": 2443
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8475858463550695e-05,
+      "loss": 0.8116,
+      "step": 2444
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.847443242240739e-05,
+      "loss": 0.9139,
+      "step": 2445
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.847300576953594e-05,
+      "loss": 0.8621,
+      "step": 2446
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8471578505039323e-05,
+      "loss": 0.8125,
+      "step": 2447
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.847015062902057e-05,
+      "loss": 0.86,
+      "step": 2448
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.846872214158275e-05,
+      "loss": 0.8726,
+      "step": 2449
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.846729304282898e-05,
+      "loss": 0.8786,
+      "step": 2450
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.846586333286242e-05,
+      "loss": 0.3854,
+      "step": 2451
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8464433011786273e-05,
+      "loss": 0.8883,
+      "step": 2452
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8463002079703793e-05,
+      "loss": 0.8658,
+      "step": 2453
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8461570536718257e-05,
+      "loss": 0.9006,
+      "step": 2454
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8460138382933018e-05,
+      "loss": 0.8029,
+      "step": 2455
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8458705618451444e-05,
+      "loss": 0.7942,
+      "step": 2456
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8457272243376964e-05,
+      "loss": 0.8787,
+      "step": 2457
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.845583825781305e-05,
+      "loss": 0.8571,
+      "step": 2458
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.845440366186321e-05,
+      "loss": 0.8502,
+      "step": 2459
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8452968455631e-05,
+      "loss": 0.8801,
+      "step": 2460
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.845153263922002e-05,
+      "loss": 0.8425,
+      "step": 2461
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8450096212733918e-05,
+      "loss": 0.8884,
+      "step": 2462
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8448659176276385e-05,
+      "loss": 0.9007,
+      "step": 2463
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8447221529951143e-05,
+      "loss": 0.8521,
+      "step": 2464
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.844578327386198e-05,
+      "loss": 0.8437,
+      "step": 2465
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8444344408112715e-05,
+      "loss": 0.8989,
+      "step": 2466
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8442904932807207e-05,
+      "loss": 0.8398,
+      "step": 2467
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8441464848049368e-05,
+      "loss": 0.929,
+      "step": 2468
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.844002415394315e-05,
+      "loss": 0.8981,
+      "step": 2469
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.843858285059255e-05,
+      "loss": 0.9389,
+      "step": 2470
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8437140938101613e-05,
+      "loss": 0.3949,
+      "step": 2471
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8435698416574417e-05,
+      "loss": 0.8815,
+      "step": 2472
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8434255286115093e-05,
+      "loss": 0.8955,
+      "step": 2473
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8432811546827812e-05,
+      "loss": 0.3679,
+      "step": 2474
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8431367198816794e-05,
+      "loss": 0.9019,
+      "step": 2475
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8429922242186298e-05,
+      "loss": 0.9089,
+      "step": 2476
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.842847667704063e-05,
+      "loss": 0.9005,
+      "step": 2477
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8427030503484133e-05,
+      "loss": 0.8978,
+      "step": 2478
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8425583721621207e-05,
+      "loss": 0.858,
+      "step": 2479
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8424136331556282e-05,
+      "loss": 0.8517,
+      "step": 2480
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8422688333393838e-05,
+      "loss": 0.865,
+      "step": 2481
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8421239727238405e-05,
+      "loss": 0.7903,
+      "step": 2482
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8419790513194544e-05,
+      "loss": 0.4035,
+      "step": 2483
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.841834069136687e-05,
+      "loss": 0.8366,
+      "step": 2484
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8416890261860038e-05,
+      "loss": 0.8663,
+      "step": 2485
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.841543922477875e-05,
+      "loss": 0.8993,
+      "step": 2486
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8413987580227745e-05,
+      "loss": 0.8864,
+      "step": 2487
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8412535328311813e-05,
+      "loss": 0.7397,
+      "step": 2488
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8411082469135782e-05,
+      "loss": 0.8924,
+      "step": 2489
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8409629002804533e-05,
+      "loss": 0.8439,
+      "step": 2490
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8408174929422978e-05,
+      "loss": 0.8748,
+      "step": 2491
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8406720249096085e-05,
+      "loss": 0.8128,
+      "step": 2492
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8405264961928856e-05,
+      "loss": 0.8131,
+      "step": 2493
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.840380906802634e-05,
+      "loss": 0.814,
+      "step": 2494
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8402352567493637e-05,
+      "loss": 0.8734,
+      "step": 2495
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.840089546043588e-05,
+      "loss": 0.7894,
+      "step": 2496
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8399437746958254e-05,
+      "loss": 0.8209,
+      "step": 2497
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8397979427165978e-05,
+      "loss": 0.875,
+      "step": 2498
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8396520501164327e-05,
+      "loss": 0.8719,
+      "step": 2499
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.839506096905861e-05,
+      "loss": 0.8519,
+      "step": 2500
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8393600830954188e-05,
+      "loss": 0.8539,
+      "step": 2501
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.839214008695646e-05,
+      "loss": 0.7835,
+      "step": 2502
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8390678737170863e-05,
+      "loss": 0.9104,
+      "step": 2503
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8389216781702896e-05,
+      "loss": 0.3985,
+      "step": 2504
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8387754220658084e-05,
+      "loss": 0.8998,
+      "step": 2505
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8386291054142003e-05,
+      "loss": 0.8522,
+      "step": 2506
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.838482728226027e-05,
+      "loss": 0.8367,
+      "step": 2507
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.838336290511855e-05,
+      "loss": 0.8144,
+      "step": 2508
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8381897922822548e-05,
+      "loss": 0.8466,
+      "step": 2509
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.838043233547802e-05,
+      "loss": 0.891,
+      "step": 2510
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.837896614319075e-05,
+      "loss": 0.869,
+      "step": 2511
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.837749934606658e-05,
+      "loss": 0.8891,
+      "step": 2512
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.837603194421139e-05,
+      "loss": 0.8792,
+      "step": 2513
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8374563937731108e-05,
+      "loss": 0.8951,
+      "step": 2514
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.83730953267317e-05,
+      "loss": 0.803,
+      "step": 2515
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8371626111319174e-05,
+      "loss": 0.8717,
+      "step": 2516
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.837015629159959e-05,
+      "loss": 0.8988,
+      "step": 2517
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8368685867679048e-05,
+      "loss": 0.8771,
+      "step": 2518
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8367214839663687e-05,
+      "loss": 0.9044,
+      "step": 2519
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8365743207659695e-05,
+      "loss": 0.8395,
+      "step": 2520
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8364270971773305e-05,
+      "loss": 0.8154,
+      "step": 2521
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8362798132110788e-05,
+      "loss": 0.8646,
+      "step": 2522
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.836132468877846e-05,
+      "loss": 0.8534,
+      "step": 2523
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8359850641882682e-05,
+      "loss": 0.8317,
+      "step": 2524
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8358375991529858e-05,
+      "loss": 0.8464,
+      "step": 2525
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8356900737826436e-05,
+      "loss": 0.9057,
+      "step": 2526
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8355424880878913e-05,
+      "loss": 0.8291,
+      "step": 2527
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8353948420793817e-05,
+      "loss": 0.8566,
+      "step": 2528
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.835247135767773e-05,
+      "loss": 0.8065,
+      "step": 2529
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8350993691637267e-05,
+      "loss": 0.8716,
+      "step": 2530
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8349515422779106e-05,
+      "loss": 0.8417,
+      "step": 2531
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8348036551209946e-05,
+      "loss": 0.8235,
+      "step": 2532
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8346557077036543e-05,
+      "loss": 0.8725,
+      "step": 2533
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8345077000365697e-05,
+      "loss": 0.845,
+      "step": 2534
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.834359632130424e-05,
+      "loss": 0.8454,
+      "step": 2535
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.834211503995906e-05,
+      "loss": 0.7614,
+      "step": 2536
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8340633156437077e-05,
+      "loss": 0.8415,
+      "step": 2537
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.833915067084527e-05,
+      "loss": 0.8076,
+      "step": 2538
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8337667583290647e-05,
+      "loss": 0.8566,
+      "step": 2539
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.833618389388027e-05,
+      "loss": 0.3868,
+      "step": 2540
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.833469960272123e-05,
+      "loss": 0.8475,
+      "step": 2541
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8333214709920682e-05,
+      "loss": 0.384,
+      "step": 2542
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8331729215585805e-05,
+      "loss": 0.8727,
+      "step": 2543
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8330243119823832e-05,
+      "loss": 0.8412,
+      "step": 2544
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.832875642274204e-05,
+      "loss": 0.8633,
+      "step": 2545
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8327269124447743e-05,
+      "loss": 0.904,
+      "step": 2546
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8325781225048304e-05,
+      "loss": 0.8298,
+      "step": 2547
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.832429272465112e-05,
+      "loss": 0.847,
+      "step": 2548
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.832280362336365e-05,
+      "loss": 0.8539,
+      "step": 2549
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8321313921293375e-05,
+      "loss": 0.8165,
+      "step": 2550
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8319823618547836e-05,
+      "loss": 0.88,
+      "step": 2551
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8318332715234612e-05,
+      "loss": 0.8948,
+      "step": 2552
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8316841211461317e-05,
+      "loss": 0.9343,
+      "step": 2553
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.831534910733562e-05,
+      "loss": 0.8524,
+      "step": 2554
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8313856402965226e-05,
+      "loss": 0.7895,
+      "step": 2555
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8312363098457886e-05,
+      "loss": 0.8572,
+      "step": 2556
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8310869193921403e-05,
+      "loss": 0.7821,
+      "step": 2557
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8309374689463606e-05,
+      "loss": 0.8132,
+      "step": 2558
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8307879585192377e-05,
+      "loss": 0.8267,
+      "step": 2559
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8306383881215638e-05,
+      "loss": 0.8097,
+      "step": 2560
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8304887577641365e-05,
+      "loss": 0.8785,
+      "step": 2561
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8303390674577562e-05,
+      "loss": 0.8403,
+      "step": 2562
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8301893172132286e-05,
+      "loss": 0.7586,
+      "step": 2563
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8300395070413634e-05,
+      "loss": 0.8911,
+      "step": 2564
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8298896369529746e-05,
+      "loss": 0.759,
+      "step": 2565
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8297397069588802e-05,
+      "loss": 0.8813,
+      "step": 2566
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8295897170699037e-05,
+      "loss": 0.8228,
+      "step": 2567
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8294396672968718e-05,
+      "loss": 0.8994,
+      "step": 2568
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8292895576506157e-05,
+      "loss": 0.8369,
+      "step": 2569
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8291393881419713e-05,
+      "loss": 0.8474,
+      "step": 2570
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8289891587817787e-05,
+      "loss": 0.8202,
+      "step": 2571
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8288388695808814e-05,
+      "loss": 0.8542,
+      "step": 2572
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.828688520550129e-05,
+      "loss": 0.7682,
+      "step": 2573
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8285381117003745e-05,
+      "loss": 0.8982,
+      "step": 2574
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8283876430424746e-05,
+      "loss": 0.8592,
+      "step": 2575
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8282371145872914e-05,
+      "loss": 0.8766,
+      "step": 2576
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8280865263456907e-05,
+      "loss": 0.8639,
+      "step": 2577
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8279358783285422e-05,
+      "loss": 0.8776,
+      "step": 2578
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.827785170546721e-05,
+      "loss": 0.899,
+      "step": 2579
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.827634403011106e-05,
+      "loss": 0.8224,
+      "step": 2580
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.82748357573258e-05,
+      "loss": 0.8328,
+      "step": 2581
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8273326887220308e-05,
+      "loss": 0.8295,
+      "step": 2582
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.82718174199035e-05,
+      "loss": 0.8397,
+      "step": 2583
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8270307355484342e-05,
+      "loss": 0.8559,
+      "step": 2584
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8268796694071827e-05,
+      "loss": 0.8089,
+      "step": 2585
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8267285435775015e-05,
+      "loss": 0.8777,
+      "step": 2586
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8265773580702986e-05,
+      "loss": 0.8067,
+      "step": 2587
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8264261128964883e-05,
+      "loss": 0.8678,
+      "step": 2588
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8262748080669878e-05,
+      "loss": 0.885,
+      "step": 2589
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.826123443592719e-05,
+      "loss": 0.8563,
+      "step": 2590
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.825972019484608e-05,
+      "loss": 0.8296,
+      "step": 2591
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.825820535753586e-05,
+      "loss": 0.8333,
+      "step": 2592
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8256689924105875e-05,
+      "loss": 0.8235,
+      "step": 2593
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8255173894665513e-05,
+      "loss": 0.8433,
+      "step": 2594
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8253657269324212e-05,
+      "loss": 0.87,
+      "step": 2595
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8252140048191452e-05,
+      "loss": 0.9238,
+      "step": 2596
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8250622231376754e-05,
+      "loss": 0.7958,
+      "step": 2597
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8249103818989678e-05,
+      "loss": 0.9068,
+      "step": 2598
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8247584811139832e-05,
+      "loss": 0.9013,
+      "step": 2599
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8246065207936867e-05,
+      "loss": 0.7839,
+      "step": 2600
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8244545009490477e-05,
+      "loss": 0.8429,
+      "step": 2601
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8243024215910393e-05,
+      "loss": 0.8906,
+      "step": 2602
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8241502827306397e-05,
+      "loss": 0.8281,
+      "step": 2603
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8239980843788313e-05,
+      "loss": 0.8267,
+      "step": 2604
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8238458265466e-05,
+      "loss": 0.8365,
+      "step": 2605
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.823693509244937e-05,
+      "loss": 0.9109,
+      "step": 2606
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8235411324848372e-05,
+      "loss": 0.8672,
+      "step": 2607
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8233886962773e-05,
+      "loss": 0.7932,
+      "step": 2608
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8232362006333287e-05,
+      "loss": 0.9139,
+      "step": 2609
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.823083645563932e-05,
+      "loss": 0.9048,
+      "step": 2610
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8229310310801212e-05,
+      "loss": 0.8553,
+      "step": 2611
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8227783571929133e-05,
+      "loss": 0.8156,
+      "step": 2612
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.822625623913329e-05,
+      "loss": 0.8422,
+      "step": 2613
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8224728312523938e-05,
+      "loss": 0.837,
+      "step": 2614
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8223199792211358e-05,
+      "loss": 0.7788,
+      "step": 2615
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.82216706783059e-05,
+      "loss": 0.7947,
+      "step": 2616
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8220140970917934e-05,
+      "loss": 0.8618,
+      "step": 2617
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8218610670157888e-05,
+      "loss": 0.8514,
+      "step": 2618
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8217079776136227e-05,
+      "loss": 0.888,
+      "step": 2619
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8215548288963454e-05,
+      "loss": 0.8069,
+      "step": 2620
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8214016208750124e-05,
+      "loss": 0.8459,
+      "step": 2621
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.821248353560683e-05,
+      "loss": 0.8177,
+      "step": 2622
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8210950269644206e-05,
+      "loss": 0.8161,
+      "step": 2623
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.820941641097293e-05,
+      "loss": 0.8333,
+      "step": 2624
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8207881959703732e-05,
+      "loss": 0.7975,
+      "step": 2625
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8206346915947366e-05,
+      "loss": 0.8602,
+      "step": 2626
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8204811279814646e-05,
+      "loss": 0.7831,
+      "step": 2627
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.820327505141642e-05,
+      "loss": 0.7947,
+      "step": 2628
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.820173823086358e-05,
+      "loss": 0.8935,
+      "step": 2629
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.820020081826706e-05,
+      "loss": 0.8769,
+      "step": 2630
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8198662813737846e-05,
+      "loss": 0.8938,
+      "step": 2631
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.819712421738695e-05,
+      "loss": 0.8474,
+      "step": 2632
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8195585029325444e-05,
+      "loss": 0.8686,
+      "step": 2633
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8194045249664427e-05,
+      "loss": 0.8551,
+      "step": 2634
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8192504878515054e-05,
+      "loss": 0.888,
+      "step": 2635
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.819096391598851e-05,
+      "loss": 0.7957,
+      "step": 2636
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8189422362196038e-05,
+      "loss": 0.8373,
+      "step": 2637
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8187880217248912e-05,
+      "loss": 0.8248,
+      "step": 2638
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8186337481258452e-05,
+      "loss": 0.7969,
+      "step": 2639
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8184794154336016e-05,
+      "loss": 0.8887,
+      "step": 2640
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8183250236593015e-05,
+      "loss": 0.8995,
+      "step": 2641
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8181705728140895e-05,
+      "loss": 0.8416,
+      "step": 2642
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.818016062909115e-05,
+      "loss": 0.8455,
+      "step": 2643
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8178614939555305e-05,
+      "loss": 0.8345,
+      "step": 2644
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8177068659644944e-05,
+      "loss": 0.8849,
+      "step": 2645
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8175521789471678e-05,
+      "loss": 0.9331,
+      "step": 2646
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8173974329147176e-05,
+      "loss": 0.7938,
+      "step": 2647
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.817242627878314e-05,
+      "loss": 0.8605,
+      "step": 2648
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.817087763849131e-05,
+      "loss": 0.8592,
+      "step": 2649
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8169328408383484e-05,
+      "loss": 0.8284,
+      "step": 2650
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8167778588571484e-05,
+      "loss": 0.8654,
+      "step": 2651
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.816622817916719e-05,
+      "loss": 0.8498,
+      "step": 2652
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8164677180282516e-05,
+      "loss": 0.852,
+      "step": 2653
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8163125592029424e-05,
+      "loss": 0.8074,
+      "step": 2654
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8161573414519908e-05,
+      "loss": 0.8007,
+      "step": 2655
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8160020647866025e-05,
+      "loss": 0.7877,
+      "step": 2656
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.815846729217985e-05,
+      "loss": 0.8603,
+      "step": 2657
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.815691334757352e-05,
+      "loss": 0.8257,
+      "step": 2658
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.81553588141592e-05,
+      "loss": 0.8886,
+      "step": 2659
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8153803692049107e-05,
+      "loss": 0.8953,
+      "step": 2660
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.81522479813555e-05,
+      "loss": 0.8547,
+      "step": 2661
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8150691682190677e-05,
+      "loss": 0.8831,
+      "step": 2662
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8149134794666977e-05,
+      "loss": 0.8773,
+      "step": 2663
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.814757731889679e-05,
+      "loss": 0.878,
+      "step": 2664
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.814601925499254e-05,
+      "loss": 0.8686,
+      "step": 2665
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.814446060306669e-05,
+      "loss": 0.8588,
+      "step": 2666
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.814290136323176e-05,
+      "loss": 0.8751,
+      "step": 2667
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.81413415356003e-05,
+      "loss": 0.8863,
+      "step": 2668
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8139781120284906e-05,
+      "loss": 0.8704,
+      "step": 2669
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8138220117398218e-05,
+      "loss": 0.925,
+      "step": 2670
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8136658527052916e-05,
+      "loss": 0.8814,
+      "step": 2671
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8135096349361725e-05,
+      "loss": 0.8115,
+      "step": 2672
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.813353358443741e-05,
+      "loss": 0.8795,
+      "step": 2673
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8131970232392783e-05,
+      "loss": 0.8733,
+      "step": 2674
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.813040629334069e-05,
+      "loss": 0.7794,
+      "step": 2675
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.812884176739403e-05,
+      "loss": 0.8389,
+      "step": 2676
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.812727665466573e-05,
+      "loss": 0.8158,
+      "step": 2677
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.812571095526877e-05,
+      "loss": 0.8711,
+      "step": 2678
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8124144669316177e-05,
+      "loss": 0.4394,
+      "step": 2679
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.812257779692101e-05,
+      "loss": 0.881,
+      "step": 2680
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.812101033819637e-05,
+      "loss": 0.3705,
+      "step": 2681
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.811944229325541e-05,
+      "loss": 0.8259,
+      "step": 2682
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8117873662211315e-05,
+      "loss": 0.8497,
+      "step": 2683
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8116304445177317e-05,
+      "loss": 0.8506,
+      "step": 2684
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8114734642266697e-05,
+      "loss": 0.8389,
+      "step": 2685
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.811316425359276e-05,
+      "loss": 0.9551,
+      "step": 2686
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8111593279268876e-05,
+      "loss": 0.844,
+      "step": 2687
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8110021719408442e-05,
+      "loss": 0.8769,
+      "step": 2688
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8108449574124896e-05,
+      "loss": 0.812,
+      "step": 2689
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8106876843531732e-05,
+      "loss": 0.8665,
+      "step": 2690
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8105303527742474e-05,
+      "loss": 0.8454,
+      "step": 2691
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8103729626870687e-05,
+      "loss": 0.7726,
+      "step": 2692
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.810215514102999e-05,
+      "loss": 0.8463,
+      "step": 2693
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.810058007033404e-05,
+      "loss": 0.8238,
+      "step": 2694
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8099004414896523e-05,
+      "loss": 0.8101,
+      "step": 2695
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8097428174831185e-05,
+      "loss": 0.7752,
+      "step": 2696
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8095851350251813e-05,
+      "loss": 0.9361,
+      "step": 2697
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8094273941272217e-05,
+      "loss": 0.8741,
+      "step": 2698
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.809269594800627e-05,
+      "loss": 0.8305,
+      "step": 2699
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8091117370567882e-05,
+      "loss": 0.8748,
+      "step": 2700
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8089538209070995e-05,
+      "loss": 0.8312,
+      "step": 2701
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.808795846362961e-05,
+      "loss": 0.8123,
+      "step": 2702
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8086378134357755e-05,
+      "loss": 0.7806,
+      "step": 2703
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.808479722136951e-05,
+      "loss": 0.8268,
+      "step": 2704
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.808321572477899e-05,
+      "loss": 0.8994,
+      "step": 2705
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8081633644700358e-05,
+      "loss": 0.8378,
+      "step": 2706
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.808005098124781e-05,
+      "loss": 0.8645,
+      "step": 2707
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8078467734535604e-05,
+      "loss": 0.8531,
+      "step": 2708
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8076883904678015e-05,
+      "loss": 0.8781,
+      "step": 2709
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.807529949178938e-05,
+      "loss": 0.7962,
+      "step": 2710
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8073714495984063e-05,
+      "loss": 0.8237,
+      "step": 2711
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.807212891737648e-05,
+      "loss": 0.8558,
+      "step": 2712
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8070542756081095e-05,
+      "loss": 0.8909,
+      "step": 2713
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8068956012212388e-05,
+      "loss": 0.4259,
+      "step": 2714
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.806736868588491e-05,
+      "loss": 0.8962,
+      "step": 2715
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8065780777213242e-05,
+      "loss": 0.889,
+      "step": 2716
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8064192286312007e-05,
+      "loss": 0.8482,
+      "step": 2717
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8062603213295866e-05,
+      "loss": 0.8878,
+      "step": 2718
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8061013558279528e-05,
+      "loss": 0.8896,
+      "step": 2719
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8059423321377746e-05,
+      "loss": 0.8483,
+      "step": 2720
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.805783250270531e-05,
+      "loss": 0.8267,
+      "step": 2721
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8056241102377048e-05,
+      "loss": 0.8491,
+      "step": 2722
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8054649120507844e-05,
+      "loss": 0.7497,
+      "step": 2723
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8053056557212613e-05,
+      "loss": 0.8737,
+      "step": 2724
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.805146341260631e-05,
+      "loss": 0.8454,
+      "step": 2725
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.804986968680394e-05,
+      "loss": 0.438,
+      "step": 2726
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8048275379920545e-05,
+      "loss": 0.831,
+      "step": 2727
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.804668049207121e-05,
+      "loss": 0.9425,
+      "step": 2728
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.804508502337107e-05,
+      "loss": 0.8351,
+      "step": 2729
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8043488973935283e-05,
+      "loss": 0.8207,
+      "step": 2730
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.804189234387906e-05,
+      "loss": 0.8158,
+      "step": 2731
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8040295133317664e-05,
+      "loss": 0.8607,
+      "step": 2732
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8038697342366383e-05,
+      "loss": 0.8912,
+      "step": 2733
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8037098971140555e-05,
+      "loss": 0.824,
+      "step": 2734
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.803550001975556e-05,
+      "loss": 0.9147,
+      "step": 2735
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8033900488326814e-05,
+      "loss": 0.8239,
+      "step": 2736
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8032300376969787e-05,
+      "loss": 0.822,
+      "step": 2737
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8030699685799976e-05,
+      "loss": 0.8187,
+      "step": 2738
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.802909841493293e-05,
+      "loss": 0.8189,
+      "step": 2739
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8027496564484236e-05,
+      "loss": 0.8789,
+      "step": 2740
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8025894134569527e-05,
+      "loss": 0.901,
+      "step": 2741
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8024291125304476e-05,
+      "loss": 0.8433,
+      "step": 2742
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8022687536804786e-05,
+      "loss": 0.8653,
+      "step": 2743
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.802108336918622e-05,
+      "loss": 0.8522,
+      "step": 2744
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.801947862256458e-05,
+      "loss": 0.8402,
+      "step": 2745
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8017873297055693e-05,
+      "loss": 0.8818,
+      "step": 2746
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.801626739277545e-05,
+      "loss": 0.8532,
+      "step": 2747
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.801466090983977e-05,
+      "loss": 0.8813,
+      "step": 2748
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8013053848364612e-05,
+      "loss": 0.8155,
+      "step": 2749
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.801144620846599e-05,
+      "loss": 0.7928,
+      "step": 2750
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.800983799025995e-05,
+      "loss": 0.9233,
+      "step": 2751
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8008229193862573e-05,
+      "loss": 0.7989,
+      "step": 2752
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.800661981939e-05,
+      "loss": 0.8329,
+      "step": 2753
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8005009866958404e-05,
+      "loss": 0.9026,
+      "step": 2754
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8003399336683994e-05,
+      "loss": 0.8199,
+      "step": 2755
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8001788228683027e-05,
+      "loss": 0.8863,
+      "step": 2756
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8000176543071805e-05,
+      "loss": 0.8436,
+      "step": 2757
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7998564279966665e-05,
+      "loss": 0.9035,
+      "step": 2758
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7996951439483987e-05,
+      "loss": 0.8633,
+      "step": 2759
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.79953380217402e-05,
+      "loss": 0.8064,
+      "step": 2760
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7993724026851764e-05,
+      "loss": 0.8852,
+      "step": 2761
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7992109454935184e-05,
+      "loss": 0.8324,
+      "step": 2762
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.799049430610701e-05,
+      "loss": 0.8934,
+      "step": 2763
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.798887858048384e-05,
+      "loss": 0.7742,
+      "step": 2764
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.798726227818229e-05,
+      "loss": 0.9174,
+      "step": 2765
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7985645399319042e-05,
+      "loss": 0.8236,
+      "step": 2766
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.798402794401081e-05,
+      "loss": 0.8585,
+      "step": 2767
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7982409912374347e-05,
+      "loss": 0.8368,
+      "step": 2768
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.798079130452646e-05,
+      "loss": 0.908,
+      "step": 2769
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7979172120583975e-05,
+      "loss": 0.8468,
+      "step": 2770
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7977552360663787e-05,
+      "loss": 0.8689,
+      "step": 2771
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7975932024882805e-05,
+      "loss": 0.8448,
+      "step": 2772
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7974311113358004e-05,
+      "loss": 0.886,
+      "step": 2773
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7972689626206384e-05,
+      "loss": 0.8159,
+      "step": 2774
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7971067563544994e-05,
+      "loss": 0.8356,
+      "step": 2775
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.796944492549092e-05,
+      "loss": 0.9134,
+      "step": 2776
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.79678217121613e-05,
+      "loss": 0.8141,
+      "step": 2777
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7966197923673297e-05,
+      "loss": 0.7931,
+      "step": 2778
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7964573560144128e-05,
+      "loss": 0.8143,
+      "step": 2779
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.796294862169105e-05,
+      "loss": 0.8804,
+      "step": 2780
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7961323108431357e-05,
+      "loss": 0.8812,
+      "step": 2781
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7959697020482387e-05,
+      "loss": 0.767,
+      "step": 2782
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7958070357961522e-05,
+      "loss": 0.883,
+      "step": 2783
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.795644312098618e-05,
+      "loss": 0.8228,
+      "step": 2784
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.795481530967382e-05,
+      "loss": 0.8559,
+      "step": 2785
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7953186924141956e-05,
+      "loss": 0.8874,
+      "step": 2786
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7951557964508125e-05,
+      "loss": 0.8067,
+      "step": 2787
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7949928430889915e-05,
+      "loss": 0.8417,
+      "step": 2788
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7948298323404955e-05,
+      "loss": 0.8509,
+      "step": 2789
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7946667642170912e-05,
+      "loss": 0.9168,
+      "step": 2790
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7945036387305504e-05,
+      "loss": 0.865,
+      "step": 2791
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7943404558926477e-05,
+      "loss": 0.8448,
+      "step": 2792
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7941772157151627e-05,
+      "loss": 0.8187,
+      "step": 2793
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7940139182098787e-05,
+      "loss": 0.8412,
+      "step": 2794
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7938505633885835e-05,
+      "loss": 0.4306,
+      "step": 2795
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.793687151263069e-05,
+      "loss": 0.8568,
+      "step": 2796
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7935236818451314e-05,
+      "loss": 0.9619,
+      "step": 2797
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7933601551465703e-05,
+      "loss": 0.8028,
+      "step": 2798
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7931965711791897e-05,
+      "loss": 0.9359,
+      "step": 2799
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7930329299547983e-05,
+      "loss": 0.8497,
+      "step": 2800
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.792869231485209e-05,
+      "loss": 0.8471,
+      "step": 2801
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7927054757822374e-05,
+      "loss": 0.773,
+      "step": 2802
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7925416628577048e-05,
+      "loss": 0.8186,
+      "step": 2803
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.792377792723436e-05,
+      "loss": 0.9035,
+      "step": 2804
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7922138653912603e-05,
+      "loss": 0.8677,
+      "step": 2805
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7920498808730104e-05,
+      "loss": 0.833,
+      "step": 2806
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7918858391805234e-05,
+      "loss": 0.7298,
+      "step": 2807
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.791721740325641e-05,
+      "loss": 0.7938,
+      "step": 2808
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7915575843202088e-05,
+      "loss": 0.8033,
+      "step": 2809
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.791393371176076e-05,
+      "loss": 0.801,
+      "step": 2810
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7912291009050967e-05,
+      "loss": 0.821,
+      "step": 2811
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.791064773519129e-05,
+      "loss": 0.8986,
+      "step": 2812
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.790900389030034e-05,
+      "loss": 0.8646,
+      "step": 2813
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.790735947449679e-05,
+      "loss": 0.7622,
+      "step": 2814
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.790571448789933e-05,
+      "loss": 0.8573,
+      "step": 2815
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7904068930626716e-05,
+      "loss": 0.8395,
+      "step": 2816
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7902422802797726e-05,
+      "loss": 0.7889,
+      "step": 2817
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7900776104531185e-05,
+      "loss": 0.8575,
+      "step": 2818
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.789912883594596e-05,
+      "loss": 0.8459,
+      "step": 2819
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7897480997160963e-05,
+      "loss": 0.8298,
+      "step": 2820
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7895832588295143e-05,
+      "loss": 0.8801,
+      "step": 2821
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7894183609467486e-05,
+      "loss": 0.8421,
+      "step": 2822
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7892534060797028e-05,
+      "loss": 0.7985,
+      "step": 2823
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7890883942402845e-05,
+      "loss": 0.8824,
+      "step": 2824
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.788923325440404e-05,
+      "loss": 0.8125,
+      "step": 2825
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7887581996919784e-05,
+      "loss": 0.8612,
+      "step": 2826
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7885930170069257e-05,
+      "loss": 0.8462,
+      "step": 2827
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.788427777397171e-05,
+      "loss": 0.7974,
+      "step": 2828
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7882624808746407e-05,
+      "loss": 0.9012,
+      "step": 2829
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7880971274512683e-05,
+      "loss": 0.8339,
+      "step": 2830
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7879317171389887e-05,
+      "loss": 0.825,
+      "step": 2831
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.787766249949743e-05,
+      "loss": 0.4108,
+      "step": 2832
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7876007258954745e-05,
+      "loss": 0.8616,
+      "step": 2833
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.787435144988132e-05,
+      "loss": 0.8709,
+      "step": 2834
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7872695072396684e-05,
+      "loss": 0.7831,
+      "step": 2835
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7871038126620397e-05,
+      "loss": 0.8567,
+      "step": 2836
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.786938061267207e-05,
+      "loss": 0.8523,
+      "step": 2837
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.786772253067135e-05,
+      "loss": 0.8597,
+      "step": 2838
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7866063880737923e-05,
+      "loss": 0.868,
+      "step": 2839
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.786440466299152e-05,
+      "loss": 0.8538,
+      "step": 2840
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7862744877551915e-05,
+      "loss": 0.8171,
+      "step": 2841
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7861084524538916e-05,
+      "loss": 0.8688,
+      "step": 2842
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7859423604072378e-05,
+      "loss": 0.8462,
+      "step": 2843
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7857762116272192e-05,
+      "loss": 0.8975,
+      "step": 2844
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.78561000612583e-05,
+      "loss": 0.3614,
+      "step": 2845
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7854437439150666e-05,
+      "loss": 0.4374,
+      "step": 2846
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.785277425006932e-05,
+      "loss": 0.8514,
+      "step": 2847
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.785111049413431e-05,
+      "loss": 0.8795,
+      "step": 2848
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7849446171465735e-05,
+      "loss": 0.9118,
+      "step": 2849
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7847781282183738e-05,
+      "loss": 0.829,
+      "step": 2850
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7846115826408497e-05,
+      "loss": 0.8466,
+      "step": 2851
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7844449804260232e-05,
+      "loss": 0.8641,
+      "step": 2852
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7842783215859213e-05,
+      "loss": 0.812,
+      "step": 2853
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7841116061325732e-05,
+      "loss": 0.8302,
+      "step": 2854
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7839448340780144e-05,
+      "loss": 0.8762,
+      "step": 2855
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.783778005434282e-05,
+      "loss": 0.9094,
+      "step": 2856
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7836111202134198e-05,
+      "loss": 0.8196,
+      "step": 2857
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.783444178427474e-05,
+      "loss": 0.8645,
+      "step": 2858
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.783277180088495e-05,
+      "loss": 0.8268,
+      "step": 2859
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7831101252085376e-05,
+      "loss": 0.8623,
+      "step": 2860
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7829430137996617e-05,
+      "loss": 0.8459,
+      "step": 2861
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.782775845873929e-05,
+      "loss": 0.367,
+      "step": 2862
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.782608621443407e-05,
+      "loss": 0.8261,
+      "step": 2863
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7824413405201667e-05,
+      "loss": 0.8304,
+      "step": 2864
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7822740031162834e-05,
+      "loss": 0.8516,
+      "step": 2865
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7821066092438367e-05,
+      "loss": 0.9125,
+      "step": 2866
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7819391589149097e-05,
+      "loss": 0.8583,
+      "step": 2867
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.78177165214159e-05,
+      "loss": 0.8254,
+      "step": 2868
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7816040889359685e-05,
+      "loss": 0.9021,
+      "step": 2869
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7814364693101414e-05,
+      "loss": 0.8152,
+      "step": 2870
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7812687932762078e-05,
+      "loss": 0.9411,
+      "step": 2871
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7811010608462723e-05,
+      "loss": 0.8292,
+      "step": 2872
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7809332720324418e-05,
+      "loss": 0.8294,
+      "step": 2873
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7807654268468285e-05,
+      "loss": 0.8789,
+      "step": 2874
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7805975253015484e-05,
+      "loss": 0.8621,
+      "step": 2875
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.780429567408721e-05,
+      "loss": 0.8273,
+      "step": 2876
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7802615531804715e-05,
+      "loss": 0.8932,
+      "step": 2877
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.780093482628927e-05,
+      "loss": 0.8651,
+      "step": 2878
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.77992535576622e-05,
+      "loss": 0.7962,
+      "step": 2879
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.779757172604487e-05,
+      "loss": 0.8405,
+      "step": 2880
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7795889331558677e-05,
+      "loss": 0.8426,
+      "step": 2881
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.779420637432507e-05,
+      "loss": 0.8267,
+      "step": 2882
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7792522854465532e-05,
+      "loss": 0.8057,
+      "step": 2883
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7790838772101594e-05,
+      "loss": 0.8012,
+      "step": 2884
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7789154127354812e-05,
+      "loss": 0.8197,
+      "step": 2885
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7787468920346794e-05,
+      "loss": 0.9242,
+      "step": 2886
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.778578315119919e-05,
+      "loss": 0.858,
+      "step": 2887
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7784096820033695e-05,
+      "loss": 0.8743,
+      "step": 2888
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7782409926972023e-05,
+      "loss": 0.9522,
+      "step": 2889
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.778072247213595e-05,
+      "loss": 0.3781,
+      "step": 2890
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7779034455647283e-05,
+      "loss": 0.3715,
+      "step": 2891
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7777345877627872e-05,
+      "loss": 0.8542,
+      "step": 2892
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7775656738199608e-05,
+      "loss": 0.7869,
+      "step": 2893
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7773967037484423e-05,
+      "loss": 0.8893,
+      "step": 2894
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7772276775604285e-05,
+      "loss": 0.8226,
+      "step": 2895
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.777058595268121e-05,
+      "loss": 0.8039,
+      "step": 2896
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.776889456883725e-05,
+      "loss": 0.8377,
+      "step": 2897
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.776720262419449e-05,
+      "loss": 0.8973,
+      "step": 2898
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7765510118875072e-05,
+      "loss": 0.7599,
+      "step": 2899
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.776381705300117e-05,
+      "loss": 0.7959,
+      "step": 2900
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7762123426694992e-05,
+      "loss": 0.8703,
+      "step": 2901
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7760429240078797e-05,
+      "loss": 0.8211,
+      "step": 2902
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7758734493274878e-05,
+      "loss": 0.819,
+      "step": 2903
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7757039186405568e-05,
+      "loss": 0.3969,
+      "step": 2904
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.775534331959325e-05,
+      "loss": 0.8533,
+      "step": 2905
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.775364689296034e-05,
+      "loss": 0.8106,
+      "step": 2906
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7751949906629287e-05,
+      "loss": 0.7985,
+      "step": 2907
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.775025236072259e-05,
+      "loss": 0.8783,
+      "step": 2908
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7748554255362794e-05,
+      "loss": 0.8732,
+      "step": 2909
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.774685559067247e-05,
+      "loss": 0.8967,
+      "step": 2910
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7745156366774242e-05,
+      "loss": 0.815,
+      "step": 2911
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7743456583790767e-05,
+      "loss": 0.9106,
+      "step": 2912
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7741756241844735e-05,
+      "loss": 0.8679,
+      "step": 2913
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.77400553410589e-05,
+      "loss": 0.7637,
+      "step": 2914
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7738353881556032e-05,
+      "loss": 0.828,
+      "step": 2915
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7736651863458954e-05,
+      "loss": 0.8431,
+      "step": 2916
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7734949286890526e-05,
+      "loss": 0.9022,
+      "step": 2917
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.773324615197365e-05,
+      "loss": 0.8226,
+      "step": 2918
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7731542458831262e-05,
+      "loss": 0.8863,
+      "step": 2919
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.772983820758635e-05,
+      "loss": 0.8504,
+      "step": 2920
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7728133398361937e-05,
+      "loss": 0.8749,
+      "step": 2921
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7726428031281076e-05,
+      "loss": 0.8721,
+      "step": 2922
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7724722106466875e-05,
+      "loss": 0.7071,
+      "step": 2923
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7723015624042478e-05,
+      "loss": 0.838,
+      "step": 2924
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7721308584131062e-05,
+      "loss": 0.9093,
+      "step": 2925
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7719600986855856e-05,
+      "loss": 0.8417,
+      "step": 2926
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.771789283234012e-05,
+      "loss": 0.8203,
+      "step": 2927
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7716184120707156e-05,
+      "loss": 0.8674,
+      "step": 2928
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7714474852080308e-05,
+      "loss": 0.8211,
+      "step": 2929
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7712765026582962e-05,
+      "loss": 0.8542,
+      "step": 2930
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7711054644338544e-05,
+      "loss": 0.8982,
+      "step": 2931
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7709343705470514e-05,
+      "loss": 0.8337,
+      "step": 2932
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7707632210102376e-05,
+      "loss": 0.7859,
+      "step": 2933
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7705920158357676e-05,
+      "loss": 0.867,
+      "step": 2934
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.770420755036e-05,
+      "loss": 0.8174,
+      "step": 2935
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7702494386232972e-05,
+      "loss": 0.8708,
+      "step": 2936
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7700780666100256e-05,
+      "loss": 0.8643,
+      "step": 2937
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.769906639008556e-05,
+      "loss": 0.8143,
+      "step": 2938
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7697351558312626e-05,
+      "loss": 0.8167,
+      "step": 2939
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.769563617090524e-05,
+      "loss": 0.887,
+      "step": 2940
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.769392022798723e-05,
+      "loss": 0.862,
+      "step": 2941
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7692203729682456e-05,
+      "loss": 0.8295,
+      "step": 2942
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.769048667611483e-05,
+      "loss": 0.9049,
+      "step": 2943
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7688769067408292e-05,
+      "loss": 0.8952,
+      "step": 2944
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7687050903686836e-05,
+      "loss": 0.8363,
+      "step": 2945
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.768533218507448e-05,
+      "loss": 0.8762,
+      "step": 2946
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.768361291169529e-05,
+      "loss": 0.8319,
+      "step": 2947
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.768189308367338e-05,
+      "loss": 0.8588,
+      "step": 2948
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7680172701132888e-05,
+      "loss": 0.8484,
+      "step": 2949
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7678451764198003e-05,
+      "loss": 0.8587,
+      "step": 2950
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7676730272992954e-05,
+      "loss": 0.8203,
+      "step": 2951
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7675008227642e-05,
+      "loss": 0.8884,
+      "step": 2952
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.767328562826945e-05,
+      "loss": 0.866,
+      "step": 2953
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7671562474999656e-05,
+      "loss": 0.8233,
+      "step": 2954
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7669838767957e-05,
+      "loss": 0.8955,
+      "step": 2955
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7668114507265904e-05,
+      "loss": 0.8261,
+      "step": 2956
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.766638969305084e-05,
+      "loss": 0.8424,
+      "step": 2957
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7664664325436308e-05,
+      "loss": 0.3875,
+      "step": 2958
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7662938404546856e-05,
+      "loss": 0.8162,
+      "step": 2959
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7661211930507075e-05,
+      "loss": 0.859,
+      "step": 2960
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7659484903441586e-05,
+      "loss": 0.7956,
+      "step": 2961
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7657757323475057e-05,
+      "loss": 0.9053,
+      "step": 2962
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.765602919073219e-05,
+      "loss": 0.8116,
+      "step": 2963
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.765430050533773e-05,
+      "loss": 0.8721,
+      "step": 2964
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.765257126741647e-05,
+      "loss": 0.8155,
+      "step": 2965
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.765084147709323e-05,
+      "loss": 0.883,
+      "step": 2966
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7649111134492874e-05,
+      "loss": 0.8392,
+      "step": 2967
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7647380239740307e-05,
+      "loss": 0.8157,
+      "step": 2968
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7645648792960475e-05,
+      "loss": 0.8635,
+      "step": 2969
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.764391679427837e-05,
+      "loss": 0.8917,
+      "step": 2970
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7642184243819005e-05,
+      "loss": 0.9112,
+      "step": 2971
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7640451141707447e-05,
+      "loss": 0.3662,
+      "step": 2972
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7638717488068808e-05,
+      "loss": 0.8653,
+      "step": 2973
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7636983283028223e-05,
+      "loss": 0.8286,
+      "step": 2974
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7635248526710882e-05,
+      "loss": 0.8359,
+      "step": 2975
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7633513219242005e-05,
+      "loss": 0.8034,
+      "step": 2976
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7631777360746856e-05,
+      "loss": 0.8466,
+      "step": 2977
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.763004095135074e-05,
+      "loss": 0.9138,
+      "step": 2978
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7628303991178995e-05,
+      "loss": 0.8538,
+      "step": 2979
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7626566480357008e-05,
+      "loss": 0.3804,
+      "step": 2980
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7624828419010202e-05,
+      "loss": 0.8651,
+      "step": 2981
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.762308980726404e-05,
+      "loss": 0.8395,
+      "step": 2982
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.762135064524402e-05,
+      "loss": 0.8988,
+      "step": 2983
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7619610933075683e-05,
+      "loss": 0.8734,
+      "step": 2984
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7617870670884617e-05,
+      "loss": 0.8258,
+      "step": 2985
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7616129858796437e-05,
+      "loss": 0.8308,
+      "step": 2986
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.761438849693681e-05,
+      "loss": 0.8214,
+      "step": 2987
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7612646585431427e-05,
+      "loss": 0.8105,
+      "step": 2988
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7610904124406035e-05,
+      "loss": 0.8729,
+      "step": 2989
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7609161113986414e-05,
+      "loss": 0.8826,
+      "step": 2990
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.760741755429838e-05,
+      "loss": 0.8491,
+      "step": 2991
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7605673445467796e-05,
+      "loss": 0.8452,
+      "step": 2992
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7603928787620558e-05,
+      "loss": 0.8353,
+      "step": 2993
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.76021835808826e-05,
+      "loss": 0.8119,
+      "step": 2994
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.760043782537991e-05,
+      "loss": 0.8548,
+      "step": 2995
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.75986915212385e-05,
+      "loss": 0.8949,
+      "step": 2996
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7596944668584428e-05,
+      "loss": 0.8253,
+      "step": 2997
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.759519726754379e-05,
+      "loss": 0.8412,
+      "step": 2998
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.759344931824272e-05,
+      "loss": 0.8008,
+      "step": 2999
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7591700820807404e-05,
+      "loss": 0.8006,
+      "step": 3000
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7589951775364045e-05,
+      "loss": 0.8168,
+      "step": 3001
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7588202182038904e-05,
+      "loss": 0.8177,
+      "step": 3002
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.758645204095828e-05,
+      "loss": 0.8988,
+      "step": 3003
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7584701352248498e-05,
+      "loss": 0.8719,
+      "step": 3004
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7582950116035935e-05,
+      "loss": 0.7919,
+      "step": 3005
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7581198332447006e-05,
+      "loss": 0.7884,
+      "step": 3006
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7579446001608166e-05,
+      "loss": 0.3828,
+      "step": 3007
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7577693123645906e-05,
+      "loss": 0.8185,
+      "step": 3008
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7575939698686754e-05,
+      "loss": 0.8118,
+      "step": 3009
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7574185726857282e-05,
+      "loss": 0.8617,
+      "step": 3010
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7572431208284104e-05,
+      "loss": 0.7825,
+      "step": 3011
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7570676143093865e-05,
+      "loss": 0.8044,
+      "step": 3012
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7568920531413264e-05,
+      "loss": 0.8088,
+      "step": 3013
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7567164373369018e-05,
+      "loss": 0.8199,
+      "step": 3014
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7565407669087905e-05,
+      "loss": 0.8756,
+      "step": 3015
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.756365041869673e-05,
+      "loss": 0.8277,
+      "step": 3016
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.756189262232234e-05,
+      "loss": 0.8373,
+      "step": 3017
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7560134280091618e-05,
+      "loss": 0.8488,
+      "step": 3018
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7558375392131493e-05,
+      "loss": 0.8392,
+      "step": 3019
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7556615958568936e-05,
+      "loss": 0.8203,
+      "step": 3020
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.755485597953095e-05,
+      "loss": 0.9012,
+      "step": 3021
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7553095455144575e-05,
+      "loss": 0.7751,
+      "step": 3022
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.755133438553689e-05,
+      "loss": 0.8073,
+      "step": 3023
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.754957277083503e-05,
+      "loss": 0.8259,
+      "step": 3024
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7547810611166153e-05,
+      "loss": 0.8603,
+      "step": 3025
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7546047906657456e-05,
+      "loss": 0.8371,
+      "step": 3026
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.754428465743619e-05,
+      "loss": 0.92,
+      "step": 3027
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.754252086362962e-05,
+      "loss": 0.848,
+      "step": 3028
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7540756525365083e-05,
+      "loss": 0.8499,
+      "step": 3029
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7538991642769927e-05,
+      "loss": 0.813,
+      "step": 3030
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7537226215971555e-05,
+      "loss": 0.8524,
+      "step": 3031
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.75354602450974e-05,
+      "loss": 0.3992,
+      "step": 3032
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7533693730274943e-05,
+      "loss": 0.8078,
+      "step": 3033
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.75319266716317e-05,
+      "loss": 0.8606,
+      "step": 3034
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.753015906929522e-05,
+      "loss": 0.8089,
+      "step": 3035
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7528390923393112e-05,
+      "loss": 0.8337,
+      "step": 3036
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7526622234052994e-05,
+      "loss": 0.8633,
+      "step": 3037
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7524853001402548e-05,
+      "loss": 0.9293,
+      "step": 3038
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7523083225569487e-05,
+      "loss": 0.855,
+      "step": 3039
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7521312906681557e-05,
+      "loss": 0.7524,
+      "step": 3040
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7519542044866548e-05,
+      "loss": 0.8346,
+      "step": 3041
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7517770640252298e-05,
+      "loss": 0.8276,
+      "step": 3042
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.751599869296667e-05,
+      "loss": 0.8957,
+      "step": 3043
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7514226203137576e-05,
+      "loss": 0.8647,
+      "step": 3044
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7512453170892956e-05,
+      "loss": 0.9092,
+      "step": 3045
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7510679596360808e-05,
+      "loss": 0.8356,
+      "step": 3046
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.750890547966915e-05,
+      "loss": 0.889,
+      "step": 3047
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7507130820946043e-05,
+      "loss": 0.7863,
+      "step": 3048
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.75053556203196e-05,
+      "loss": 0.8847,
+      "step": 3049
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7503579877917963e-05,
+      "loss": 0.8418,
+      "step": 3050
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7501803593869308e-05,
+      "loss": 0.9006,
+      "step": 3051
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.750002676830186e-05,
+      "loss": 0.8449,
+      "step": 3052
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7498249401343884e-05,
+      "loss": 0.8804,
+      "step": 3053
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.749647149312367e-05,
+      "loss": 0.8565,
+      "step": 3054
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7494693043769565e-05,
+      "loss": 0.7979,
+      "step": 3055
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7492914053409942e-05,
+      "loss": 0.7582,
+      "step": 3056
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.749113452217322e-05,
+      "loss": 0.8853,
+      "step": 3057
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.748935445018785e-05,
+      "loss": 0.8143,
+      "step": 3058
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.748757383758233e-05,
+      "loss": 0.8586,
+      "step": 3059
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.74857926844852e-05,
+      "loss": 0.8195,
+      "step": 3060
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7484010991025023e-05,
+      "loss": 0.8788,
+      "step": 3061
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7482228757330417e-05,
+      "loss": 0.782,
+      "step": 3062
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.748044598353003e-05,
+      "loss": 0.8156,
+      "step": 3063
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7478662669752548e-05,
+      "loss": 0.8558,
+      "step": 3064
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.747687881612671e-05,
+      "loss": 0.8948,
+      "step": 3065
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.747509442278127e-05,
+      "loss": 0.8698,
+      "step": 3066
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.747330948984505e-05,
+      "loss": 0.8783,
+      "step": 3067
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7471524017446886e-05,
+      "loss": 0.9314,
+      "step": 3068
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7469738005715664e-05,
+      "loss": 0.8558,
+      "step": 3069
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7467951454780306e-05,
+      "loss": 0.881,
+      "step": 3070
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7466164364769782e-05,
+      "loss": 0.7995,
+      "step": 3071
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7464376735813086e-05,
+      "loss": 0.8149,
+      "step": 3072
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.746258856803926e-05,
+      "loss": 0.8633,
+      "step": 3073
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7460799861577382e-05,
+      "loss": 0.8266,
+      "step": 3074
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.745901061655657e-05,
+      "loss": 0.8868,
+      "step": 3075
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.745722083310599e-05,
+      "loss": 0.9055,
+      "step": 3076
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.745543051135482e-05,
+      "loss": 0.879,
+      "step": 3077
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7453639651432313e-05,
+      "loss": 0.8238,
+      "step": 3078
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.745184825346773e-05,
+      "loss": 0.8024,
+      "step": 3079
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.745005631759039e-05,
+      "loss": 0.8194,
+      "step": 3080
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.744826384392964e-05,
+      "loss": 0.8312,
+      "step": 3081
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7446470832614872e-05,
+      "loss": 0.3866,
+      "step": 3082
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7444677283775515e-05,
+      "loss": 0.8368,
+      "step": 3083
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7442883197541038e-05,
+      "loss": 0.8087,
+      "step": 3084
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7441088574040942e-05,
+      "loss": 0.7876,
+      "step": 3085
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.743929341340478e-05,
+      "loss": 0.926,
+      "step": 3086
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7437497715762128e-05,
+      "loss": 0.8805,
+      "step": 3087
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7435701481242612e-05,
+      "loss": 0.87,
+      "step": 3088
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7433904709975898e-05,
+      "loss": 0.8307,
+      "step": 3089
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7432107402091678e-05,
+      "loss": 0.829,
+      "step": 3090
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7430309557719696e-05,
+      "loss": 0.907,
+      "step": 3091
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7428511176989727e-05,
+      "loss": 0.8662,
+      "step": 3092
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.742671226003159e-05,
+      "loss": 0.8695,
+      "step": 3093
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.742491280697514e-05,
+      "loss": 0.8779,
+      "step": 3094
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.742311281795027e-05,
+      "loss": 0.8433,
+      "step": 3095
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7421312293086913e-05,
+      "loss": 0.8283,
+      "step": 3096
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7419511232515035e-05,
+      "loss": 0.8354,
+      "step": 3097
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7417709636364654e-05,
+      "loss": 0.8958,
+      "step": 3098
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7415907504765815e-05,
+      "loss": 0.8198,
+      "step": 3099
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7414104837848604e-05,
+      "loss": 0.8465,
+      "step": 3100
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.741230163574315e-05,
+      "loss": 0.7664,
+      "step": 3101
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.741049789857961e-05,
+      "loss": 0.847,
+      "step": 3102
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7408693626488196e-05,
+      "loss": 0.829,
+      "step": 3103
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.740688881959914e-05,
+      "loss": 0.8628,
+      "step": 3104
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7405083478042736e-05,
+      "loss": 0.799,
+      "step": 3105
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.740327760194929e-05,
+      "loss": 0.789,
+      "step": 3106
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7401471191449165e-05,
+      "loss": 0.8907,
+      "step": 3107
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7399664246672756e-05,
+      "loss": 0.8159,
+      "step": 3108
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.73978567677505e-05,
+      "loss": 0.857,
+      "step": 3109
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7396048754812867e-05,
+      "loss": 0.8706,
+      "step": 3110
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.739424020799037e-05,
+      "loss": 0.8678,
+      "step": 3111
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7392431127413557e-05,
+      "loss": 0.865,
+      "step": 3112
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7390621513213017e-05,
+      "loss": 0.7831,
+      "step": 3113
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.738881136551938e-05,
+      "loss": 0.8342,
+      "step": 3114
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7387000684463314e-05,
+      "loss": 0.8336,
+      "step": 3115
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7385189470175518e-05,
+      "loss": 0.7898,
+      "step": 3116
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7383377722786737e-05,
+      "loss": 0.8578,
+      "step": 3117
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.738156544242775e-05,
+      "loss": 0.8818,
+      "step": 3118
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.737975262922938e-05,
+      "loss": 0.8325,
+      "step": 3119
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7377939283322484e-05,
+      "loss": 0.8202,
+      "step": 3120
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.737612540483796e-05,
+      "loss": 0.8592,
+      "step": 3121
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.737431099390674e-05,
+      "loss": 0.8708,
+      "step": 3122
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.73724960506598e-05,
+      "loss": 0.806,
+      "step": 3123
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.737068057522815e-05,
+      "loss": 0.8046,
+      "step": 3124
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.736886456774284e-05,
+      "loss": 0.8152,
+      "step": 3125
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7367048028334963e-05,
+      "loss": 0.4257,
+      "step": 3126
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7365230957135643e-05,
+      "loss": 0.915,
+      "step": 3127
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7363413354276043e-05,
+      "loss": 0.8509,
+      "step": 3128
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7361595219887374e-05,
+      "loss": 0.8948,
+      "step": 3129
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.735977655410087e-05,
+      "loss": 0.7875,
+      "step": 3130
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7357957357047817e-05,
+      "loss": 0.7761,
+      "step": 3131
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7356137628859536e-05,
+      "loss": 0.8862,
+      "step": 3132
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7354317369667378e-05,
+      "loss": 0.859,
+      "step": 3133
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.735249657960274e-05,
+      "loss": 0.8401,
+      "step": 3134
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7350675258797064e-05,
+      "loss": 0.8629,
+      "step": 3135
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.734885340738181e-05,
+      "loss": 0.8523,
+      "step": 3136
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.73470310254885e-05,
+      "loss": 0.8323,
+      "step": 3137
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7345208113248674e-05,
+      "loss": 0.8616,
+      "step": 3138
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7343384670793923e-05,
+      "loss": 0.7845,
+      "step": 3139
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7341560698255873e-05,
+      "loss": 0.8069,
+      "step": 3140
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7339736195766185e-05,
+      "loss": 0.9072,
+      "step": 3141
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7337911163456565e-05,
+      "loss": 0.8817,
+      "step": 3142
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.733608560145875e-05,
+      "loss": 0.7877,
+      "step": 3143
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7334259509904523e-05,
+      "loss": 0.8767,
+      "step": 3144
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7332432888925694e-05,
+      "loss": 0.8406,
+      "step": 3145
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7330605738654123e-05,
+      "loss": 0.9234,
+      "step": 3146
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7328778059221697e-05,
+      "loss": 0.883,
+      "step": 3147
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.732694985076036e-05,
+      "loss": 0.7991,
+      "step": 3148
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7325121113402067e-05,
+      "loss": 0.7665,
+      "step": 3149
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7323291847278833e-05,
+      "loss": 0.8467,
+      "step": 3150
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7321462052522703e-05,
+      "loss": 0.7999,
+      "step": 3151
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.731963172926576e-05,
+      "loss": 0.8502,
+      "step": 3152
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7317800877640126e-05,
+      "loss": 0.8433,
+      "step": 3153
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7315969497777963e-05,
+      "loss": 0.7905,
+      "step": 3154
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.731413758981147e-05,
+      "loss": 0.8219,
+      "step": 3155
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.731230515387288e-05,
+      "loss": 0.8501,
+      "step": 3156
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7310472190094474e-05,
+      "loss": 0.7866,
+      "step": 3157
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7308638698608556e-05,
+      "loss": 0.8501,
+      "step": 3158
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7306804679547485e-05,
+      "loss": 0.8713,
+      "step": 3159
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.730497013304364e-05,
+      "loss": 0.8409,
+      "step": 3160
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7303135059229463e-05,
+      "loss": 0.8081,
+      "step": 3161
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7301299458237405e-05,
+      "loss": 0.9025,
+      "step": 3162
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7299463330199973e-05,
+      "loss": 0.8337,
+      "step": 3163
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7297626675249715e-05,
+      "loss": 0.8781,
+      "step": 3164
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.72957894935192e-05,
+      "loss": 0.8364,
+      "step": 3165
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7293951785141053e-05,
+      "loss": 0.8065,
+      "step": 3166
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7292113550247926e-05,
+      "loss": 0.8855,
+      "step": 3167
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7290274788972513e-05,
+      "loss": 0.8005,
+      "step": 3168
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.728843550144754e-05,
+      "loss": 0.8146,
+      "step": 3169
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7286595687805787e-05,
+      "loss": 0.7944,
+      "step": 3170
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.728475534818005e-05,
+      "loss": 0.8487,
+      "step": 3171
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7282914482703187e-05,
+      "loss": 0.7732,
+      "step": 3172
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.728107309150807e-05,
+      "loss": 0.8454,
+      "step": 3173
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7279231174727624e-05,
+      "loss": 0.8629,
+      "step": 3174
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7277388732494803e-05,
+      "loss": 0.7767,
+      "step": 3175
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7275545764942614e-05,
+      "loss": 0.8324,
+      "step": 3176
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7273702272204086e-05,
+      "loss": 0.8274,
+      "step": 3177
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7271858254412293e-05,
+      "loss": 0.8308,
+      "step": 3178
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7270013711700343e-05,
+      "loss": 0.8338,
+      "step": 3179
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7268168644201385e-05,
+      "loss": 0.8671,
+      "step": 3180
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7266323052048613e-05,
+      "loss": 0.8,
+      "step": 3181
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7264476935375243e-05,
+      "loss": 0.8446,
+      "step": 3182
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.726263029431454e-05,
+      "loss": 0.819,
+      "step": 3183
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.72607831289998e-05,
+      "loss": 0.813,
+      "step": 3184
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.725893543956437e-05,
+      "loss": 0.8371,
+      "step": 3185
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7257087226141614e-05,
+      "loss": 0.8621,
+      "step": 3186
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7255238488864958e-05,
+      "loss": 0.8072,
+      "step": 3187
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7253389227867848e-05,
+      "loss": 0.8521,
+      "step": 3188
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.725153944328377e-05,
+      "loss": 0.8354,
+      "step": 3189
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7249689135246254e-05,
+      "loss": 0.8836,
+      "step": 3190
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.724783830388886e-05,
+      "loss": 0.8012,
+      "step": 3191
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7245986949345202e-05,
+      "loss": 0.9021,
+      "step": 3192
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7244135071748907e-05,
+      "loss": 0.8908,
+      "step": 3193
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7242282671233665e-05,
+      "loss": 0.8145,
+      "step": 3194
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.724042974793318e-05,
+      "loss": 0.8448,
+      "step": 3195
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7238576301981214e-05,
+      "loss": 0.9103,
+      "step": 3196
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7236722333511556e-05,
+      "loss": 0.8304,
+      "step": 3197
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7234867842658035e-05,
+      "loss": 0.8867,
+      "step": 3198
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7233012829554516e-05,
+      "loss": 0.7704,
+      "step": 3199
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7231157294334907e-05,
+      "loss": 0.8387,
+      "step": 3200
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7229301237133146e-05,
+      "loss": 0.8924,
+      "step": 3201
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7227444658083215e-05,
+      "loss": 0.835,
+      "step": 3202
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7225587557319132e-05,
+      "loss": 0.787,
+      "step": 3203
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7223729934974952e-05,
+      "loss": 0.8558,
+      "step": 3204
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7221871791184764e-05,
+      "loss": 0.8403,
+      "step": 3205
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.72200131260827e-05,
+      "loss": 0.8507,
+      "step": 3206
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7218153939802935e-05,
+      "loss": 0.8292,
+      "step": 3207
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7216294232479668e-05,
+      "loss": 0.8049,
+      "step": 3208
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.721443400424714e-05,
+      "loss": 0.8918,
+      "step": 3209
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7212573255239634e-05,
+      "loss": 0.8342,
+      "step": 3210
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7210711985591472e-05,
+      "loss": 0.8484,
+      "step": 3211
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.720885019543701e-05,
+      "loss": 0.8701,
+      "step": 3212
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.720698788491064e-05,
+      "loss": 0.8128,
+      "step": 3213
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7205125054146785e-05,
+      "loss": 0.422,
+      "step": 3214
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7203261703279925e-05,
+      "loss": 0.7733,
+      "step": 3215
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7201397832444563e-05,
+      "loss": 0.8248,
+      "step": 3216
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.719953344177524e-05,
+      "loss": 0.7791,
+      "step": 3217
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7197668531406542e-05,
+      "loss": 0.842,
+      "step": 3218
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.719580310147308e-05,
+      "loss": 0.8476,
+      "step": 3219
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7193937152109522e-05,
+      "loss": 0.8346,
+      "step": 3220
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7192070683450552e-05,
+      "loss": 0.7514,
+      "step": 3221
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7190203695630905e-05,
+      "loss": 0.8687,
+      "step": 3222
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.718833618878535e-05,
+      "loss": 0.8682,
+      "step": 3223
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.718646816304869e-05,
+      "loss": 0.8456,
+      "step": 3224
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7184599618555773e-05,
+      "loss": 0.8319,
+      "step": 3225
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7182730555441482e-05,
+      "loss": 0.755,
+      "step": 3226
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7180860973840726e-05,
+      "loss": 0.7645,
+      "step": 3227
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.717899087388847e-05,
+      "loss": 0.8265,
+      "step": 3228
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7177120255719708e-05,
+      "loss": 0.8918,
+      "step": 3229
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7175249119469462e-05,
+      "loss": 0.8183,
+      "step": 3230
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.717337746527281e-05,
+      "loss": 0.9189,
+      "step": 3231
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7171505293264853e-05,
+      "loss": 0.8565,
+      "step": 3232
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.716963260358073e-05,
+      "loss": 0.8356,
+      "step": 3233
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.716775939635563e-05,
+      "loss": 0.8531,
+      "step": 3234
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7165885671724764e-05,
+      "loss": 0.9038,
+      "step": 3235
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7164011429823393e-05,
+      "loss": 0.8754,
+      "step": 3236
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7162136670786802e-05,
+      "loss": 0.8867,
+      "step": 3237
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7160261394750326e-05,
+      "loss": 0.8403,
+      "step": 3238
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.715838560184933e-05,
+      "loss": 0.8418,
+      "step": 3239
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7156509292219223e-05,
+      "loss": 0.8622,
+      "step": 3240
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.715463246599544e-05,
+      "loss": 0.8249,
+      "step": 3241
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7152755123313462e-05,
+      "loss": 0.8302,
+      "step": 3242
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7150877264308807e-05,
+      "loss": 0.838,
+      "step": 3243
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.714899888911703e-05,
+      "loss": 0.9047,
+      "step": 3244
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7147119997873715e-05,
+      "loss": 0.8307,
+      "step": 3245
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.71452405907145e-05,
+      "loss": 0.9039,
+      "step": 3246
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.714336066777504e-05,
+      "loss": 0.8187,
+      "step": 3247
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7141480229191043e-05,
+      "loss": 0.8661,
+      "step": 3248
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.713959927509825e-05,
+      "loss": 0.8105,
+      "step": 3249
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7137717805632433e-05,
+      "loss": 0.8488,
+      "step": 3250
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7135835820929413e-05,
+      "loss": 0.8529,
+      "step": 3251
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7133953321125034e-05,
+      "loss": 0.7941,
+      "step": 3252
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7132070306355188e-05,
+      "loss": 0.9813,
+      "step": 3253
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7130186776755805e-05,
+      "loss": 0.875,
+      "step": 3254
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.712830273246284e-05,
+      "loss": 0.8148,
+      "step": 3255
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7126418173612296e-05,
+      "loss": 0.9261,
+      "step": 3256
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7124533100340213e-05,
+      "loss": 0.8516,
+      "step": 3257
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7122647512782663e-05,
+      "loss": 0.7593,
+      "step": 3258
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7120761411075752e-05,
+      "loss": 0.8584,
+      "step": 3259
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.711887479535564e-05,
+      "loss": 0.8544,
+      "step": 3260
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7116987665758505e-05,
+      "loss": 0.8598,
+      "step": 3261
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.711510002242057e-05,
+      "loss": 0.7962,
+      "step": 3262
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7113211865478097e-05,
+      "loss": 0.8851,
+      "step": 3263
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.711132319506738e-05,
+      "loss": 0.8281,
+      "step": 3264
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7109434011324758e-05,
+      "loss": 0.8615,
+      "step": 3265
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.71075443143866e-05,
+      "loss": 0.8439,
+      "step": 3266
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7105654104389308e-05,
+      "loss": 0.869,
+      "step": 3267
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7103763381469337e-05,
+      "loss": 0.8455,
+      "step": 3268
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7101872145763162e-05,
+      "loss": 0.9113,
+      "step": 3269
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7099980397407304e-05,
+      "loss": 0.8107,
+      "step": 3270
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.709808813653832e-05,
+      "loss": 0.8971,
+      "step": 3271
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.70961953632928e-05,
+      "loss": 0.7869,
+      "step": 3272
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7094302077807385e-05,
+      "loss": 0.8676,
+      "step": 3273
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7092408280218724e-05,
+      "loss": 0.8712,
+      "step": 3274
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7090513970663536e-05,
+      "loss": 0.8581,
+      "step": 3275
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7088619149278555e-05,
+      "loss": 0.8354,
+      "step": 3276
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.708672381620056e-05,
+      "loss": 0.805,
+      "step": 3277
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.708482797156637e-05,
+      "loss": 0.7947,
+      "step": 3278
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7082931615512832e-05,
+      "loss": 0.8065,
+      "step": 3279
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7081034748176834e-05,
+      "loss": 0.8185,
+      "step": 3280
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7079137369695307e-05,
+      "loss": 0.8469,
+      "step": 3281
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.707723948020521e-05,
+      "loss": 0.8696,
+      "step": 3282
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7075341079843543e-05,
+      "loss": 0.7923,
+      "step": 3283
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.707344216874734e-05,
+      "loss": 0.8782,
+      "step": 3284
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.707154274705368e-05,
+      "loss": 0.8541,
+      "step": 3285
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7069642814899666e-05,
+      "loss": 0.8545,
+      "step": 3286
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7067742372422448e-05,
+      "loss": 0.8009,
+      "step": 3287
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7065841419759212e-05,
+      "loss": 0.7931,
+      "step": 3288
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7063939957047177e-05,
+      "loss": 0.8601,
+      "step": 3289
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.70620379844236e-05,
+      "loss": 0.8226,
+      "step": 3290
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7060135502025774e-05,
+      "loss": 0.8429,
+      "step": 3291
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.705823250999103e-05,
+      "loss": 0.8253,
+      "step": 3292
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7056329008456737e-05,
+      "loss": 0.8353,
+      "step": 3293
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.70544249975603e-05,
+      "loss": 0.9064,
+      "step": 3294
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.705252047743916e-05,
+      "loss": 0.8303,
+      "step": 3295
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7050615448230798e-05,
+      "loss": 0.9163,
+      "step": 3296
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.704870991007272e-05,
+      "loss": 0.8487,
+      "step": 3297
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7046803863102486e-05,
+      "loss": 0.8633,
+      "step": 3298
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.704489730745768e-05,
+      "loss": 0.7798,
+      "step": 3299
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.704299024327593e-05,
+      "loss": 0.848,
+      "step": 3300
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7041082670694897e-05,
+      "loss": 0.781,
+      "step": 3301
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7039174589852277e-05,
+      "loss": 0.95,
+      "step": 3302
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7037266000885807e-05,
+      "loss": 0.9179,
+      "step": 3303
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.703535690393326e-05,
+      "loss": 0.8045,
+      "step": 3304
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7033447299132437e-05,
+      "loss": 0.8119,
+      "step": 3305
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.703153718662119e-05,
+      "loss": 0.814,
+      "step": 3306
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7029626566537406e-05,
+      "loss": 0.9036,
+      "step": 3307
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.702771543901899e-05,
+      "loss": 0.8201,
+      "step": 3308
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.702580380420391e-05,
+      "loss": 0.8313,
+      "step": 3309
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.702389166223015e-05,
+      "loss": 0.8965,
+      "step": 3310
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7021979013235737e-05,
+      "loss": 0.907,
+      "step": 3311
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.702006585735874e-05,
+      "loss": 0.8456,
+      "step": 3312
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.701815219473726e-05,
+      "loss": 0.856,
+      "step": 3313
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7016238025509436e-05,
+      "loss": 0.8245,
+      "step": 3314
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7014323349813437e-05,
+      "loss": 0.8493,
+      "step": 3315
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.701240816778748e-05,
+      "loss": 0.8419,
+      "step": 3316
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.701049247956981e-05,
+      "loss": 0.851,
+      "step": 3317
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7008576285298714e-05,
+      "loss": 0.8466,
+      "step": 3318
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.700665958511251e-05,
+      "loss": 0.8386,
+      "step": 3319
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7004742379149557e-05,
+      "loss": 0.885,
+      "step": 3320
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7002824667548243e-05,
+      "loss": 0.8863,
+      "step": 3321
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7000906450447007e-05,
+      "loss": 0.8731,
+      "step": 3322
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6998987727984312e-05,
+      "loss": 0.846,
+      "step": 3323
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.699706850029866e-05,
+      "loss": 0.8665,
+      "step": 3324
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6995148767528592e-05,
+      "loss": 0.8726,
+      "step": 3325
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6993228529812683e-05,
+      "loss": 0.8378,
+      "step": 3326
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6991307787289552e-05,
+      "loss": 0.7897,
+      "step": 3327
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6989386540097835e-05,
+      "loss": 0.8667,
+      "step": 3328
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.698746478837623e-05,
+      "loss": 0.8325,
+      "step": 3329
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6985542532263453e-05,
+      "loss": 0.7831,
+      "step": 3330
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.698361977189826e-05,
+      "loss": 0.8379,
+      "step": 3331
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6981696507419457e-05,
+      "loss": 0.8911,
+      "step": 3332
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6979772738965865e-05,
+      "loss": 0.8279,
+      "step": 3333
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6977848466676346e-05,
+      "loss": 0.8267,
+      "step": 3334
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.697592369068982e-05,
+      "loss": 0.8088,
+      "step": 3335
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6973998411145217e-05,
+      "loss": 0.8072,
+      "step": 3336
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6972072628181514e-05,
+      "loss": 0.7825,
+      "step": 3337
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.697014634193773e-05,
+      "loss": 0.4134,
+      "step": 3338
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6968219552552904e-05,
+      "loss": 0.3765,
+      "step": 3339
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6966292260166134e-05,
+      "loss": 0.8841,
+      "step": 3340
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6964364464916528e-05,
+      "loss": 0.8106,
+      "step": 3341
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6962436166943256e-05,
+      "loss": 0.823,
+      "step": 3342
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6960507366385508e-05,
+      "loss": 0.8852,
+      "step": 3343
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.695857806338251e-05,
+      "loss": 0.882,
+      "step": 3344
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6956648258073542e-05,
+      "loss": 0.7357,
+      "step": 3345
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6954717950597892e-05,
+      "loss": 0.81,
+      "step": 3346
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.695278714109491e-05,
+      "loss": 0.8685,
+      "step": 3347
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6950855829703965e-05,
+      "loss": 0.8642,
+      "step": 3348
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6948924016564473e-05,
+      "loss": 0.8745,
+      "step": 3349
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6946991701815885e-05,
+      "loss": 0.9116,
+      "step": 3350
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.694505888559768e-05,
+      "loss": 0.8211,
+      "step": 3351
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6943125568049377e-05,
+      "loss": 0.8529,
+      "step": 3352
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6941191749310538e-05,
+      "loss": 0.857,
+      "step": 3353
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6939257429520756e-05,
+      "loss": 0.8993,
+      "step": 3354
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.693732260881966e-05,
+      "loss": 0.8955,
+      "step": 3355
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6935387287346913e-05,
+      "loss": 0.8419,
+      "step": 3356
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6933451465242213e-05,
+      "loss": 0.8469,
+      "step": 3357
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.693151514264531e-05,
+      "loss": 0.9072,
+      "step": 3358
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.692957831969596e-05,
+      "loss": 0.8464,
+      "step": 3359
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6927640996533988e-05,
+      "loss": 0.8347,
+      "step": 3360
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6925703173299232e-05,
+      "loss": 0.8627,
+      "step": 3361
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6923764850131577e-05,
+      "loss": 0.854,
+      "step": 3362
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6921826027170946e-05,
+      "loss": 0.8443,
+      "step": 3363
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.691988670455728e-05,
+      "loss": 0.8887,
+      "step": 3364
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.691794688243058e-05,
+      "loss": 0.8977,
+      "step": 3365
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6916006560930868e-05,
+      "loss": 0.8595,
+      "step": 3366
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.691406574019821e-05,
+      "loss": 0.7755,
+      "step": 3367
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6912124420372703e-05,
+      "loss": 0.861,
+      "step": 3368
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6910182601594482e-05,
+      "loss": 0.8151,
+      "step": 3369
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.690824028400371e-05,
+      "loss": 0.7541,
+      "step": 3370
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6906297467740605e-05,
+      "loss": 0.8487,
+      "step": 3371
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.69043541529454e-05,
+      "loss": 0.8662,
+      "step": 3372
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6902410339758383e-05,
+      "loss": 0.8405,
+      "step": 3373
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.690046602831986e-05,
+      "loss": 0.8262,
+      "step": 3374
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6898521218770184e-05,
+      "loss": 0.8743,
+      "step": 3375
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.689657591124974e-05,
+      "loss": 0.948,
+      "step": 3376
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6894630105898954e-05,
+      "loss": 0.8117,
+      "step": 3377
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6892683802858284e-05,
+      "loss": 0.8483,
+      "step": 3378
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.689073700226822e-05,
+      "loss": 0.8279,
+      "step": 3379
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6888789704269298e-05,
+      "loss": 0.7979,
+      "step": 3380
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6886841909002075e-05,
+      "loss": 0.7862,
+      "step": 3381
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6884893616607164e-05,
+      "loss": 0.8621,
+      "step": 3382
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6882944827225195e-05,
+      "loss": 0.8698,
+      "step": 3383
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6880995540996846e-05,
+      "loss": 0.8645,
+      "step": 3384
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.687904575806282e-05,
+      "loss": 0.8873,
+      "step": 3385
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6877095478563872e-05,
+      "loss": 0.8584,
+      "step": 3386
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6875144702640776e-05,
+      "loss": 0.778,
+      "step": 3387
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6873193430434348e-05,
+      "loss": 0.8304,
+      "step": 3388
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.687124166208545e-05,
+      "loss": 0.8544,
+      "step": 3389
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.686928939773496e-05,
+      "loss": 0.8938,
+      "step": 3390
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.686733663752381e-05,
+      "loss": 0.8548,
+      "step": 3391
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6865383381592954e-05,
+      "loss": 0.7972,
+      "step": 3392
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6863429630083392e-05,
+      "loss": 0.8669,
+      "step": 3393
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6861475383136156e-05,
+      "loss": 0.8452,
+      "step": 3394
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6859520640892312e-05,
+      "loss": 0.7547,
+      "step": 3395
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6857565403492967e-05,
+      "loss": 0.8655,
+      "step": 3396
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6855609671079257e-05,
+      "loss": 0.8212,
+      "step": 3397
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6853653443792354e-05,
+      "loss": 0.8667,
+      "step": 3398
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6851696721773473e-05,
+      "loss": 0.768,
+      "step": 3399
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6849739505163862e-05,
+      "loss": 0.8468,
+      "step": 3400
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6847781794104795e-05,
+      "loss": 0.8446,
+      "step": 3401
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.68458235887376e-05,
+      "loss": 0.8817,
+      "step": 3402
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.684386488920362e-05,
+      "loss": 0.9429,
+      "step": 3403
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6841905695644252e-05,
+      "loss": 0.8301,
+      "step": 3404
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.683994600820092e-05,
+      "loss": 0.8387,
+      "step": 3405
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6837985827015078e-05,
+      "loss": 0.7634,
+      "step": 3406
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6836025152228227e-05,
+      "loss": 0.8063,
+      "step": 3407
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6834063983981896e-05,
+      "loss": 0.7767,
+      "step": 3408
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6832102322417653e-05,
+      "loss": 0.8838,
+      "step": 3409
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6830140167677107e-05,
+      "loss": 0.7798,
+      "step": 3410
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.682817751990189e-05,
+      "loss": 0.8018,
+      "step": 3411
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6826214379233672e-05,
+      "loss": 0.904,
+      "step": 3412
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.682425074581417e-05,
+      "loss": 0.8273,
+      "step": 3413
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.682228661978513e-05,
+      "loss": 0.9177,
+      "step": 3414
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6820322001288323e-05,
+      "loss": 0.8582,
+      "step": 3415
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6818356890465576e-05,
+      "loss": 0.783,
+      "step": 3416
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6816391287458736e-05,
+      "loss": 0.8341,
+      "step": 3417
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6814425192409695e-05,
+      "loss": 0.7003,
+      "step": 3418
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6812458605460364e-05,
+      "loss": 0.8246,
+      "step": 3419
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6810491526752713e-05,
+      "loss": 0.8218,
+      "step": 3420
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6808523956428736e-05,
+      "loss": 0.7814,
+      "step": 3421
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6806555894630456e-05,
+      "loss": 0.7924,
+      "step": 3422
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6804587341499936e-05,
+      "loss": 0.8465,
+      "step": 3423
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6802618297179283e-05,
+      "loss": 0.8911,
+      "step": 3424
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6800648761810632e-05,
+      "loss": 0.8303,
+      "step": 3425
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6798678735536153e-05,
+      "loss": 0.7909,
+      "step": 3426
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6796708218498047e-05,
+      "loss": 0.8159,
+      "step": 3427
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6794737210838564e-05,
+      "loss": 0.7314,
+      "step": 3428
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.679276571269998e-05,
+      "loss": 0.8458,
+      "step": 3429
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6790793724224604e-05,
+      "loss": 0.7845,
+      "step": 3430
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6788821245554792e-05,
+      "loss": 0.8486,
+      "step": 3431
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6786848276832913e-05,
+      "loss": 0.8538,
+      "step": 3432
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6784874818201404e-05,
+      "loss": 0.8421,
+      "step": 3433
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.678290086980271e-05,
+      "loss": 0.8218,
+      "step": 3434
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.678092643177932e-05,
+      "loss": 0.7958,
+      "step": 3435
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.677895150427376e-05,
+      "loss": 0.873,
+      "step": 3436
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6776976087428593e-05,
+      "loss": 0.767,
+      "step": 3437
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6775000181386408e-05,
+      "loss": 0.8739,
+      "step": 3438
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.677302378628985e-05,
+      "loss": 0.8367,
+      "step": 3439
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.677104690228157e-05,
+      "loss": 0.7817,
+      "step": 3440
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6769069529504277e-05,
+      "loss": 0.8616,
+      "step": 3441
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6767091668100706e-05,
+      "loss": 0.8754,
+      "step": 3442
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.676511331821363e-05,
+      "loss": 0.8326,
+      "step": 3443
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.676313447998586e-05,
+      "loss": 0.8321,
+      "step": 3444
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6761155153560234e-05,
+      "loss": 0.8833,
+      "step": 3445
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.675917533907963e-05,
+      "loss": 0.8322,
+      "step": 3446
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6757195036686965e-05,
+      "loss": 0.8086,
+      "step": 3447
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6755214246525182e-05,
+      "loss": 0.7995,
+      "step": 3448
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6753232968737272e-05,
+      "loss": 0.757,
+      "step": 3449
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6751251203466244e-05,
+      "loss": 0.8608,
+      "step": 3450
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6749268950855163e-05,
+      "loss": 0.7692,
+      "step": 3451
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.674728621104711e-05,
+      "loss": 0.8659,
+      "step": 3452
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.674530298418521e-05,
+      "loss": 0.8559,
+      "step": 3453
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6743319270412627e-05,
+      "loss": 0.8627,
+      "step": 3454
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.674133506987255e-05,
+      "loss": 0.7879,
+      "step": 3455
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6739350382708216e-05,
+      "loss": 0.7938,
+      "step": 3456
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6737365209062884e-05,
+      "loss": 0.8679,
+      "step": 3457
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6735379549079857e-05,
+      "loss": 0.8376,
+      "step": 3458
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.673339340290247e-05,
+      "loss": 0.8017,
+      "step": 3459
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.673140677067409e-05,
+      "loss": 0.8546,
+      "step": 3460
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6729419652538128e-05,
+      "loss": 0.8329,
+      "step": 3461
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6727432048638017e-05,
+      "loss": 0.8303,
+      "step": 3462
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.672544395911724e-05,
+      "loss": 0.9034,
+      "step": 3463
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.67234553841193e-05,
+      "loss": 0.8447,
+      "step": 3464
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6721466323787752e-05,
+      "loss": 0.8278,
+      "step": 3465
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.671947677826617e-05,
+      "loss": 0.8657,
+      "step": 3466
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.671748674769817e-05,
+      "loss": 0.7729,
+      "step": 3467
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6715496232227403e-05,
+      "loss": 0.868,
+      "step": 3468
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.671350523199756e-05,
+      "loss": 0.8094,
+      "step": 3469
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.671151374715235e-05,
+      "loss": 0.8489,
+      "step": 3470
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.670952177783554e-05,
+      "loss": 0.8607,
+      "step": 3471
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6707529324190915e-05,
+      "loss": 0.8216,
+      "step": 3472
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.67055363863623e-05,
+      "loss": 0.8647,
+      "step": 3473
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.670354296449356e-05,
+      "loss": 0.8517,
+      "step": 3474
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6701549058728586e-05,
+      "loss": 0.8715,
+      "step": 3475
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6699554669211312e-05,
+      "loss": 0.8378,
+      "step": 3476
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.66975597960857e-05,
+      "loss": 0.8497,
+      "step": 3477
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.669556443949575e-05,
+      "loss": 0.8293,
+      "step": 3478
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.66935685995855e-05,
+      "loss": 0.8783,
+      "step": 3479
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.669157227649902e-05,
+      "loss": 0.8459,
+      "step": 3480
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6689575470380413e-05,
+      "loss": 0.4347,
+      "step": 3481
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.668757818137382e-05,
+      "loss": 0.8479,
+      "step": 3482
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6685580409623416e-05,
+      "loss": 0.9123,
+      "step": 3483
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6683582155273406e-05,
+      "loss": 0.8054,
+      "step": 3484
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.668158341846804e-05,
+      "loss": 0.8184,
+      "step": 3485
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6679584199351594e-05,
+      "loss": 0.8499,
+      "step": 3486
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.667758449806838e-05,
+      "loss": 0.866,
+      "step": 3487
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.667558431476275e-05,
+      "loss": 0.8601,
+      "step": 3488
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.667358364957909e-05,
+      "loss": 0.8238,
+      "step": 3489
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6671582502661815e-05,
+      "loss": 0.8494,
+      "step": 3490
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.666958087415538e-05,
+      "loss": 0.8366,
+      "step": 3491
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6667578764204268e-05,
+      "loss": 0.8405,
+      "step": 3492
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6665576172953e-05,
+      "loss": 0.8863,
+      "step": 3493
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6663573100546145e-05,
+      "loss": 0.8681,
+      "step": 3494
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6661569547128286e-05,
+      "loss": 0.8674,
+      "step": 3495
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6659565512844053e-05,
+      "loss": 0.8605,
+      "step": 3496
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6657560997838105e-05,
+      "loss": 0.7775,
+      "step": 3497
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6655556002255144e-05,
+      "loss": 0.8499,
+      "step": 3498
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6653550526239892e-05,
+      "loss": 0.8112,
+      "step": 3499
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6651544569937118e-05,
+      "loss": 0.8163,
+      "step": 3500
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6649538133491628e-05,
+      "loss": 0.878,
+      "step": 3501
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6647531217048253e-05,
+      "loss": 0.841,
+      "step": 3502
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6645523820751857e-05,
+      "loss": 0.822,
+      "step": 3503
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6643515944747354e-05,
+      "loss": 0.8673,
+      "step": 3504
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.664150758917967e-05,
+      "loss": 0.8307,
+      "step": 3505
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6639498754193796e-05,
+      "loss": 0.8223,
+      "step": 3506
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6637489439934724e-05,
+      "loss": 0.8887,
+      "step": 3507
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6635479646547505e-05,
+      "loss": 0.862,
+      "step": 3508
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6633469374177216e-05,
+      "loss": 0.7642,
+      "step": 3509
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6631458622968965e-05,
+      "loss": 0.8225,
+      "step": 3510
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6629447393067898e-05,
+      "loss": 0.8514,
+      "step": 3511
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6627435684619203e-05,
+      "loss": 0.84,
+      "step": 3512
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6625423497768088e-05,
+      "loss": 0.862,
+      "step": 3513
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.662341083265981e-05,
+      "loss": 0.8146,
+      "step": 3514
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6621397689439643e-05,
+      "loss": 0.7905,
+      "step": 3515
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6619384068252917e-05,
+      "loss": 0.8072,
+      "step": 3516
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.661736996924498e-05,
+      "loss": 0.8826,
+      "step": 3517
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6615355392561225e-05,
+      "loss": 0.4099,
+      "step": 3518
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6613340338347065e-05,
+      "loss": 0.8342,
+      "step": 3519
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6611324806747962e-05,
+      "loss": 0.8079,
+      "step": 3520
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6609308797909413e-05,
+      "loss": 0.8924,
+      "step": 3521
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6607292311976933e-05,
+      "loss": 0.812,
+      "step": 3522
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.660527534909609e-05,
+      "loss": 0.8377,
+      "step": 3523
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.660325790941248e-05,
+      "loss": 0.8751,
+      "step": 3524
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6601239993071727e-05,
+      "loss": 0.8213,
+      "step": 3525
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6599221600219496e-05,
+      "loss": 0.8173,
+      "step": 3526
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6597202731001485e-05,
+      "loss": 0.7536,
+      "step": 3527
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6595183385563428e-05,
+      "loss": 0.9225,
+      "step": 3528
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6593163564051092e-05,
+      "loss": 0.8259,
+      "step": 3529
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6591143266610277e-05,
+      "loss": 0.8254,
+      "step": 3530
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.658912249338682e-05,
+      "loss": 0.815,
+      "step": 3531
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6587101244526582e-05,
+      "loss": 0.7761,
+      "step": 3532
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6585079520175483e-05,
+      "loss": 0.8304,
+      "step": 3533
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6583057320479445e-05,
+      "loss": 0.8165,
+      "step": 3534
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6581034645584452e-05,
+      "loss": 0.8391,
+      "step": 3535
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.657901149563651e-05,
+      "loss": 0.8325,
+      "step": 3536
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6576987870781656e-05,
+      "loss": 0.3676,
+      "step": 3537
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.657496377116597e-05,
+      "loss": 0.823,
+      "step": 3538
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6572939196935555e-05,
+      "loss": 0.8311,
+      "step": 3539
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6570914148236562e-05,
+      "loss": 0.7704,
+      "step": 3540
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6568888625215168e-05,
+      "loss": 0.7716,
+      "step": 3541
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6566862628017585e-05,
+      "loss": 0.8014,
+      "step": 3542
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.656483615679006e-05,
+      "loss": 0.9216,
+      "step": 3543
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.656280921167887e-05,
+      "loss": 0.8222,
+      "step": 3544
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.656078179283034e-05,
+      "loss": 0.8485,
+      "step": 3545
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.655875390039081e-05,
+      "loss": 0.8287,
+      "step": 3546
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.655672553450667e-05,
+      "loss": 0.8401,
+      "step": 3547
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6554696695324334e-05,
+      "loss": 0.9075,
+      "step": 3548
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6552667382990258e-05,
+      "loss": 0.7719,
+      "step": 3549
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.655063759765092e-05,
+      "loss": 0.8034,
+      "step": 3550
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6548607339452853e-05,
+      "loss": 0.8295,
+      "step": 3551
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.65465766085426e-05,
+      "loss": 0.8186,
+      "step": 3552
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6544545405066755e-05,
+      "loss": 0.8286,
+      "step": 3553
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.654251372917194e-05,
+      "loss": 0.8338,
+      "step": 3554
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.654048158100481e-05,
+      "loss": 0.8443,
+      "step": 3555
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6538448960712062e-05,
+      "loss": 0.8323,
+      "step": 3556
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6536415868440415e-05,
+      "loss": 0.7664,
+      "step": 3557
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6534382304336627e-05,
+      "loss": 0.8172,
+      "step": 3558
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6532348268547495e-05,
+      "loss": 0.8605,
+      "step": 3559
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.653031376121985e-05,
+      "loss": 0.8321,
+      "step": 3560
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6528278782500538e-05,
+      "loss": 0.7888,
+      "step": 3561
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.652624333253647e-05,
+      "loss": 0.8713,
+      "step": 3562
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.652420741147457e-05,
+      "loss": 0.7981,
+      "step": 3563
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6522171019461805e-05,
+      "loss": 0.7803,
+      "step": 3564
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.652013415664516e-05,
+      "loss": 0.4344,
+      "step": 3565
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.651809682317168e-05,
+      "loss": 0.8083,
+      "step": 3566
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.651605901918842e-05,
+      "loss": 0.7553,
+      "step": 3567
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6514020744842487e-05,
+      "loss": 0.8169,
+      "step": 3568
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6511982000281013e-05,
+      "loss": 0.7886,
+      "step": 3569
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6509942785651157e-05,
+      "loss": 0.8692,
+      "step": 3570
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.650790310110013e-05,
+      "loss": 0.9065,
+      "step": 3571
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6505862946775163e-05,
+      "loss": 0.9022,
+      "step": 3572
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6503822322823524e-05,
+      "loss": 0.836,
+      "step": 3573
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6501781229392517e-05,
+      "loss": 0.8195,
+      "step": 3574
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.649973966662948e-05,
+      "loss": 0.8155,
+      "step": 3575
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.649769763468178e-05,
+      "loss": 0.8805,
+      "step": 3576
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.649565513369682e-05,
+      "loss": 0.8508,
+      "step": 3577
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.649361216382204e-05,
+      "loss": 0.8444,
+      "step": 3578
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.649156872520492e-05,
+      "loss": 0.8528,
+      "step": 3579
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6489524817992956e-05,
+      "loss": 0.812,
+      "step": 3580
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6487480442333693e-05,
+      "loss": 0.83,
+      "step": 3581
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6485435598374698e-05,
+      "loss": 0.7296,
+      "step": 3582
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.648339028626359e-05,
+      "loss": 0.8946,
+      "step": 3583
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6481344506147995e-05,
+      "loss": 0.867,
+      "step": 3584
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6479298258175603e-05,
+      "loss": 0.8165,
+      "step": 3585
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.647725154249411e-05,
+      "loss": 0.816,
+      "step": 3586
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.647520435925127e-05,
+      "loss": 0.8359,
+      "step": 3587
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6473156708594848e-05,
+      "loss": 0.8207,
+      "step": 3588
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.647110859067266e-05,
+      "loss": 0.8258,
+      "step": 3589
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6469060005632554e-05,
+      "loss": 0.8216,
+      "step": 3590
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.64670109536224e-05,
+      "loss": 0.7832,
+      "step": 3591
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.646496143479011e-05,
+      "loss": 0.9007,
+      "step": 3592
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.646291144928363e-05,
+      "loss": 0.3961,
+      "step": 3593
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.646086099725094e-05,
+      "loss": 0.3911,
+      "step": 3594
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6458810078840054e-05,
+      "loss": 0.8405,
+      "step": 3595
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6456758694199008e-05,
+      "loss": 0.8722,
+      "step": 3596
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6454706843475893e-05,
+      "loss": 0.8158,
+      "step": 3597
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.645265452681882e-05,
+      "loss": 0.4083,
+      "step": 3598
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.645060174437593e-05,
+      "loss": 0.3896,
+      "step": 3599
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6448548496295406e-05,
+      "loss": 0.7896,
+      "step": 3600
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6446494782725463e-05,
+      "loss": 0.8265,
+      "step": 3601
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6444440603814352e-05,
+      "loss": 0.8439,
+      "step": 3602
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6442385959710348e-05,
+      "loss": 0.808,
+      "step": 3603
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.644033085056177e-05,
+      "loss": 0.8052,
+      "step": 3604
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.643827527651696e-05,
+      "loss": 0.8544,
+      "step": 3605
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6436219237724312e-05,
+      "loss": 0.7456,
+      "step": 3606
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6434162734332227e-05,
+      "loss": 0.8089,
+      "step": 3607
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6432105766489162e-05,
+      "loss": 0.8337,
+      "step": 3608
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6430048334343604e-05,
+      "loss": 0.8046,
+      "step": 3609
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6427990438044063e-05,
+      "loss": 0.8488,
+      "step": 3610
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6425932077739087e-05,
+      "loss": 0.8352,
+      "step": 3611
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.642387325357726e-05,
+      "loss": 0.7702,
+      "step": 3612
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6421813965707198e-05,
+      "loss": 0.8172,
+      "step": 3613
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.641975421427756e-05,
+      "loss": 0.8391,
+      "step": 3614
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6417693999437015e-05,
+      "loss": 0.7972,
+      "step": 3615
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.641563332133429e-05,
+      "loss": 0.8799,
+      "step": 3616
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6413572180118132e-05,
+      "loss": 0.8559,
+      "step": 3617
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.641151057593733e-05,
+      "loss": 0.8656,
+      "step": 3618
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6409448508940692e-05,
+      "loss": 0.873,
+      "step": 3619
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6407385979277075e-05,
+      "loss": 0.876,
+      "step": 3620
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6405322987095358e-05,
+      "loss": 0.8393,
+      "step": 3621
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6403259532544465e-05,
+      "loss": 0.8189,
+      "step": 3622
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.640119561577334e-05,
+      "loss": 0.8749,
+      "step": 3623
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6399131236930974e-05,
+      "loss": 0.7838,
+      "step": 3624
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.639706639616638e-05,
+      "loss": 0.8483,
+      "step": 3625
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6395001093628607e-05,
+      "loss": 0.814,
+      "step": 3626
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6392935329466742e-05,
+      "loss": 0.8323,
+      "step": 3627
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6390869103829903e-05,
+      "loss": 0.8246,
+      "step": 3628
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6388802416867245e-05,
+      "loss": 0.8656,
+      "step": 3629
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.638673526872794e-05,
+      "loss": 0.7272,
+      "step": 3630
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6384667659561217e-05,
+      "loss": 0.9005,
+      "step": 3631
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6382599589516322e-05,
+      "loss": 0.8422,
+      "step": 3632
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.638053105874254e-05,
+      "loss": 0.7937,
+      "step": 3633
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6378462067389188e-05,
+      "loss": 0.8231,
+      "step": 3634
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6376392615605615e-05,
+      "loss": 0.7579,
+      "step": 3635
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6374322703541202e-05,
+      "loss": 0.8509,
+      "step": 3636
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6372252331345375e-05,
+      "loss": 0.7875,
+      "step": 3637
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6370181499167574e-05,
+      "loss": 0.8408,
+      "step": 3638
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.636811020715729e-05,
+      "loss": 0.8007,
+      "step": 3639
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.636603845546404e-05,
+      "loss": 0.846,
+      "step": 3640
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6363966244237364e-05,
+      "loss": 0.8589,
+      "step": 3641
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6361893573626854e-05,
+      "loss": 0.8372,
+      "step": 3642
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.635982044378212e-05,
+      "loss": 0.7993,
+      "step": 3643
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6357746854852817e-05,
+      "loss": 0.8106,
+      "step": 3644
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6355672806988624e-05,
+      "loss": 0.8019,
+      "step": 3645
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6353598300339255e-05,
+      "loss": 0.7386,
+      "step": 3646
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.635152333505446e-05,
+      "loss": 0.8865,
+      "step": 3647
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.634944791128403e-05,
+      "loss": 0.9401,
+      "step": 3648
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6347372029177763e-05,
+      "loss": 0.8294,
+      "step": 3649
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6345295688885517e-05,
+      "loss": 0.7789,
+      "step": 3650
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6343218890557172e-05,
+      "loss": 0.8321,
+      "step": 3651
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.634114163434264e-05,
+      "loss": 0.7937,
+      "step": 3652
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6339063920391865e-05,
+      "loss": 0.8552,
+      "step": 3653
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6336985748854838e-05,
+      "loss": 0.8346,
+      "step": 3654
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6334907119881563e-05,
+      "loss": 0.8701,
+      "step": 3655
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6332828033622082e-05,
+      "loss": 0.814,
+      "step": 3656
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6330748490226485e-05,
+      "loss": 0.7745,
+      "step": 3657
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.632866848984488e-05,
+      "loss": 0.9097,
+      "step": 3658
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6326588032627412e-05,
+      "loss": 0.829,
+      "step": 3659
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.632450711872426e-05,
+      "loss": 0.7938,
+      "step": 3660
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6322425748285633e-05,
+      "loss": 0.8706,
+      "step": 3661
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6320343921461774e-05,
+      "loss": 0.7302,
+      "step": 3662
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6318261638402965e-05,
+      "loss": 0.8384,
+      "step": 3663
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6316178899259506e-05,
+      "loss": 0.8108,
+      "step": 3664
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6314095704181752e-05,
+      "loss": 0.757,
+      "step": 3665
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6312012053320075e-05,
+      "loss": 0.8831,
+      "step": 3666
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.630992794682488e-05,
+      "loss": 0.8102,
+      "step": 3667
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.630784338484661e-05,
+      "loss": 0.8426,
+      "step": 3668
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6305758367535735e-05,
+      "loss": 0.8895,
+      "step": 3669
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6303672895042773e-05,
+      "loss": 0.8175,
+      "step": 3670
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.630158696751826e-05,
+      "loss": 0.8125,
+      "step": 3671
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.629950058511276e-05,
+      "loss": 0.7932,
+      "step": 3672
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6297413747976892e-05,
+      "loss": 0.8255,
+      "step": 3673
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6295326456261288e-05,
+      "loss": 0.806,
+      "step": 3674
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.629323871011662e-05,
+      "loss": 0.8562,
+      "step": 3675
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6291150509693593e-05,
+      "loss": 0.8175,
+      "step": 3676
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6289061855142942e-05,
+      "loss": 0.82,
+      "step": 3677
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6286972746615438e-05,
+      "loss": 0.8032,
+      "step": 3678
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6284883184261885e-05,
+      "loss": 0.8373,
+      "step": 3679
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6282793168233117e-05,
+      "loss": 0.8181,
+      "step": 3680
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6280702698680002e-05,
+      "loss": 0.8491,
+      "step": 3681
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6278611775753445e-05,
+      "loss": 0.6983,
+      "step": 3682
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.627652039960437e-05,
+      "loss": 0.8147,
+      "step": 3683
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6274428570383755e-05,
+      "loss": 0.8239,
+      "step": 3684
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.627233628824259e-05,
+      "loss": 0.8522,
+      "step": 3685
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.627024355333191e-05,
+      "loss": 0.8108,
+      "step": 3686
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.626815036580278e-05,
+      "loss": 0.8384,
+      "step": 3687
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6266056725806296e-05,
+      "loss": 0.3643,
+      "step": 3688
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6263962633493588e-05,
+      "loss": 0.8535,
+      "step": 3689
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.626186808901582e-05,
+      "loss": 0.8177,
+      "step": 3690
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6259773092524183e-05,
+      "loss": 0.8576,
+      "step": 3691
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.625767764416991e-05,
+      "loss": 0.8243,
+      "step": 3692
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6255581744104253e-05,
+      "loss": 0.7976,
+      "step": 3693
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.625348539247851e-05,
+      "loss": 0.8598,
+      "step": 3694
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6251388589444013e-05,
+      "loss": 0.8454,
+      "step": 3695
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.624929133515211e-05,
+      "loss": 0.828,
+      "step": 3696
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6247193629754197e-05,
+      "loss": 0.8151,
+      "step": 3697
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6245095473401694e-05,
+      "loss": 0.8114,
+      "step": 3698
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6242996866246056e-05,
+      "loss": 0.8815,
+      "step": 3699
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6240897808438776e-05,
+      "loss": 0.8199,
+      "step": 3700
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6238798300131372e-05,
+      "loss": 0.8498,
+      "step": 3701
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6236698341475398e-05,
+      "loss": 0.8465,
+      "step": 3702
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6234597932622435e-05,
+      "loss": 0.858,
+      "step": 3703
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.623249707372411e-05,
+      "loss": 0.7701,
+      "step": 3704
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.623039576493207e-05,
+      "loss": 0.8261,
+      "step": 3705
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6228294006397994e-05,
+      "loss": 0.7714,
+      "step": 3706
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6226191798273603e-05,
+      "loss": 0.8282,
+      "step": 3707
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6224089140710647e-05,
+      "loss": 0.8191,
+      "step": 3708
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6221986033860898e-05,
+      "loss": 0.8654,
+      "step": 3709
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.621988247787618e-05,
+      "loss": 0.8379,
+      "step": 3710
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.621777847290833e-05,
+      "loss": 0.8936,
+      "step": 3711
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6215674019109226e-05,
+      "loss": 0.8317,
+      "step": 3712
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6213569116630784e-05,
+      "loss": 0.8271,
+      "step": 3713
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.621146376562494e-05,
+      "loss": 0.8356,
+      "step": 3714
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.620935796624368e-05,
+      "loss": 0.8739,
+      "step": 3715
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6207251718638998e-05,
+      "loss": 0.3976,
+      "step": 3716
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6205145022962946e-05,
+      "loss": 0.4047,
+      "step": 3717
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6203037879367585e-05,
+      "loss": 0.7582,
+      "step": 3718
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6200930288005027e-05,
+      "loss": 0.8326,
+      "step": 3719
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6198822249027407e-05,
+      "loss": 0.853,
+      "step": 3720
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6196713762586892e-05,
+      "loss": 0.8003,
+      "step": 3721
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6194604828835684e-05,
+      "loss": 0.7506,
+      "step": 3722
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6192495447926023e-05,
+      "loss": 0.8414,
+      "step": 3723
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6190385620010163e-05,
+      "loss": 0.8319,
+      "step": 3724
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.618827534524041e-05,
+      "loss": 0.8509,
+      "step": 3725
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.61861646237691e-05,
+      "loss": 0.8067,
+      "step": 3726
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6184053455748582e-05,
+      "loss": 0.7996,
+      "step": 3727
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.618194184133126e-05,
+      "loss": 0.7928,
+      "step": 3728
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.617982978066956e-05,
+      "loss": 0.9063,
+      "step": 3729
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.617771727391594e-05,
+      "loss": 0.8293,
+      "step": 3730
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6175604321222892e-05,
+      "loss": 0.8547,
+      "step": 3731
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6173490922742944e-05,
+      "loss": 0.8356,
+      "step": 3732
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6171377078628645e-05,
+      "loss": 0.8556,
+      "step": 3733
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6169262789032587e-05,
+      "loss": 0.8324,
+      "step": 3734
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.616714805410739e-05,
+      "loss": 0.847,
+      "step": 3735
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6165032874005705e-05,
+      "loss": 0.8798,
+      "step": 3736
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6162917248880215e-05,
+      "loss": 0.8082,
+      "step": 3737
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.616080117888364e-05,
+      "loss": 0.8394,
+      "step": 3738
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.615868466416873e-05,
+      "loss": 0.8073,
+      "step": 3739
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6156567704888266e-05,
+      "loss": 0.8968,
+      "step": 3740
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6154450301195056e-05,
+      "loss": 0.4494,
+      "step": 3741
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.615233245324195e-05,
+      "loss": 0.832,
+      "step": 3742
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.615021416118182e-05,
+      "loss": 0.7934,
+      "step": 3743
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6148095425167583e-05,
+      "loss": 0.7974,
+      "step": 3744
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6145976245352172e-05,
+      "loss": 0.8202,
+      "step": 3745
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6143856621888567e-05,
+      "loss": 0.8155,
+      "step": 3746
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.614173655492977e-05,
+      "loss": 0.8076,
+      "step": 3747
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6139616044628815e-05,
+      "loss": 0.7781,
+      "step": 3748
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6137495091138782e-05,
+      "loss": 0.7823,
+      "step": 3749
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.613537369461276e-05,
+      "loss": 0.8496,
+      "step": 3750
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.613325185520389e-05,
+      "loss": 0.8248,
+      "step": 3751
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6131129573065333e-05,
+      "loss": 0.7883,
+      "step": 3752
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.612900684835029e-05,
+      "loss": 0.8525,
+      "step": 3753
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.612688368121199e-05,
+      "loss": 0.8667,
+      "step": 3754
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.612476007180369e-05,
+      "loss": 0.8499,
+      "step": 3755
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.612263602027869e-05,
+      "loss": 0.8403,
+      "step": 3756
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6120511526790304e-05,
+      "loss": 0.8895,
+      "step": 3757
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6118386591491902e-05,
+      "loss": 0.8172,
+      "step": 3758
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.611626121453686e-05,
+      "loss": 0.834,
+      "step": 3759
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.611413539607861e-05,
+      "loss": 0.7673,
+      "step": 3760
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.61120091362706e-05,
+      "loss": 0.7664,
+      "step": 3761
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6109882435266312e-05,
+      "loss": 0.8692,
+      "step": 3762
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6107755293219264e-05,
+      "loss": 0.8557,
+      "step": 3763
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6105627710283003e-05,
+      "loss": 0.8504,
+      "step": 3764
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.610349968661111e-05,
+      "loss": 0.8008,
+      "step": 3765
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.61013712223572e-05,
+      "loss": 0.8659,
+      "step": 3766
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.609924231767491e-05,
+      "loss": 0.8526,
+      "step": 3767
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.609711297271792e-05,
+      "loss": 0.7996,
+      "step": 3768
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6094983187639934e-05,
+      "loss": 0.8639,
+      "step": 3769
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6092852962594694e-05,
+      "loss": 0.8345,
+      "step": 3770
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.609072229773597e-05,
+      "loss": 0.8495,
+      "step": 3771
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.608859119321756e-05,
+      "loss": 0.8681,
+      "step": 3772
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6086459649193305e-05,
+      "loss": 0.892,
+      "step": 3773
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6084327665817063e-05,
+      "loss": 0.8163,
+      "step": 3774
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6082195243242742e-05,
+      "loss": 0.8275,
+      "step": 3775
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6080062381624262e-05,
+      "loss": 0.8498,
+      "step": 3776
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6077929081115585e-05,
+      "loss": 0.8438,
+      "step": 3777
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.607579534187071e-05,
+      "loss": 0.7566,
+      "step": 3778
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6073661164043653e-05,
+      "loss": 0.8333,
+      "step": 3779
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6071526547788473e-05,
+      "loss": 0.8421,
+      "step": 3780
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6069391493259267e-05,
+      "loss": 0.8285,
+      "step": 3781
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.606725600061014e-05,
+      "loss": 0.794,
+      "step": 3782
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6065120069995246e-05,
+      "loss": 0.8253,
+      "step": 3783
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.606298370156877e-05,
+      "loss": 0.824,
+      "step": 3784
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.606084689548493e-05,
+      "loss": 0.9444,
+      "step": 3785
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6058709651897966e-05,
+      "loss": 0.8705,
+      "step": 3786
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6056571970962157e-05,
+      "loss": 0.8468,
+      "step": 3787
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6054433852831813e-05,
+      "loss": 0.8876,
+      "step": 3788
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.605229529766127e-05,
+      "loss": 0.7879,
+      "step": 3789
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6050156305604904e-05,
+      "loss": 0.8242,
+      "step": 3790
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.604801687681712e-05,
+      "loss": 0.8254,
+      "step": 3791
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.604587701145235e-05,
+      "loss": 0.8283,
+      "step": 3792
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.604373670966506e-05,
+      "loss": 0.8047,
+      "step": 3793
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6041595971609747e-05,
+      "loss": 0.8292,
+      "step": 3794
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6039454797440946e-05,
+      "loss": 0.7891,
+      "step": 3795
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6037313187313214e-05,
+      "loss": 0.7793,
+      "step": 3796
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.603517114138114e-05,
+      "loss": 0.7622,
+      "step": 3797
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6033028659799357e-05,
+      "loss": 0.8523,
+      "step": 3798
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6030885742722512e-05,
+      "loss": 0.817,
+      "step": 3799
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6028742390305297e-05,
+      "loss": 0.7352,
+      "step": 3800
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6026598602702426e-05,
+      "loss": 0.8826,
+      "step": 3801
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6024454380068653e-05,
+      "loss": 0.8913,
+      "step": 3802
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6022309722558754e-05,
+      "loss": 0.8831,
+      "step": 3803
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6020164630327544e-05,
+      "loss": 0.8006,
+      "step": 3804
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.601801910352987e-05,
+      "loss": 0.7987,
+      "step": 3805
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6015873142320603e-05,
+      "loss": 0.8655,
+      "step": 3806
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6013726746854652e-05,
+      "loss": 0.77,
+      "step": 3807
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.601157991728695e-05,
+      "loss": 0.817,
+      "step": 3808
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6009432653772474e-05,
+      "loss": 0.4165,
+      "step": 3809
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6007284956466217e-05,
+      "loss": 0.8224,
+      "step": 3810
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6005136825523215e-05,
+      "loss": 0.8704,
+      "step": 3811
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6002988261098532e-05,
+      "loss": 0.8036,
+      "step": 3812
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6000839263347257e-05,
+      "loss": 0.81,
+      "step": 3813
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5998689832424522e-05,
+      "loss": 0.8475,
+      "step": 3814
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5996539968485483e-05,
+      "loss": 0.8799,
+      "step": 3815
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5994389671685323e-05,
+      "loss": 0.829,
+      "step": 3816
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.599223894217927e-05,
+      "loss": 0.8102,
+      "step": 3817
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5990087780122565e-05,
+      "loss": 0.8103,
+      "step": 3818
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5987936185670498e-05,
+      "loss": 0.8547,
+      "step": 3819
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.598578415897838e-05,
+      "loss": 0.8276,
+      "step": 3820
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.598363170020155e-05,
+      "loss": 0.887,
+      "step": 3821
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.598147880949539e-05,
+      "loss": 0.7792,
+      "step": 3822
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5979325487015304e-05,
+      "loss": 0.8289,
+      "step": 3823
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5977171732916732e-05,
+      "loss": 0.8408,
+      "step": 3824
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5975017547355142e-05,
+      "loss": 0.7888,
+      "step": 3825
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5972862930486032e-05,
+      "loss": 0.7994,
+      "step": 3826
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5970707882464934e-05,
+      "loss": 0.8232,
+      "step": 3827
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5968552403447417e-05,
+      "loss": 0.8259,
+      "step": 3828
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5966396493589066e-05,
+      "loss": 0.7055,
+      "step": 3829
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.596424015304551e-05,
+      "loss": 0.8078,
+      "step": 3830
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.59620833819724e-05,
+      "loss": 0.8804,
+      "step": 3831
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.595992618052543e-05,
+      "loss": 0.8753,
+      "step": 3832
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.595776854886031e-05,
+      "loss": 0.784,
+      "step": 3833
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.59556104871328e-05,
+      "loss": 0.7964,
+      "step": 3834
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5953451995498668e-05,
+      "loss": 0.8072,
+      "step": 3835
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.595129307411373e-05,
+      "loss": 0.807,
+      "step": 3836
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5949133723133833e-05,
+      "loss": 0.8056,
+      "step": 3837
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.594697394271484e-05,
+      "loss": 0.7855,
+      "step": 3838
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5944813733012666e-05,
+      "loss": 0.7864,
+      "step": 3839
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5942653094183237e-05,
+      "loss": 0.8616,
+      "step": 3840
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5940492026382524e-05,
+      "loss": 0.8611,
+      "step": 3841
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5938330529766523e-05,
+      "loss": 0.8182,
+      "step": 3842
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5936168604491257e-05,
+      "loss": 0.8671,
+      "step": 3843
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5934006250712794e-05,
+      "loss": 0.8492,
+      "step": 3844
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.593184346858722e-05,
+      "loss": 0.7402,
+      "step": 3845
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5929680258270655e-05,
+      "loss": 0.8074,
+      "step": 3846
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5927516619919247e-05,
+      "loss": 0.8143,
+      "step": 3847
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5925352553689185e-05,
+      "loss": 0.8904,
+      "step": 3848
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5923188059736678e-05,
+      "loss": 0.4032,
+      "step": 3849
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5921023138217974e-05,
+      "loss": 0.8015,
+      "step": 3850
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5918857789289346e-05,
+      "loss": 0.8112,
+      "step": 3851
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5916692013107097e-05,
+      "loss": 0.8385,
+      "step": 3852
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.591452580982757e-05,
+      "loss": 0.8277,
+      "step": 3853
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.591235917960713e-05,
+      "loss": 0.8794,
+      "step": 3854
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.591019212260217e-05,
+      "loss": 0.8209,
+      "step": 3855
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5908024638969136e-05,
+      "loss": 0.8515,
+      "step": 3856
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5905856728864468e-05,
+      "loss": 0.8775,
+      "step": 3857
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5903688392444668e-05,
+      "loss": 0.9191,
+      "step": 3858
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5901519629866255e-05,
+      "loss": 0.8402,
+      "step": 3859
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.589935044128578e-05,
+      "loss": 0.7709,
+      "step": 3860
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5897180826859827e-05,
+      "loss": 0.84,
+      "step": 3861
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5895010786745015e-05,
+      "loss": 0.8519,
+      "step": 3862
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5892840321097984e-05,
+      "loss": 0.743,
+      "step": 3863
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.589066943007541e-05,
+      "loss": 0.8747,
+      "step": 3864
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5888498113833998e-05,
+      "loss": 0.8479,
+      "step": 3865
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5886326372530483e-05,
+      "loss": 0.7534,
+      "step": 3866
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.588415420632164e-05,
+      "loss": 0.8317,
+      "step": 3867
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.588198161536426e-05,
+      "loss": 0.4143,
+      "step": 3868
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5879808599815176e-05,
+      "loss": 0.8117,
+      "step": 3869
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5877635159831242e-05,
+      "loss": 0.737,
+      "step": 3870
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5875461295569356e-05,
+      "loss": 0.8655,
+      "step": 3871
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.587328700718643e-05,
+      "loss": 0.8375,
+      "step": 3872
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5871112294839424e-05,
+      "loss": 0.8988,
+      "step": 3873
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5868937158685315e-05,
+      "loss": 0.8264,
+      "step": 3874
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5866761598881115e-05,
+      "loss": 0.8199,
+      "step": 3875
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5864585615583864e-05,
+      "loss": 0.7918,
+      "step": 3876
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5862409208950648e-05,
+      "loss": 0.7629,
+      "step": 3877
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.586023237913856e-05,
+      "loss": 0.822,
+      "step": 3878
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.585805512630474e-05,
+      "loss": 0.8161,
+      "step": 3879
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5855877450606348e-05,
+      "loss": 0.8498,
+      "step": 3880
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5853699352200583e-05,
+      "loss": 0.8206,
+      "step": 3881
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5851520831244674e-05,
+      "loss": 0.8216,
+      "step": 3882
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5849341887895873e-05,
+      "loss": 0.8523,
+      "step": 3883
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.584716252231147e-05,
+      "loss": 0.8702,
+      "step": 3884
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5844982734648784e-05,
+      "loss": 0.8677,
+      "step": 3885
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5842802525065162e-05,
+      "loss": 0.78,
+      "step": 3886
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.584062189371798e-05,
+      "loss": 0.848,
+      "step": 3887
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5838440840764656e-05,
+      "loss": 0.8546,
+      "step": 3888
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5836259366362618e-05,
+      "loss": 0.7956,
+      "step": 3889
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5834077470669343e-05,
+      "loss": 0.8005,
+      "step": 3890
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5831895153842328e-05,
+      "loss": 0.8517,
+      "step": 3891
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5829712416039105e-05,
+      "loss": 0.8905,
+      "step": 3892
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.582752925741724e-05,
+      "loss": 0.8439,
+      "step": 3893
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.582534567813432e-05,
+      "loss": 0.7767,
+      "step": 3894
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5823161678347967e-05,
+      "loss": 0.7477,
+      "step": 3895
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5820977258215833e-05,
+      "loss": 0.7799,
+      "step": 3896
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5818792417895603e-05,
+      "loss": 0.8362,
+      "step": 3897
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5816607157544987e-05,
+      "loss": 0.8496,
+      "step": 3898
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5814421477321734e-05,
+      "loss": 0.7868,
+      "step": 3899
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5812235377383607e-05,
+      "loss": 0.8617,
+      "step": 3900
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.581004885788842e-05,
+      "loss": 0.8133,
+      "step": 3901
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5807861918994005e-05,
+      "loss": 0.8284,
+      "step": 3902
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5805674560858227e-05,
+      "loss": 0.8007,
+      "step": 3903
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5803486783638976e-05,
+      "loss": 0.8417,
+      "step": 3904
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.580129858749418e-05,
+      "loss": 0.7968,
+      "step": 3905
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5799109972581798e-05,
+      "loss": 0.7851,
+      "step": 3906
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5796920939059807e-05,
+      "loss": 0.7775,
+      "step": 3907
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.579473148708623e-05,
+      "loss": 0.8511,
+      "step": 3908
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.579254161681911e-05,
+      "loss": 0.794,
+      "step": 3909
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5790351328416523e-05,
+      "loss": 0.8399,
+      "step": 3910
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5788160622036578e-05,
+      "loss": 0.8247,
+      "step": 3911
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5785969497837407e-05,
+      "loss": 0.8513,
+      "step": 3912
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5783777955977183e-05,
+      "loss": 0.9572,
+      "step": 3913
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5781585996614097e-05,
+      "loss": 0.7674,
+      "step": 3914
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5779393619906373e-05,
+      "loss": 0.8063,
+      "step": 3915
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.577720082601228e-05,
+      "loss": 0.7845,
+      "step": 3916
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.577500761509009e-05,
+      "loss": 0.816,
+      "step": 3917
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5772813987298135e-05,
+      "loss": 0.8188,
+      "step": 3918
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5770619942794755e-05,
+      "loss": 0.7784,
+      "step": 3919
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.576842548173832e-05,
+      "loss": 0.8075,
+      "step": 3920
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5766230604287254e-05,
+      "loss": 0.8003,
+      "step": 3921
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5764035310599985e-05,
+      "loss": 0.9466,
+      "step": 3922
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.576183960083498e-05,
+      "loss": 0.8193,
+      "step": 3923
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5759643475150736e-05,
+      "loss": 0.7931,
+      "step": 3924
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5757446933705785e-05,
+      "loss": 0.813,
+      "step": 3925
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.575524997665868e-05,
+      "loss": 0.8054,
+      "step": 3926
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5753052604168017e-05,
+      "loss": 0.8338,
+      "step": 3927
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5750854816392404e-05,
+      "loss": 0.8745,
+      "step": 3928
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.574865661349049e-05,
+      "loss": 0.8648,
+      "step": 3929
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.574645799562096e-05,
+      "loss": 0.8078,
+      "step": 3930
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.574425896294251e-05,
+      "loss": 0.8714,
+      "step": 3931
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.574205951561389e-05,
+      "loss": 0.7802,
+      "step": 3932
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.573985965379386e-05,
+      "loss": 0.8368,
+      "step": 3933
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5737659377641218e-05,
+      "loss": 0.8761,
+      "step": 3934
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5735458687314796e-05,
+      "loss": 0.8806,
+      "step": 3935
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.573325758297344e-05,
+      "loss": 0.7882,
+      "step": 3936
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.573105606477605e-05,
+      "loss": 0.7826,
+      "step": 3937
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5728854132881535e-05,
+      "loss": 0.7689,
+      "step": 3938
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5726651787448844e-05,
+      "loss": 0.8392,
+      "step": 3939
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5724449028636953e-05,
+      "loss": 0.8489,
+      "step": 3940
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5722245856604873e-05,
+      "loss": 0.8347,
+      "step": 3941
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.572004227151163e-05,
+      "loss": 0.8362,
+      "step": 3942
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.57178382735163e-05,
+      "loss": 0.7888,
+      "step": 3943
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5715633862777978e-05,
+      "loss": 0.8434,
+      "step": 3944
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5713429039455784e-05,
+      "loss": 0.8536,
+      "step": 3945
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5711223803708878e-05,
+      "loss": 0.8187,
+      "step": 3946
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.570901815569644e-05,
+      "loss": 0.7575,
+      "step": 3947
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5706812095577696e-05,
+      "loss": 0.8487,
+      "step": 3948
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5704605623511878e-05,
+      "loss": 0.8649,
+      "step": 3949
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.570239873965827e-05,
+      "loss": 0.7925,
+      "step": 3950
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.570019144417617e-05,
+      "loss": 0.8214,
+      "step": 3951
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5697983737224912e-05,
+      "loss": 0.416,
+      "step": 3952
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5695775618963866e-05,
+      "loss": 0.8589,
+      "step": 3953
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.569356708955242e-05,
+      "loss": 0.77,
+      "step": 3954
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.569135814915e-05,
+      "loss": 0.8501,
+      "step": 3955
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5689148797916054e-05,
+      "loss": 0.8127,
+      "step": 3956
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5686939036010068e-05,
+      "loss": 0.7772,
+      "step": 3957
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5684728863591556e-05,
+      "loss": 0.8337,
+      "step": 3958
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5682518280820055e-05,
+      "loss": 0.8858,
+      "step": 3959
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.568030728785514e-05,
+      "loss": 0.7867,
+      "step": 3960
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.567809588485641e-05,
+      "loss": 0.8223,
+      "step": 3961
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5675884071983492e-05,
+      "loss": 0.8332,
+      "step": 3962
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5673671849396053e-05,
+      "loss": 0.802,
+      "step": 3963
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5671459217253778e-05,
+      "loss": 0.772,
+      "step": 3964
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5669246175716387e-05,
+      "loss": 0.8395,
+      "step": 3965
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5667032724943633e-05,
+      "loss": 0.7782,
+      "step": 3966
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5664818865095287e-05,
+      "loss": 0.8247,
+      "step": 3967
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.566260459633116e-05,
+      "loss": 0.8243,
+      "step": 3968
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.566038991881109e-05,
+      "loss": 0.7582,
+      "step": 3969
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.565817483269494e-05,
+      "loss": 0.8378,
+      "step": 3970
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5655959338142615e-05,
+      "loss": 0.8128,
+      "step": 3971
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5653743435314033e-05,
+      "loss": 0.9001,
+      "step": 3972
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.565152712436915e-05,
+      "loss": 0.8172,
+      "step": 3973
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5649310405467957e-05,
+      "loss": 0.8248,
+      "step": 3974
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.564709327877046e-05,
+      "loss": 0.7694,
+      "step": 3975
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.564487574443671e-05,
+      "loss": 0.8594,
+      "step": 3976
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.564265780262677e-05,
+      "loss": 0.8061,
+      "step": 3977
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5640439453500756e-05,
+      "loss": 0.8042,
+      "step": 3978
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.563822069721879e-05,
+      "loss": 0.7447,
+      "step": 3979
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5636001533941035e-05,
+      "loss": 0.8302,
+      "step": 3980
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.563378196382768e-05,
+      "loss": 0.795,
+      "step": 3981
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.563156198703895e-05,
+      "loss": 0.8269,
+      "step": 3982
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.562934160373509e-05,
+      "loss": 0.7916,
+      "step": 3983
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5627120814076386e-05,
+      "loss": 0.8322,
+      "step": 3984
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5624899618223135e-05,
+      "loss": 0.7876,
+      "step": 3985
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5622678016335683e-05,
+      "loss": 0.8717,
+      "step": 3986
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5620456008574393e-05,
+      "loss": 0.8674,
+      "step": 3987
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.561823359509966e-05,
+      "loss": 0.8187,
+      "step": 3988
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.561601077607191e-05,
+      "loss": 0.8939,
+      "step": 3989
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.56137875516516e-05,
+      "loss": 0.8858,
+      "step": 3990
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5611563921999212e-05,
+      "loss": 0.8037,
+      "step": 3991
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5609339887275262e-05,
+      "loss": 0.7773,
+      "step": 3992
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5607115447640283e-05,
+      "loss": 0.8526,
+      "step": 3993
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5604890603254855e-05,
+      "loss": 0.822,
+      "step": 3994
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.560266535427958e-05,
+      "loss": 0.7693,
+      "step": 3995
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5600439700875082e-05,
+      "loss": 0.8215,
+      "step": 3996
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5598213643202022e-05,
+      "loss": 0.815,
+      "step": 3997
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.559598718142109e-05,
+      "loss": 0.9057,
+      "step": 3998
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5593760315693e-05,
+      "loss": 0.8809,
+      "step": 3999
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5591533046178503e-05,
+      "loss": 0.8596,
+      "step": 4000
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5589305373038372e-05,
+      "loss": 0.8298,
+      "step": 4001
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5587077296433416e-05,
+      "loss": 0.8551,
+      "step": 4002
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5584848816524462e-05,
+      "loss": 0.848,
+      "step": 4003
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.558261993347238e-05,
+      "loss": 0.8546,
+      "step": 4004
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5580390647438056e-05,
+      "loss": 0.8215,
+      "step": 4005
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.557816095858242e-05,
+      "loss": 0.8893,
+      "step": 4006
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5575930867066408e-05,
+      "loss": 0.872,
+      "step": 4007
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5573700373051016e-05,
+      "loss": 0.7813,
+      "step": 4008
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.557146947669724e-05,
+      "loss": 0.8483,
+      "step": 4009
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5569238178166127e-05,
+      "loss": 0.8665,
+      "step": 4010
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5567006477618733e-05,
+      "loss": 0.8908,
+      "step": 4011
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5564774375216164e-05,
+      "loss": 0.8064,
+      "step": 4012
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.556254187111954e-05,
+      "loss": 0.7566,
+      "step": 4013
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5560308965490013e-05,
+      "loss": 0.8258,
+      "step": 4014
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.555807565848877e-05,
+      "loss": 0.8277,
+      "step": 4015
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.555584195027702e-05,
+      "loss": 0.8201,
+      "step": 4016
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5553607841016003e-05,
+      "loss": 0.8481,
+      "step": 4017
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.555137333086699e-05,
+      "loss": 0.8244,
+      "step": 4018
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.554913841999128e-05,
+      "loss": 0.8353,
+      "step": 4019
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5546903108550196e-05,
+      "loss": 0.8331,
+      "step": 4020
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.55446673967051e-05,
+      "loss": 0.8986,
+      "step": 4021
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5542431284617372e-05,
+      "loss": 0.7525,
+      "step": 4022
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.554019477244843e-05,
+      "loss": 0.8693,
+      "step": 4023
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5537957860359715e-05,
+      "loss": 0.8763,
+      "step": 4024
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.55357205485127e-05,
+      "loss": 0.8793,
+      "step": 4025
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5533482837068886e-05,
+      "loss": 0.8494,
+      "step": 4026
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.55312447261898e-05,
+      "loss": 0.8808,
+      "step": 4027
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5529006216037003e-05,
+      "loss": 0.8187,
+      "step": 4028
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5526767306772082e-05,
+      "loss": 0.7925,
+      "step": 4029
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5524527998556653e-05,
+      "loss": 0.8229,
+      "step": 4030
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5522288291552355e-05,
+      "loss": 0.9135,
+      "step": 4031
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.552004818592087e-05,
+      "loss": 0.8186,
+      "step": 4032
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.55178076818239e-05,
+      "loss": 0.8083,
+      "step": 4033
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.551556677942317e-05,
+      "loss": 0.7316,
+      "step": 4034
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.551332547888044e-05,
+      "loss": 0.7571,
+      "step": 4035
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5511083780357507e-05,
+      "loss": 0.8251,
+      "step": 4036
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5508841684016176e-05,
+      "loss": 0.7976,
+      "step": 4037
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5506599190018305e-05,
+      "loss": 0.8455,
+      "step": 4038
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.550435629852576e-05,
+      "loss": 0.8436,
+      "step": 4039
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5502113009700452e-05,
+      "loss": 0.8582,
+      "step": 4040
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.549986932370431e-05,
+      "loss": 0.8495,
+      "step": 4041
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5497625240699293e-05,
+      "loss": 0.7597,
+      "step": 4042
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5495380760847388e-05,
+      "loss": 0.8401,
+      "step": 4043
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5493135884310618e-05,
+      "loss": 0.874,
+      "step": 4044
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.549089061125103e-05,
+      "loss": 0.7526,
+      "step": 4045
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5488644941830696e-05,
+      "loss": 0.7838,
+      "step": 4046
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5486398876211724e-05,
+      "loss": 0.8731,
+      "step": 4047
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5484152414556243e-05,
+      "loss": 0.825,
+      "step": 4048
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5481905557026415e-05,
+      "loss": 0.8201,
+      "step": 4049
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.547965830378443e-05,
+      "loss": 0.8129,
+      "step": 4050
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.547741065499251e-05,
+      "loss": 0.8684,
+      "step": 4051
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5475162610812893e-05,
+      "loss": 0.8076,
+      "step": 4052
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5472914171407865e-05,
+      "loss": 0.8175,
+      "step": 4053
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.547066533693972e-05,
+      "loss": 0.8663,
+      "step": 4054
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5468416107570795e-05,
+      "loss": 0.8955,
+      "step": 4055
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5466166483463453e-05,
+      "loss": 0.8203,
+      "step": 4056
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5463916464780083e-05,
+      "loss": 0.8324,
+      "step": 4057
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5461666051683103e-05,
+      "loss": 0.8362,
+      "step": 4058
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5459415244334952e-05,
+      "loss": 0.8823,
+      "step": 4059
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5457164042898116e-05,
+      "loss": 0.8047,
+      "step": 4060
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.545491244753509e-05,
+      "loss": 0.8394,
+      "step": 4061
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.545266045840841e-05,
+      "loss": 0.7955,
+      "step": 4062
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.545040807568063e-05,
+      "loss": 0.8807,
+      "step": 4063
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.544815529951435e-05,
+      "loss": 0.841,
+      "step": 4064
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5445902130072178e-05,
+      "loss": 0.774,
+      "step": 4065
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.544364856751676e-05,
+      "loss": 0.9152,
+      "step": 4066
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5441394612010772e-05,
+      "loss": 0.8214,
+      "step": 4067
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.543914026371692e-05,
+      "loss": 0.9429,
+      "step": 4068
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5436885522797927e-05,
+      "loss": 0.7702,
+      "step": 4069
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5434630389416555e-05,
+      "loss": 0.8682,
+      "step": 4070
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5432374863735594e-05,
+      "loss": 0.8299,
+      "step": 4071
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5430118945917853e-05,
+      "loss": 0.886,
+      "step": 4072
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.542786263612618e-05,
+      "loss": 0.772,
+      "step": 4073
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5425605934523445e-05,
+      "loss": 0.8368,
+      "step": 4074
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.542334884127255e-05,
+      "loss": 0.7801,
+      "step": 4075
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5421091356536428e-05,
+      "loss": 0.8393,
+      "step": 4076
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5418833480478026e-05,
+      "loss": 0.8111,
+      "step": 4077
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5416575213260337e-05,
+      "loss": 0.8826,
+      "step": 4078
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5414316555046368e-05,
+      "loss": 0.8524,
+      "step": 4079
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5412057505999167e-05,
+      "loss": 0.7983,
+      "step": 4080
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.54097980662818e-05,
+      "loss": 0.7737,
+      "step": 4081
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5407538236057368e-05,
+      "loss": 0.869,
+      "step": 4082
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5405278015488993e-05,
+      "loss": 0.8554,
+      "step": 4083
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.540301740473983e-05,
+      "loss": 0.8698,
+      "step": 4084
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.540075640397307e-05,
+      "loss": 0.8412,
+      "step": 4085
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.539849501335191e-05,
+      "loss": 0.8083,
+      "step": 4086
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5396233233039598e-05,
+      "loss": 0.7706,
+      "step": 4087
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.53939710631994e-05,
+      "loss": 0.8465,
+      "step": 4088
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5391708503994608e-05,
+      "loss": 0.8218,
+      "step": 4089
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5389445555588546e-05,
+      "loss": 0.7927,
+      "step": 4090
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5387182218144565e-05,
+      "loss": 0.8121,
+      "step": 4091
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5384918491826047e-05,
+      "loss": 0.8999,
+      "step": 4092
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5382654376796398e-05,
+      "loss": 0.74,
+      "step": 4093
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.538038987321905e-05,
+      "loss": 0.8606,
+      "step": 4094
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.537812498125747e-05,
+      "loss": 0.8808,
+      "step": 4095
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5375859701075155e-05,
+      "loss": 0.81,
+      "step": 4096
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5373594032835613e-05,
+      "loss": 0.8124,
+      "step": 4097
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.53713279767024e-05,
+      "loss": 0.8642,
+      "step": 4098
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5369061532839088e-05,
+      "loss": 0.8186,
+      "step": 4099
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5366794701409285e-05,
+      "loss": 0.8069,
+      "step": 4100
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5364527482576617e-05,
+      "loss": 0.8001,
+      "step": 4101
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5362259876504745e-05,
+      "loss": 0.8009,
+      "step": 4102
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5359991883357356e-05,
+      "loss": 0.804,
+      "step": 4103
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5357723503298165e-05,
+      "loss": 0.8192,
+      "step": 4104
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.535545473649092e-05,
+      "loss": 0.8651,
+      "step": 4105
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.535318558309939e-05,
+      "loss": 0.8067,
+      "step": 4106
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5350916043287374e-05,
+      "loss": 0.7748,
+      "step": 4107
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5348646117218694e-05,
+      "loss": 0.8304,
+      "step": 4108
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5346375805057214e-05,
+      "loss": 0.8722,
+      "step": 4109
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5344105106966812e-05,
+      "loss": 0.8015,
+      "step": 4110
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5341834023111398e-05,
+      "loss": 0.8805,
+      "step": 4111
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.533956255365491e-05,
+      "loss": 0.8773,
+      "step": 4112
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5337290698761316e-05,
+      "loss": 0.8801,
+      "step": 4113
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.533501845859461e-05,
+      "loss": 0.8722,
+      "step": 4114
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5332745833318813e-05,
+      "loss": 0.8286,
+      "step": 4115
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5330472823097977e-05,
+      "loss": 0.8317,
+      "step": 4116
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.532819942809618e-05,
+      "loss": 0.8283,
+      "step": 4117
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.532592564847752e-05,
+      "loss": 0.8861,
+      "step": 4118
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.532365148440614e-05,
+      "loss": 0.9534,
+      "step": 4119
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.53213769360462e-05,
+      "loss": 0.7781,
+      "step": 4120
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.531910200356188e-05,
+      "loss": 0.7482,
+      "step": 4121
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5316826687117403e-05,
+      "loss": 0.8626,
+      "step": 4122
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5314550986877013e-05,
+      "loss": 0.8542,
+      "step": 4123
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.531227490300498e-05,
+      "loss": 0.8581,
+      "step": 4124
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.53099984356656e-05,
+      "loss": 0.8352,
+      "step": 4125
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5307721585023207e-05,
+      "loss": 0.8105,
+      "step": 4126
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5305444351242152e-05,
+      "loss": 0.7996,
+      "step": 4127
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5303166734486822e-05,
+      "loss": 0.7838,
+      "step": 4128
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5300888734921616e-05,
+      "loss": 0.8085,
+      "step": 4129
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5298610352710984e-05,
+      "loss": 0.8299,
+      "step": 4130
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.529633158801939e-05,
+      "loss": 0.8336,
+      "step": 4131
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5294052441011314e-05,
+      "loss": 0.8767,
+      "step": 4132
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.529177291185129e-05,
+      "loss": 0.7885,
+      "step": 4133
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5289493000703865e-05,
+      "loss": 0.8032,
+      "step": 4134
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.528721270773361e-05,
+      "loss": 0.8145,
+      "step": 4135
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5284932033105128e-05,
+      "loss": 0.8297,
+      "step": 4136
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5282650976983053e-05,
+      "loss": 0.8716,
+      "step": 4137
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5280369539532044e-05,
+      "loss": 0.7533,
+      "step": 4138
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5278087720916785e-05,
+      "loss": 0.8254,
+      "step": 4139
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5275805521301986e-05,
+      "loss": 0.8529,
+      "step": 4140
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5273522940852396e-05,
+      "loss": 0.8074,
+      "step": 4141
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5271239979732778e-05,
+      "loss": 0.8481,
+      "step": 4142
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.526895663810793e-05,
+      "loss": 0.8426,
+      "step": 4143
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5266672916142668e-05,
+      "loss": 0.8201,
+      "step": 4144
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5264388814001854e-05,
+      "loss": 0.8049,
+      "step": 4145
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.526210433185036e-05,
+      "loss": 0.836,
+      "step": 4146
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5259819469853094e-05,
+      "loss": 0.8038,
+      "step": 4147
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5257534228174986e-05,
+      "loss": 0.8073,
+      "step": 4148
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5255248606980998e-05,
+      "loss": 0.823,
+      "step": 4149
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5252962606436119e-05,
+      "loss": 0.7793,
+      "step": 4150
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5250676226705365e-05,
+      "loss": 0.8722,
+      "step": 4151
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5248389467953776e-05,
+      "loss": 0.7909,
+      "step": 4152
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5246102330346424e-05,
+      "loss": 0.8244,
+      "step": 4153
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5243814814048408e-05,
+      "loss": 0.7996,
+      "step": 4154
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5241526919224848e-05,
+      "loss": 0.7468,
+      "step": 4155
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5239238646040898e-05,
+      "loss": 0.8562,
+      "step": 4156
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5236949994661736e-05,
+      "loss": 0.8401,
+      "step": 4157
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5234660965252572e-05,
+      "loss": 0.8598,
+      "step": 4158
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.523237155797864e-05,
+      "loss": 0.8685,
+      "step": 4159
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5230081773005196e-05,
+      "loss": 0.8031,
+      "step": 4160
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5227791610497537e-05,
+      "loss": 0.8692,
+      "step": 4161
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.522550107062097e-05,
+      "loss": 0.8293,
+      "step": 4162
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5223210153540842e-05,
+      "loss": 0.819,
+      "step": 4163
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5220918859422526e-05,
+      "loss": 0.768,
+      "step": 4164
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5218627188431411e-05,
+      "loss": 0.8709,
+      "step": 4165
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.521633514073293e-05,
+      "loss": 0.7975,
+      "step": 4166
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.521404271649253e-05,
+      "loss": 0.7625,
+      "step": 4167
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5211749915875693e-05,
+      "loss": 0.8548,
+      "step": 4168
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5209456739047926e-05,
+      "loss": 0.4338,
+      "step": 4169
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5207163186174756e-05,
+      "loss": 0.8559,
+      "step": 4170
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5204869257421748e-05,
+      "loss": 0.8943,
+      "step": 4171
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5202574952954491e-05,
+      "loss": 0.7922,
+      "step": 4172
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5200280272938596e-05,
+      "loss": 0.8446,
+      "step": 4173
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5197985217539707e-05,
+      "loss": 0.7722,
+      "step": 4174
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5195689786923496e-05,
+      "loss": 0.849,
+      "step": 4175
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.519339398125565e-05,
+      "loss": 0.8283,
+      "step": 4176
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5191097800701901e-05,
+      "loss": 0.8822,
+      "step": 4177
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5188801245427995e-05,
+      "loss": 0.4135,
+      "step": 4178
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5186504315599711e-05,
+      "loss": 0.7884,
+      "step": 4179
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5184207011382852e-05,
+      "loss": 0.8466,
+      "step": 4180
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5181909332943248e-05,
+      "loss": 0.3879,
+      "step": 4181
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.517961128044676e-05,
+      "loss": 0.8566,
+      "step": 4182
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.517731285405927e-05,
+      "loss": 0.801,
+      "step": 4183
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5175014053946693e-05,
+      "loss": 0.7988,
+      "step": 4184
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5172714880274968e-05,
+      "loss": 0.8442,
+      "step": 4185
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5170415333210058e-05,
+      "loss": 0.7826,
+      "step": 4186
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.516811541291796e-05,
+      "loss": 0.7946,
+      "step": 4187
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5165815119564693e-05,
+      "loss": 0.8231,
+      "step": 4188
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5163514453316304e-05,
+      "loss": 0.8116,
+      "step": 4189
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5161213414338867e-05,
+      "loss": 0.8253,
+      "step": 4190
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5158912002798483e-05,
+      "loss": 0.8043,
+      "step": 4191
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5156610218861278e-05,
+      "loss": 0.8761,
+      "step": 4192
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.515430806269341e-05,
+      "loss": 0.8795,
+      "step": 4193
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5152005534461056e-05,
+      "loss": 0.8715,
+      "step": 4194
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.514970263433043e-05,
+      "loss": 0.8317,
+      "step": 4195
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5147399362467759e-05,
+      "loss": 0.7965,
+      "step": 4196
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.514509571903931e-05,
+      "loss": 0.8208,
+      "step": 4197
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5142791704211375e-05,
+      "loss": 0.7901,
+      "step": 4198
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5140487318150265e-05,
+      "loss": 0.8612,
+      "step": 4199
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5138182561022326e-05,
+      "loss": 0.7794,
+      "step": 4200
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5135877432993923e-05,
+      "loss": 0.8188,
+      "step": 4201
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5133571934231453e-05,
+      "loss": 0.8441,
+      "step": 4202
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5131266064901344e-05,
+      "loss": 0.7691,
+      "step": 4203
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5128959825170037e-05,
+      "loss": 0.8187,
+      "step": 4204
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5126653215204013e-05,
+      "loss": 0.7559,
+      "step": 4205
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5124346235169776e-05,
+      "loss": 0.8366,
+      "step": 4206
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5122038885233851e-05,
+      "loss": 0.7914,
+      "step": 4207
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5119731165562805e-05,
+      "loss": 0.7595,
+      "step": 4208
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5117423076323208e-05,
+      "loss": 0.8392,
+      "step": 4209
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5115114617681677e-05,
+      "loss": 0.7908,
+      "step": 4210
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5112805789804847e-05,
+      "loss": 0.8144,
+      "step": 4211
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5110496592859381e-05,
+      "loss": 0.8082,
+      "step": 4212
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5108187027011969e-05,
+      "loss": 0.8427,
+      "step": 4213
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5105877092429327e-05,
+      "loss": 0.867,
+      "step": 4214
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5103566789278196e-05,
+      "loss": 0.8422,
+      "step": 4215
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5101256117725349e-05,
+      "loss": 0.7933,
+      "step": 4216
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5098945077937584e-05,
+      "loss": 0.856,
+      "step": 4217
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5096633670081716e-05,
+      "loss": 0.8361,
+      "step": 4218
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5094321894324599e-05,
+      "loss": 0.8728,
+      "step": 4219
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5092009750833112e-05,
+      "loss": 0.8332,
+      "step": 4220
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5089697239774155e-05,
+      "loss": 0.8188,
+      "step": 4221
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5087384361314652e-05,
+      "loss": 0.8502,
+      "step": 4222
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5085071115621564e-05,
+      "loss": 0.8536,
+      "step": 4223
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5082757502861873e-05,
+      "loss": 0.8392,
+      "step": 4224
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5080443523202583e-05,
+      "loss": 0.8191,
+      "step": 4225
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5078129176810735e-05,
+      "loss": 0.7728,
+      "step": 4226
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5075814463853384e-05,
+      "loss": 0.7298,
+      "step": 4227
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5073499384497623e-05,
+      "loss": 0.8656,
+      "step": 4228
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5071183938910564e-05,
+      "loss": 0.4232,
+      "step": 4229
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5068868127259351e-05,
+      "loss": 0.8224,
+      "step": 4230
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5066551949711148e-05,
+      "loss": 0.7726,
+      "step": 4231
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.506423540643315e-05,
+      "loss": 0.8149,
+      "step": 4232
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5061918497592571e-05,
+      "loss": 0.8678,
+      "step": 4233
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5059601223356666e-05,
+      "loss": 0.8187,
+      "step": 4234
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5057283583892708e-05,
+      "loss": 0.7999,
+      "step": 4235
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5054965579367988e-05,
+      "loss": 0.806,
+      "step": 4236
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5052647209949837e-05,
+      "loss": 0.8017,
+      "step": 4237
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5050328475805605e-05,
+      "loss": 0.7833,
+      "step": 4238
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5048009377102673e-05,
+      "loss": 0.7276,
+      "step": 4239
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5045689914008443e-05,
+      "loss": 0.8661,
+      "step": 4240
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5043370086690346e-05,
+      "loss": 0.7534,
+      "step": 4241
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5041049895315842e-05,
+      "loss": 0.8095,
+      "step": 4242
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.503872934005241e-05,
+      "loss": 0.7893,
+      "step": 4243
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5036408421067558e-05,
+      "loss": 0.9016,
+      "step": 4244
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.503408713852883e-05,
+      "loss": 0.8453,
+      "step": 4245
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.503176549260378e-05,
+      "loss": 0.8261,
+      "step": 4246
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5029443483460004e-05,
+      "loss": 0.7878,
+      "step": 4247
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5027121111265109e-05,
+      "loss": 0.8291,
+      "step": 4248
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5024798376186737e-05,
+      "loss": 0.766,
+      "step": 4249
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5022475278392563e-05,
+      "loss": 0.884,
+      "step": 4250
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.502015181805027e-05,
+      "loss": 0.7536,
+      "step": 4251
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5017827995327584e-05,
+      "loss": 0.831,
+      "step": 4252
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5015503810392247e-05,
+      "loss": 0.7556,
+      "step": 4253
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5013179263412032e-05,
+      "loss": 0.8717,
+      "step": 4254
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5010854354554735e-05,
+      "loss": 0.7935,
+      "step": 4255
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5008529083988182e-05,
+      "loss": 0.8014,
+      "step": 4256
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5006203451880224e-05,
+      "loss": 0.802,
+      "step": 4257
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5003877458398733e-05,
+      "loss": 0.8026,
+      "step": 4258
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5001551103711613e-05,
+      "loss": 0.7893,
+      "step": 4259
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4999224387986795e-05,
+      "loss": 0.8237,
+      "step": 4260
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4996897311392231e-05,
+      "loss": 0.8208,
+      "step": 4261
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4994569874095901e-05,
+      "loss": 0.7913,
+      "step": 4262
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4992242076265813e-05,
+      "loss": 0.7986,
+      "step": 4263
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4989913918069996e-05,
+      "loss": 0.7815,
+      "step": 4264
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4987585399676512e-05,
+      "loss": 0.8842,
+      "step": 4265
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4985256521253445e-05,
+      "loss": 0.9393,
+      "step": 4266
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4982927282968907e-05,
+      "loss": 0.8231,
+      "step": 4267
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4980597684991027e-05,
+      "loss": 0.8354,
+      "step": 4268
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4978267727487974e-05,
+      "loss": 0.8606,
+      "step": 4269
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4975937410627937e-05,
+      "loss": 0.8497,
+      "step": 4270
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4973606734579127e-05,
+      "loss": 0.8385,
+      "step": 4271
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4971275699509784e-05,
+      "loss": 0.8798,
+      "step": 4272
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4968944305588176e-05,
+      "loss": 0.7869,
+      "step": 4273
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4966612552982596e-05,
+      "loss": 0.8068,
+      "step": 4274
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4964280441861356e-05,
+      "loss": 0.8511,
+      "step": 4275
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4961947972392809e-05,
+      "loss": 0.8212,
+      "step": 4276
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.495961514474532e-05,
+      "loss": 0.8073,
+      "step": 4277
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4957281959087278e-05,
+      "loss": 0.8288,
+      "step": 4278
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.4954948415587115e-05,
+      "loss": 0.8741,
+      "step": 4279
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4952614514413273e-05,
+      "loss": 0.765,
+      "step": 4280
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4950280255734226e-05,
+      "loss": 0.8486,
+      "step": 4281
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4947945639718473e-05,
+      "loss": 0.8022,
+      "step": 4282
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4945610666534536e-05,
+      "loss": 0.8243,
+      "step": 4283
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4943275336350969e-05,
+      "loss": 0.8759,
+      "step": 4284
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4940939649336348e-05,
+      "loss": 0.7859,
+      "step": 4285
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4938603605659272e-05,
+      "loss": 0.8041,
+      "step": 4286
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4936267205488371e-05,
+      "loss": 0.8024,
+      "step": 4287
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4933930448992295e-05,
+      "loss": 0.789,
+      "step": 4288
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.493159333633973e-05,
+      "loss": 0.8045,
+      "step": 4289
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4929255867699372e-05,
+      "loss": 0.7215,
+      "step": 4290
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4926918043239958e-05,
+      "loss": 0.8173,
+      "step": 4291
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4924579863130245e-05,
+      "loss": 0.8284,
+      "step": 4292
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4922241327539009e-05,
+      "loss": 0.7983,
+      "step": 4293
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.491990243663506e-05,
+      "loss": 0.8518,
+      "step": 4294
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4917563190587234e-05,
+      "loss": 0.8561,
+      "step": 4295
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4915223589564386e-05,
+      "loss": 0.8063,
+      "step": 4296
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4912883633735407e-05,
+      "loss": 0.8206,
+      "step": 4297
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4910543323269198e-05,
+      "loss": 0.7878,
+      "step": 4298
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4908202658334699e-05,
+      "loss": 0.773,
+      "step": 4299
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4905861639100869e-05,
+      "loss": 0.7768,
+      "step": 4300
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4903520265736699e-05,
+      "loss": 0.8216,
+      "step": 4301
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4901178538411201e-05,
+      "loss": 0.8116,
+      "step": 4302
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4898836457293409e-05,
+      "loss": 0.7965,
+      "step": 4303
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4896494022552388e-05,
+      "loss": 0.8295,
+      "step": 4304
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.489415123435723e-05,
+      "loss": 0.8005,
+      "step": 4305
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4891808092877046e-05,
+      "loss": 0.8101,
+      "step": 4306
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4889464598280979e-05,
+      "loss": 0.7721,
+      "step": 4307
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4887120750738191e-05,
+      "loss": 0.8374,
+      "step": 4308
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4884776550417874e-05,
+      "loss": 0.86,
+      "step": 4309
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4882431997489244e-05,
+      "loss": 0.7779,
+      "step": 4310
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4880087092121546e-05,
+      "loss": 0.8698,
+      "step": 4311
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4877741834484044e-05,
+      "loss": 0.8522,
+      "step": 4312
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4875396224746034e-05,
+      "loss": 0.8105,
+      "step": 4313
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4873050263076834e-05,
+      "loss": 0.8772,
+      "step": 4314
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4870703949645784e-05,
+      "loss": 0.8223,
+      "step": 4315
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4868357284622254e-05,
+      "loss": 0.795,
+      "step": 4316
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4866010268175639e-05,
+      "loss": 0.817,
+      "step": 4317
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4863662900475362e-05,
+      "loss": 0.8649,
+      "step": 4318
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4861315181690861e-05,
+      "loss": 0.8365,
+      "step": 4319
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.485896711199161e-05,
+      "loss": 0.8249,
+      "step": 4320
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4856618691547109e-05,
+      "loss": 0.7666,
+      "step": 4321
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4854269920526872e-05,
+      "loss": 0.8221,
+      "step": 4322
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4851920799100453e-05,
+      "loss": 0.7981,
+      "step": 4323
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4849571327437415e-05,
+      "loss": 0.87,
+      "step": 4324
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4847221505707362e-05,
+      "loss": 0.8316,
+      "step": 4325
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4844871334079911e-05,
+      "loss": 0.8012,
+      "step": 4326
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4842520812724712e-05,
+      "loss": 0.8131,
+      "step": 4327
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4840169941811442e-05,
+      "loss": 0.8242,
+      "step": 4328
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4837818721509791e-05,
+      "loss": 0.8071,
+      "step": 4329
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4835467151989487e-05,
+      "loss": 0.8031,
+      "step": 4330
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4833115233420277e-05,
+      "loss": 0.7721,
+      "step": 4331
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4830762965971933e-05,
+      "loss": 0.8195,
+      "step": 4332
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4828410349814257e-05,
+      "loss": 0.8245,
+      "step": 4333
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4826057385117074e-05,
+      "loss": 0.8437,
+      "step": 4334
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4823704072050226e-05,
+      "loss": 0.8352,
+      "step": 4335
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4821350410783597e-05,
+      "loss": 0.7931,
+      "step": 4336
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4818996401487079e-05,
+      "loss": 0.8817,
+      "step": 4337
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4816642044330599e-05,
+      "loss": 0.7644,
+      "step": 4338
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4814287339484108e-05,
+      "loss": 0.8221,
+      "step": 4339
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4811932287117576e-05,
+      "loss": 0.7592,
+      "step": 4340
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4809576887401007e-05,
+      "loss": 0.8287,
+      "step": 4341
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4807221140504425e-05,
+      "loss": 0.8711,
+      "step": 4342
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4804865046597878e-05,
+      "loss": 0.8162,
+      "step": 4343
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4802508605851447e-05,
+      "loss": 0.8679,
+      "step": 4344
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4800151818435225e-05,
+      "loss": 0.7877,
+      "step": 4345
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.479779468451934e-05,
+      "loss": 0.8413,
+      "step": 4346
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4795437204273944e-05,
+      "loss": 0.8212,
+      "step": 4347
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4793079377869206e-05,
+      "loss": 0.8716,
+      "step": 4348
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4790721205475334e-05,
+      "loss": 0.8286,
+      "step": 4349
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4788362687262545e-05,
+      "loss": 0.8813,
+      "step": 4350
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4786003823401096e-05,
+      "loss": 0.8224,
+      "step": 4351
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4783644614061257e-05,
+      "loss": 0.8586,
+      "step": 4352
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4781285059413328e-05,
+      "loss": 0.7892,
+      "step": 4353
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.477892515962764e-05,
+      "loss": 0.7935,
+      "step": 4354
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4776564914874537e-05,
+      "loss": 0.8611,
+      "step": 4355
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4774204325324394e-05,
+      "loss": 0.8604,
+      "step": 4356
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4771843391147612e-05,
+      "loss": 0.7777,
+      "step": 4357
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4769482112514614e-05,
+      "loss": 0.7815,
+      "step": 4358
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.476712048959585e-05,
+      "loss": 0.7363,
+      "step": 4359
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4764758522561798e-05,
+      "loss": 0.739,
+      "step": 4360
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4762396211582947e-05,
+      "loss": 0.8268,
+      "step": 4361
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.476003355682983e-05,
+      "loss": 0.8329,
+      "step": 4362
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4757670558472992e-05,
+      "loss": 0.8816,
+      "step": 4363
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4755307216683007e-05,
+      "loss": 0.7713,
+      "step": 4364
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4752943531630473e-05,
+      "loss": 0.8623,
+      "step": 4365
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4750579503486011e-05,
+      "loss": 0.9108,
+      "step": 4366
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4748215132420272e-05,
+      "loss": 0.8111,
+      "step": 4367
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4745850418603929e-05,
+      "loss": 0.8103,
+      "step": 4368
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4743485362207674e-05,
+      "loss": 0.8083,
+      "step": 4369
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4741119963402237e-05,
+      "loss": 0.8676,
+      "step": 4370
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4738754222358355e-05,
+      "loss": 0.7902,
+      "step": 4371
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4736388139246806e-05,
+      "loss": 0.8679,
+      "step": 4372
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4734021714238385e-05,
+      "loss": 0.4018,
+      "step": 4373
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.473165494750391e-05,
+      "loss": 0.8545,
+      "step": 4374
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4729287839214233e-05,
+      "loss": 0.8416,
+      "step": 4375
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4726920389540216e-05,
+      "loss": 0.8544,
+      "step": 4376
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4724552598652756e-05,
+      "loss": 0.8332,
+      "step": 4377
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4722184466722778e-05,
+      "loss": 0.7573,
+      "step": 4378
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4719815993921219e-05,
+      "loss": 0.8001,
+      "step": 4379
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4717447180419049e-05,
+      "loss": 0.8443,
+      "step": 4380
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4715078026387264e-05,
+      "loss": 0.7707,
+      "step": 4381
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4712708531996875e-05,
+      "loss": 0.7892,
+      "step": 4382
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4710338697418935e-05,
+      "loss": 0.8478,
+      "step": 4383
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4707968522824505e-05,
+      "loss": 0.7845,
+      "step": 4384
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4705598008384674e-05,
+      "loss": 0.81,
+      "step": 4385
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4703227154270564e-05,
+      "loss": 0.4261,
+      "step": 4386
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4700855960653309e-05,
+      "loss": 0.8686,
+      "step": 4387
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4698484427704079e-05,
+      "loss": 0.7912,
+      "step": 4388
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4696112555594058e-05,
+      "loss": 0.8207,
+      "step": 4389
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4693740344494465e-05,
+      "loss": 0.771,
+      "step": 4390
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4691367794576538e-05,
+      "loss": 0.8062,
+      "step": 4391
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4688994906011536e-05,
+      "loss": 0.8057,
+      "step": 4392
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4686621678970747e-05,
+      "loss": 0.396,
+      "step": 4393
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.468424811362549e-05,
+      "loss": 0.806,
+      "step": 4394
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4681874210147093e-05,
+      "loss": 0.8008,
+      "step": 4395
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.467949996870692e-05,
+      "loss": 0.8114,
+      "step": 4396
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4677125389476355e-05,
+      "loss": 0.8385,
+      "step": 4397
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4674750472626806e-05,
+      "loss": 0.8238,
+      "step": 4398
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4672375218329714e-05,
+      "loss": 0.8154,
+      "step": 4399
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4669999626756526e-05,
+      "loss": 0.8124,
+      "step": 4400
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.466762369807873e-05,
+      "loss": 0.8077,
+      "step": 4401
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4665247432467833e-05,
+      "loss": 0.8022,
+      "step": 4402
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4662870830095366e-05,
+      "loss": 0.8291,
+      "step": 4403
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4660493891132885e-05,
+      "loss": 0.8023,
+      "step": 4404
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.465811661575197e-05,
+      "loss": 0.7432,
+      "step": 4405
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4655739004124221e-05,
+      "loss": 0.7659,
+      "step": 4406
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4653361056421272e-05,
+      "loss": 0.8342,
+      "step": 4407
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4650982772814768e-05,
+      "loss": 0.831,
+      "step": 4408
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4648604153476395e-05,
+      "loss": 0.7465,
+      "step": 4409
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4646225198577847e-05,
+      "loss": 0.8403,
+      "step": 4410
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.464384590829085e-05,
+      "loss": 0.8287,
+      "step": 4411
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4641466282787158e-05,
+      "loss": 0.8207,
+      "step": 4412
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4639086322238538e-05,
+      "loss": 0.8825,
+      "step": 4413
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4636706026816793e-05,
+      "loss": 0.8603,
+      "step": 4414
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4634325396693745e-05,
+      "loss": 0.7972,
+      "step": 4415
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4631944432041237e-05,
+      "loss": 0.8516,
+      "step": 4416
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4629563133031139e-05,
+      "loss": 0.8368,
+      "step": 4417
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4627181499835351e-05,
+      "loss": 0.805,
+      "step": 4418
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4624799532625784e-05,
+      "loss": 0.8407,
+      "step": 4419
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4622417231574387e-05,
+      "loss": 0.7999,
+      "step": 4420
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4620034596853122e-05,
+      "loss": 0.8587,
+      "step": 4421
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4617651628633982e-05,
+      "loss": 0.8919,
+      "step": 4422
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.461526832708898e-05,
+      "loss": 0.8221,
+      "step": 4423
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4612884692390159e-05,
+      "loss": 0.7468,
+      "step": 4424
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.461050072470958e-05,
+      "loss": 0.853,
+      "step": 4425
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.460811642421933e-05,
+      "loss": 0.8305,
+      "step": 4426
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4605731791091518e-05,
+      "loss": 0.7821,
+      "step": 4427
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4603346825498284e-05,
+      "loss": 0.8242,
+      "step": 4428
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4600961527611781e-05,
+      "loss": 0.7646,
+      "step": 4429
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4598575897604198e-05,
+      "loss": 0.8421,
+      "step": 4430
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.459618993564774e-05,
+      "loss": 0.772,
+      "step": 4431
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4593803641914633e-05,
+      "loss": 0.9044,
+      "step": 4432
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4591417016577138e-05,
+      "loss": 0.7817,
+      "step": 4433
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4589030059807532e-05,
+      "loss": 0.8382,
+      "step": 4434
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4586642771778119e-05,
+      "loss": 0.8394,
+      "step": 4435
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4584255152661225e-05,
+      "loss": 0.8425,
+      "step": 4436
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4581867202629198e-05,
+      "loss": 0.8484,
+      "step": 4437
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4579478921854418e-05,
+      "loss": 0.8189,
+      "step": 4438
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4577090310509277e-05,
+      "loss": 0.7493,
+      "step": 4439
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4574701368766203e-05,
+      "loss": 0.8329,
+      "step": 4440
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4572312096797638e-05,
+      "loss": 0.8927,
+      "step": 4441
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4569922494776054e-05,
+      "loss": 0.7712,
+      "step": 4442
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4567532562873942e-05,
+      "loss": 0.8275,
+      "step": 4443
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4565142301263826e-05,
+      "loss": 0.3947,
+      "step": 4444
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4562751710118242e-05,
+      "loss": 0.8277,
+      "step": 4445
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4560360789609756e-05,
+      "loss": 0.8625,
+      "step": 4446
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4557969539910956e-05,
+      "loss": 0.8057,
+      "step": 4447
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4555577961194458e-05,
+      "loss": 0.79,
+      "step": 4448
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4553186053632898e-05,
+      "loss": 0.7942,
+      "step": 4449
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4550793817398931e-05,
+      "loss": 0.8591,
+      "step": 4450
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4548401252665248e-05,
+      "loss": 0.747,
+      "step": 4451
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.454600835960455e-05,
+      "loss": 0.8315,
+      "step": 4452
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4543615138389573e-05,
+      "loss": 0.7968,
+      "step": 4453
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.454122158919307e-05,
+      "loss": 0.7872,
+      "step": 4454
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4538827712187822e-05,
+      "loss": 0.8417,
+      "step": 4455
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.453643350754663e-05,
+      "loss": 0.7769,
+      "step": 4456
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.453403897544232e-05,
+      "loss": 0.7642,
+      "step": 4457
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4531644116047739e-05,
+      "loss": 0.8199,
+      "step": 4458
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4529248929535767e-05,
+      "loss": 0.7967,
+      "step": 4459
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4526853416079292e-05,
+      "loss": 0.8348,
+      "step": 4460
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4524457575851243e-05,
+      "loss": 0.8386,
+      "step": 4461
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4522061409024558e-05,
+      "loss": 0.8268,
+      "step": 4462
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4519664915772204e-05,
+      "loss": 0.7901,
+      "step": 4463
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4517268096267178e-05,
+      "loss": 0.8147,
+      "step": 4464
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4514870950682492e-05,
+      "loss": 0.7984,
+      "step": 4465
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4512473479191183e-05,
+      "loss": 0.7983,
+      "step": 4466
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4510075681966316e-05,
+      "loss": 0.8273,
+      "step": 4467
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.450767755918097e-05,
+      "loss": 0.8127,
+      "step": 4468
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4505279111008263e-05,
+      "loss": 0.8275,
+      "step": 4469
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.450288033762132e-05,
+      "loss": 0.7827,
+      "step": 4470
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4500481239193301e-05,
+      "loss": 0.7843,
+      "step": 4471
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4498081815897382e-05,
+      "loss": 0.7794,
+      "step": 4472
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4495682067906769e-05,
+      "loss": 0.7473,
+      "step": 4473
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4493281995394683e-05,
+      "loss": 0.8582,
+      "step": 4474
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4490881598534379e-05,
+      "loss": 0.8068,
+      "step": 4475
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4488480877499127e-05,
+      "loss": 0.7576,
+      "step": 4476
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4486079832462228e-05,
+      "loss": 0.8226,
+      "step": 4477
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4483678463596993e-05,
+      "loss": 0.868,
+      "step": 4478
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4481276771076773e-05,
+      "loss": 0.8596,
+      "step": 4479
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4478874755074932e-05,
+      "loss": 0.8524,
+      "step": 4480
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.447647241576486e-05,
+      "loss": 0.8311,
+      "step": 4481
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.447406975331997e-05,
+      "loss": 0.8532,
+      "step": 4482
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.44716667679137e-05,
+      "loss": 0.4321,
+      "step": 4483
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4469263459719503e-05,
+      "loss": 0.824,
+      "step": 4484
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.446685982891087e-05,
+      "loss": 0.8307,
+      "step": 4485
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4464455875661308e-05,
+      "loss": 0.8272,
+      "step": 4486
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4462051600144338e-05,
+      "loss": 0.8023,
+      "step": 4487
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4459647002533523e-05,
+      "loss": 0.7835,
+      "step": 4488
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4457242083002437e-05,
+      "loss": 0.8479,
+      "step": 4489
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4454836841724676e-05,
+      "loss": 0.802,
+      "step": 4490
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.445243127887386e-05,
+      "loss": 0.8289,
+      "step": 4491
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4450025394623642e-05,
+      "loss": 0.8772,
+      "step": 4492
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4447619189147687e-05,
+      "loss": 0.8754,
+      "step": 4493
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.444521266261969e-05,
+      "loss": 0.8855,
+      "step": 4494
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.444280581521336e-05,
+      "loss": 0.7758,
+      "step": 4495
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4440398647102444e-05,
+      "loss": 0.7506,
+      "step": 4496
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4437991158460703e-05,
+      "loss": 0.7794,
+      "step": 4497
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4435583349461918e-05,
+      "loss": 0.8525,
+      "step": 4498
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4433175220279896e-05,
+      "loss": 0.7693,
+      "step": 4499
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4430766771088469e-05,
+      "loss": 0.8231,
+      "step": 4500
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4428358002061498e-05,
+      "loss": 0.7654,
+      "step": 4501
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4425948913372852e-05,
+      "loss": 0.8191,
+      "step": 4502
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4423539505196433e-05,
+      "loss": 0.8381,
+      "step": 4503
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.442112977770617e-05,
+      "loss": 0.8457,
+      "step": 4504
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4418719731075999e-05,
+      "loss": 0.8633,
+      "step": 4505
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.44163093654799e-05,
+      "loss": 0.8252,
+      "step": 4506
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.441389868109186e-05,
+      "loss": 0.8435,
+      "step": 4507
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4411487678085899e-05,
+      "loss": 0.8266,
+      "step": 4508
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4409076356636051e-05,
+      "loss": 0.854,
+      "step": 4509
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.440666471691638e-05,
+      "loss": 0.9075,
+      "step": 4510
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.440425275910097e-05,
+      "loss": 0.4238,
+      "step": 4511
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4401840483363926e-05,
+      "loss": 0.7657,
+      "step": 4512
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4399427889879382e-05,
+      "loss": 0.7559,
+      "step": 4513
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4397014978821494e-05,
+      "loss": 0.7772,
+      "step": 4514
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4394601750364432e-05,
+      "loss": 0.7843,
+      "step": 4515
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4392188204682396e-05,
+      "loss": 0.8111,
+      "step": 4516
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4389774341949611e-05,
+      "loss": 0.8102,
+      "step": 4517
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4387360162340323e-05,
+      "loss": 0.8541,
+      "step": 4518
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4384945666028798e-05,
+      "loss": 0.7864,
+      "step": 4519
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4382530853189325e-05,
+      "loss": 0.8301,
+      "step": 4520
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4380115723996224e-05,
+      "loss": 0.7917,
+      "step": 4521
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4377700278623824e-05,
+      "loss": 0.7912,
+      "step": 4522
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4375284517246484e-05,
+      "loss": 0.8024,
+      "step": 4523
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4372868440038595e-05,
+      "loss": 0.7811,
+      "step": 4524
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.437045204717455e-05,
+      "loss": 0.8445,
+      "step": 4525
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4368035338828787e-05,
+      "loss": 0.8511,
+      "step": 4526
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4365618315175751e-05,
+      "loss": 0.8367,
+      "step": 4527
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4363200976389915e-05,
+      "loss": 0.7759,
+      "step": 4528
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.436078332264578e-05,
+      "loss": 0.8141,
+      "step": 4529
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4358365354117859e-05,
+      "loss": 0.8275,
+      "step": 4530
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4355947070980696e-05,
+      "loss": 0.8562,
+      "step": 4531
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4353528473408856e-05,
+      "loss": 0.7952,
+      "step": 4532
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4351109561576922e-05,
+      "loss": 0.8617,
+      "step": 4533
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4348690335659507e-05,
+      "loss": 0.7674,
+      "step": 4534
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.434627079583124e-05,
+      "loss": 0.7986,
+      "step": 4535
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.434385094226678e-05,
+      "loss": 0.8439,
+      "step": 4536
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4341430775140804e-05,
+      "loss": 0.8038,
+      "step": 4537
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4339010294628009e-05,
+      "loss": 0.8298,
+      "step": 4538
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4336589500903117e-05,
+      "loss": 0.8547,
+      "step": 4539
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.433416839414088e-05,
+      "loss": 0.7886,
+      "step": 4540
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.433174697451606e-05,
+      "loss": 0.845,
+      "step": 4541
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.432932524220345e-05,
+      "loss": 0.8534,
+      "step": 4542
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4326903197377862e-05,
+      "loss": 0.8108,
+      "step": 4543
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.432448084021413e-05,
+      "loss": 0.8588,
+      "step": 4544
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4322058170887116e-05,
+      "loss": 0.8261,
+      "step": 4545
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.43196351895717e-05,
+      "loss": 0.8494,
+      "step": 4546
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4317211896442782e-05,
+      "loss": 0.7973,
+      "step": 4547
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4314788291675288e-05,
+      "loss": 0.8279,
+      "step": 4548
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.431236437544417e-05,
+      "loss": 0.8113,
+      "step": 4549
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.43099401479244e-05,
+      "loss": 0.8015,
+      "step": 4550
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4307515609290964e-05,
+      "loss": 0.7294,
+      "step": 4551
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4305090759718882e-05,
+      "loss": 0.7853,
+      "step": 4552
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4302665599383196e-05,
+      "loss": 0.8774,
+      "step": 4553
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.430024012845896e-05,
+      "loss": 0.7819,
+      "step": 4554
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4297814347121261e-05,
+      "loss": 0.7815,
+      "step": 4555
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4295388255545201e-05,
+      "loss": 0.8292,
+      "step": 4556
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4292961853905908e-05,
+      "loss": 0.8456,
+      "step": 4557
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4290535142378536e-05,
+      "loss": 0.8293,
+      "step": 4558
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4288108121138256e-05,
+      "loss": 0.8188,
+      "step": 4559
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4285680790360261e-05,
+      "loss": 0.8157,
+      "step": 4560
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4283253150219771e-05,
+      "loss": 0.8281,
+      "step": 4561
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4280825200892021e-05,
+      "loss": 0.7959,
+      "step": 4562
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.427839694255228e-05,
+      "loss": 0.7675,
+      "step": 4563
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4275968375375825e-05,
+      "loss": 0.8568,
+      "step": 4564
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4273539499537965e-05,
+      "loss": 0.7708,
+      "step": 4565
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4271110315214033e-05,
+      "loss": 0.7781,
+      "step": 4566
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4268680822579373e-05,
+      "loss": 0.7177,
+      "step": 4567
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4266251021809364e-05,
+      "loss": 0.7863,
+      "step": 4568
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4263820913079396e-05,
+      "loss": 0.8171,
+      "step": 4569
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4261390496564891e-05,
+      "loss": 0.858,
+      "step": 4570
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.425895977244129e-05,
+      "loss": 0.8022,
+      "step": 4571
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.425652874088405e-05,
+      "loss": 0.815,
+      "step": 4572
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4254097402068659e-05,
+      "loss": 0.4359,
+      "step": 4573
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4251665756170626e-05,
+      "loss": 0.7985,
+      "step": 4574
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4249233803365474e-05,
+      "loss": 0.8036,
+      "step": 4575
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4246801543828759e-05,
+      "loss": 0.8493,
+      "step": 4576
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4244368977736048e-05,
+      "loss": 0.8372,
+      "step": 4577
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.424193610526294e-05,
+      "loss": 0.8843,
+      "step": 4578
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4239502926585054e-05,
+      "loss": 0.8527,
+      "step": 4579
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4237069441878023e-05,
+      "loss": 0.8978,
+      "step": 4580
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4234635651317521e-05,
+      "loss": 0.786,
+      "step": 4581
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4232201555079217e-05,
+      "loss": 0.8094,
+      "step": 4582
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4229767153338822e-05,
+      "loss": 0.7778,
+      "step": 4583
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.422733244627207e-05,
+      "loss": 0.7637,
+      "step": 4584
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4224897434054704e-05,
+      "loss": 0.7686,
+      "step": 4585
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4222462116862492e-05,
+      "loss": 0.7868,
+      "step": 4586
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4220026494871236e-05,
+      "loss": 0.7968,
+      "step": 4587
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4217590568256747e-05,
+      "loss": 0.7876,
+      "step": 4588
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4215154337194868e-05,
+      "loss": 0.8488,
+      "step": 4589
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4212717801861451e-05,
+      "loss": 0.8209,
+      "step": 4590
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4210280962432381e-05,
+      "loss": 0.7985,
+      "step": 4591
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4207843819083566e-05,
+      "loss": 0.835,
+      "step": 4592
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4205406371990923e-05,
+      "loss": 0.8037,
+      "step": 4593
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.420296862133041e-05,
+      "loss": 0.8111,
+      "step": 4594
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4200530567277986e-05,
+      "loss": 0.833,
+      "step": 4595
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4198092210009646e-05,
+      "loss": 0.7935,
+      "step": 4596
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4195653549701406e-05,
+      "loss": 0.7844,
+      "step": 4597
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4193214586529297e-05,
+      "loss": 0.8462,
+      "step": 4598
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.419077532066938e-05,
+      "loss": 0.7856,
+      "step": 4599
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4188335752297733e-05,
+      "loss": 0.8178,
+      "step": 4600
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4185895881590455e-05,
+      "loss": 0.8575,
+      "step": 4601
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4183455708723667e-05,
+      "loss": 0.8109,
+      "step": 4602
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4181015233873515e-05,
+      "loss": 0.7658,
+      "step": 4603
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4178574457216166e-05,
+      "loss": 0.7851,
+      "step": 4604
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4176133378927805e-05,
+      "loss": 0.7787,
+      "step": 4605
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4173691999184644e-05,
+      "loss": 0.7564,
+      "step": 4606
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4171250318162915e-05,
+      "loss": 0.8004,
+      "step": 4607
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4168808336038869e-05,
+      "loss": 0.7979,
+      "step": 4608
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4166366052988781e-05,
+      "loss": 0.7479,
+      "step": 4609
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4163923469188949e-05,
+      "loss": 0.8077,
+      "step": 4610
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4161480584815692e-05,
+      "loss": 0.7288,
+      "step": 4611
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4159037400045346e-05,
+      "loss": 0.822,
+      "step": 4612
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4156593915054277e-05,
+      "loss": 0.8662,
+      "step": 4613
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4154150130018867e-05,
+      "loss": 0.8489,
+      "step": 4614
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.415170604511552e-05,
+      "loss": 0.7831,
+      "step": 4615
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4149261660520661e-05,
+      "loss": 0.435,
+      "step": 4616
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4146816976410742e-05,
+      "loss": 0.7522,
+      "step": 4617
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.414437199296223e-05,
+      "loss": 0.7716,
+      "step": 4618
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4141926710351617e-05,
+      "loss": 0.822,
+      "step": 4619
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4139481128755419e-05,
+      "loss": 0.8335,
+      "step": 4620
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4137035248350168e-05,
+      "loss": 0.8409,
+      "step": 4621
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4134589069312422e-05,
+      "loss": 0.8344,
+      "step": 4622
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4132142591818755e-05,
+      "loss": 0.7781,
+      "step": 4623
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4129695816045769e-05,
+      "loss": 0.8173,
+      "step": 4624
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4127248742170086e-05,
+      "loss": 0.807,
+      "step": 4625
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4124801370368348e-05,
+      "loss": 0.7604,
+      "step": 4626
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4122353700817216e-05,
+      "loss": 0.8449,
+      "step": 4627
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4119905733693378e-05,
+      "loss": 0.8287,
+      "step": 4628
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4117457469173539e-05,
+      "loss": 0.7524,
+      "step": 4629
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.411500890743443e-05,
+      "loss": 0.7891,
+      "step": 4630
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.41125600486528e-05,
+      "loss": 0.8042,
+      "step": 4631
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4110110893005418e-05,
+      "loss": 0.8348,
+      "step": 4632
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4107661440669078e-05,
+      "loss": 0.8581,
+      "step": 4633
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4105211691820597e-05,
+      "loss": 0.7786,
+      "step": 4634
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4102761646636805e-05,
+      "loss": 0.7986,
+      "step": 4635
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4100311305294563e-05,
+      "loss": 0.7742,
+      "step": 4636
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4097860667970748e-05,
+      "loss": 0.8015,
+      "step": 4637
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4095409734842258e-05,
+      "loss": 0.7905,
+      "step": 4638
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4092958506086017e-05,
+      "loss": 0.8,
+      "step": 4639
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4090506981878964e-05,
+      "loss": 0.7886,
+      "step": 4640
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4088055162398066e-05,
+      "loss": 0.8083,
+      "step": 4641
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4085603047820308e-05,
+      "loss": 0.8918,
+      "step": 4642
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4083150638322692e-05,
+      "loss": 0.8688,
+      "step": 4643
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.408069793408225e-05,
+      "loss": 0.8291,
+      "step": 4644
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4078244935276027e-05,
+      "loss": 0.8041,
+      "step": 4645
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4075791642081097e-05,
+      "loss": 0.7976,
+      "step": 4646
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.407333805467455e-05,
+      "loss": 0.7934,
+      "step": 4647
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4070884173233495e-05,
+      "loss": 0.8336,
+      "step": 4648
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4068429997935069e-05,
+      "loss": 0.8202,
+      "step": 4649
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4065975528956428e-05,
+      "loss": 0.8369,
+      "step": 4650
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4063520766474747e-05,
+      "loss": 0.8224,
+      "step": 4651
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4061065710667223e-05,
+      "loss": 0.8564,
+      "step": 4652
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4058610361711075e-05,
+      "loss": 0.7202,
+      "step": 4653
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4056154719783545e-05,
+      "loss": 0.7797,
+      "step": 4654
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.405369878506189e-05,
+      "loss": 0.7572,
+      "step": 4655
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4051242557723393e-05,
+      "loss": 0.8972,
+      "step": 4656
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4048786037945357e-05,
+      "loss": 0.8066,
+      "step": 4657
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4046329225905107e-05,
+      "loss": 0.8007,
+      "step": 4658
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.404387212177999e-05,
+      "loss": 0.8085,
+      "step": 4659
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4041414725747366e-05,
+      "loss": 0.8612,
+      "step": 4660
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4038957037984631e-05,
+      "loss": 0.7969,
+      "step": 4661
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.403649905866919e-05,
+      "loss": 0.8099,
+      "step": 4662
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4034040787978472e-05,
+      "loss": 0.8034,
+      "step": 4663
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4031582226089926e-05,
+      "loss": 0.8931,
+      "step": 4664
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4029123373181028e-05,
+      "loss": 0.8261,
+      "step": 4665
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4026664229429267e-05,
+      "loss": 0.8572,
+      "step": 4666
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.402420479501216e-05,
+      "loss": 0.7398,
+      "step": 4667
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4021745070107238e-05,
+      "loss": 0.825,
+      "step": 4668
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4019285054892055e-05,
+      "loss": 0.853,
+      "step": 4669
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4016824749544195e-05,
+      "loss": 0.4094,
+      "step": 4670
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.401436415424125e-05,
+      "loss": 0.8039,
+      "step": 4671
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4011903269160838e-05,
+      "loss": 0.835,
+      "step": 4672
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4009442094480603e-05,
+      "loss": 0.8559,
+      "step": 4673
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.40069806303782e-05,
+      "loss": 0.8661,
+      "step": 4674
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4004518877031314e-05,
+      "loss": 0.7519,
+      "step": 4675
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4002056834617644e-05,
+      "loss": 0.7772,
+      "step": 4676
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3999594503314913e-05,
+      "loss": 0.7265,
+      "step": 4677
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.399713188330087e-05,
+      "loss": 0.7966,
+      "step": 4678
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3994668974753272e-05,
+      "loss": 0.8605,
+      "step": 4679
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3992205777849908e-05,
+      "loss": 0.8092,
+      "step": 4680
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3989742292768584e-05,
+      "loss": 0.7891,
+      "step": 4681
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3987278519687131e-05,
+      "loss": 0.8864,
+      "step": 4682
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.398481445878339e-05,
+      "loss": 0.7596,
+      "step": 4683
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3982350110235234e-05,
+      "loss": 0.8312,
+      "step": 4684
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.397988547422055e-05,
+      "loss": 0.8391,
+      "step": 4685
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3977420550917252e-05,
+      "loss": 0.8486,
+      "step": 4686
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3974955340503265e-05,
+      "loss": 0.7795,
+      "step": 4687
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3972489843156547e-05,
+      "loss": 0.7921,
+      "step": 4688
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3970024059055063e-05,
+      "loss": 0.7975,
+      "step": 4689
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3967557988376812e-05,
+      "loss": 0.8053,
+      "step": 4690
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3965091631299803e-05,
+      "loss": 0.7636,
+      "step": 4691
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3962624988002076e-05,
+      "loss": 0.7767,
+      "step": 4692
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3960158058661679e-05,
+      "loss": 0.7783,
+      "step": 4693
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3957690843456697e-05,
+      "loss": 0.7956,
+      "step": 4694
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3955223342565219e-05,
+      "loss": 0.795,
+      "step": 4695
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3952755556165363e-05,
+      "loss": 0.8512,
+      "step": 4696
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3950287484435268e-05,
+      "loss": 0.778,
+      "step": 4697
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3947819127553088e-05,
+      "loss": 0.8262,
+      "step": 4698
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3945350485697012e-05,
+      "loss": 0.8098,
+      "step": 4699
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3942881559045228e-05,
+      "loss": 0.7745,
+      "step": 4700
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3940412347775957e-05,
+      "loss": 0.8163,
+      "step": 4701
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3937942852067448e-05,
+      "loss": 0.7162,
+      "step": 4702
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3935473072097954e-05,
+      "loss": 0.8413,
+      "step": 4703
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.393300300804576e-05,
+      "loss": 0.7805,
+      "step": 4704
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3930532660089166e-05,
+      "loss": 0.7608,
+      "step": 4705
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3928062028406497e-05,
+      "loss": 0.8657,
+      "step": 4706
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3925591113176093e-05,
+      "loss": 0.8054,
+      "step": 4707
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.392311991457632e-05,
+      "loss": 0.8155,
+      "step": 4708
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.392064843278556e-05,
+      "loss": 0.7755,
+      "step": 4709
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3918176667982218e-05,
+      "loss": 0.8829,
+      "step": 4710
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3915704620344715e-05,
+      "loss": 0.7862,
+      "step": 4711
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3913232290051501e-05,
+      "loss": 0.7958,
+      "step": 4712
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3910759677281041e-05,
+      "loss": 0.8736,
+      "step": 4713
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3908286782211822e-05,
+      "loss": 0.7906,
+      "step": 4714
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3905813605022345e-05,
+      "loss": 0.8275,
+      "step": 4715
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3903340145891143e-05,
+      "loss": 0.824,
+      "step": 4716
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3900866404996759e-05,
+      "loss": 0.7997,
+      "step": 4717
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3898392382517759e-05,
+      "loss": 0.8164,
+      "step": 4718
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3895918078632733e-05,
+      "loss": 0.842,
+      "step": 4719
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3893443493520293e-05,
+      "loss": 0.803,
+      "step": 4720
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3890968627359058e-05,
+      "loss": 0.8329,
+      "step": 4721
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3888493480327685e-05,
+      "loss": 0.7977,
+      "step": 4722
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3886018052604839e-05,
+      "loss": 0.8169,
+      "step": 4723
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.388354234436921e-05,
+      "loss": 0.8552,
+      "step": 4724
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3881066355799508e-05,
+      "loss": 0.8437,
+      "step": 4725
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3878590087074462e-05,
+      "loss": 0.7547,
+      "step": 4726
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.387611353837282e-05,
+      "loss": 0.808,
+      "step": 4727
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3873636709873355e-05,
+      "loss": 0.7916,
+      "step": 4728
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3871159601754853e-05,
+      "loss": 0.8838,
+      "step": 4729
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3868682214196132e-05,
+      "loss": 0.8005,
+      "step": 4730
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3866204547376015e-05,
+      "loss": 0.7778,
+      "step": 4731
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3863726601473353e-05,
+      "loss": 0.7783,
+      "step": 4732
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3861248376667022e-05,
+      "loss": 0.8254,
+      "step": 4733
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3858769873135911e-05,
+      "loss": 0.8432,
+      "step": 4734
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.385629109105893e-05,
+      "loss": 0.8351,
+      "step": 4735
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3853812030615012e-05,
+      "loss": 0.7909,
+      "step": 4736
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3851332691983105e-05,
+      "loss": 0.8477,
+      "step": 4737
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3848853075342185e-05,
+      "loss": 0.8617,
+      "step": 4738
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.384637318087124e-05,
+      "loss": 0.6764,
+      "step": 4739
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3843893008749283e-05,
+      "loss": 0.7493,
+      "step": 4740
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3841412559155345e-05,
+      "loss": 0.8385,
+      "step": 4741
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3838931832268479e-05,
+      "loss": 0.8177,
+      "step": 4742
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3836450828267754e-05,
+      "loss": 0.8441,
+      "step": 4743
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3833969547332262e-05,
+      "loss": 0.849,
+      "step": 4744
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3831487989641118e-05,
+      "loss": 0.7869,
+      "step": 4745
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.382900615537345e-05,
+      "loss": 0.7757,
+      "step": 4746
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3826524044708411e-05,
+      "loss": 0.7873,
+      "step": 4747
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3824041657825176e-05,
+      "loss": 0.8353,
+      "step": 4748
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3821558994902928e-05,
+      "loss": 0.8787,
+      "step": 4749
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3819076056120885e-05,
+      "loss": 0.781,
+      "step": 4750
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3816592841658278e-05,
+      "loss": 0.9039,
+      "step": 4751
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3814109351694356e-05,
+      "loss": 0.7983,
+      "step": 4752
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.381162558640839e-05,
+      "loss": 0.8518,
+      "step": 4753
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3809141545979674e-05,
+      "loss": 0.8008,
+      "step": 4754
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3806657230587515e-05,
+      "loss": 0.7721,
+      "step": 4755
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3804172640411246e-05,
+      "loss": 0.818,
+      "step": 4756
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3801687775630218e-05,
+      "loss": 0.849,
+      "step": 4757
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3799202636423796e-05,
+      "loss": 0.8257,
+      "step": 4758
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3796717222971379e-05,
+      "loss": 0.7901,
+      "step": 4759
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3794231535452371e-05,
+      "loss": 0.7732,
+      "step": 4760
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3791745574046202e-05,
+      "loss": 0.8401,
+      "step": 4761
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3789259338932324e-05,
+      "loss": 0.8518,
+      "step": 4762
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3786772830290201e-05,
+      "loss": 0.8095,
+      "step": 4763
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.378428604829933e-05,
+      "loss": 0.7827,
+      "step": 4764
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3781798993139212e-05,
+      "loss": 0.9193,
+      "step": 4765
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.377931166498938e-05,
+      "loss": 0.8041,
+      "step": 4766
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3776824064029386e-05,
+      "loss": 0.8682,
+      "step": 4767
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3774336190438788e-05,
+      "loss": 0.8332,
+      "step": 4768
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.377184804439718e-05,
+      "loss": 0.4434,
+      "step": 4769
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3769359626084166e-05,
+      "loss": 0.8056,
+      "step": 4770
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3766870935679375e-05,
+      "loss": 0.7955,
+      "step": 4771
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3764381973362454e-05,
+      "loss": 0.7517,
+      "step": 4772
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3761892739313065e-05,
+      "loss": 0.4061,
+      "step": 4773
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3759403233710899e-05,
+      "loss": 0.7773,
+      "step": 4774
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3756913456735658e-05,
+      "loss": 0.7089,
+      "step": 4775
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.375442340856707e-05,
+      "loss": 0.7662,
+      "step": 4776
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3751933089384876e-05,
+      "loss": 0.7425,
+      "step": 4777
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3749442499368844e-05,
+      "loss": 0.7862,
+      "step": 4778
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3746951638698751e-05,
+      "loss": 0.8206,
+      "step": 4779
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.374446050755441e-05,
+      "loss": 0.7793,
+      "step": 4780
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3741969106115632e-05,
+      "loss": 0.8409,
+      "step": 4781
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3739477434562268e-05,
+      "loss": 0.8755,
+      "step": 4782
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3736985493074175e-05,
+      "loss": 0.8353,
+      "step": 4783
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3734493281831236e-05,
+      "loss": 0.8319,
+      "step": 4784
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3732000801013354e-05,
+      "loss": 0.8093,
+      "step": 4785
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3729508050800443e-05,
+      "loss": 0.7409,
+      "step": 4786
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3727015031372452e-05,
+      "loss": 0.7874,
+      "step": 4787
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3724521742909331e-05,
+      "loss": 0.8097,
+      "step": 4788
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3722028185591061e-05,
+      "loss": 0.8466,
+      "step": 4789
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3719534359597644e-05,
+      "loss": 0.8441,
+      "step": 4790
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.371704026510909e-05,
+      "loss": 0.7761,
+      "step": 4791
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3714545902305439e-05,
+      "loss": 0.7589,
+      "step": 4792
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3712051271366747e-05,
+      "loss": 0.9052,
+      "step": 4793
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3709556372473087e-05,
+      "loss": 0.8545,
+      "step": 4794
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3707061205804562e-05,
+      "loss": 0.7893,
+      "step": 4795
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3704565771541277e-05,
+      "loss": 0.7684,
+      "step": 4796
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3702070069863369e-05,
+      "loss": 0.8031,
+      "step": 4797
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.369957410095099e-05,
+      "loss": 0.8959,
+      "step": 4798
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3697077864984311e-05,
+      "loss": 0.8534,
+      "step": 4799
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3694581362143525e-05,
+      "loss": 0.8693,
+      "step": 4800
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3692084592608841e-05,
+      "loss": 0.8779,
+      "step": 4801
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3689587556560489e-05,
+      "loss": 0.8047,
+      "step": 4802
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3687090254178717e-05,
+      "loss": 0.8175,
+      "step": 4803
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3684592685643793e-05,
+      "loss": 0.84,
+      "step": 4804
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3682094851136008e-05,
+      "loss": 0.7679,
+      "step": 4805
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3679596750835667e-05,
+      "loss": 0.788,
+      "step": 4806
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3677098384923094e-05,
+      "loss": 0.8882,
+      "step": 4807
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3674599753578632e-05,
+      "loss": 0.8295,
+      "step": 4808
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3672100856982651e-05,
+      "loss": 0.7478,
+      "step": 4809
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.366960169531553e-05,
+      "loss": 0.7859,
+      "step": 4810
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3667102268757671e-05,
+      "loss": 0.842,
+      "step": 4811
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.36646025774895e-05,
+      "loss": 0.8056,
+      "step": 4812
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.366210262169145e-05,
+      "loss": 0.8237,
+      "step": 4813
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3659602401543986e-05,
+      "loss": 0.829,
+      "step": 4814
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3657101917227587e-05,
+      "loss": 0.8669,
+      "step": 4815
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3654601168922748e-05,
+      "loss": 0.8437,
+      "step": 4816
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3652100156809989e-05,
+      "loss": 0.7843,
+      "step": 4817
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3649598881069843e-05,
+      "loss": 0.8359,
+      "step": 4818
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3647097341882869e-05,
+      "loss": 0.8398,
+      "step": 4819
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3644595539429637e-05,
+      "loss": 0.875,
+      "step": 4820
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3642093473890743e-05,
+      "loss": 0.6928,
+      "step": 4821
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3639591145446795e-05,
+      "loss": 0.8295,
+      "step": 4822
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3637088554278426e-05,
+      "loss": 0.8191,
+      "step": 4823
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3634585700566287e-05,
+      "loss": 0.8163,
+      "step": 4824
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3632082584491043e-05,
+      "loss": 0.8213,
+      "step": 4825
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3629579206233388e-05,
+      "loss": 0.8172,
+      "step": 4826
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.362707556597403e-05,
+      "loss": 0.8204,
+      "step": 4827
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3624571663893683e-05,
+      "loss": 0.7779,
+      "step": 4828
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3622067500173102e-05,
+      "loss": 0.7934,
+      "step": 4829
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3619563074993046e-05,
+      "loss": 0.4105,
+      "step": 4830
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3617058388534298e-05,
+      "loss": 0.8729,
+      "step": 4831
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3614553440977663e-05,
+      "loss": 0.859,
+      "step": 4832
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3612048232503953e-05,
+      "loss": 0.8378,
+      "step": 4833
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3609542763294012e-05,
+      "loss": 0.7957,
+      "step": 4834
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3607037033528698e-05,
+      "loss": 0.8055,
+      "step": 4835
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3604531043388888e-05,
+      "loss": 0.8151,
+      "step": 4836
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3602024793055475e-05,
+      "loss": 0.8606,
+      "step": 4837
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3599518282709374e-05,
+      "loss": 0.7625,
+      "step": 4838
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3597011512531517e-05,
+      "loss": 0.861,
+      "step": 4839
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.359450448270286e-05,
+      "loss": 0.7823,
+      "step": 4840
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3591997193404364e-05,
+      "loss": 0.7644,
+      "step": 4841
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3589489644817029e-05,
+      "loss": 0.7746,
+      "step": 4842
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3586981837121854e-05,
+      "loss": 0.7996,
+      "step": 4843
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.358447377049987e-05,
+      "loss": 0.834,
+      "step": 4844
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.358196544513212e-05,
+      "loss": 0.7816,
+      "step": 4845
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3579456861199667e-05,
+      "loss": 0.861,
+      "step": 4846
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3576948018883598e-05,
+      "loss": 0.7911,
+      "step": 4847
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3574438918365011e-05,
+      "loss": 0.7819,
+      "step": 4848
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3571929559825026e-05,
+      "loss": 0.7838,
+      "step": 4849
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3569419943444783e-05,
+      "loss": 0.7522,
+      "step": 4850
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3566910069405439e-05,
+      "loss": 0.7464,
+      "step": 4851
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3564399937888163e-05,
+      "loss": 0.8025,
+      "step": 4852
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3561889549074161e-05,
+      "loss": 0.8597,
+      "step": 4853
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3559378903144635e-05,
+      "loss": 0.7855,
+      "step": 4854
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3556868000280823e-05,
+      "loss": 0.8133,
+      "step": 4855
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.355435684066397e-05,
+      "loss": 0.8599,
+      "step": 4856
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.355184542447535e-05,
+      "loss": 0.8175,
+      "step": 4857
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3549333751896246e-05,
+      "loss": 0.8205,
+      "step": 4858
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3546821823107963e-05,
+      "loss": 0.7681,
+      "step": 4859
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3544309638291828e-05,
+      "loss": 0.7769,
+      "step": 4860
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3541797197629183e-05,
+      "loss": 0.8157,
+      "step": 4861
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3539284501301383e-05,
+      "loss": 0.7556,
+      "step": 4862
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3536771549489819e-05,
+      "loss": 0.7587,
+      "step": 4863
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3534258342375876e-05,
+      "loss": 0.8667,
+      "step": 4864
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3531744880140978e-05,
+      "loss": 0.7457,
+      "step": 4865
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3529231162966556e-05,
+      "loss": 0.8136,
+      "step": 4866
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3526717191034067e-05,
+      "loss": 0.8569,
+      "step": 4867
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3524202964524981e-05,
+      "loss": 0.7932,
+      "step": 4868
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3521688483620785e-05,
+      "loss": 0.8133,
+      "step": 4869
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3519173748502989e-05,
+      "loss": 0.7909,
+      "step": 4870
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3516658759353123e-05,
+      "loss": 0.7591,
+      "step": 4871
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3514143516352727e-05,
+      "loss": 0.8029,
+      "step": 4872
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3511628019683363e-05,
+      "loss": 0.8669,
+      "step": 4873
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.350911226952662e-05,
+      "loss": 0.8061,
+      "step": 4874
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3506596266064088e-05,
+      "loss": 0.7788,
+      "step": 4875
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3504080009477391e-05,
+      "loss": 0.8008,
+      "step": 4876
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3501563499948166e-05,
+      "loss": 0.7994,
+      "step": 4877
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3499046737658066e-05,
+      "loss": 0.7733,
+      "step": 4878
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3496529722788765e-05,
+      "loss": 0.8273,
+      "step": 4879
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3494012455521952e-05,
+      "loss": 0.7997,
+      "step": 4880
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3491494936039339e-05,
+      "loss": 0.7985,
+      "step": 4881
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3488977164522648e-05,
+      "loss": 0.8138,
+      "step": 4882
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.348645914115363e-05,
+      "loss": 0.3792,
+      "step": 4883
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.348394086611405e-05,
+      "loss": 0.7428,
+      "step": 4884
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3481422339585684e-05,
+      "loss": 0.8012,
+      "step": 4885
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3478903561750339e-05,
+      "loss": 0.8606,
+      "step": 4886
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3476384532789826e-05,
+      "loss": 0.8546,
+      "step": 4887
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3473865252885987e-05,
+      "loss": 0.7881,
+      "step": 4888
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3471345722220676e-05,
+      "loss": 0.7882,
+      "step": 4889
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3468825940975764e-05,
+      "loss": 0.8011,
+      "step": 4890
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3466305909333144e-05,
+      "loss": 0.8047,
+      "step": 4891
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3463785627474723e-05,
+      "loss": 0.767,
+      "step": 4892
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3461265095582427e-05,
+      "loss": 0.8236,
+      "step": 4893
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3458744313838205e-05,
+      "loss": 0.8363,
+      "step": 4894
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3456223282424017e-05,
+      "loss": 0.7835,
+      "step": 4895
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3453702001521842e-05,
+      "loss": 0.7752,
+      "step": 4896
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.345118047131368e-05,
+      "loss": 0.8293,
+      "step": 4897
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3448658691981553e-05,
+      "loss": 0.8087,
+      "step": 4898
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3446136663707489e-05,
+      "loss": 0.8329,
+      "step": 4899
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3443614386673549e-05,
+      "loss": 0.7732,
+      "step": 4900
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3441091861061796e-05,
+      "loss": 0.3881,
+      "step": 4901
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3438569087054325e-05,
+      "loss": 0.8155,
+      "step": 4902
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3436046064833237e-05,
+      "loss": 0.4061,
+      "step": 4903
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.343352279458066e-05,
+      "loss": 0.812,
+      "step": 4904
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.343099927647874e-05,
+      "loss": 0.8087,
+      "step": 4905
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.342847551070963e-05,
+      "loss": 0.8278,
+      "step": 4906
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3425951497455514e-05,
+      "loss": 0.7736,
+      "step": 4907
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3423427236898583e-05,
+      "loss": 0.7718,
+      "step": 4908
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3420902729221057e-05,
+      "loss": 0.8911,
+      "step": 4909
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3418377974605168e-05,
+      "loss": 0.775,
+      "step": 4910
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.341585297323316e-05,
+      "loss": 0.8696,
+      "step": 4911
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3413327725287309e-05,
+      "loss": 0.3999,
+      "step": 4912
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.341080223094989e-05,
+      "loss": 0.8486,
+      "step": 4913
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3408276490403216e-05,
+      "loss": 0.8342,
+      "step": 4914
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3405750503829603e-05,
+      "loss": 0.8489,
+      "step": 4915
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3403224271411386e-05,
+      "loss": 0.773,
+      "step": 4916
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3400697793330928e-05,
+      "loss": 0.7765,
+      "step": 4917
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.33981710697706e-05,
+      "loss": 0.8523,
+      "step": 4918
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3395644100912796e-05,
+      "loss": 0.8173,
+      "step": 4919
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3393116886939925e-05,
+      "loss": 0.7416,
+      "step": 4920
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3390589428034414e-05,
+      "loss": 0.8823,
+      "step": 4921
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3388061724378707e-05,
+      "loss": 0.8044,
+      "step": 4922
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3385533776155268e-05,
+      "loss": 0.8154,
+      "step": 4923
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3383005583546575e-05,
+      "loss": 0.8488,
+      "step": 4924
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.338047714673513e-05,
+      "loss": 0.8444,
+      "step": 4925
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3377948465903445e-05,
+      "loss": 0.8222,
+      "step": 4926
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3375419541234054e-05,
+      "loss": 0.7773,
+      "step": 4927
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3372890372909509e-05,
+      "loss": 0.8024,
+      "step": 4928
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3370360961112377e-05,
+      "loss": 0.7818,
+      "step": 4929
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3367831306025242e-05,
+      "loss": 0.7556,
+      "step": 4930
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3365301407830716e-05,
+      "loss": 0.833,
+      "step": 4931
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.336277126671141e-05,
+      "loss": 0.7916,
+      "step": 4932
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3360240882849967e-05,
+      "loss": 0.7907,
+      "step": 4933
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3357710256429047e-05,
+      "loss": 0.8989,
+      "step": 4934
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3355179387631314e-05,
+      "loss": 0.8124,
+      "step": 4935
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3352648276639468e-05,
+      "loss": 0.826,
+      "step": 4936
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3350116923636213e-05,
+      "loss": 0.7915,
+      "step": 4937
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3347585328804278e-05,
+      "loss": 0.8014,
+      "step": 4938
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3345053492326403e-05,
+      "loss": 0.837,
+      "step": 4939
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.334252141438535e-05,
+      "loss": 0.8018,
+      "step": 4940
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3339989095163902e-05,
+      "loss": 0.7429,
+      "step": 4941
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3337456534844848e-05,
+      "loss": 0.865,
+      "step": 4942
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3334923733611005e-05,
+      "loss": 0.7964,
+      "step": 4943
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3332390691645208e-05,
+      "loss": 0.7954,
+      "step": 4944
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3329857409130293e-05,
+      "loss": 0.7887,
+      "step": 4945
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3327323886249134e-05,
+      "loss": 0.8208,
+      "step": 4946
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3324790123184617e-05,
+      "loss": 0.4191,
+      "step": 4947
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3322256120119632e-05,
+      "loss": 0.7643,
+      "step": 4948
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3319721877237103e-05,
+      "loss": 0.7183,
+      "step": 4949
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3317187394719964e-05,
+      "loss": 0.741,
+      "step": 4950
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3314652672751165e-05,
+      "loss": 0.7715,
+      "step": 4951
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3312117711513678e-05,
+      "loss": 0.7987,
+      "step": 4952
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3309582511190487e-05,
+      "loss": 0.8182,
+      "step": 4953
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3307047071964599e-05,
+      "loss": 0.7887,
+      "step": 4954
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.330451139401903e-05,
+      "loss": 0.8403,
+      "step": 4955
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3301975477536824e-05,
+      "loss": 0.779,
+      "step": 4956
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3299439322701033e-05,
+      "loss": 0.7669,
+      "step": 4957
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3296902929694728e-05,
+      "loss": 0.723,
+      "step": 4958
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3294366298701003e-05,
+      "loss": 0.7233,
+      "step": 4959
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3291829429902965e-05,
+      "loss": 0.7758,
+      "step": 4960
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3289292323483734e-05,
+      "loss": 0.8344,
+      "step": 4961
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3286754979626457e-05,
+      "loss": 0.8622,
+      "step": 4962
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3284217398514288e-05,
+      "loss": 0.7767,
+      "step": 4963
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3281679580330402e-05,
+      "loss": 0.8528,
+      "step": 4964
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3279141525257999e-05,
+      "loss": 0.8325,
+      "step": 4965
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.327660323348028e-05,
+      "loss": 0.8124,
+      "step": 4966
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.327406470518048e-05,
+      "loss": 0.7942,
+      "step": 4967
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3271525940541832e-05,
+      "loss": 0.7593,
+      "step": 4968
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3268986939747608e-05,
+      "loss": 0.8542,
+      "step": 4969
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3266447702981084e-05,
+      "loss": 0.7827,
+      "step": 4970
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3263908230425552e-05,
+      "loss": 0.7637,
+      "step": 4971
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3261368522264326e-05,
+      "loss": 0.8801,
+      "step": 4972
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3258828578680735e-05,
+      "loss": 0.7645,
+      "step": 4973
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3256288399858124e-05,
+      "loss": 0.8023,
+      "step": 4974
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.325374798597986e-05,
+      "loss": 0.4005,
+      "step": 4975
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3251207337229316e-05,
+      "loss": 0.7943,
+      "step": 4976
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3248666453789893e-05,
+      "loss": 0.8457,
+      "step": 4977
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3246125335845008e-05,
+      "loss": 0.7934,
+      "step": 4978
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3243583983578087e-05,
+      "loss": 0.8531,
+      "step": 4979
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3241042397172586e-05,
+      "loss": 0.8577,
+      "step": 4980
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.323850057681196e-05,
+      "loss": 0.8595,
+      "step": 4981
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3235958522679693e-05,
+      "loss": 0.7622,
+      "step": 4982
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3233416234959289e-05,
+      "loss": 0.8395,
+      "step": 4983
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3230873713834258e-05,
+      "loss": 0.7781,
+      "step": 4984
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3228330959488133e-05,
+      "loss": 0.7746,
+      "step": 4985
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3225787972104468e-05,
+      "loss": 0.841,
+      "step": 4986
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3223244751866821e-05,
+      "loss": 0.797,
+      "step": 4987
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3220701298958781e-05,
+      "loss": 0.8556,
+      "step": 4988
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3218157613563942e-05,
+      "loss": 0.83,
+      "step": 4989
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3215613695865925e-05,
+      "loss": 0.8265,
+      "step": 4990
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3213069546048366e-05,
+      "loss": 0.8211,
+      "step": 4991
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3210525164294905e-05,
+      "loss": 0.8138,
+      "step": 4992
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3207980550789217e-05,
+      "loss": 0.7745,
+      "step": 4993
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3205435705714981e-05,
+      "loss": 0.792,
+      "step": 4994
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.32028906292559e-05,
+      "loss": 0.8161,
+      "step": 4995
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3200345321595689e-05,
+      "loss": 0.7581,
+      "step": 4996
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3197799782918079e-05,
+      "loss": 0.7916,
+      "step": 4997
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3195254013406824e-05,
+      "loss": 0.7013,
+      "step": 4998
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3192708013245688e-05,
+      "loss": 0.8074,
+      "step": 4999
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3190161782618456e-05,
+      "loss": 0.7719,
+      "step": 5000
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.318761532170893e-05,
+      "loss": 0.87,
+      "step": 5001
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3185068630700923e-05,
+      "loss": 0.7615,
+      "step": 5002
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3182521709778269e-05,
+      "loss": 0.7808,
+      "step": 5003
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3179974559124822e-05,
+      "loss": 0.8011,
+      "step": 5004
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.317742717892444e-05,
+      "loss": 0.7479,
+      "step": 5005
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3174879569361013e-05,
+      "loss": 0.8278,
+      "step": 5006
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3172331730618439e-05,
+      "loss": 0.8532,
+      "step": 5007
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3169783662880633e-05,
+      "loss": 0.7316,
+      "step": 5008
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3167235366331528e-05,
+      "loss": 0.713,
+      "step": 5009
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3164686841155075e-05,
+      "loss": 0.7605,
+      "step": 5010
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3162138087535234e-05,
+      "loss": 0.8007,
+      "step": 5011
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3159589105655996e-05,
+      "loss": 0.8345,
+      "step": 5012
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3157039895701354e-05,
+      "loss": 0.4332,
+      "step": 5013
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3154490457855325e-05,
+      "loss": 0.8837,
+      "step": 5014
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3151940792301937e-05,
+      "loss": 0.8298,
+      "step": 5015
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3149390899225241e-05,
+      "loss": 0.8612,
+      "step": 5016
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.31468407788093e-05,
+      "loss": 0.832,
+      "step": 5017
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3144290431238199e-05,
+      "loss": 0.8523,
+      "step": 5018
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3141739856696028e-05,
+      "loss": 0.8295,
+      "step": 5019
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3139189055366906e-05,
+      "loss": 0.8546,
+      "step": 5020
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.313663802743496e-05,
+      "loss": 0.8033,
+      "step": 5021
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3134086773084342e-05,
+      "loss": 0.826,
+      "step": 5022
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3131535292499207e-05,
+      "loss": 0.8246,
+      "step": 5023
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3128983585863739e-05,
+      "loss": 0.8369,
+      "step": 5024
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3126431653362133e-05,
+      "loss": 0.7844,
+      "step": 5025
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3123879495178597e-05,
+      "loss": 0.7702,
+      "step": 5026
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3121327111497366e-05,
+      "loss": 0.7962,
+      "step": 5027
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3118774502502673e-05,
+      "loss": 0.8324,
+      "step": 5028
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.311622166837879e-05,
+      "loss": 0.8998,
+      "step": 5029
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3113668609309986e-05,
+      "loss": 0.8015,
+      "step": 5030
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3111115325480557e-05,
+      "loss": 0.7509,
+      "step": 5031
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.310856181707481e-05,
+      "loss": 0.8318,
+      "step": 5032
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3106008084277076e-05,
+      "loss": 0.7644,
+      "step": 5033
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3103454127271691e-05,
+      "loss": 0.8744,
+      "step": 5034
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3100899946243017e-05,
+      "loss": 0.8005,
+      "step": 5035
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.309834554137542e-05,
+      "loss": 0.8058,
+      "step": 5036
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3095790912853301e-05,
+      "loss": 0.8554,
+      "step": 5037
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.309323606086106e-05,
+      "loss": 0.7827,
+      "step": 5038
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3090680985583118e-05,
+      "loss": 0.82,
+      "step": 5039
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3088125687203916e-05,
+      "loss": 0.803,
+      "step": 5040
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3085570165907908e-05,
+      "loss": 0.8217,
+      "step": 5041
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3083014421879565e-05,
+      "loss": 0.7305,
+      "step": 5042
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3080458455303375e-05,
+      "loss": 0.8445,
+      "step": 5043
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.307790226636384e-05,
+      "loss": 0.8778,
+      "step": 5044
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3075345855245475e-05,
+      "loss": 0.76,
+      "step": 5045
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3072789222132823e-05,
+      "loss": 0.8092,
+      "step": 5046
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3070232367210428e-05,
+      "loss": 0.8269,
+      "step": 5047
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.306767529066286e-05,
+      "loss": 0.8362,
+      "step": 5048
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3065117992674699e-05,
+      "loss": 0.8238,
+      "step": 5049
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3062560473430548e-05,
+      "loss": 0.7751,
+      "step": 5050
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3060002733115016e-05,
+      "loss": 0.8279,
+      "step": 5051
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3057444771912741e-05,
+      "loss": 0.8039,
+      "step": 5052
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3054886590008366e-05,
+      "loss": 0.7536,
+      "step": 5053
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3052328187586556e-05,
+      "loss": 0.7948,
+      "step": 5054
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3049769564831986e-05,
+      "loss": 0.7676,
+      "step": 5055
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3047210721929356e-05,
+      "loss": 0.8286,
+      "step": 5056
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.304465165906337e-05,
+      "loss": 0.7908,
+      "step": 5057
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3042092376418756e-05,
+      "loss": 0.8682,
+      "step": 5058
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.303953287418026e-05,
+      "loss": 0.7663,
+      "step": 5059
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3036973152532636e-05,
+      "loss": 0.7479,
+      "step": 5060
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3034413211660661e-05,
+      "loss": 0.7691,
+      "step": 5061
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3031853051749122e-05,
+      "loss": 0.8063,
+      "step": 5062
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3029292672982825e-05,
+      "loss": 0.8152,
+      "step": 5063
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3026732075546595e-05,
+      "loss": 0.8465,
+      "step": 5064
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3024171259625265e-05,
+      "loss": 0.8207,
+      "step": 5065
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3021610225403687e-05,
+      "loss": 0.7857,
+      "step": 5066
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3019048973066735e-05,
+      "loss": 0.8752,
+      "step": 5067
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3016487502799287e-05,
+      "loss": 0.7657,
+      "step": 5068
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3013925814786248e-05,
+      "loss": 0.7714,
+      "step": 5069
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3011363909212529e-05,
+      "loss": 0.829,
+      "step": 5070
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3008801786263066e-05,
+      "loss": 0.8443,
+      "step": 5071
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3006239446122806e-05,
+      "loss": 0.8029,
+      "step": 5072
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.300367688897671e-05,
+      "loss": 0.7337,
+      "step": 5073
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3001114115009756e-05,
+      "loss": 0.7962,
+      "step": 5074
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.299855112440694e-05,
+      "loss": 0.8341,
+      "step": 5075
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.299598791735327e-05,
+      "loss": 0.8508,
+      "step": 5076
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2993424494033775e-05,
+      "loss": 0.7949,
+      "step": 5077
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2990860854633494e-05,
+      "loss": 0.834,
+      "step": 5078
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.298829699933748e-05,
+      "loss": 0.8134,
+      "step": 5079
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2985732928330811e-05,
+      "loss": 0.8442,
+      "step": 5080
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2983168641798574e-05,
+      "loss": 0.7607,
+      "step": 5081
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2980604139925867e-05,
+      "loss": 0.7767,
+      "step": 5082
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2978039422897816e-05,
+      "loss": 0.7856,
+      "step": 5083
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.297547449089955e-05,
+      "loss": 0.7874,
+      "step": 5084
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2972909344116227e-05,
+      "loss": 0.3883,
+      "step": 5085
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2970343982733e-05,
+      "loss": 0.3991,
+      "step": 5086
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2967778406935061e-05,
+      "loss": 0.7765,
+      "step": 5087
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2965212616907601e-05,
+      "loss": 0.8754,
+      "step": 5088
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2962646612835835e-05,
+      "loss": 0.8027,
+      "step": 5089
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2960080394904987e-05,
+      "loss": 0.819,
+      "step": 5090
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2957513963300301e-05,
+      "loss": 0.8126,
+      "step": 5091
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2954947318207035e-05,
+      "loss": 0.9157,
+      "step": 5092
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2952380459810465e-05,
+      "loss": 0.8193,
+      "step": 5093
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2949813388295876e-05,
+      "loss": 0.8363,
+      "step": 5094
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2947246103848578e-05,
+      "loss": 0.7608,
+      "step": 5095
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2944678606653887e-05,
+      "loss": 0.8801,
+      "step": 5096
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2942110896897138e-05,
+      "loss": 0.8694,
+      "step": 5097
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2939542974763683e-05,
+      "loss": 0.7961,
+      "step": 5098
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2936974840438884e-05,
+      "loss": 0.7727,
+      "step": 5099
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.293440649410813e-05,
+      "loss": 0.8769,
+      "step": 5100
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2931837935956812e-05,
+      "loss": 0.8368,
+      "step": 5101
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.292926916617034e-05,
+      "loss": 0.7635,
+      "step": 5102
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2926700184934144e-05,
+      "loss": 0.7038,
+      "step": 5103
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2924130992433665e-05,
+      "loss": 0.7723,
+      "step": 5104
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2921561588854364e-05,
+      "loss": 0.8404,
+      "step": 5105
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.291899197438171e-05,
+      "loss": 0.8029,
+      "step": 5106
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2916422149201192e-05,
+      "loss": 0.7589,
+      "step": 5107
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2913852113498316e-05,
+      "loss": 0.8001,
+      "step": 5108
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2911281867458594e-05,
+      "loss": 0.8261,
+      "step": 5109
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2908711411267563e-05,
+      "loss": 0.795,
+      "step": 5110
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2906140745110775e-05,
+      "loss": 0.7886,
+      "step": 5111
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2903569869173789e-05,
+      "loss": 0.8234,
+      "step": 5112
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2900998783642184e-05,
+      "loss": 0.7432,
+      "step": 5113
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.289842748870156e-05,
+      "loss": 0.7946,
+      "step": 5114
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.289585598453752e-05,
+      "loss": 0.7838,
+      "step": 5115
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2893284271335694e-05,
+      "loss": 0.7469,
+      "step": 5116
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2890712349281716e-05,
+      "loss": 0.7583,
+      "step": 5117
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2888140218561242e-05,
+      "loss": 0.8032,
+      "step": 5118
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2885567879359947e-05,
+      "loss": 0.7781,
+      "step": 5119
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2882995331863505e-05,
+      "loss": 0.8642,
+      "step": 5120
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.288042257625763e-05,
+      "loss": 0.7894,
+      "step": 5121
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2877849612728021e-05,
+      "loss": 0.8367,
+      "step": 5122
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.287527644146042e-05,
+      "loss": 0.8027,
+      "step": 5123
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2872703062640566e-05,
+      "loss": 0.7914,
+      "step": 5124
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.287012947645422e-05,
+      "loss": 0.8045,
+      "step": 5125
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2867555683087156e-05,
+      "loss": 0.8365,
+      "step": 5126
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2864981682725167e-05,
+      "loss": 0.7917,
+      "step": 5127
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2862407475554051e-05,
+      "loss": 0.7508,
+      "step": 5128
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2859833061759637e-05,
+      "loss": 0.7271,
+      "step": 5129
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.285725844152775e-05,
+      "loss": 0.8781,
+      "step": 5130
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.285468361504424e-05,
+      "loss": 0.7486,
+      "step": 5131
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2852108582494982e-05,
+      "loss": 0.7833,
+      "step": 5132
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2849533344065843e-05,
+      "loss": 0.7624,
+      "step": 5133
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2846957899942723e-05,
+      "loss": 0.8219,
+      "step": 5134
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2844382250311527e-05,
+      "loss": 0.8199,
+      "step": 5135
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2841806395358182e-05,
+      "loss": 0.8031,
+      "step": 5136
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2839230335268628e-05,
+      "loss": 0.823,
+      "step": 5137
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2836654070228814e-05,
+      "loss": 0.7934,
+      "step": 5138
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2834077600424709e-05,
+      "loss": 0.7699,
+      "step": 5139
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2831500926042298e-05,
+      "loss": 0.8284,
+      "step": 5140
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2828924047267575e-05,
+      "loss": 0.8003,
+      "step": 5141
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2826346964286558e-05,
+      "loss": 0.8838,
+      "step": 5142
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2823769677285269e-05,
+      "loss": 0.8076,
+      "step": 5143
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.282119218644975e-05,
+      "loss": 0.786,
+      "step": 5144
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.281861449196606e-05,
+      "loss": 0.828,
+      "step": 5145
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.281603659402027e-05,
+      "loss": 0.8375,
+      "step": 5146
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2813458492798467e-05,
+      "loss": 0.7572,
+      "step": 5147
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2810880188486747e-05,
+      "loss": 0.8824,
+      "step": 5148
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2808301681271229e-05,
+      "loss": 0.7844,
+      "step": 5149
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2805722971338046e-05,
+      "loss": 0.7654,
+      "step": 5150
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2803144058873333e-05,
+      "loss": 0.7918,
+      "step": 5151
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2800564944063259e-05,
+      "loss": 0.7764,
+      "step": 5152
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2797985627093994e-05,
+      "loss": 0.8133,
+      "step": 5153
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.279540610815172e-05,
+      "loss": 0.8016,
+      "step": 5154
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2792826387422651e-05,
+      "loss": 0.7798,
+      "step": 5155
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2790246465092999e-05,
+      "loss": 0.709,
+      "step": 5156
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2787666341348997e-05,
+      "loss": 0.8769,
+      "step": 5157
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2785086016376891e-05,
+      "loss": 0.7185,
+      "step": 5158
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2782505490362942e-05,
+      "loss": 0.7919,
+      "step": 5159
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2779924763493425e-05,
+      "loss": 0.8229,
+      "step": 5160
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2777343835954632e-05,
+      "loss": 0.8541,
+      "step": 5161
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2774762707932866e-05,
+      "loss": 0.7747,
+      "step": 5162
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.277218137961445e-05,
+      "loss": 0.8261,
+      "step": 5163
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.276959985118571e-05,
+      "loss": 0.7881,
+      "step": 5164
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2767018122832999e-05,
+      "loss": 0.7675,
+      "step": 5165
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2764436194742683e-05,
+      "loss": 0.7453,
+      "step": 5166
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2761854067101132e-05,
+      "loss": 0.8417,
+      "step": 5167
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.275927174009474e-05,
+      "loss": 0.7913,
+      "step": 5168
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2756689213909912e-05,
+      "loss": 0.77,
+      "step": 5169
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2754106488733069e-05,
+      "loss": 0.825,
+      "step": 5170
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2751523564750647e-05,
+      "loss": 0.7795,
+      "step": 5171
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2748940442149091e-05,
+      "loss": 0.7902,
+      "step": 5172
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2746357121114866e-05,
+      "loss": 0.842,
+      "step": 5173
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2743773601834449e-05,
+      "loss": 0.7931,
+      "step": 5174
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2741189884494332e-05,
+      "loss": 0.8306,
+      "step": 5175
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2738605969281024e-05,
+      "loss": 0.7659,
+      "step": 5176
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2736021856381037e-05,
+      "loss": 0.8075,
+      "step": 5177
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2733437545980916e-05,
+      "loss": 0.8255,
+      "step": 5178
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2730853038267204e-05,
+      "loss": 0.8472,
+      "step": 5179
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2728268333426463e-05,
+      "loss": 0.7785,
+      "step": 5180
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2725683431645272e-05,
+      "loss": 0.7746,
+      "step": 5181
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2723098333110224e-05,
+      "loss": 0.7814,
+      "step": 5182
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2720513038007922e-05,
+      "loss": 0.8128,
+      "step": 5183
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2717927546524988e-05,
+      "loss": 0.8219,
+      "step": 5184
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2715341858848055e-05,
+      "loss": 0.8413,
+      "step": 5185
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.271275597516377e-05,
+      "loss": 0.7556,
+      "step": 5186
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2710169895658799e-05,
+      "loss": 0.8297,
+      "step": 5187
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2707583620519817e-05,
+      "loss": 0.7462,
+      "step": 5188
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2704997149933515e-05,
+      "loss": 0.8199,
+      "step": 5189
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2702410484086593e-05,
+      "loss": 0.7494,
+      "step": 5190
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2699823623165776e-05,
+      "loss": 0.8012,
+      "step": 5191
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2697236567357797e-05,
+      "loss": 0.7399,
+      "step": 5192
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2694649316849399e-05,
+      "loss": 0.7728,
+      "step": 5193
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2692061871827343e-05,
+      "loss": 0.7392,
+      "step": 5194
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2689474232478408e-05,
+      "loss": 0.8435,
+      "step": 5195
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.268688639898938e-05,
+      "loss": 0.8368,
+      "step": 5196
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2684298371547063e-05,
+      "loss": 0.8188,
+      "step": 5197
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2681710150338273e-05,
+      "loss": 0.8047,
+      "step": 5198
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2679121735549845e-05,
+      "loss": 0.877,
+      "step": 5199
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2676533127368622e-05,
+      "loss": 0.7935,
+      "step": 5200
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.267394432598146e-05,
+      "loss": 0.8659,
+      "step": 5201
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2671355331575237e-05,
+      "loss": 0.7364,
+      "step": 5202
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2668766144336835e-05,
+      "loss": 0.797,
+      "step": 5203
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2666176764453157e-05,
+      "loss": 0.8632,
+      "step": 5204
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.266358719211112e-05,
+      "loss": 0.8034,
+      "step": 5205
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.266099742749765e-05,
+      "loss": 0.8121,
+      "step": 5206
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2658407470799692e-05,
+      "loss": 0.7906,
+      "step": 5207
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2655817322204198e-05,
+      "loss": 0.7977,
+      "step": 5208
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2653226981898144e-05,
+      "loss": 0.8598,
+      "step": 5209
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.265063645006851e-05,
+      "loss": 0.7464,
+      "step": 5210
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2648045726902294e-05,
+      "loss": 0.7982,
+      "step": 5211
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2645454812586508e-05,
+      "loss": 0.7916,
+      "step": 5212
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2642863707308179e-05,
+      "loss": 0.7724,
+      "step": 5213
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2640272411254347e-05,
+      "loss": 0.8309,
+      "step": 5214
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2637680924612061e-05,
+      "loss": 0.8578,
+      "step": 5215
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2635089247568393e-05,
+      "loss": 0.7842,
+      "step": 5216
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.263249738031042e-05,
+      "loss": 0.7929,
+      "step": 5217
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2629905323025238e-05,
+      "loss": 0.8175,
+      "step": 5218
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2627313075899953e-05,
+      "loss": 0.8767,
+      "step": 5219
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2624720639121693e-05,
+      "loss": 0.828,
+      "step": 5220
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2622128012877583e-05,
+      "loss": 0.8052,
+      "step": 5221
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.261953519735478e-05,
+      "loss": 0.8836,
+      "step": 5222
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2616942192740447e-05,
+      "loss": 0.8235,
+      "step": 5223
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2614348999221752e-05,
+      "loss": 0.829,
+      "step": 5224
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2611755616985894e-05,
+      "loss": 0.8101,
+      "step": 5225
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2609162046220074e-05,
+      "loss": 0.8661,
+      "step": 5226
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2606568287111507e-05,
+      "loss": 0.8026,
+      "step": 5227
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2603974339847429e-05,
+      "loss": 0.7972,
+      "step": 5228
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2601380204615079e-05,
+      "loss": 0.7526,
+      "step": 5229
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2598785881601718e-05,
+      "loss": 0.7832,
+      "step": 5230
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2596191370994619e-05,
+      "loss": 0.8571,
+      "step": 5231
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2593596672981061e-05,
+      "loss": 0.7569,
+      "step": 5232
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2591001787748352e-05,
+      "loss": 0.7981,
+      "step": 5233
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2588406715483793e-05,
+      "loss": 0.8009,
+      "step": 5234
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2585811456374719e-05,
+      "loss": 0.813,
+      "step": 5235
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2583216010608463e-05,
+      "loss": 0.7825,
+      "step": 5236
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2580620378372381e-05,
+      "loss": 0.8884,
+      "step": 5237
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2578024559853841e-05,
+      "loss": 0.8514,
+      "step": 5238
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2575428555240219e-05,
+      "loss": 0.8205,
+      "step": 5239
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2572832364718908e-05,
+      "loss": 0.773,
+      "step": 5240
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2570235988477317e-05,
+      "loss": 0.8505,
+      "step": 5241
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2567639426702862e-05,
+      "loss": 0.7864,
+      "step": 5242
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.256504267958298e-05,
+      "loss": 0.8275,
+      "step": 5243
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2562445747305114e-05,
+      "loss": 0.7801,
+      "step": 5244
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2559848630056728e-05,
+      "loss": 0.8102,
+      "step": 5245
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2557251328025291e-05,
+      "loss": 0.7868,
+      "step": 5246
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2554653841398293e-05,
+      "loss": 0.7394,
+      "step": 5247
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2552056170363231e-05,
+      "loss": 0.7951,
+      "step": 5248
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2549458315107623e-05,
+      "loss": 0.8053,
+      "step": 5249
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.254686027581899e-05,
+      "loss": 0.8261,
+      "step": 5250
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2544262052684874e-05,
+      "loss": 0.8254,
+      "step": 5251
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2541663645892832e-05,
+      "loss": 0.7712,
+      "step": 5252
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2539065055630425e-05,
+      "loss": 0.7599,
+      "step": 5253
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2536466282085234e-05,
+      "loss": 0.8881,
+      "step": 5254
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2533867325444851e-05,
+      "loss": 0.7687,
+      "step": 5255
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2531268185896885e-05,
+      "loss": 0.8216,
+      "step": 5256
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2528668863628952e-05,
+      "loss": 0.7914,
+      "step": 5257
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2526069358828688e-05,
+      "loss": 0.8417,
+      "step": 5258
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2523469671683734e-05,
+      "loss": 0.7196,
+      "step": 5259
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2520869802381757e-05,
+      "loss": 0.8124,
+      "step": 5260
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2518269751110419e-05,
+      "loss": 0.8164,
+      "step": 5261
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2515669518057414e-05,
+      "loss": 0.7908,
+      "step": 5262
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2513069103410431e-05,
+      "loss": 0.8943,
+      "step": 5263
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2510468507357188e-05,
+      "loss": 0.7155,
+      "step": 5264
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2507867730085408e-05,
+      "loss": 0.806,
+      "step": 5265
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2505266771782827e-05,
+      "loss": 0.776,
+      "step": 5266
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2502665632637196e-05,
+      "loss": 0.8078,
+      "step": 5267
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.250006431283628e-05,
+      "loss": 0.9147,
+      "step": 5268
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2497462812567854e-05,
+      "loss": 0.8314,
+      "step": 5269
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2494861132019712e-05,
+      "loss": 0.821,
+      "step": 5270
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2492259271379648e-05,
+      "loss": 0.7511,
+      "step": 5271
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2489657230835486e-05,
+      "loss": 0.7548,
+      "step": 5272
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.248705501057505e-05,
+      "loss": 0.8363,
+      "step": 5273
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2484452610786186e-05,
+      "loss": 0.442,
+      "step": 5274
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2481850031656747e-05,
+      "loss": 0.8212,
+      "step": 5275
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2479247273374596e-05,
+      "loss": 0.7751,
+      "step": 5276
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2476644336127614e-05,
+      "loss": 0.7316,
+      "step": 5277
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.24740412201037e-05,
+      "loss": 0.4312,
+      "step": 5278
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.247143792549076e-05,
+      "loss": 0.8559,
+      "step": 5279
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2468834452476708e-05,
+      "loss": 0.8505,
+      "step": 5280
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.246623080124948e-05,
+      "loss": 0.7667,
+      "step": 5281
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2463626971997019e-05,
+      "loss": 0.7875,
+      "step": 5282
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2461022964907285e-05,
+      "loss": 0.7542,
+      "step": 5283
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2458418780168246e-05,
+      "loss": 0.8354,
+      "step": 5284
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2455814417967886e-05,
+      "loss": 0.7816,
+      "step": 5285
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2453209878494207e-05,
+      "loss": 0.8851,
+      "step": 5286
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2450605161935208e-05,
+      "loss": 0.8133,
+      "step": 5287
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2448000268478917e-05,
+      "loss": 0.7374,
+      "step": 5288
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2445395198313367e-05,
+      "loss": 0.7541,
+      "step": 5289
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2442789951626606e-05,
+      "loss": 0.842,
+      "step": 5290
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2440184528606696e-05,
+      "loss": 0.7835,
+      "step": 5291
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2437578929441705e-05,
+      "loss": 0.7997,
+      "step": 5292
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2434973154319723e-05,
+      "loss": 0.7654,
+      "step": 5293
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2432367203428846e-05,
+      "loss": 0.7987,
+      "step": 5294
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2429761076957186e-05,
+      "loss": 0.8301,
+      "step": 5295
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2427154775092866e-05,
+      "loss": 0.7675,
+      "step": 5296
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2424548298024023e-05,
+      "loss": 0.7513,
+      "step": 5297
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2421941645938804e-05,
+      "loss": 0.8087,
+      "step": 5298
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2419334819025372e-05,
+      "loss": 0.7679,
+      "step": 5299
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2416727817471904e-05,
+      "loss": 0.4036,
+      "step": 5300
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2414120641466582e-05,
+      "loss": 0.8592,
+      "step": 5301
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2411513291197608e-05,
+      "loss": 0.8479,
+      "step": 5302
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2408905766853192e-05,
+      "loss": 0.823,
+      "step": 5303
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2406298068621562e-05,
+      "loss": 0.8193,
+      "step": 5304
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2403690196690951e-05,
+      "loss": 0.7712,
+      "step": 5305
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2401082151249611e-05,
+      "loss": 0.4176,
+      "step": 5306
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2398473932485806e-05,
+      "loss": 0.7686,
+      "step": 5307
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2395865540587803e-05,
+      "loss": 0.7397,
+      "step": 5308
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.23932569757439e-05,
+      "loss": 0.6922,
+      "step": 5309
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2390648238142386e-05,
+      "loss": 0.7848,
+      "step": 5310
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2388039327971581e-05,
+      "loss": 0.7441,
+      "step": 5311
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.238543024541981e-05,
+      "loss": 0.8212,
+      "step": 5312
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.23828209906754e-05,
+      "loss": 0.7842,
+      "step": 5313
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2380211563926714e-05,
+      "loss": 0.7644,
+      "step": 5314
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2377601965362102e-05,
+      "loss": 0.8434,
+      "step": 5315
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2374992195169946e-05,
+      "loss": 0.7835,
+      "step": 5316
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2372382253538631e-05,
+      "loss": 0.8049,
+      "step": 5317
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2369772140656551e-05,
+      "loss": 0.813,
+      "step": 5318
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2367161856712124e-05,
+      "loss": 0.8241,
+      "step": 5319
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.236455140189377e-05,
+      "loss": 0.6903,
+      "step": 5320
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2361940776389927e-05,
+      "loss": 0.7807,
+      "step": 5321
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2359329980389043e-05,
+      "loss": 0.7645,
+      "step": 5322
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2356719014079577e-05,
+      "loss": 0.8049,
+      "step": 5323
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2354107877650003e-05,
+      "loss": 0.71,
+      "step": 5324
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2351496571288811e-05,
+      "loss": 0.8059,
+      "step": 5325
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.234888509518449e-05,
+      "loss": 0.8529,
+      "step": 5326
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2346273449525554e-05,
+      "loss": 0.863,
+      "step": 5327
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2343661634500524e-05,
+      "loss": 0.7831,
+      "step": 5328
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2341049650297936e-05,
+      "loss": 0.8215,
+      "step": 5329
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2338437497106335e-05,
+      "loss": 0.4044,
+      "step": 5330
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.233582517511428e-05,
+      "loss": 0.7553,
+      "step": 5331
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2333212684510342e-05,
+      "loss": 0.811,
+      "step": 5332
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2330600025483108e-05,
+      "loss": 0.8079,
+      "step": 5333
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2327987198221165e-05,
+      "loss": 0.8354,
+      "step": 5334
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2325374202913127e-05,
+      "loss": 0.829,
+      "step": 5335
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.232276103974761e-05,
+      "loss": 0.7651,
+      "step": 5336
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2320147708913245e-05,
+      "loss": 0.8256,
+      "step": 5337
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2317534210598682e-05,
+      "loss": 0.7927,
+      "step": 5338
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2314920544992568e-05,
+      "loss": 0.7983,
+      "step": 5339
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2312306712283575e-05,
+      "loss": 0.8026,
+      "step": 5340
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2309692712660384e-05,
+      "loss": 0.7893,
+      "step": 5341
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2307078546311685e-05,
+      "loss": 0.7062,
+      "step": 5342
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2304464213426184e-05,
+      "loss": 0.8077,
+      "step": 5343
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2301849714192593e-05,
+      "loss": 0.7758,
+      "step": 5344
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2299235048799645e-05,
+      "loss": 0.8346,
+      "step": 5345
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2296620217436076e-05,
+      "loss": 0.7951,
+      "step": 5346
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2294005220290642e-05,
+      "loss": 0.8158,
+      "step": 5347
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2291390057552104e-05,
+      "loss": 0.7162,
+      "step": 5348
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2288774729409236e-05,
+      "loss": 0.8353,
+      "step": 5349
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2286159236050829e-05,
+      "loss": 0.8016,
+      "step": 5350
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2283543577665683e-05,
+      "loss": 0.7863,
+      "step": 5351
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.228092775444261e-05,
+      "loss": 0.7947,
+      "step": 5352
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2278311766570434e-05,
+      "loss": 0.8137,
+      "step": 5353
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2275695614237986e-05,
+      "loss": 0.8737,
+      "step": 5354
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2273079297634117e-05,
+      "loss": 0.7983,
+      "step": 5355
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2270462816947685e-05,
+      "loss": 0.8221,
+      "step": 5356
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2267846172367563e-05,
+      "loss": 0.7614,
+      "step": 5357
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2265229364082632e-05,
+      "loss": 0.8054,
+      "step": 5358
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2262612392281783e-05,
+      "loss": 0.7427,
+      "step": 5359
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.225999525715393e-05,
+      "loss": 0.8917,
+      "step": 5360
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.225737795888799e-05,
+      "loss": 0.8428,
+      "step": 5361
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2254760497672886e-05,
+      "loss": 0.7968,
+      "step": 5362
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2252142873697567e-05,
+      "loss": 0.7477,
+      "step": 5363
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2249525087150986e-05,
+      "loss": 0.4202,
+      "step": 5364
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2246907138222106e-05,
+      "loss": 0.8345,
+      "step": 5365
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2244289027099903e-05,
+      "loss": 0.7802,
+      "step": 5366
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2241670753973369e-05,
+      "loss": 0.7545,
+      "step": 5367
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2239052319031502e-05,
+      "loss": 0.7634,
+      "step": 5368
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2236433722463312e-05,
+      "loss": 0.7958,
+      "step": 5369
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2233814964457829e-05,
+      "loss": 0.849,
+      "step": 5370
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2231196045204083e-05,
+      "loss": 0.388,
+      "step": 5371
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2228576964891125e-05,
+      "loss": 0.8961,
+      "step": 5372
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2225957723708013e-05,
+      "loss": 0.8147,
+      "step": 5373
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2223338321843817e-05,
+      "loss": 0.8082,
+      "step": 5374
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2220718759487614e-05,
+      "loss": 0.8825,
+      "step": 5375
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2218099036828503e-05,
+      "loss": 0.7634,
+      "step": 5376
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2215479154055589e-05,
+      "loss": 0.7812,
+      "step": 5377
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2212859111357984e-05,
+      "loss": 0.8218,
+      "step": 5378
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2210238908924822e-05,
+      "loss": 0.7632,
+      "step": 5379
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.220761854694524e-05,
+      "loss": 0.7695,
+      "step": 5380
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2204998025608388e-05,
+      "loss": 0.7903,
+      "step": 5381
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2202377345103433e-05,
+      "loss": 0.8878,
+      "step": 5382
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2199756505619543e-05,
+      "loss": 0.7941,
+      "step": 5383
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2197135507345908e-05,
+      "loss": 0.7639,
+      "step": 5384
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2194514350471725e-05,
+      "loss": 0.7919,
+      "step": 5385
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2191893035186203e-05,
+      "loss": 0.7213,
+      "step": 5386
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2189271561678558e-05,
+      "loss": 0.7703,
+      "step": 5387
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2186649930138023e-05,
+      "loss": 0.7086,
+      "step": 5388
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2184028140753846e-05,
+      "loss": 0.8066,
+      "step": 5389
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2181406193715273e-05,
+      "loss": 0.753,
+      "step": 5390
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2178784089211576e-05,
+      "loss": 0.8306,
+      "step": 5391
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.217616182743203e-05,
+      "loss": 0.8107,
+      "step": 5392
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2173539408565925e-05,
+      "loss": 0.7633,
+      "step": 5393
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2170916832802554e-05,
+      "loss": 0.8331,
+      "step": 5394
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.216829410033124e-05,
+      "loss": 0.7965,
+      "step": 5395
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2165671211341292e-05,
+      "loss": 0.7288,
+      "step": 5396
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.216304816602205e-05,
+      "loss": 0.8398,
+      "step": 5397
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2160424964562865e-05,
+      "loss": 0.7508,
+      "step": 5398
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2157801607153081e-05,
+      "loss": 0.8131,
+      "step": 5399
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2155178093982075e-05,
+      "loss": 0.7774,
+      "step": 5400
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2152554425239219e-05,
+      "loss": 0.7786,
+      "step": 5401
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2149930601113906e-05,
+      "loss": 0.8197,
+      "step": 5402
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2147306621795541e-05,
+      "loss": 0.8251,
+      "step": 5403
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.214468248747353e-05,
+      "loss": 0.7936,
+      "step": 5404
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2142058198337298e-05,
+      "loss": 0.78,
+      "step": 5405
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2139433754576284e-05,
+      "loss": 0.8485,
+      "step": 5406
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2136809156379927e-05,
+      "loss": 0.7767,
+      "step": 5407
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2134184403937693e-05,
+      "loss": 0.7965,
+      "step": 5408
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.213155949743904e-05,
+      "loss": 0.7682,
+      "step": 5409
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2128934437073452e-05,
+      "loss": 0.7325,
+      "step": 5410
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.212630922303042e-05,
+      "loss": 0.3938,
+      "step": 5411
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2123683855499444e-05,
+      "loss": 0.6972,
+      "step": 5412
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.212105833467004e-05,
+      "loss": 0.8292,
+      "step": 5413
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2118432660731727e-05,
+      "loss": 0.8333,
+      "step": 5414
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2115806833874041e-05,
+      "loss": 0.8232,
+      "step": 5415
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.211318085428653e-05,
+      "loss": 0.8196,
+      "step": 5416
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2110554722158749e-05,
+      "loss": 0.8066,
+      "step": 5417
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2107928437680263e-05,
+      "loss": 0.7376,
+      "step": 5418
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2105302001040657e-05,
+      "loss": 0.8015,
+      "step": 5419
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2102675412429517e-05,
+      "loss": 0.8711,
+      "step": 5420
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2100048672036442e-05,
+      "loss": 0.8233,
+      "step": 5421
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2097421780051043e-05,
+      "loss": 0.7614,
+      "step": 5422
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.209479473666295e-05,
+      "loss": 0.8403,
+      "step": 5423
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.209216754206179e-05,
+      "loss": 0.8141,
+      "step": 5424
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2089540196437208e-05,
+      "loss": 0.7922,
+      "step": 5425
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2086912699978863e-05,
+      "loss": 0.8738,
+      "step": 5426
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2084285052876417e-05,
+      "loss": 0.8013,
+      "step": 5427
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2081657255319548e-05,
+      "loss": 0.8628,
+      "step": 5428
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2079029307497947e-05,
+      "loss": 0.8707,
+      "step": 5429
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.207640120960131e-05,
+      "loss": 0.798,
+      "step": 5430
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2073772961819343e-05,
+      "loss": 0.7662,
+      "step": 5431
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2071144564341775e-05,
+      "loss": 0.7539,
+      "step": 5432
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.206851601735833e-05,
+      "loss": 0.7615,
+      "step": 5433
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2065887321058755e-05,
+      "loss": 0.8148,
+      "step": 5434
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.20632584756328e-05,
+      "loss": 0.8144,
+      "step": 5435
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2060629481270228e-05,
+      "loss": 0.8444,
+      "step": 5436
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2058000338160815e-05,
+      "loss": 0.8114,
+      "step": 5437
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2055371046494347e-05,
+      "loss": 0.8267,
+      "step": 5438
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2052741606460616e-05,
+      "loss": 0.7201,
+      "step": 5439
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2050112018249436e-05,
+      "loss": 0.7922,
+      "step": 5440
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2047482282050615e-05,
+      "loss": 0.7588,
+      "step": 5441
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2044852398053985e-05,
+      "loss": 0.833,
+      "step": 5442
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2042222366449385e-05,
+      "loss": 0.8114,
+      "step": 5443
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2039592187426665e-05,
+      "loss": 0.7893,
+      "step": 5444
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2036961861175684e-05,
+      "loss": 0.8263,
+      "step": 5445
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2034331387886311e-05,
+      "loss": 0.7592,
+      "step": 5446
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2031700767748431e-05,
+      "loss": 0.8407,
+      "step": 5447
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.202907000095193e-05,
+      "loss": 0.7641,
+      "step": 5448
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2026439087686715e-05,
+      "loss": 0.7832,
+      "step": 5449
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2023808028142695e-05,
+      "loss": 0.7906,
+      "step": 5450
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2021176822509797e-05,
+      "loss": 0.7575,
+      "step": 5451
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2018545470977954e-05,
+      "loss": 0.8306,
+      "step": 5452
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.201591397373711e-05,
+      "loss": 0.8418,
+      "step": 5453
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.201328233097722e-05,
+      "loss": 0.769,
+      "step": 5454
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2010650542888249e-05,
+      "loss": 0.8478,
+      "step": 5455
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2008018609660173e-05,
+      "loss": 0.7524,
+      "step": 5456
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2005386531482978e-05,
+      "loss": 0.8077,
+      "step": 5457
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2002754308546666e-05,
+      "loss": 0.8022,
+      "step": 5458
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2000121941041239e-05,
+      "loss": 0.8051,
+      "step": 5459
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1997489429156717e-05,
+      "loss": 0.8369,
+      "step": 5460
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1994856773083126e-05,
+      "loss": 0.8074,
+      "step": 5461
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1992223973010506e-05,
+      "loss": 0.4192,
+      "step": 5462
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1989591029128906e-05,
+      "loss": 0.7935,
+      "step": 5463
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1986957941628385e-05,
+      "loss": 0.768,
+      "step": 5464
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1984324710699015e-05,
+      "loss": 0.829,
+      "step": 5465
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1981691336530877e-05,
+      "loss": 0.8017,
+      "step": 5466
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1979057819314054e-05,
+      "loss": 0.8004,
+      "step": 5467
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1976424159238657e-05,
+      "loss": 0.8376,
+      "step": 5468
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.197379035649479e-05,
+      "loss": 0.4011,
+      "step": 5469
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1971156411272575e-05,
+      "loss": 0.7839,
+      "step": 5470
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1968522323762148e-05,
+      "loss": 0.7608,
+      "step": 5471
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1965888094153645e-05,
+      "loss": 0.7729,
+      "step": 5472
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1963253722637225e-05,
+      "loss": 0.8243,
+      "step": 5473
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1960619209403044e-05,
+      "loss": 0.7605,
+      "step": 5474
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.195798455464128e-05,
+      "loss": 0.8231,
+      "step": 5475
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1955349758542114e-05,
+      "loss": 0.8139,
+      "step": 5476
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1952714821295736e-05,
+      "loss": 0.8472,
+      "step": 5477
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1950079743092354e-05,
+      "loss": 0.8065,
+      "step": 5478
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.194744452412218e-05,
+      "loss": 0.828,
+      "step": 5479
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1944809164575438e-05,
+      "loss": 0.8155,
+      "step": 5480
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1942173664642361e-05,
+      "loss": 0.7956,
+      "step": 5481
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1939538024513191e-05,
+      "loss": 0.8455,
+      "step": 5482
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1936902244378185e-05,
+      "loss": 0.8397,
+      "step": 5483
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.1934266324427606e-05,
+      "loss": 0.8841,
+      "step": 5484
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.193163026485173e-05,
+      "loss": 0.8068,
+      "step": 5485
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1928994065840837e-05,
+      "loss": 0.7595,
+      "step": 5486
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1926357727585229e-05,
+      "loss": 0.8161,
+      "step": 5487
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1923721250275202e-05,
+      "loss": 0.8069,
+      "step": 5488
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1921084634101078e-05,
+      "loss": 0.8049,
+      "step": 5489
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1918447879253174e-05,
+      "loss": 0.8402,
+      "step": 5490
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1915810985921832e-05,
+      "loss": 0.7548,
+      "step": 5491
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1913173954297392e-05,
+      "loss": 0.7461,
+      "step": 5492
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1910536784570208e-05,
+      "loss": 0.8165,
+      "step": 5493
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1907899476930645e-05,
+      "loss": 0.8208,
+      "step": 5494
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1905262031569081e-05,
+      "loss": 0.7838,
+      "step": 5495
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1902624448675894e-05,
+      "loss": 0.8229,
+      "step": 5496
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1899986728441487e-05,
+      "loss": 0.8493,
+      "step": 5497
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1897348871056255e-05,
+      "loss": 0.8134,
+      "step": 5498
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1894710876710617e-05,
+      "loss": 0.8457,
+      "step": 5499
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1892072745595e-05,
+      "loss": 0.7822,
+      "step": 5500
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1889434477899832e-05,
+      "loss": 0.8163,
+      "step": 5501
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.188679607381556e-05,
+      "loss": 0.7386,
+      "step": 5502
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1884157533532634e-05,
+      "loss": 0.7466,
+      "step": 5503
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1881518857241523e-05,
+      "loss": 0.7929,
+      "step": 5504
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1878880045132695e-05,
+      "loss": 0.8016,
+      "step": 5505
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1876241097396634e-05,
+      "loss": 0.7716,
+      "step": 5506
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.187360201422384e-05,
+      "loss": 0.8373,
+      "step": 5507
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1870962795804806e-05,
+      "loss": 0.8422,
+      "step": 5508
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1868323442330047e-05,
+      "loss": 0.7887,
+      "step": 5509
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1865683953990089e-05,
+      "loss": 0.7431,
+      "step": 5510
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1863044330975458e-05,
+      "loss": 0.8213,
+      "step": 5511
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1860404573476701e-05,
+      "loss": 0.7797,
+      "step": 5512
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1857764681684367e-05,
+      "loss": 0.7909,
+      "step": 5513
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1855124655789015e-05,
+      "loss": 0.8407,
+      "step": 5514
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1852484495981218e-05,
+      "loss": 0.7763,
+      "step": 5515
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1849844202451557e-05,
+      "loss": 0.7786,
+      "step": 5516
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1847203775390618e-05,
+      "loss": 0.8036,
+      "step": 5517
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1844563214989009e-05,
+      "loss": 0.8106,
+      "step": 5518
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.184192252143733e-05,
+      "loss": 0.7972,
+      "step": 5519
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1839281694926206e-05,
+      "loss": 0.8154,
+      "step": 5520
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.183664073564626e-05,
+      "loss": 0.797,
+      "step": 5521
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1833999643788135e-05,
+      "loss": 0.7994,
+      "step": 5522
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1831358419542476e-05,
+      "loss": 0.8081,
+      "step": 5523
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1828717063099942e-05,
+      "loss": 0.7875,
+      "step": 5524
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1826075574651195e-05,
+      "loss": 0.752,
+      "step": 5525
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1823433954386916e-05,
+      "loss": 0.776,
+      "step": 5526
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.182079220249779e-05,
+      "loss": 0.7519,
+      "step": 5527
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1818150319174515e-05,
+      "loss": 0.8371,
+      "step": 5528
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.181550830460779e-05,
+      "loss": 0.8008,
+      "step": 5529
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.181286615898833e-05,
+      "loss": 0.8486,
+      "step": 5530
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1810223882506864e-05,
+      "loss": 0.831,
+      "step": 5531
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1807581475354119e-05,
+      "loss": 0.7676,
+      "step": 5532
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1804938937720841e-05,
+      "loss": 0.8074,
+      "step": 5533
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.180229626979778e-05,
+      "loss": 0.8105,
+      "step": 5534
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1799653471775696e-05,
+      "loss": 0.8177,
+      "step": 5535
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1797010543845366e-05,
+      "loss": 0.6981,
+      "step": 5536
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1794367486197564e-05,
+      "loss": 0.7461,
+      "step": 5537
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1791724299023084e-05,
+      "loss": 0.8953,
+      "step": 5538
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1789080982512724e-05,
+      "loss": 0.8509,
+      "step": 5539
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1786437536857287e-05,
+      "loss": 0.7747,
+      "step": 5540
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1783793962247598e-05,
+      "loss": 0.7679,
+      "step": 5541
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1781150258874477e-05,
+      "loss": 0.8028,
+      "step": 5542
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1778506426928762e-05,
+      "loss": 0.7304,
+      "step": 5543
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1775862466601304e-05,
+      "loss": 0.7827,
+      "step": 5544
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.177321837808295e-05,
+      "loss": 0.8379,
+      "step": 5545
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.177057416156457e-05,
+      "loss": 0.4015,
+      "step": 5546
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.176792981723703e-05,
+      "loss": 0.7854,
+      "step": 5547
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.176528534529122e-05,
+      "loss": 0.766,
+      "step": 5548
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1762640745918031e-05,
+      "loss": 0.7485,
+      "step": 5549
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1759996019308355e-05,
+      "loss": 0.773,
+      "step": 5550
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1757351165653113e-05,
+      "loss": 0.728,
+      "step": 5551
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1754706185143219e-05,
+      "loss": 0.7482,
+      "step": 5552
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1752061077969598e-05,
+      "loss": 0.769,
+      "step": 5553
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1749415844323193e-05,
+      "loss": 0.885,
+      "step": 5554
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.174677048439495e-05,
+      "loss": 0.8812,
+      "step": 5555
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1744124998375821e-05,
+      "loss": 0.8631,
+      "step": 5556
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1741479386456775e-05,
+      "loss": 0.829,
+      "step": 5557
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1738833648828783e-05,
+      "loss": 0.7991,
+      "step": 5558
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.173618778568283e-05,
+      "loss": 0.756,
+      "step": 5559
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1733541797209908e-05,
+      "loss": 0.7499,
+      "step": 5560
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1730895683601014e-05,
+      "loss": 0.8388,
+      "step": 5561
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1728249445047165e-05,
+      "loss": 0.7484,
+      "step": 5562
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1725603081739372e-05,
+      "loss": 0.8055,
+      "step": 5563
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1722956593868668e-05,
+      "loss": 0.8669,
+      "step": 5564
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1720309981626093e-05,
+      "loss": 0.8087,
+      "step": 5565
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1717663245202686e-05,
+      "loss": 0.78,
+      "step": 5566
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1715016384789508e-05,
+      "loss": 0.8498,
+      "step": 5567
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1712369400577619e-05,
+      "loss": 0.8325,
+      "step": 5568
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1709722292758094e-05,
+      "loss": 0.7741,
+      "step": 5569
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1707075061522017e-05,
+      "loss": 0.7759,
+      "step": 5570
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1704427707060473e-05,
+      "loss": 0.7564,
+      "step": 5571
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1701780229564566e-05,
+      "loss": 0.7896,
+      "step": 5572
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1699132629225405e-05,
+      "loss": 0.773,
+      "step": 5573
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1696484906234104e-05,
+      "loss": 0.8125,
+      "step": 5574
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.169383706078179e-05,
+      "loss": 0.7629,
+      "step": 5575
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.16911890930596e-05,
+      "loss": 0.7836,
+      "step": 5576
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1688541003258679e-05,
+      "loss": 0.4341,
+      "step": 5577
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1685892791570178e-05,
+      "loss": 0.7675,
+      "step": 5578
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1683244458185257e-05,
+      "loss": 0.7896,
+      "step": 5579
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.168059600329509e-05,
+      "loss": 0.7798,
+      "step": 5580
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1677947427090852e-05,
+      "loss": 0.8168,
+      "step": 5581
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.167529872976373e-05,
+      "loss": 0.843,
+      "step": 5582
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.167264991150493e-05,
+      "loss": 0.7964,
+      "step": 5583
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1670000972505645e-05,
+      "loss": 0.7904,
+      "step": 5584
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1667351912957098e-05,
+      "loss": 0.7726,
+      "step": 5585
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1664702733050508e-05,
+      "loss": 0.8875,
+      "step": 5586
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1662053432977104e-05,
+      "loss": 0.7797,
+      "step": 5587
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1659404012928135e-05,
+      "loss": 0.8509,
+      "step": 5588
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1656754473094841e-05,
+      "loss": 0.7736,
+      "step": 5589
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.165410481366848e-05,
+      "loss": 0.7846,
+      "step": 5590
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1651455034840327e-05,
+      "loss": 0.8778,
+      "step": 5591
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1648805136801645e-05,
+      "loss": 0.7739,
+      "step": 5592
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1646155119743724e-05,
+      "loss": 0.7729,
+      "step": 5593
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1643504983857852e-05,
+      "loss": 0.7472,
+      "step": 5594
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1640854729335332e-05,
+      "loss": 0.7638,
+      "step": 5595
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1638204356367473e-05,
+      "loss": 0.7913,
+      "step": 5596
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.163555386514559e-05,
+      "loss": 0.7983,
+      "step": 5597
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1632903255861015e-05,
+      "loss": 0.7518,
+      "step": 5598
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1630252528705079e-05,
+      "loss": 0.8332,
+      "step": 5599
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1627601683869122e-05,
+      "loss": 0.7968,
+      "step": 5600
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1624950721544504e-05,
+      "loss": 0.7388,
+      "step": 5601
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1622299641922573e-05,
+      "loss": 0.7402,
+      "step": 5602
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1619648445194709e-05,
+      "loss": 0.7509,
+      "step": 5603
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1616997131552283e-05,
+      "loss": 0.8135,
+      "step": 5604
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1614345701186682e-05,
+      "loss": 0.781,
+      "step": 5605
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1611694154289299e-05,
+      "loss": 0.7447,
+      "step": 5606
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1609042491051539e-05,
+      "loss": 0.7661,
+      "step": 5607
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1606390711664808e-05,
+      "loss": 0.7792,
+      "step": 5608
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1603738816320531e-05,
+      "loss": 0.8131,
+      "step": 5609
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.160108680521013e-05,
+      "loss": 0.7429,
+      "step": 5610
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1598434678525041e-05,
+      "loss": 0.7888,
+      "step": 5611
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1595782436456715e-05,
+      "loss": 0.7673,
+      "step": 5612
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1593130079196598e-05,
+      "loss": 0.8224,
+      "step": 5613
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1590477606936153e-05,
+      "loss": 0.8143,
+      "step": 5614
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1587825019866847e-05,
+      "loss": 0.709,
+      "step": 5615
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.158517231818016e-05,
+      "loss": 0.8607,
+      "step": 5616
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1582519502067575e-05,
+      "loss": 0.7762,
+      "step": 5617
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1579866571720588e-05,
+      "loss": 0.8293,
+      "step": 5618
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1577213527330702e-05,
+      "loss": 0.8035,
+      "step": 5619
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1574560369089428e-05,
+      "loss": 0.7605,
+      "step": 5620
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.157190709718828e-05,
+      "loss": 0.7702,
+      "step": 5621
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1569253711818788e-05,
+      "loss": 0.838,
+      "step": 5622
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1566600213172487e-05,
+      "loss": 0.8382,
+      "step": 5623
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1563946601440918e-05,
+      "loss": 0.8254,
+      "step": 5624
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1561292876815639e-05,
+      "loss": 0.8798,
+      "step": 5625
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.15586390394882e-05,
+      "loss": 0.7646,
+      "step": 5626
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1555985089650175e-05,
+      "loss": 0.7829,
+      "step": 5627
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1553331027493137e-05,
+      "loss": 0.7906,
+      "step": 5628
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1550676853208672e-05,
+      "loss": 0.8135,
+      "step": 5629
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.154802256698837e-05,
+      "loss": 0.8315,
+      "step": 5630
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1545368169023831e-05,
+      "loss": 0.8044,
+      "step": 5631
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1542713659506666e-05,
+      "loss": 0.8013,
+      "step": 5632
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1540059038628491e-05,
+      "loss": 0.7747,
+      "step": 5633
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1537404306580924e-05,
+      "loss": 0.8064,
+      "step": 5634
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1534749463555606e-05,
+      "loss": 0.7904,
+      "step": 5635
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.153209450974417e-05,
+      "loss": 0.876,
+      "step": 5636
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1529439445338267e-05,
+      "loss": 0.8238,
+      "step": 5637
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1526784270529553e-05,
+      "loss": 0.8066,
+      "step": 5638
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1524128985509694e-05,
+      "loss": 0.8102,
+      "step": 5639
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.152147359047036e-05,
+      "loss": 0.835,
+      "step": 5640
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1518818085603231e-05,
+      "loss": 0.8012,
+      "step": 5641
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1516162471099994e-05,
+      "loss": 0.8275,
+      "step": 5642
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1513506747152352e-05,
+      "loss": 0.7417,
+      "step": 5643
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1510850913952e-05,
+      "loss": 0.7783,
+      "step": 5644
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1508194971690652e-05,
+      "loss": 0.833,
+      "step": 5645
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1505538920560031e-05,
+      "loss": 0.7641,
+      "step": 5646
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1502882760751861e-05,
+      "loss": 0.9005,
+      "step": 5647
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1500226492457878e-05,
+      "loss": 0.8115,
+      "step": 5648
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1497570115869826e-05,
+      "loss": 0.8369,
+      "step": 5649
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1494913631179455e-05,
+      "loss": 0.8382,
+      "step": 5650
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1492257038578526e-05,
+      "loss": 0.7814,
+      "step": 5651
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1489600338258802e-05,
+      "loss": 0.7768,
+      "step": 5652
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1486943530412063e-05,
+      "loss": 0.7667,
+      "step": 5653
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1484286615230082e-05,
+      "loss": 0.8512,
+      "step": 5654
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1481629592904655e-05,
+      "loss": 0.7426,
+      "step": 5655
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.147897246362758e-05,
+      "loss": 0.93,
+      "step": 5656
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.147631522759066e-05,
+      "loss": 0.8432,
+      "step": 5657
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1473657884985708e-05,
+      "loss": 0.8337,
+      "step": 5658
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1471000436004547e-05,
+      "loss": 0.8201,
+      "step": 5659
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1468342880839001e-05,
+      "loss": 0.8478,
+      "step": 5660
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1465685219680914e-05,
+      "loss": 0.7575,
+      "step": 5661
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.146302745272212e-05,
+      "loss": 0.7631,
+      "step": 5662
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1460369580154473e-05,
+      "loss": 0.4261,
+      "step": 5663
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.145771160216984e-05,
+      "loss": 0.8066,
+      "step": 5664
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1455053518960075e-05,
+      "loss": 0.8328,
+      "step": 5665
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1452395330717064e-05,
+      "loss": 0.8321,
+      "step": 5666
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1449737037632677e-05,
+      "loss": 0.7998,
+      "step": 5667
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1447078639898811e-05,
+      "loss": 0.819,
+      "step": 5668
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1444420137707364e-05,
+      "loss": 0.7659,
+      "step": 5669
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1441761531250233e-05,
+      "loss": 0.8116,
+      "step": 5670
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1439102820719336e-05,
+      "loss": 0.8163,
+      "step": 5671
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1436444006306596e-05,
+      "loss": 0.8152,
+      "step": 5672
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1433785088203927e-05,
+      "loss": 0.8434,
+      "step": 5673
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1431126066603277e-05,
+      "loss": 0.7841,
+      "step": 5674
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1428466941696578e-05,
+      "loss": 0.7967,
+      "step": 5675
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1425807713675784e-05,
+      "loss": 0.7717,
+      "step": 5676
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1423148382732854e-05,
+      "loss": 0.7615,
+      "step": 5677
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1420488949059747e-05,
+      "loss": 0.7897,
+      "step": 5678
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1417829412848436e-05,
+      "loss": 0.7598,
+      "step": 5679
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.14151697742909e-05,
+      "loss": 0.8271,
+      "step": 5680
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1412510033579127e-05,
+      "loss": 0.7891,
+      "step": 5681
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1409850190905115e-05,
+      "loss": 0.7385,
+      "step": 5682
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1407190246460852e-05,
+      "loss": 0.8303,
+      "step": 5683
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1404530200438359e-05,
+      "loss": 0.9063,
+      "step": 5684
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.140187005302965e-05,
+      "loss": 0.7898,
+      "step": 5685
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.139920980442674e-05,
+      "loss": 0.8767,
+      "step": 5686
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.139654945482167e-05,
+      "loss": 0.8293,
+      "step": 5687
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1393889004406468e-05,
+      "loss": 0.7701,
+      "step": 5688
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1391228453373186e-05,
+      "loss": 0.7602,
+      "step": 5689
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1388567801913875e-05,
+      "loss": 0.7599,
+      "step": 5690
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1385907050220593e-05,
+      "loss": 0.7027,
+      "step": 5691
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1383246198485406e-05,
+      "loss": 0.8256,
+      "step": 5692
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.138058524690039e-05,
+      "loss": 0.7937,
+      "step": 5693
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1377924195657625e-05,
+      "loss": 0.7928,
+      "step": 5694
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1375263044949202e-05,
+      "loss": 0.7621,
+      "step": 5695
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1372601794967212e-05,
+      "loss": 0.7976,
+      "step": 5696
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.136994044590376e-05,
+      "loss": 0.7763,
+      "step": 5697
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.136727899795096e-05,
+      "loss": 0.766,
+      "step": 5698
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1364617451300919e-05,
+      "loss": 0.7736,
+      "step": 5699
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.136195580614577e-05,
+      "loss": 0.805,
+      "step": 5700
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.135929406267764e-05,
+      "loss": 0.7771,
+      "step": 5701
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.135663222108867e-05,
+      "loss": 0.7324,
+      "step": 5702
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1353970281571005e-05,
+      "loss": 0.8204,
+      "step": 5703
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1351308244316794e-05,
+      "loss": 0.789,
+      "step": 5704
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.13486461095182e-05,
+      "loss": 0.7727,
+      "step": 5705
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.134598387736739e-05,
+      "loss": 0.7855,
+      "step": 5706
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1343321548056535e-05,
+      "loss": 0.8065,
+      "step": 5707
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.134065912177782e-05,
+      "loss": 0.7158,
+      "step": 5708
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1337996598723426e-05,
+      "loss": 0.8161,
+      "step": 5709
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1335333979085551e-05,
+      "loss": 0.827,
+      "step": 5710
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.13326712630564e-05,
+      "loss": 0.8505,
+      "step": 5711
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1330008450828175e-05,
+      "loss": 0.7743,
+      "step": 5712
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1327345542593098e-05,
+      "loss": 0.8017,
+      "step": 5713
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1324682538543388e-05,
+      "loss": 0.7424,
+      "step": 5714
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1322019438871273e-05,
+      "loss": 0.7199,
+      "step": 5715
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1319356243768991e-05,
+      "loss": 0.7907,
+      "step": 5716
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1316692953428786e-05,
+      "loss": 0.8218,
+      "step": 5717
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1314029568042907e-05,
+      "loss": 0.7649,
+      "step": 5718
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.131136608780361e-05,
+      "loss": 0.7393,
+      "step": 5719
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.130870251290316e-05,
+      "loss": 0.8168,
+      "step": 5720
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1306038843533826e-05,
+      "loss": 0.833,
+      "step": 5721
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1303375079887886e-05,
+      "loss": 0.7512,
+      "step": 5722
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1300711222157625e-05,
+      "loss": 0.7631,
+      "step": 5723
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1298047270535336e-05,
+      "loss": 0.7795,
+      "step": 5724
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1295383225213315e-05,
+      "loss": 0.8215,
+      "step": 5725
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1292719086383864e-05,
+      "loss": 0.8078,
+      "step": 5726
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1290054854239297e-05,
+      "loss": 0.791,
+      "step": 5727
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1287390528971929e-05,
+      "loss": 0.7569,
+      "step": 5728
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.128472611077409e-05,
+      "loss": 0.7536,
+      "step": 5729
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1282061599838101e-05,
+      "loss": 0.7819,
+      "step": 5730
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1279396996356313e-05,
+      "loss": 0.8508,
+      "step": 5731
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1276732300521065e-05,
+      "loss": 0.7617,
+      "step": 5732
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1274067512524709e-05,
+      "loss": 0.7609,
+      "step": 5733
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1271402632559603e-05,
+      "loss": 0.8124,
+      "step": 5734
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.126873766081811e-05,
+      "loss": 0.846,
+      "step": 5735
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1266072597492601e-05,
+      "loss": 0.7987,
+      "step": 5736
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.126340744277546e-05,
+      "loss": 0.7917,
+      "step": 5737
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1260742196859065e-05,
+      "loss": 0.778,
+      "step": 5738
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.125807685993581e-05,
+      "loss": 0.7503,
+      "step": 5739
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1255411432198091e-05,
+      "loss": 0.3865,
+      "step": 5740
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1252745913838315e-05,
+      "loss": 0.7932,
+      "step": 5741
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1250080305048892e-05,
+      "loss": 0.7937,
+      "step": 5742
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1247414606022238e-05,
+      "loss": 0.7663,
+      "step": 5743
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1244748816950776e-05,
+      "loss": 0.7283,
+      "step": 5744
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1242082938026942e-05,
+      "loss": 0.7658,
+      "step": 5745
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1239416969443167e-05,
+      "loss": 0.7809,
+      "step": 5746
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1236750911391897e-05,
+      "loss": 0.7124,
+      "step": 5747
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1234084764065578e-05,
+      "loss": 0.7629,
+      "step": 5748
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1231418527656672e-05,
+      "loss": 0.8416,
+      "step": 5749
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1228752202357638e-05,
+      "loss": 0.736,
+      "step": 5750
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1226085788360945e-05,
+      "loss": 0.754,
+      "step": 5751
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.122341928585907e-05,
+      "loss": 0.763,
+      "step": 5752
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1220752695044497e-05,
+      "loss": 0.7983,
+      "step": 5753
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1218086016109708e-05,
+      "loss": 0.8159,
+      "step": 5754
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1215419249247207e-05,
+      "loss": 0.7069,
+      "step": 5755
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1212752394649481e-05,
+      "loss": 0.84,
+      "step": 5756
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.121008545250905e-05,
+      "loss": 0.8084,
+      "step": 5757
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1207418423018423e-05,
+      "loss": 0.8392,
+      "step": 5758
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.120475130637012e-05,
+      "loss": 0.759,
+      "step": 5759
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1202084102756667e-05,
+      "loss": 0.8099,
+      "step": 5760
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1199416812370596e-05,
+      "loss": 0.7991,
+      "step": 5761
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1196749435404445e-05,
+      "loss": 0.7226,
+      "step": 5762
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1194081972050765e-05,
+      "loss": 0.8308,
+      "step": 5763
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.11914144225021e-05,
+      "loss": 0.7168,
+      "step": 5764
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.118874678695101e-05,
+      "loss": 0.8031,
+      "step": 5765
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1186079065590061e-05,
+      "loss": 0.7761,
+      "step": 5766
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1183411258611819e-05,
+      "loss": 0.7565,
+      "step": 5767
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1180743366208861e-05,
+      "loss": 0.7954,
+      "step": 5768
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.117807538857377e-05,
+      "loss": 0.4293,
+      "step": 5769
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1175407325899134e-05,
+      "loss": 0.8276,
+      "step": 5770
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1172739178377548e-05,
+      "loss": 0.7708,
+      "step": 5771
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1170070946201613e-05,
+      "loss": 0.7673,
+      "step": 5772
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1167402629563934e-05,
+      "loss": 0.8424,
+      "step": 5773
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1164734228657127e-05,
+      "loss": 0.8027,
+      "step": 5774
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1162065743673806e-05,
+      "loss": 0.8449,
+      "step": 5775
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1159397174806603e-05,
+      "loss": 0.824,
+      "step": 5776
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1156728522248141e-05,
+      "loss": 0.7557,
+      "step": 5777
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1154059786191062e-05,
+      "loss": 0.8552,
+      "step": 5778
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1151390966828008e-05,
+      "loss": 0.7973,
+      "step": 5779
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.114872206435163e-05,
+      "loss": 0.7326,
+      "step": 5780
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1146053078954577e-05,
+      "loss": 0.8104,
+      "step": 5781
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1143384010829516e-05,
+      "loss": 0.8483,
+      "step": 5782
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1140714860169115e-05,
+      "loss": 0.7248,
+      "step": 5783
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1138045627166043e-05,
+      "loss": 0.7471,
+      "step": 5784
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.113537631201298e-05,
+      "loss": 0.8359,
+      "step": 5785
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1132706914902613e-05,
+      "loss": 0.8219,
+      "step": 5786
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.113003743602763e-05,
+      "loss": 0.6916,
+      "step": 5787
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1127367875580728e-05,
+      "loss": 0.7711,
+      "step": 5788
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1124698233754613e-05,
+      "loss": 0.8325,
+      "step": 5789
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.112202851074199e-05,
+      "loss": 0.8136,
+      "step": 5790
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1119358706735572e-05,
+      "loss": 0.7945,
+      "step": 5791
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1116688821928084e-05,
+      "loss": 0.7659,
+      "step": 5792
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1114018856512247e-05,
+      "loss": 0.7593,
+      "step": 5793
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1111348810680798e-05,
+      "loss": 0.8328,
+      "step": 5794
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1108678684626471e-05,
+      "loss": 0.8306,
+      "step": 5795
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.110600847854201e-05,
+      "loss": 0.838,
+      "step": 5796
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1103338192620168e-05,
+      "loss": 0.8049,
+      "step": 5797
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1100667827053693e-05,
+      "loss": 0.8131,
+      "step": 5798
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1097997382035352e-05,
+      "loss": 0.8292,
+      "step": 5799
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1095326857757907e-05,
+      "loss": 0.8056,
+      "step": 5800
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1092656254414131e-05,
+      "loss": 0.8179,
+      "step": 5801
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1089985572196807e-05,
+      "loss": 0.7569,
+      "step": 5802
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.108731481129871e-05,
+      "loss": 0.7515,
+      "step": 5803
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1084643971912636e-05,
+      "loss": 0.7469,
+      "step": 5804
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1081973054231379e-05,
+      "loss": 0.8194,
+      "step": 5805
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1079302058447736e-05,
+      "loss": 0.8142,
+      "step": 5806
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.107663098475452e-05,
+      "loss": 0.7305,
+      "step": 5807
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1073959833344535e-05,
+      "loss": 0.8046,
+      "step": 5808
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.10712886044106e-05,
+      "loss": 0.8264,
+      "step": 5809
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1068617298145542e-05,
+      "loss": 0.86,
+      "step": 5810
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.106594591474219e-05,
+      "loss": 0.871,
+      "step": 5811
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1063274454393371e-05,
+      "loss": 0.788,
+      "step": 5812
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1060602917291932e-05,
+      "loss": 0.8033,
+      "step": 5813
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1057931303630713e-05,
+      "loss": 0.8007,
+      "step": 5814
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1055259613602573e-05,
+      "loss": 0.7901,
+      "step": 5815
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.105258784740036e-05,
+      "loss": 0.7776,
+      "step": 5816
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1049916005216938e-05,
+      "loss": 0.8168,
+      "step": 5817
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.104724408724518e-05,
+      "loss": 0.8302,
+      "step": 5818
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1044572093677947e-05,
+      "loss": 0.8605,
+      "step": 5819
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1041900024708129e-05,
+      "loss": 0.7405,
+      "step": 5820
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1039227880528602e-05,
+      "loss": 0.7839,
+      "step": 5821
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.103655566133226e-05,
+      "loss": 0.8007,
+      "step": 5822
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1033883367311995e-05,
+      "loss": 0.8453,
+      "step": 5823
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1031210998660706e-05,
+      "loss": 0.7674,
+      "step": 5824
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.10285385555713e-05,
+      "loss": 0.7735,
+      "step": 5825
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.102586603823669e-05,
+      "loss": 0.7798,
+      "step": 5826
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1023193446849787e-05,
+      "loss": 0.848,
+      "step": 5827
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1020520781603518e-05,
+      "loss": 0.7995,
+      "step": 5828
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1017848042690804e-05,
+      "loss": 0.8874,
+      "step": 5829
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.101517523030458e-05,
+      "loss": 0.7705,
+      "step": 5830
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1012502344637786e-05,
+      "loss": 0.8119,
+      "step": 5831
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1009829385883357e-05,
+      "loss": 0.8459,
+      "step": 5832
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1007156354234247e-05,
+      "loss": 0.8051,
+      "step": 5833
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1004483249883408e-05,
+      "loss": 0.7803,
+      "step": 5834
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1001810073023799e-05,
+      "loss": 0.787,
+      "step": 5835
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0999136823848385e-05,
+      "loss": 0.7804,
+      "step": 5836
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.099646350255013e-05,
+      "loss": 0.8237,
+      "step": 5837
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0993790109322008e-05,
+      "loss": 0.7742,
+      "step": 5838
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0991116644357009e-05,
+      "loss": 0.7966,
+      "step": 5839
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0988443107848103e-05,
+      "loss": 0.8325,
+      "step": 5840
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.098576949998829e-05,
+      "loss": 0.7821,
+      "step": 5841
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.098309582097056e-05,
+      "loss": 0.8281,
+      "step": 5842
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0980422070987913e-05,
+      "loss": 0.795,
+      "step": 5843
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0977748250233356e-05,
+      "loss": 0.8146,
+      "step": 5844
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0975074358899896e-05,
+      "loss": 0.8648,
+      "step": 5845
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.0972400397180554e-05,
+      "loss": 0.853,
+      "step": 5846
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0969726365268343e-05,
+      "loss": 0.8235,
+      "step": 5847
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0967052263356293e-05,
+      "loss": 0.8015,
+      "step": 5848
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0964378091637436e-05,
+      "loss": 0.8215,
+      "step": 5849
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0961703850304804e-05,
+      "loss": 0.819,
+      "step": 5850
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0959029539551436e-05,
+      "loss": 0.8105,
+      "step": 5851
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0956355159570381e-05,
+      "loss": 0.7661,
+      "step": 5852
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0953680710554691e-05,
+      "loss": 0.783,
+      "step": 5853
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0951006192697416e-05,
+      "loss": 0.7988,
+      "step": 5854
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.094833160619162e-05,
+      "loss": 0.7168,
+      "step": 5855
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0945656951230369e-05,
+      "loss": 0.814,
+      "step": 5856
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0942982228006733e-05,
+      "loss": 0.8016,
+      "step": 5857
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0940307436713784e-05,
+      "loss": 0.9044,
+      "step": 5858
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0937632577544608e-05,
+      "loss": 0.7818,
+      "step": 5859
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0934957650692284e-05,
+      "loss": 0.8547,
+      "step": 5860
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0932282656349904e-05,
+      "loss": 0.7806,
+      "step": 5861
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0929607594710567e-05,
+      "loss": 0.8161,
+      "step": 5862
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0926932465967367e-05,
+      "loss": 0.7643,
+      "step": 5863
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0924257270313411e-05,
+      "loss": 0.8355,
+      "step": 5864
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.092158200794181e-05,
+      "loss": 0.871,
+      "step": 5865
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0918906679045676e-05,
+      "loss": 0.7975,
+      "step": 5866
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0916231283818129e-05,
+      "loss": 0.7686,
+      "step": 5867
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.091355582245229e-05,
+      "loss": 0.7689,
+      "step": 5868
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0910880295141293e-05,
+      "loss": 0.7702,
+      "step": 5869
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0908204702078267e-05,
+      "loss": 0.83,
+      "step": 5870
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0905529043456351e-05,
+      "loss": 0.7873,
+      "step": 5871
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0902853319468686e-05,
+      "loss": 0.7516,
+      "step": 5872
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0900177530308427e-05,
+      "loss": 0.7959,
+      "step": 5873
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0897501676168719e-05,
+      "loss": 0.8625,
+      "step": 5874
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0894825757242718e-05,
+      "loss": 0.7731,
+      "step": 5875
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0892149773723588e-05,
+      "loss": 0.7727,
+      "step": 5876
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0889473725804498e-05,
+      "loss": 0.7912,
+      "step": 5877
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0886797613678618e-05,
+      "loss": 0.8478,
+      "step": 5878
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0884121437539121e-05,
+      "loss": 0.6901,
+      "step": 5879
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.088144519757919e-05,
+      "loss": 0.7725,
+      "step": 5880
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0878768893992004e-05,
+      "loss": 0.7788,
+      "step": 5881
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.087609252697076e-05,
+      "loss": 0.8144,
+      "step": 5882
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0873416096708649e-05,
+      "loss": 0.8554,
+      "step": 5883
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0870739603398866e-05,
+      "loss": 0.7248,
+      "step": 5884
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0868063047234617e-05,
+      "loss": 0.8193,
+      "step": 5885
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.086538642840911e-05,
+      "loss": 0.8451,
+      "step": 5886
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0862709747115557e-05,
+      "loss": 0.8017,
+      "step": 5887
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0860033003547175e-05,
+      "loss": 0.7894,
+      "step": 5888
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0857356197897185e-05,
+      "loss": 0.8298,
+      "step": 5889
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0854679330358808e-05,
+      "loss": 0.8205,
+      "step": 5890
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0852002401125284e-05,
+      "loss": 0.767,
+      "step": 5891
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.084932541038984e-05,
+      "loss": 0.7793,
+      "step": 5892
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0846648358345715e-05,
+      "loss": 0.8352,
+      "step": 5893
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0843971245186155e-05,
+      "loss": 0.763,
+      "step": 5894
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0841294071104407e-05,
+      "loss": 0.7501,
+      "step": 5895
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.083861683629372e-05,
+      "loss": 0.8349,
+      "step": 5896
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0835939540947357e-05,
+      "loss": 0.7866,
+      "step": 5897
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0833262185258575e-05,
+      "loss": 0.8164,
+      "step": 5898
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.083058476942064e-05,
+      "loss": 0.717,
+      "step": 5899
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.082790729362682e-05,
+      "loss": 0.8254,
+      "step": 5900
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0825229758070395e-05,
+      "loss": 0.7915,
+      "step": 5901
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0822552162944635e-05,
+      "loss": 0.8576,
+      "step": 5902
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0819874508442828e-05,
+      "loss": 0.7743,
+      "step": 5903
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0817196794758258e-05,
+      "loss": 0.7645,
+      "step": 5904
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0814519022084217e-05,
+      "loss": 0.8224,
+      "step": 5905
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0811841190614003e-05,
+      "loss": 0.6942,
+      "step": 5906
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.080916330054091e-05,
+      "loss": 0.7984,
+      "step": 5907
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.080648535205825e-05,
+      "loss": 0.7467,
+      "step": 5908
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0803807345359325e-05,
+      "loss": 0.4117,
+      "step": 5909
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0801129280637448e-05,
+      "loss": 0.8387,
+      "step": 5910
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0798451158085936e-05,
+      "loss": 0.7993,
+      "step": 5911
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.079577297789811e-05,
+      "loss": 0.7784,
+      "step": 5912
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0793094740267297e-05,
+      "loss": 0.404,
+      "step": 5913
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0790416445386825e-05,
+      "loss": 0.8198,
+      "step": 5914
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0787738093450022e-05,
+      "loss": 0.8446,
+      "step": 5915
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0785059684650228e-05,
+      "loss": 0.746,
+      "step": 5916
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0782381219180788e-05,
+      "loss": 0.8067,
+      "step": 5917
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0779702697235047e-05,
+      "loss": 0.8004,
+      "step": 5918
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.077702411900635e-05,
+      "loss": 0.7325,
+      "step": 5919
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0774345484688054e-05,
+      "loss": 0.81,
+      "step": 5920
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0771666794473514e-05,
+      "loss": 0.7855,
+      "step": 5921
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0768988048556094e-05,
+      "loss": 0.8456,
+      "step": 5922
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0766309247129159e-05,
+      "loss": 0.807,
+      "step": 5923
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.076363039038608e-05,
+      "loss": 0.781,
+      "step": 5924
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0760951478520227e-05,
+      "loss": 0.7839,
+      "step": 5925
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0758272511724979e-05,
+      "loss": 0.779,
+      "step": 5926
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0755593490193723e-05,
+      "loss": 0.7793,
+      "step": 5927
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0752914414119836e-05,
+      "loss": 0.7218,
+      "step": 5928
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0750235283696713e-05,
+      "loss": 0.8065,
+      "step": 5929
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0747556099117746e-05,
+      "loss": 0.8125,
+      "step": 5930
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0744876860576332e-05,
+      "loss": 0.746,
+      "step": 5931
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0742197568265873e-05,
+      "loss": 0.7817,
+      "step": 5932
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0739518222379775e-05,
+      "loss": 0.8117,
+      "step": 5933
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0736838823111442e-05,
+      "loss": 0.7876,
+      "step": 5934
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0734159370654293e-05,
+      "loss": 0.7969,
+      "step": 5935
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0731479865201741e-05,
+      "loss": 0.7982,
+      "step": 5936
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0728800306947206e-05,
+      "loss": 0.716,
+      "step": 5937
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0726120696084119e-05,
+      "loss": 0.8008,
+      "step": 5938
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.07234410328059e-05,
+      "loss": 0.7712,
+      "step": 5939
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0720761317305985e-05,
+      "loss": 0.7893,
+      "step": 5940
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0718081549777808e-05,
+      "loss": 0.8207,
+      "step": 5941
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0715401730414807e-05,
+      "loss": 0.4286,
+      "step": 5942
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.071272185941043e-05,
+      "loss": 0.7339,
+      "step": 5943
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.071004193695812e-05,
+      "loss": 0.7456,
+      "step": 5944
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0707361963251328e-05,
+      "loss": 0.7495,
+      "step": 5945
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.070468193848351e-05,
+      "loss": 0.7884,
+      "step": 5946
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0702001862848121e-05,
+      "loss": 0.7725,
+      "step": 5947
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0699321736538628e-05,
+      "loss": 0.7137,
+      "step": 5948
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0696641559748488e-05,
+      "loss": 0.7072,
+      "step": 5949
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0693961332671178e-05,
+      "loss": 0.745,
+      "step": 5950
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0691281055500169e-05,
+      "loss": 0.771,
+      "step": 5951
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0688600728428932e-05,
+      "loss": 0.4267,
+      "step": 5952
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0685920351650953e-05,
+      "loss": 0.8669,
+      "step": 5953
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0683239925359712e-05,
+      "loss": 0.7954,
+      "step": 5954
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0680559449748695e-05,
+      "loss": 0.8511,
+      "step": 5955
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0677878925011396e-05,
+      "loss": 0.7616,
+      "step": 5956
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0675198351341306e-05,
+      "loss": 0.7714,
+      "step": 5957
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0672517728931926e-05,
+      "loss": 0.8004,
+      "step": 5958
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0669837057976756e-05,
+      "loss": 0.7642,
+      "step": 5959
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0667156338669295e-05,
+      "loss": 0.8551,
+      "step": 5960
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.066447557120306e-05,
+      "loss": 0.8421,
+      "step": 5961
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0661794755771558e-05,
+      "loss": 0.8111,
+      "step": 5962
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0659113892568302e-05,
+      "loss": 0.7802,
+      "step": 5963
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0656432981786817e-05,
+      "loss": 0.829,
+      "step": 5964
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0653752023620618e-05,
+      "loss": 0.7732,
+      "step": 5965
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0651071018263233e-05,
+      "loss": 0.8318,
+      "step": 5966
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.064838996590819e-05,
+      "loss": 0.8333,
+      "step": 5967
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0645708866749025e-05,
+      "loss": 0.8154,
+      "step": 5968
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0643027720979271e-05,
+      "loss": 0.8533,
+      "step": 5969
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0640346528792465e-05,
+      "loss": 0.81,
+      "step": 5970
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0637665290382148e-05,
+      "loss": 0.7624,
+      "step": 5971
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0634984005941875e-05,
+      "loss": 0.7859,
+      "step": 5972
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0632302675665185e-05,
+      "loss": 0.8015,
+      "step": 5973
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0629621299745634e-05,
+      "loss": 0.8296,
+      "step": 5974
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0626939878376776e-05,
+      "loss": 0.8129,
+      "step": 5975
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0624258411752171e-05,
+      "loss": 0.7534,
+      "step": 5976
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0621576900065383e-05,
+      "loss": 0.7987,
+      "step": 5977
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0618895343509972e-05,
+      "loss": 0.7382,
+      "step": 5978
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0616213742279513e-05,
+      "loss": 0.8405,
+      "step": 5979
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0613532096567574e-05,
+      "loss": 0.7774,
+      "step": 5980
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0610850406567727e-05,
+      "loss": 0.7746,
+      "step": 5981
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.060816867247356e-05,
+      "loss": 0.7977,
+      "step": 5982
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0605486894478644e-05,
+      "loss": 0.7898,
+      "step": 5983
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0602805072776568e-05,
+      "loss": 0.7731,
+      "step": 5984
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0600123207560923e-05,
+      "loss": 0.792,
+      "step": 5985
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.059744129902529e-05,
+      "loss": 0.8965,
+      "step": 5986
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0594759347363272e-05,
+      "loss": 0.807,
+      "step": 5987
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0592077352768464e-05,
+      "loss": 0.7462,
+      "step": 5988
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0589395315434465e-05,
+      "loss": 0.7993,
+      "step": 5989
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0586713235554879e-05,
+      "loss": 0.7676,
+      "step": 5990
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0584031113323312e-05,
+      "loss": 0.8086,
+      "step": 5991
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0581348948933372e-05,
+      "loss": 0.7923,
+      "step": 5992
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0578666742578677e-05,
+      "loss": 0.7121,
+      "step": 5993
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0575984494452833e-05,
+      "loss": 0.7808,
+      "step": 5994
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.057330220474947e-05,
+      "loss": 0.8855,
+      "step": 5995
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0570619873662198e-05,
+      "loss": 0.7736,
+      "step": 5996
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0567937501384647e-05,
+      "loss": 0.7734,
+      "step": 5997
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0565255088110445e-05,
+      "loss": 0.7629,
+      "step": 5998
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.056257263403322e-05,
+      "loss": 0.7644,
+      "step": 5999
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0559890139346611e-05,
+      "loss": 0.8566,
+      "step": 6000
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0557207604244247e-05,
+      "loss": 0.7426,
+      "step": 6001
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.055452502891977e-05,
+      "loss": 0.7522,
+      "step": 6002
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0551842413566826e-05,
+      "loss": 0.8419,
+      "step": 6003
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0549159758379052e-05,
+      "loss": 0.7885,
+      "step": 6004
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0546477063550102e-05,
+      "loss": 0.8388,
+      "step": 6005
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0543794329273629e-05,
+      "loss": 0.7924,
+      "step": 6006
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0541111555743277e-05,
+      "loss": 0.8846,
+      "step": 6007
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0538428743152709e-05,
+      "loss": 0.8132,
+      "step": 6008
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0535745891695583e-05,
+      "loss": 0.8358,
+      "step": 6009
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0533063001565562e-05,
+      "loss": 0.7641,
+      "step": 6010
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0530380072956312e-05,
+      "loss": 0.8151,
+      "step": 6011
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0527697106061496e-05,
+      "loss": 0.7786,
+      "step": 6012
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0525014101074792e-05,
+      "loss": 0.8735,
+      "step": 6013
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0522331058189861e-05,
+      "loss": 0.7742,
+      "step": 6014
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.051964797760039e-05,
+      "loss": 0.8293,
+      "step": 6015
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0516964859500052e-05,
+      "loss": 0.8419,
+      "step": 6016
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0514281704082529e-05,
+      "loss": 0.7912,
+      "step": 6017
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0511598511541507e-05,
+      "loss": 0.8742,
+      "step": 6018
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.050891528207067e-05,
+      "loss": 0.7835,
+      "step": 6019
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0506232015863711e-05,
+      "loss": 0.7763,
+      "step": 6020
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.050354871311432e-05,
+      "loss": 0.7595,
+      "step": 6021
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0500865374016189e-05,
+      "loss": 0.8377,
+      "step": 6022
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0498181998763018e-05,
+      "loss": 0.7802,
+      "step": 6023
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.049549858754851e-05,
+      "loss": 0.7466,
+      "step": 6024
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.049281514056636e-05,
+      "loss": 0.6434,
+      "step": 6025
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0490131658010277e-05,
+      "loss": 0.764,
+      "step": 6026
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0487448140073967e-05,
+      "loss": 0.7573,
+      "step": 6027
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0484764586951144e-05,
+      "loss": 0.7463,
+      "step": 6028
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0482080998835518e-05,
+      "loss": 0.7853,
+      "step": 6029
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0479397375920801e-05,
+      "loss": 0.7735,
+      "step": 6030
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0476713718400718e-05,
+      "loss": 0.8033,
+      "step": 6031
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0474030026468984e-05,
+      "loss": 0.8451,
+      "step": 6032
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0471346300319323e-05,
+      "loss": 0.7891,
+      "step": 6033
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0468662540145463e-05,
+      "loss": 0.8357,
+      "step": 6034
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0465978746141126e-05,
+      "loss": 0.8112,
+      "step": 6035
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0463294918500045e-05,
+      "loss": 0.8053,
+      "step": 6036
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0460611057415952e-05,
+      "loss": 0.7732,
+      "step": 6037
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0457927163082581e-05,
+      "loss": 0.7973,
+      "step": 6038
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0455243235693672e-05,
+      "loss": 0.8017,
+      "step": 6039
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0452559275442965e-05,
+      "loss": 0.8106,
+      "step": 6040
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0449875282524199e-05,
+      "loss": 0.7588,
+      "step": 6041
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.044719125713112e-05,
+      "loss": 0.7898,
+      "step": 6042
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0444507199457472e-05,
+      "loss": 0.7897,
+      "step": 6043
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.044182310969701e-05,
+      "loss": 0.8541,
+      "step": 6044
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0439138988043481e-05,
+      "loss": 0.8082,
+      "step": 6045
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0436454834690637e-05,
+      "loss": 0.7817,
+      "step": 6046
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0433770649832242e-05,
+      "loss": 0.7276,
+      "step": 6047
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0431086433662045e-05,
+      "loss": 0.4407,
+      "step": 6048
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0428402186373812e-05,
+      "loss": 0.7866,
+      "step": 6049
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0425717908161301e-05,
+      "loss": 0.7764,
+      "step": 6050
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0423033599218283e-05,
+      "loss": 0.827,
+      "step": 6051
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0420349259738523e-05,
+      "loss": 0.851,
+      "step": 6052
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0417664889915791e-05,
+      "loss": 0.8026,
+      "step": 6053
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0414980489943855e-05,
+      "loss": 0.775,
+      "step": 6054
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0412296060016495e-05,
+      "loss": 0.8084,
+      "step": 6055
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.040961160032748e-05,
+      "loss": 0.7872,
+      "step": 6056
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0406927111070593e-05,
+      "loss": 0.7487,
+      "step": 6057
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0404242592439615e-05,
+      "loss": 0.8028,
+      "step": 6058
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0401558044628323e-05,
+      "loss": 0.7735,
+      "step": 6059
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0398873467830506e-05,
+      "loss": 0.7632,
+      "step": 6060
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0396188862239946e-05,
+      "loss": 0.7863,
+      "step": 6061
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0393504228050437e-05,
+      "loss": 0.7966,
+      "step": 6062
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.039081956545577e-05,
+      "loss": 0.7308,
+      "step": 6063
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0388134874649734e-05,
+      "loss": 0.4165,
+      "step": 6064
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0385450155826124e-05,
+      "loss": 0.8184,
+      "step": 6065
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.038276540917874e-05,
+      "loss": 0.7657,
+      "step": 6066
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0380080634901377e-05,
+      "loss": 0.8006,
+      "step": 6067
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.037739583318784e-05,
+      "loss": 0.8044,
+      "step": 6068
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0374711004231925e-05,
+      "loss": 0.8168,
+      "step": 6069
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0372026148227443e-05,
+      "loss": 0.8432,
+      "step": 6070
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.03693412653682e-05,
+      "loss": 0.7514,
+      "step": 6071
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0366656355848002e-05,
+      "loss": 0.7598,
+      "step": 6072
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0363971419860663e-05,
+      "loss": 0.8574,
+      "step": 6073
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0361286457599992e-05,
+      "loss": 0.7916,
+      "step": 6074
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0358601469259806e-05,
+      "loss": 0.7379,
+      "step": 6075
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0355916455033923e-05,
+      "loss": 0.7624,
+      "step": 6076
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0353231415116154e-05,
+      "loss": 0.7913,
+      "step": 6077
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0350546349700325e-05,
+      "loss": 0.805,
+      "step": 6078
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0347861258980259e-05,
+      "loss": 0.7893,
+      "step": 6079
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0345176143149774e-05,
+      "loss": 0.4341,
+      "step": 6080
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0342491002402697e-05,
+      "loss": 0.7983,
+      "step": 6081
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.033980583693286e-05,
+      "loss": 0.8305,
+      "step": 6082
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0337120646934084e-05,
+      "loss": 0.7828,
+      "step": 6083
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0334435432600212e-05,
+      "loss": 0.7742,
+      "step": 6084
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0331750194125066e-05,
+      "loss": 0.8102,
+      "step": 6085
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0329064931702483e-05,
+      "loss": 0.7601,
+      "step": 6086
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0326379645526299e-05,
+      "loss": 0.8113,
+      "step": 6087
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0323694335790353e-05,
+      "loss": 0.7506,
+      "step": 6088
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0321009002688484e-05,
+      "loss": 0.7553,
+      "step": 6089
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0318323646414534e-05,
+      "loss": 0.7397,
+      "step": 6090
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0315638267162342e-05,
+      "loss": 0.7014,
+      "step": 6091
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0312952865125759e-05,
+      "loss": 0.7991,
+      "step": 6092
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0310267440498622e-05,
+      "loss": 0.7425,
+      "step": 6093
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.030758199347479e-05,
+      "loss": 0.8028,
+      "step": 6094
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0304896524248104e-05,
+      "loss": 0.7612,
+      "step": 6095
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0302211033012419e-05,
+      "loss": 0.7899,
+      "step": 6096
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0299525519961588e-05,
+      "loss": 0.8277,
+      "step": 6097
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0296839985289459e-05,
+      "loss": 0.8049,
+      "step": 6098
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0294154429189898e-05,
+      "loss": 0.7457,
+      "step": 6099
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.029146885185675e-05,
+      "loss": 0.8072,
+      "step": 6100
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0288783253483883e-05,
+      "loss": 0.7907,
+      "step": 6101
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0286097634265154e-05,
+      "loss": 0.7854,
+      "step": 6102
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0283411994394424e-05,
+      "loss": 0.734,
+      "step": 6103
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0280726334065558e-05,
+      "loss": 0.7923,
+      "step": 6104
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0278040653472424e-05,
+      "loss": 0.7388,
+      "step": 6105
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0275354952808881e-05,
+      "loss": 0.7941,
+      "step": 6106
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0272669232268803e-05,
+      "loss": 0.7762,
+      "step": 6107
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0269983492046054e-05,
+      "loss": 0.8447,
+      "step": 6108
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.026729773233451e-05,
+      "loss": 0.7801,
+      "step": 6109
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0264611953328038e-05,
+      "loss": 0.7662,
+      "step": 6110
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0261926155220509e-05,
+      "loss": 0.7619,
+      "step": 6111
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.025924033820581e-05,
+      "loss": 0.7001,
+      "step": 6112
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0256554502477805e-05,
+      "loss": 0.7886,
+      "step": 6113
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0253868648230373e-05,
+      "loss": 0.7416,
+      "step": 6114
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0251182775657403e-05,
+      "loss": 0.7953,
+      "step": 6115
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0248496884952762e-05,
+      "loss": 0.7779,
+      "step": 6116
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0245810976310334e-05,
+      "loss": 0.7977,
+      "step": 6117
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0243125049924012e-05,
+      "loss": 0.797,
+      "step": 6118
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.024043910598767e-05,
+      "loss": 0.8097,
+      "step": 6119
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0237753144695195e-05,
+      "loss": 0.7713,
+      "step": 6120
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0235067166240471e-05,
+      "loss": 0.78,
+      "step": 6121
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0232381170817394e-05,
+      "loss": 0.7464,
+      "step": 6122
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0229695158619848e-05,
+      "loss": 0.8217,
+      "step": 6123
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0227009129841721e-05,
+      "loss": 0.7463,
+      "step": 6124
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0224323084676906e-05,
+      "loss": 0.7646,
+      "step": 6125
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.02216370233193e-05,
+      "loss": 0.7933,
+      "step": 6126
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0218950945962789e-05,
+      "loss": 0.7894,
+      "step": 6127
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0216264852801276e-05,
+      "loss": 0.7934,
+      "step": 6128
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.021357874402865e-05,
+      "loss": 0.7631,
+      "step": 6129
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.021089261983881e-05,
+      "loss": 0.7989,
+      "step": 6130
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0208206480425658e-05,
+      "loss": 0.8458,
+      "step": 6131
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0205520325983087e-05,
+      "loss": 0.8209,
+      "step": 6132
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0202834156705006e-05,
+      "loss": 0.7898,
+      "step": 6133
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0200147972785308e-05,
+      "loss": 0.821,
+      "step": 6134
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.01974617744179e-05,
+      "loss": 0.7397,
+      "step": 6135
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0194775561796685e-05,
+      "loss": 0.7675,
+      "step": 6136
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0192089335115569e-05,
+      "loss": 0.7708,
+      "step": 6137
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0189403094568453e-05,
+      "loss": 0.7481,
+      "step": 6138
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0186716840349251e-05,
+      "loss": 0.7659,
+      "step": 6139
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0184030572651864e-05,
+      "loss": 0.7876,
+      "step": 6140
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0181344291670204e-05,
+      "loss": 0.8348,
+      "step": 6141
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0178657997598179e-05,
+      "loss": 0.7371,
+      "step": 6142
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0175971690629699e-05,
+      "loss": 0.7892,
+      "step": 6143
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.017328537095868e-05,
+      "loss": 0.8079,
+      "step": 6144
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0170599038779028e-05,
+      "loss": 0.7626,
+      "step": 6145
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0167912694284661e-05,
+      "loss": 0.8307,
+      "step": 6146
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.016522633766949e-05,
+      "loss": 0.8014,
+      "step": 6147
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0162539969127432e-05,
+      "loss": 0.8184,
+      "step": 6148
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0159853588852405e-05,
+      "loss": 0.7528,
+      "step": 6149
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.015716719703832e-05,
+      "loss": 0.7897,
+      "step": 6150
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0154480793879098e-05,
+      "loss": 0.7802,
+      "step": 6151
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0151794379568657e-05,
+      "loss": 0.7486,
+      "step": 6152
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0149107954300918e-05,
+      "loss": 0.7769,
+      "step": 6153
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0146421518269801e-05,
+      "loss": 0.8058,
+      "step": 6154
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0143735071669223e-05,
+      "loss": 0.8129,
+      "step": 6155
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0141048614693107e-05,
+      "loss": 0.7398,
+      "step": 6156
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.013836214753538e-05,
+      "loss": 0.8001,
+      "step": 6157
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0135675670389957e-05,
+      "loss": 0.8019,
+      "step": 6158
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.013298918345077e-05,
+      "loss": 0.825,
+      "step": 6159
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0130302686911735e-05,
+      "loss": 0.7864,
+      "step": 6160
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0127616180966781e-05,
+      "loss": 0.8436,
+      "step": 6161
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0124929665809838e-05,
+      "loss": 0.8198,
+      "step": 6162
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0122243141634824e-05,
+      "loss": 0.7507,
+      "step": 6163
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0119556608635675e-05,
+      "loss": 0.7007,
+      "step": 6164
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0116870067006314e-05,
+      "loss": 0.8379,
+      "step": 6165
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0114183516940668e-05,
+      "loss": 0.7832,
+      "step": 6166
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0111496958632672e-05,
+      "loss": 0.781,
+      "step": 6167
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.010881039227625e-05,
+      "loss": 0.7785,
+      "step": 6168
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0106123818065333e-05,
+      "loss": 0.3969,
+      "step": 6169
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0103437236193852e-05,
+      "loss": 0.7404,
+      "step": 6170
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.010075064685574e-05,
+      "loss": 0.8197,
+      "step": 6171
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0098064050244926e-05,
+      "loss": 0.8346,
+      "step": 6172
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0095377446555345e-05,
+      "loss": 0.7768,
+      "step": 6173
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.009269083598093e-05,
+      "loss": 0.7804,
+      "step": 6174
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0090004218715614e-05,
+      "loss": 0.7571,
+      "step": 6175
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0087317594953327e-05,
+      "loss": 0.8103,
+      "step": 6176
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0084630964888009e-05,
+      "loss": 0.7684,
+      "step": 6177
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0081944328713595e-05,
+      "loss": 0.8372,
+      "step": 6178
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0079257686624014e-05,
+      "loss": 0.7233,
+      "step": 6179
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0076571038813208e-05,
+      "loss": 0.8,
+      "step": 6180
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0073884385475107e-05,
+      "loss": 0.8448,
+      "step": 6181
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0071197726803652e-05,
+      "loss": 0.7708,
+      "step": 6182
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0068511062992778e-05,
+      "loss": 0.7607,
+      "step": 6183
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0065824394236423e-05,
+      "loss": 0.7687,
+      "step": 6184
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0063137720728527e-05,
+      "loss": 0.8039,
+      "step": 6185
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0060451042663024e-05,
+      "loss": 0.8098,
+      "step": 6186
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0057764360233853e-05,
+      "loss": 0.7676,
+      "step": 6187
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0055077673634954e-05,
+      "loss": 0.7675,
+      "step": 6188
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0052390983060265e-05,
+      "loss": 0.7642,
+      "step": 6189
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0049704288703723e-05,
+      "loss": 0.7861,
+      "step": 6190
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0047017590759274e-05,
+      "loss": 0.792,
+      "step": 6191
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0044330889420848e-05,
+      "loss": 0.8278,
+      "step": 6192
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0041644184882394e-05,
+      "loss": 0.8467,
+      "step": 6193
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0038957477337846e-05,
+      "loss": 0.798,
+      "step": 6194
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0036270766981146e-05,
+      "loss": 0.777,
+      "step": 6195
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0033584054006241e-05,
+      "loss": 0.7649,
+      "step": 6196
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0030897338607061e-05,
+      "loss": 0.7817,
+      "step": 6197
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0028210620977552e-05,
+      "loss": 0.7757,
+      "step": 6198
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0025523901311661e-05,
+      "loss": 0.7604,
+      "step": 6199
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0022837179803317e-05,
+      "loss": 0.7838,
+      "step": 6200
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0020150456646471e-05,
+      "loss": 0.7607,
+      "step": 6201
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0017463732035059e-05,
+      "loss": 0.7337,
+      "step": 6202
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0014777006163028e-05,
+      "loss": 0.7552,
+      "step": 6203
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0012090279224314e-05,
+      "loss": 0.7921,
+      "step": 6204
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0009403551412863e-05,
+      "loss": 0.7582,
+      "step": 6205
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0006716822922616e-05,
+      "loss": 0.8066,
+      "step": 6206
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.000403009394751e-05,
+      "loss": 0.7985,
+      "step": 6207
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0001343364681495e-05,
+      "loss": 0.7655,
+      "step": 6208
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.998656635318506e-06,
+      "loss": 0.7482,
+      "step": 6209
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.995969906052491e-06,
+      "loss": 0.8044,
+      "step": 6210
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.99328317707739e-06,
+      "loss": 0.7281,
+      "step": 6211
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.990596448587139e-06,
+      "loss": 0.8264,
+      "step": 6212
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.98790972077569e-06,
+      "loss": 0.7769,
+      "step": 6213
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.985222993836977e-06,
+      "loss": 0.7874,
+      "step": 6214
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.982536267964943e-06,
+      "loss": 0.7853,
+      "step": 6215
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.97984954335353e-06,
+      "loss": 0.7948,
+      "step": 6216
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.977162820196685e-06,
+      "loss": 0.7889,
+      "step": 6217
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.974476098688346e-06,
+      "loss": 0.7738,
+      "step": 6218
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.971789379022448e-06,
+      "loss": 0.7925,
+      "step": 6219
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.969102661392942e-06,
+      "loss": 0.8168,
+      "step": 6220
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.966415945993764e-06,
+      "loss": 0.7775,
+      "step": 6221
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.963729233018854e-06,
+      "loss": 0.8495,
+      "step": 6222
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.961042522662156e-06,
+      "loss": 0.7739,
+      "step": 6223
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.95835581511761e-06,
+      "loss": 0.799,
+      "step": 6224
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.955669110579154e-06,
+      "loss": 0.7638,
+      "step": 6225
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.95298240924073e-06,
+      "loss": 0.8436,
+      "step": 6226
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.95029571129628e-06,
+      "loss": 0.81,
+      "step": 6227
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.947609016939736e-06,
+      "loss": 0.8352,
+      "step": 6228
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.94492232636505e-06,
+      "loss": 0.7759,
+      "step": 6229
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.942235639766152e-06,
+      "loss": 0.7608,
+      "step": 6230
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.939548957336983e-06,
+      "loss": 0.7272,
+      "step": 6231
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.936862279271475e-06,
+      "loss": 0.7732,
+      "step": 6232
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.93417560576358e-06,
+      "loss": 0.7945,
+      "step": 6233
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.931488937007225e-06,
+      "loss": 0.7713,
+      "step": 6234
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.928802273196351e-06,
+      "loss": 0.8214,
+      "step": 6235
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.926115614524894e-06,
+      "loss": 0.7463,
+      "step": 6236
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.923428961186797e-06,
+      "loss": 0.8161,
+      "step": 6237
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.920742313375986e-06,
+      "loss": 0.774,
+      "step": 6238
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.918055671286409e-06,
+      "loss": 0.8104,
+      "step": 6239
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.915369035111994e-06,
+      "loss": 0.7614,
+      "step": 6240
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.912682405046671e-06,
+      "loss": 0.4003,
+      "step": 6241
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.90999578128439e-06,
+      "loss": 0.8268,
+      "step": 6242
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.907309164019074e-06,
+      "loss": 0.7684,
+      "step": 6243
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.904622553444657e-06,
+      "loss": 0.7711,
+      "step": 6244
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.901935949755077e-06,
+      "loss": 0.8054,
+      "step": 6245
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.899249353144264e-06,
+      "loss": 0.7054,
+      "step": 6246
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.896562763806152e-06,
+      "loss": 0.7573,
+      "step": 6247
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.893876181934669e-06,
+      "loss": 0.8022,
+      "step": 6248
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.891189607723755e-06,
+      "loss": 0.8453,
+      "step": 6249
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.888503041367333e-06,
+      "loss": 0.7987,
+      "step": 6250
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.885816483059332e-06,
+      "loss": 0.7732,
+      "step": 6251
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.883129932993689e-06,
+      "loss": 0.873,
+      "step": 6252
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.88044339136433e-06,
+      "loss": 0.7894,
+      "step": 6253
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.877756858365176e-06,
+      "loss": 0.7483,
+      "step": 6254
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.875070334190167e-06,
+      "loss": 0.7725,
+      "step": 6255
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.872383819033222e-06,
+      "loss": 0.8093,
+      "step": 6256
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.86969731308827e-06,
+      "loss": 0.8322,
+      "step": 6257
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.867010816549234e-06,
+      "loss": 0.7841,
+      "step": 6258
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.864324329610046e-06,
+      "loss": 0.7886,
+      "step": 6259
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.861637852464626e-06,
+      "loss": 0.8232,
+      "step": 6260
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.858951385306893e-06,
+      "loss": 0.7784,
+      "step": 6261
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.85626492833078e-06,
+      "loss": 0.8686,
+      "step": 6262
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.853578481730204e-06,
+      "loss": 0.8207,
+      "step": 6263
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.850892045699082e-06,
+      "loss": 0.799,
+      "step": 6264
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.848205620431344e-06,
+      "loss": 0.8029,
+      "step": 6265
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.845519206120903e-06,
+      "loss": 0.8195,
+      "step": 6266
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.842832802961683e-06,
+      "loss": 0.7775,
+      "step": 6267
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.840146411147599e-06,
+      "loss": 0.8042,
+      "step": 6268
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.837460030872571e-06,
+      "loss": 0.7559,
+      "step": 6269
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.834773662330514e-06,
+      "loss": 0.8046,
+      "step": 6270
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.832087305715342e-06,
+      "loss": 0.7694,
+      "step": 6271
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.829400961220976e-06,
+      "loss": 0.8273,
+      "step": 6272
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.826714629041326e-06,
+      "loss": 0.7477,
+      "step": 6273
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.824028309370303e-06,
+      "loss": 0.7949,
+      "step": 6274
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.821342002401826e-06,
+      "loss": 0.7236,
+      "step": 6275
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.818655708329801e-06,
+      "loss": 0.7775,
+      "step": 6276
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.815969427348138e-06,
+      "loss": 0.7736,
+      "step": 6277
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.813283159650752e-06,
+      "loss": 0.7657,
+      "step": 6278
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.810596905431549e-06,
+      "loss": 0.7947,
+      "step": 6279
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.807910664884433e-06,
+      "loss": 0.7475,
+      "step": 6280
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.805224438203316e-06,
+      "loss": 0.8272,
+      "step": 6281
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.802538225582103e-06,
+      "loss": 0.8212,
+      "step": 6282
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.799852027214697e-06,
+      "loss": 0.7346,
+      "step": 6283
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.797165843294999e-06,
+      "loss": 0.8231,
+      "step": 6284
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.794479674016915e-06,
+      "loss": 0.7834,
+      "step": 6285
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.791793519574346e-06,
+      "loss": 0.7962,
+      "step": 6286
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.789107380161192e-06,
+      "loss": 0.7978,
+      "step": 6287
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.786421255971353e-06,
+      "loss": 0.7748,
+      "step": 6288
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.783735147198729e-06,
+      "loss": 0.7468,
+      "step": 6289
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.781049054037211e-06,
+      "loss": 0.7465,
+      "step": 6290
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.778362976680703e-06,
+      "loss": 0.8348,
+      "step": 6291
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.775676915323096e-06,
+      "loss": 0.7324,
+      "step": 6292
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.77299087015828e-06,
+      "loss": 0.7646,
+      "step": 6293
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.770304841380156e-06,
+      "loss": 0.8077,
+      "step": 6294
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.767618829182611e-06,
+      "loss": 0.818,
+      "step": 6295
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.76493283375953e-06,
+      "loss": 0.8231,
+      "step": 6296
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.762246855304807e-06,
+      "loss": 0.8088,
+      "step": 6297
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.759560894012335e-06,
+      "loss": 0.7668,
+      "step": 6298
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.756874950075993e-06,
+      "loss": 0.8096,
+      "step": 6299
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.754189023689664e-06,
+      "loss": 0.7471,
+      "step": 6300
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.751503115047242e-06,
+      "loss": 0.8111,
+      "step": 6301
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.748817224342604e-06,
+      "loss": 0.8169,
+      "step": 6302
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.746131351769626e-06,
+      "loss": 0.7614,
+      "step": 6303
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.7434454975222e-06,
+      "loss": 0.7766,
+      "step": 6304
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.740759661794195e-06,
+      "loss": 0.841,
+      "step": 6305
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.738073844779491e-06,
+      "loss": 0.8279,
+      "step": 6306
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.735388046671965e-06,
+      "loss": 0.8312,
+      "step": 6307
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.732702267665494e-06,
+      "loss": 0.8013,
+      "step": 6308
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.73001650795395e-06,
+      "loss": 0.7379,
+      "step": 6309
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.727330767731198e-06,
+      "loss": 0.7645,
+      "step": 6310
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.72464504719112e-06,
+      "loss": 0.7685,
+      "step": 6311
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.721959346527581e-06,
+      "loss": 0.8537,
+      "step": 6312
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.719273665934442e-06,
+      "loss": 0.7952,
+      "step": 6313
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.716588005605578e-06,
+      "loss": 0.396,
+      "step": 6314
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.713902365734851e-06,
+      "loss": 0.7401,
+      "step": 6315
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.71121674651612e-06,
+      "loss": 0.8338,
+      "step": 6316
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.708531148143252e-06,
+      "loss": 0.8057,
+      "step": 6317
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.705845570810108e-06,
+      "loss": 0.8458,
+      "step": 6318
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.703160014710541e-06,
+      "loss": 0.7649,
+      "step": 6319
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.700474480038415e-06,
+      "loss": 0.8229,
+      "step": 6320
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.697788966987584e-06,
+      "loss": 0.7224,
+      "step": 6321
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.6951034757519e-06,
+      "loss": 0.7899,
+      "step": 6322
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.692418006525211e-06,
+      "loss": 0.7489,
+      "step": 6323
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.68973255950138e-06,
+      "loss": 0.767,
+      "step": 6324
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.687047134874248e-06,
+      "loss": 0.8173,
+      "step": 6325
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.684361732837661e-06,
+      "loss": 0.7728,
+      "step": 6326
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.68167635358547e-06,
+      "loss": 0.7643,
+      "step": 6327
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.678990997311521e-06,
+      "loss": 0.7129,
+      "step": 6328
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.676305664209648e-06,
+      "loss": 0.8301,
+      "step": 6329
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.673620354473703e-06,
+      "loss": 0.7819,
+      "step": 6330
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.670935068297522e-06,
+      "loss": 0.7982,
+      "step": 6331
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.668249805874936e-06,
+      "loss": 0.7036,
+      "step": 6332
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.665564567399791e-06,
+      "loss": 0.7852,
+      "step": 6333
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.662879353065917e-06,
+      "loss": 0.7644,
+      "step": 6334
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.660194163067142e-06,
+      "loss": 0.7756,
+      "step": 6335
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.657508997597306e-06,
+      "loss": 0.7796,
+      "step": 6336
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.65482385685023e-06,
+      "loss": 0.874,
+      "step": 6337
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.652138741019746e-06,
+      "loss": 0.7773,
+      "step": 6338
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.649453650299677e-06,
+      "loss": 0.7963,
+      "step": 6339
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.64676858488385e-06,
+      "loss": 0.8341,
+      "step": 6340
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.644083544966084e-06,
+      "loss": 0.8135,
+      "step": 6341
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.641398530740194e-06,
+      "loss": 0.7059,
+      "step": 6342
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.63871354240001e-06,
+      "loss": 0.7706,
+      "step": 6343
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.63602858013934e-06,
+      "loss": 0.7308,
+      "step": 6344
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.633343644152e-06,
+      "loss": 0.7925,
+      "step": 6345
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.630658734631803e-06,
+      "loss": 0.8274,
+      "step": 6346
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.62797385177256e-06,
+      "loss": 0.828,
+      "step": 6347
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.625288995768077e-06,
+      "loss": 0.7876,
+      "step": 6348
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.622604166812164e-06,
+      "loss": 0.8038,
+      "step": 6349
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.619919365098627e-06,
+      "loss": 0.8049,
+      "step": 6350
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.617234590821266e-06,
+      "loss": 0.7556,
+      "step": 6351
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.614549844173877e-06,
+      "loss": 0.7886,
+      "step": 6352
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.611865125350269e-06,
+      "loss": 0.7999,
+      "step": 6353
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.609180434544234e-06,
+      "loss": 0.7338,
+      "step": 6354
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.606495771949561e-06,
+      "loss": 0.8033,
+      "step": 6355
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.603811137760056e-06,
+      "loss": 0.8315,
+      "step": 6356
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.601126532169498e-06,
+      "loss": 0.7532,
+      "step": 6357
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.59844195537168e-06,
+      "loss": 0.7677,
+      "step": 6358
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.595757407560388e-06,
+      "loss": 0.7597,
+      "step": 6359
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.593072888929409e-06,
+      "loss": 0.7654,
+      "step": 6360
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.59038839967252e-06,
+      "loss": 0.7579,
+      "step": 6361
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.587703939983507e-06,
+      "loss": 0.7654,
+      "step": 6362
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.585019510056147e-06,
+      "loss": 0.7975,
+      "step": 6363
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.582335110084214e-06,
+      "loss": 0.7931,
+      "step": 6364
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.579650740261479e-06,
+      "loss": 0.7334,
+      "step": 6365
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.57696640078172e-06,
+      "loss": 0.8089,
+      "step": 6366
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.574282091838702e-06,
+      "loss": 0.7954,
+      "step": 6367
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.571597813626192e-06,
+      "loss": 0.8825,
+      "step": 6368
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.568913566337958e-06,
+      "loss": 0.7322,
+      "step": 6369
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.566229350167763e-06,
+      "loss": 0.7707,
+      "step": 6370
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.563545165309363e-06,
+      "loss": 0.7785,
+      "step": 6371
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.560861011956522e-06,
+      "loss": 0.7884,
+      "step": 6372
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.558176890302996e-06,
+      "loss": 0.7502,
+      "step": 6373
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.55549280054253e-06,
+      "loss": 0.8274,
+      "step": 6374
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.552808742868884e-06,
+      "loss": 0.4029,
+      "step": 6375
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.550124717475806e-06,
+      "loss": 0.7598,
+      "step": 6376
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.547440724557039e-06,
+      "loss": 0.7437,
+      "step": 6377
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.54475676430633e-06,
+      "loss": 0.7564,
+      "step": 6378
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.54207283691742e-06,
+      "loss": 0.7903,
+      "step": 6379
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.539388942584052e-06,
+      "loss": 0.7713,
+      "step": 6380
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.536705081499957e-06,
+      "loss": 0.7513,
+      "step": 6381
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.534021253858877e-06,
+      "loss": 0.6967,
+      "step": 6382
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.531337459854542e-06,
+      "loss": 0.7956,
+      "step": 6383
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.528653699680677e-06,
+      "loss": 0.7932,
+      "step": 6384
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.525969973531018e-06,
+      "loss": 0.7795,
+      "step": 6385
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.523286281599286e-06,
+      "loss": 0.8086,
+      "step": 6386
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.520602624079199e-06,
+      "loss": 0.8483,
+      "step": 6387
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.517919001164487e-06,
+      "loss": 0.7502,
+      "step": 6388
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.515235413048858e-06,
+      "loss": 0.3829,
+      "step": 6389
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.512551859926038e-06,
+      "loss": 0.8154,
+      "step": 6390
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.509868341989725e-06,
+      "loss": 0.7544,
+      "step": 6391
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.507184859433644e-06,
+      "loss": 0.7611,
+      "step": 6392
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.504501412451497e-06,
+      "loss": 0.8002,
+      "step": 6393
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.501818001236982e-06,
+      "loss": 0.7723,
+      "step": 6394
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.499134625983814e-06,
+      "loss": 0.7555,
+      "step": 6395
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.496451286885686e-06,
+      "loss": 0.737,
+      "step": 6396
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.49376798413629e-06,
+      "loss": 0.796,
+      "step": 6397
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.491084717929332e-06,
+      "loss": 0.7569,
+      "step": 6398
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.488401488458495e-06,
+      "loss": 0.7307,
+      "step": 6399
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.485718295917473e-06,
+      "loss": 0.4212,
+      "step": 6400
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.48303514049995e-06,
+      "loss": 0.7623,
+      "step": 6401
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.480352022399615e-06,
+      "loss": 0.7483,
+      "step": 6402
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.477668941810144e-06,
+      "loss": 0.7236,
+      "step": 6403
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.474985898925213e-06,
+      "loss": 0.7993,
+      "step": 6404
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.472302893938506e-06,
+      "loss": 0.7772,
+      "step": 6405
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.469619927043693e-06,
+      "loss": 0.8338,
+      "step": 6406
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.466936998434438e-06,
+      "loss": 0.7718,
+      "step": 6407
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.464254108304419e-06,
+      "loss": 0.8467,
+      "step": 6408
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.461571256847295e-06,
+      "loss": 0.7758,
+      "step": 6409
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.458888444256725e-06,
+      "loss": 0.8451,
+      "step": 6410
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.456205670726375e-06,
+      "loss": 0.7495,
+      "step": 6411
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.453522936449901e-06,
+      "loss": 0.737,
+      "step": 6412
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.450840241620948e-06,
+      "loss": 0.7435,
+      "step": 6413
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.448157586433176e-06,
+      "loss": 0.7883,
+      "step": 6414
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.445474971080231e-06,
+      "loss": 0.7829,
+      "step": 6415
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.442792395755758e-06,
+      "loss": 0.8344,
+      "step": 6416
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.440109860653392e-06,
+      "loss": 0.7536,
+      "step": 6417
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.437427365966783e-06,
+      "loss": 0.7648,
+      "step": 6418
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.434744911889559e-06,
+      "loss": 0.8264,
+      "step": 6419
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.432062498615355e-06,
+      "loss": 0.782,
+      "step": 6420
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.429380126337805e-06,
+      "loss": 0.7678,
+      "step": 6421
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.426697795250538e-06,
+      "loss": 0.7237,
+      "step": 6422
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.424015505547165e-06,
+      "loss": 0.8083,
+      "step": 6423
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.421333257421327e-06,
+      "loss": 0.8267,
+      "step": 6424
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.41865105106663e-06,
+      "loss": 0.8191,
+      "step": 6425
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.415968886676688e-06,
+      "loss": 0.7216,
+      "step": 6426
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.413286764445123e-06,
+      "loss": 0.732,
+      "step": 6427
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.410604684565538e-06,
+      "loss": 0.7608,
+      "step": 6428
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.407922647231538e-06,
+      "loss": 0.8149,
+      "step": 6429
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.40524065263673e-06,
+      "loss": 0.8278,
+      "step": 6430
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.402558700974711e-06,
+      "loss": 0.6935,
+      "step": 6431
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.399876792439082e-06,
+      "loss": 0.8076,
+      "step": 6432
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.397194927223432e-06,
+      "loss": 0.7834,
+      "step": 6433
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.394513105521359e-06,
+      "loss": 0.8003,
+      "step": 6434
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.391831327526445e-06,
+      "loss": 0.8153,
+      "step": 6435
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.389149593432273e-06,
+      "loss": 0.7762,
+      "step": 6436
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.38646790343243e-06,
+      "loss": 0.7079,
+      "step": 6437
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.383786257720492e-06,
+      "loss": 0.7701,
+      "step": 6438
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.38110465649003e-06,
+      "loss": 0.7721,
+      "step": 6439
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.378423099934622e-06,
+      "loss": 0.7902,
+      "step": 6440
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.37574158824783e-06,
+      "loss": 0.8119,
+      "step": 6441
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.373060121623229e-06,
+      "loss": 0.8227,
+      "step": 6442
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.370378700254368e-06,
+      "loss": 0.7508,
+      "step": 6443
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.367697324334818e-06,
+      "loss": 0.7609,
+      "step": 6444
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.36501599405813e-06,
+      "loss": 0.7638,
+      "step": 6445
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.36233470961785e-06,
+      "loss": 0.7337,
+      "step": 6446
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.359653471207538e-06,
+      "loss": 0.834,
+      "step": 6447
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.356972279020734e-06,
+      "loss": 0.8298,
+      "step": 6448
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.354291133250976e-06,
+      "loss": 0.8068,
+      "step": 6449
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.351610034091812e-06,
+      "loss": 0.7925,
+      "step": 6450
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.34892898173677e-06,
+      "loss": 0.8321,
+      "step": 6451
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.346247976379385e-06,
+      "loss": 0.7687,
+      "step": 6452
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.343567018213185e-06,
+      "loss": 0.8275,
+      "step": 6453
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.3408861074317e-06,
+      "loss": 0.8309,
+      "step": 6454
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.338205244228447e-06,
+      "loss": 0.7584,
+      "step": 6455
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.335524428796941e-06,
+      "loss": 0.4066,
+      "step": 6456
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.332843661330707e-06,
+      "loss": 0.7862,
+      "step": 6457
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.330162942023249e-06,
+      "loss": 0.7878,
+      "step": 6458
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.327482271068076e-06,
+      "loss": 0.401,
+      "step": 6459
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.324801648658695e-06,
+      "loss": 0.7787,
+      "step": 6460
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.322121074988605e-06,
+      "loss": 0.7989,
+      "step": 6461
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.319440550251307e-06,
+      "loss": 0.7844,
+      "step": 6462
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.31676007464029e-06,
+      "loss": 0.7576,
+      "step": 6463
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.31407964834905e-06,
+      "loss": 0.8569,
+      "step": 6464
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.311399271571068e-06,
+      "loss": 0.7366,
+      "step": 6465
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.308718944499835e-06,
+      "loss": 0.8284,
+      "step": 6466
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.306038667328825e-06,
+      "loss": 0.809,
+      "step": 6467
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.30335844025151e-06,
+      "loss": 0.7558,
+      "step": 6468
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.300678263461376e-06,
+      "loss": 0.7452,
+      "step": 6469
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.297998137151882e-06,
+      "loss": 0.7884,
+      "step": 6470
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.295318061516494e-06,
+      "loss": 0.6777,
+      "step": 6471
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.292638036748674e-06,
+      "loss": 0.7588,
+      "step": 6472
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.289958063041882e-06,
+      "loss": 0.7479,
+      "step": 6473
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.287278140589575e-06,
+      "loss": 0.4367,
+      "step": 6474
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.284598269585193e-06,
+      "loss": 0.7692,
+      "step": 6475
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.281918450222195e-06,
+      "loss": 0.7759,
+      "step": 6476
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.27923868269402e-06,
+      "loss": 0.7955,
+      "step": 6477
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.2765589671941e-06,
+      "loss": 0.7234,
+      "step": 6478
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.273879303915884e-06,
+      "loss": 0.7185,
+      "step": 6479
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.271199693052795e-06,
+      "loss": 0.7911,
+      "step": 6480
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.26852013479826e-06,
+      "loss": 0.7821,
+      "step": 6481
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.26584062934571e-06,
+      "loss": 0.7621,
+      "step": 6482
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.26316117688856e-06,
+      "loss": 0.3727,
+      "step": 6483
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.26048177762023e-06,
+      "loss": 0.7573,
+      "step": 6484
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.257802431734128e-06,
+      "loss": 0.7805,
+      "step": 6485
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.255123139423671e-06,
+      "loss": 0.7503,
+      "step": 6486
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.252443900882258e-06,
+      "loss": 0.8118,
+      "step": 6487
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.249764716303289e-06,
+      "loss": 0.7577,
+      "step": 6488
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.247085585880166e-06,
+      "loss": 0.8129,
+      "step": 6489
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.244406509806284e-06,
+      "loss": 0.797,
+      "step": 6490
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.241727488275021e-06,
+      "loss": 0.7671,
+      "step": 6491
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.239048521479775e-06,
+      "loss": 0.7731,
+      "step": 6492
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.236369609613924e-06,
+      "loss": 0.8364,
+      "step": 6493
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.233690752870841e-06,
+      "loss": 0.803,
+      "step": 6494
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.231011951443907e-06,
+      "loss": 0.6974,
+      "step": 6495
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.22833320552649e-06,
+      "loss": 0.7557,
+      "step": 6496
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.225654515311953e-06,
+      "loss": 0.8326,
+      "step": 6497
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.222975880993652e-06,
+      "loss": 0.8109,
+      "step": 6498
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.220297302764958e-06,
+      "loss": 0.7471,
+      "step": 6499
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.217618780819217e-06,
+      "loss": 0.8142,
+      "step": 6500
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.214940315349773e-06,
+      "loss": 0.8091,
+      "step": 6501
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.212261906549981e-06,
+      "loss": 0.7921,
+      "step": 6502
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.20958355461318e-06,
+      "loss": 0.7734,
+      "step": 6503
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.206905259732704e-06,
+      "loss": 0.7773,
+      "step": 6504
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.204227022101891e-06,
+      "loss": 0.7656,
+      "step": 6505
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.201548841914068e-06,
+      "loss": 0.8116,
+      "step": 6506
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.198870719362553e-06,
+      "loss": 0.7881,
+      "step": 6507
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.196192654640677e-06,
+      "loss": 0.7453,
+      "step": 6508
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.193514647941755e-06,
+      "loss": 0.7647,
+      "step": 6509
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.190836699459093e-06,
+      "loss": 0.7963,
+      "step": 6510
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.188158809386e-06,
+      "loss": 0.7657,
+      "step": 6511
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.185480977915784e-06,
+      "loss": 0.7562,
+      "step": 6512
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.182803205241746e-06,
+      "loss": 0.8639,
+      "step": 6513
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.180125491557174e-06,
+      "loss": 0.7487,
+      "step": 6514
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.177447837055368e-06,
+      "loss": 0.8184,
+      "step": 6515
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.17477024192961e-06,
+      "loss": 0.747,
+      "step": 6516
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.172092706373178e-06,
+      "loss": 0.8076,
+      "step": 6517
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.169415230579363e-06,
+      "loss": 0.7774,
+      "step": 6518
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.166737814741428e-06,
+      "loss": 0.8048,
+      "step": 6519
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.164060459052645e-06,
+      "loss": 0.7703,
+      "step": 6520
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.161383163706282e-06,
+      "loss": 0.7459,
+      "step": 6521
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.158705928895596e-06,
+      "loss": 0.7144,
+      "step": 6522
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.156028754813851e-06,
+      "loss": 0.7399,
+      "step": 6523
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.153351641654287e-06,
+      "loss": 0.7573,
+      "step": 6524
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.150674589610164e-06,
+      "loss": 0.7911,
+      "step": 6525
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.147997598874721e-06,
+      "loss": 0.8121,
+      "step": 6526
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.145320669641192e-06,
+      "loss": 0.7183,
+      "step": 6527
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.14264380210282e-06,
+      "loss": 0.7418,
+      "step": 6528
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.13996699645283e-06,
+      "loss": 0.7456,
+      "step": 6529
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.137290252884445e-06,
+      "loss": 0.7559,
+      "step": 6530
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.134613571590894e-06,
+      "loss": 0.6891,
+      "step": 6531
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.131936952765387e-06,
+      "loss": 0.7874,
+      "step": 6532
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.129260396601137e-06,
+      "loss": 0.8045,
+      "step": 6533
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.126583903291355e-06,
+      "loss": 0.8008,
+      "step": 6534
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.123907473029242e-06,
+      "loss": 0.7633,
+      "step": 6535
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.121231106008e-06,
+      "loss": 0.772,
+      "step": 6536
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.118554802420813e-06,
+      "loss": 0.3869,
+      "step": 6537
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.115878562460882e-06,
+      "loss": 0.7772,
+      "step": 6538
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.113202386321387e-06,
+      "loss": 0.775,
+      "step": 6539
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.110526274195502e-06,
+      "loss": 0.8107,
+      "step": 6540
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.107850226276413e-06,
+      "loss": 0.749,
+      "step": 6541
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.105174242757286e-06,
+      "loss": 0.777,
+      "step": 6542
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.102498323831286e-06,
+      "loss": 0.8332,
+      "step": 6543
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.099822469691576e-06,
+      "loss": 0.745,
+      "step": 6544
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.097146680531315e-06,
+      "loss": 0.7544,
+      "step": 6545
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.09447095654365e-06,
+      "loss": 0.8054,
+      "step": 6546
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.091795297921735e-06,
+      "loss": 0.7711,
+      "step": 6547
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.08911970485871e-06,
+      "loss": 0.7856,
+      "step": 6548
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.086444177547715e-06,
+      "loss": 0.7679,
+      "step": 6549
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.083768716181875e-06,
+      "loss": 0.82,
+      "step": 6550
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.081093320954329e-06,
+      "loss": 0.856,
+      "step": 6551
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.078417992058192e-06,
+      "loss": 0.7707,
+      "step": 6552
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.07574272968659e-06,
+      "loss": 0.7733,
+      "step": 6553
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.073067534032634e-06,
+      "loss": 0.8339,
+      "step": 6554
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.070392405289436e-06,
+      "loss": 0.7188,
+      "step": 6555
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.067717343650096e-06,
+      "loss": 0.772,
+      "step": 6556
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.06504234930772e-06,
+      "loss": 0.7178,
+      "step": 6557
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.062367422455397e-06,
+      "loss": 0.7914,
+      "step": 6558
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.059692563286216e-06,
+      "loss": 0.7233,
+      "step": 6559
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.05701777199327e-06,
+      "loss": 0.7635,
+      "step": 6560
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.054343048769635e-06,
+      "loss": 0.7851,
+      "step": 6561
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.051668393808383e-06,
+      "loss": 0.7679,
+      "step": 6562
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.048993807302586e-06,
+      "loss": 0.8283,
+      "step": 6563
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.046319289445312e-06,
+      "loss": 0.8198,
+      "step": 6564
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.043644840429622e-06,
+      "loss": 0.8195,
+      "step": 6565
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.040970460448565e-06,
+      "loss": 0.7575,
+      "step": 6566
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.0382961496952e-06,
+      "loss": 0.3976,
+      "step": 6567
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.035621908362569e-06,
+      "loss": 0.9013,
+      "step": 6568
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.032947736643707e-06,
+      "loss": 0.8163,
+      "step": 6569
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.03027363473166e-06,
+      "loss": 0.8108,
+      "step": 6570
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.027599602819453e-06,
+      "loss": 0.8431,
+      "step": 6571
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.024925641100106e-06,
+      "loss": 0.7451,
+      "step": 6572
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.022251749766649e-06,
+      "loss": 0.7463,
+      "step": 6573
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.01957792901209e-06,
+      "loss": 0.7511,
+      "step": 6574
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.016904179029445e-06,
+      "loss": 0.747,
+      "step": 6575
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.014230500011712e-06,
+      "loss": 0.7976,
+      "step": 6576
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.011556892151899e-06,
+      "loss": 0.7801,
+      "step": 6577
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.008883355642996e-06,
+      "loss": 0.704,
+      "step": 6578
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.00620989067799e-06,
+      "loss": 0.7692,
+      "step": 6579
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.003536497449874e-06,
+      "loss": 0.749,
+      "step": 6580
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.00086317615162e-06,
+      "loss": 0.7982,
+      "step": 6581
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.998189926976203e-06,
+      "loss": 0.7625,
+      "step": 6582
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.995516750116594e-06,
+      "loss": 0.8446,
+      "step": 6583
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.992843645765755e-06,
+      "loss": 0.7812,
+      "step": 6584
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.990170614116646e-06,
+      "loss": 0.8116,
+      "step": 6585
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.987497655362219e-06,
+      "loss": 0.4229,
+      "step": 6586
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.984824769695422e-06,
+      "loss": 0.812,
+      "step": 6587
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.982151957309196e-06,
+      "loss": 0.7707,
+      "step": 6588
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.979479218396485e-06,
+      "loss": 0.7401,
+      "step": 6589
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.976806553150216e-06,
+      "loss": 0.795,
+      "step": 6590
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.974133961763316e-06,
+      "loss": 0.8219,
+      "step": 6591
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.971461444428701e-06,
+      "loss": 0.7836,
+      "step": 6592
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.968789001339297e-06,
+      "loss": 0.8292,
+      "step": 6593
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.966116632688008e-06,
+      "loss": 0.7761,
+      "step": 6594
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.963444338667743e-06,
+      "loss": 0.7317,
+      "step": 6595
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.9607721194714e-06,
+      "loss": 0.766,
+      "step": 6596
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.958099975291876e-06,
+      "loss": 0.7679,
+      "step": 6597
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.955427906322053e-06,
+      "loss": 0.7222,
+      "step": 6598
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.952755912754825e-06,
+      "loss": 0.7654,
+      "step": 6599
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.950083994783064e-06,
+      "loss": 0.7353,
+      "step": 6600
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.947412152599641e-06,
+      "loss": 0.394,
+      "step": 6601
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.94474038639743e-06,
+      "loss": 0.7789,
+      "step": 6602
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.942068696369289e-06,
+      "loss": 0.7286,
+      "step": 6603
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.939397082708073e-06,
+      "loss": 0.7771,
+      "step": 6604
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.936725545606632e-06,
+      "loss": 0.8196,
+      "step": 6605
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.934054085257816e-06,
+      "loss": 0.8399,
+      "step": 6606
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.931382701854463e-06,
+      "loss": 0.7607,
+      "step": 6607
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.9287113955894e-06,
+      "loss": 0.7485,
+      "step": 6608
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.92604016665547e-06,
+      "loss": 0.7559,
+      "step": 6609
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.923369015245487e-06,
+      "loss": 0.8655,
+      "step": 6610
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.920697941552264e-06,
+      "loss": 0.7622,
+      "step": 6611
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.918026945768625e-06,
+      "loss": 0.7883,
+      "step": 6612
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.915356028087368e-06,
+      "loss": 0.8121,
+      "step": 6613
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.912685188701291e-06,
+      "loss": 0.8415,
+      "step": 6614
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.910014427803198e-06,
+      "loss": 0.8407,
+      "step": 6615
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.90734374558587e-06,
+      "loss": 0.7829,
+      "step": 6616
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.904673142242098e-06,
+      "loss": 0.7528,
+      "step": 6617
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.90200261796465e-06,
+      "loss": 0.736,
+      "step": 6618
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.89933217294631e-06,
+      "loss": 0.7913,
+      "step": 6619
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.896661807379837e-06,
+      "loss": 0.7654,
+      "step": 6620
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.893991521457989e-06,
+      "loss": 0.8016,
+      "step": 6621
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.89132131537353e-06,
+      "loss": 0.8101,
+      "step": 6622
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.888651189319207e-06,
+      "loss": 0.7387,
+      "step": 6623
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.885981143487755e-06,
+      "loss": 0.7529,
+      "step": 6624
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.883311178071921e-06,
+      "loss": 0.7601,
+      "step": 6625
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.880641293264431e-06,
+      "loss": 0.7305,
+      "step": 6626
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.877971489258014e-06,
+      "loss": 0.7996,
+      "step": 6627
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.87530176624539e-06,
+      "loss": 0.8463,
+      "step": 6628
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.872632124419275e-06,
+      "loss": 0.7012,
+      "step": 6629
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.869962563972376e-06,
+      "loss": 0.7863,
+      "step": 6630
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.867293085097388e-06,
+      "loss": 0.751,
+      "step": 6631
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.864623687987022e-06,
+      "loss": 0.8002,
+      "step": 6632
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.86195437283396e-06,
+      "loss": 0.7699,
+      "step": 6633
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.859285139830886e-06,
+      "loss": 0.8261,
+      "step": 6634
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.856615989170485e-06,
+      "loss": 0.7543,
+      "step": 6635
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.853946921045425e-06,
+      "loss": 0.691,
+      "step": 6636
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.851277935648374e-06,
+      "loss": 0.7618,
+      "step": 6637
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.848609033171994e-06,
+      "loss": 0.7498,
+      "step": 6638
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.845940213808941e-06,
+      "loss": 0.7417,
+      "step": 6639
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.84327147775186e-06,
+      "loss": 0.7716,
+      "step": 6640
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.8406028251934e-06,
+      "loss": 0.7778,
+      "step": 6641
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.837934256326195e-06,
+      "loss": 0.8069,
+      "step": 6642
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.835265771342878e-06,
+      "loss": 0.7988,
+      "step": 6643
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.832597370436067e-06,
+      "loss": 0.7425,
+      "step": 6644
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.82992905379839e-06,
+      "loss": 0.8053,
+      "step": 6645
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.827260821622455e-06,
+      "loss": 0.8185,
+      "step": 6646
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.82459267410087e-06,
+      "loss": 0.8097,
+      "step": 6647
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.821924611426232e-06,
+      "loss": 0.8043,
+      "step": 6648
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.819256633791144e-06,
+      "loss": 0.7567,
+      "step": 6649
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.816588741388183e-06,
+      "loss": 0.8173,
+      "step": 6650
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.813920934409942e-06,
+      "loss": 0.807,
+      "step": 6651
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.811253213048994e-06,
+      "loss": 0.8377,
+      "step": 6652
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.808585577497902e-06,
+      "loss": 0.8134,
+      "step": 6653
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.805918027949237e-06,
+      "loss": 0.7703,
+      "step": 6654
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.803250564595556e-06,
+      "loss": 0.7676,
+      "step": 6655
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.800583187629407e-06,
+      "loss": 0.8304,
+      "step": 6656
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.797915897243336e-06,
+      "loss": 0.7826,
+      "step": 6657
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.795248693629881e-06,
+      "loss": 0.8338,
+      "step": 6658
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.79258157698158e-06,
+      "loss": 0.8087,
+      "step": 6659
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.78991454749095e-06,
+      "loss": 0.6884,
+      "step": 6660
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.78724760535052e-06,
+      "loss": 0.7888,
+      "step": 6661
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.7845807507528e-06,
+      "loss": 0.7429,
+      "step": 6662
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.781913983890292e-06,
+      "loss": 0.7363,
+      "step": 6663
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.779247304955506e-06,
+      "loss": 0.8318,
+      "step": 6664
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.776580714140934e-06,
+      "loss": 0.8058,
+      "step": 6665
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.773914211639057e-06,
+      "loss": 0.776,
+      "step": 6666
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.771247797642367e-06,
+      "loss": 0.679,
+      "step": 6667
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.768581472343331e-06,
+      "loss": 0.8167,
+      "step": 6668
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.765915235934425e-06,
+      "loss": 0.7709,
+      "step": 6669
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.763249088608105e-06,
+      "loss": 0.7945,
+      "step": 6670
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.760583030556838e-06,
+      "loss": 0.8018,
+      "step": 6671
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.757917061973062e-06,
+      "loss": 0.7808,
+      "step": 6672
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.755251183049224e-06,
+      "loss": 0.7458,
+      "step": 6673
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.752585393977765e-06,
+      "loss": 0.7596,
+      "step": 6674
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.749919694951113e-06,
+      "loss": 0.7571,
+      "step": 6675
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.747254086161686e-06,
+      "loss": 0.6805,
+      "step": 6676
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.744588567801912e-06,
+      "loss": 0.7017,
+      "step": 6677
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.741923140064194e-06,
+      "loss": 0.8066,
+      "step": 6678
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.739257803140935e-06,
+      "loss": 0.7791,
+      "step": 6679
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.736592557224543e-06,
+      "loss": 0.7891,
+      "step": 6680
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.7339274025074e-06,
+      "loss": 0.7123,
+      "step": 6681
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.731262339181896e-06,
+      "loss": 0.7662,
+      "step": 6682
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.7285973674404e-06,
+      "loss": 0.8007,
+      "step": 6683
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.725932487475294e-06,
+      "loss": 0.767,
+      "step": 6684
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.723267699478938e-06,
+      "loss": 0.726,
+      "step": 6685
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.720603003643689e-06,
+      "loss": 0.7361,
+      "step": 6686
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.717938400161902e-06,
+      "loss": 0.7689,
+      "step": 6687
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.715273889225917e-06,
+      "loss": 0.8016,
+      "step": 6688
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.712609471028073e-06,
+      "loss": 0.7,
+      "step": 6689
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.709945145760708e-06,
+      "loss": 0.8004,
+      "step": 6690
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.707280913616141e-06,
+      "loss": 0.7833,
+      "step": 6691
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.704616774786687e-06,
+      "loss": 0.7947,
+      "step": 6692
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.701952729464665e-06,
+      "loss": 0.7285,
+      "step": 6693
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.699288777842377e-06,
+      "loss": 0.7654,
+      "step": 6694
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.696624920112119e-06,
+      "loss": 0.8111,
+      "step": 6695
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.693961156466178e-06,
+      "loss": 0.7549,
+      "step": 6696
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.691297487096843e-06,
+      "loss": 0.8193,
+      "step": 6697
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.688633912196395e-06,
+      "loss": 0.6833,
+      "step": 6698
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.685970431957095e-06,
+      "loss": 0.7768,
+      "step": 6699
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.683307046571216e-06,
+      "loss": 0.8543,
+      "step": 6700
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.680643756231012e-06,
+      "loss": 0.7616,
+      "step": 6701
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.677980561128729e-06,
+      "loss": 0.7993,
+      "step": 6702
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.675317461456615e-06,
+      "loss": 0.7332,
+      "step": 6703
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.672654457406906e-06,
+      "loss": 0.7206,
+      "step": 6704
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.669991549171825e-06,
+      "loss": 0.841,
+      "step": 6705
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.667328736943603e-06,
+      "loss": 0.7802,
+      "step": 6706
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.66466602091445e-06,
+      "loss": 0.7721,
+      "step": 6707
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.662003401276578e-06,
+      "loss": 0.6956,
+      "step": 6708
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.659340878222182e-06,
+      "loss": 0.8091,
+      "step": 6709
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.656678451943466e-06,
+      "loss": 0.7859,
+      "step": 6710
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.654016122632613e-06,
+      "loss": 0.7237,
+      "step": 6711
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.6513538904818e-06,
+      "loss": 0.8155,
+      "step": 6712
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.648691755683208e-06,
+      "loss": 0.7039,
+      "step": 6713
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.646029718429001e-06,
+      "loss": 0.7479,
+      "step": 6714
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.643367778911331e-06,
+      "loss": 0.8202,
+      "step": 6715
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.640705937322362e-06,
+      "loss": 0.7242,
+      "step": 6716
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.638044193854233e-06,
+      "loss": 0.7597,
+      "step": 6717
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.635382548699083e-06,
+      "loss": 0.8144,
+      "step": 6718
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.632721002049044e-06,
+      "loss": 0.7857,
+      "step": 6719
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.630059554096243e-06,
+      "loss": 0.7799,
+      "step": 6720
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.62739820503279e-06,
+      "loss": 0.7979,
+      "step": 6721
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.624736955050801e-06,
+      "loss": 0.7925,
+      "step": 6722
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.622075804342378e-06,
+      "loss": 0.7367,
+      "step": 6723
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.619414753099615e-06,
+      "loss": 0.843,
+      "step": 6724
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.616753801514596e-06,
+      "loss": 0.7345,
+      "step": 6725
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.614092949779412e-06,
+      "loss": 0.8042,
+      "step": 6726
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.611432198086128e-06,
+      "loss": 0.8654,
+      "step": 6727
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.608771546626816e-06,
+      "loss": 0.7469,
+      "step": 6728
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.606110995593534e-06,
+      "loss": 0.8085,
+      "step": 6729
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.603450545178335e-06,
+      "loss": 0.7707,
+      "step": 6730
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.60079019557326e-06,
+      "loss": 0.7735,
+      "step": 6731
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.598129946970354e-06,
+      "loss": 0.8002,
+      "step": 6732
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.595469799561642e-06,
+      "loss": 0.8062,
+      "step": 6733
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.592809753539148e-06,
+      "loss": 0.6951,
+      "step": 6734
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.59014980909489e-06,
+      "loss": 0.8111,
+      "step": 6735
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.587489966420874e-06,
+      "loss": 0.7996,
+      "step": 6736
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.584830225709101e-06,
+      "loss": 0.7029,
+      "step": 6737
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.582170587151567e-06,
+      "loss": 0.7926,
+      "step": 6738
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.579511050940256e-06,
+      "loss": 0.7503,
+      "step": 6739
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.576851617267151e-06,
+      "loss": 0.7438,
+      "step": 6740
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.574192286324216e-06,
+      "loss": 0.7246,
+      "step": 6741
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.571533058303424e-06,
+      "loss": 0.8296,
+      "step": 6742
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.568873933396728e-06,
+      "loss": 0.7617,
+      "step": 6743
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.566214911796073e-06,
+      "loss": 0.8125,
+      "step": 6744
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.56355599369341e-06,
+      "loss": 0.7104,
+      "step": 6745
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.560897179280666e-06,
+      "loss": 0.8438,
+      "step": 6746
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.558238468749767e-06,
+      "loss": 0.8354,
+      "step": 6747
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.55557986229264e-06,
+      "loss": 0.7977,
+      "step": 6748
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.55292136010119e-06,
+      "loss": 0.7884,
+      "step": 6749
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.550262962367328e-06,
+      "loss": 0.8084,
+      "step": 6750
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.54760466928294e-06,
+      "loss": 0.7097,
+      "step": 6751
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.544946481039927e-06,
+      "loss": 0.7706,
+      "step": 6752
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.542288397830166e-06,
+      "loss": 0.7984,
+      "step": 6753
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.539630419845527e-06,
+      "loss": 0.7762,
+      "step": 6754
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.536972547277884e-06,
+      "loss": 0.7779,
+      "step": 6755
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.534314780319093e-06,
+      "loss": 0.7664,
+      "step": 6756
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.531657119161e-06,
+      "loss": 0.8483,
+      "step": 6757
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.528999563995456e-06,
+      "loss": 0.7871,
+      "step": 6758
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.526342115014294e-06,
+      "loss": 0.81,
+      "step": 6759
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.523684772409342e-06,
+      "loss": 0.7583,
+      "step": 6760
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.521027536372422e-06,
+      "loss": 0.7754,
+      "step": 6761
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.518370407095348e-06,
+      "loss": 0.814,
+      "step": 6762
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.515713384769923e-06,
+      "loss": 0.8345,
+      "step": 6763
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.513056469587942e-06,
+      "loss": 0.7795,
+      "step": 6764
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.5103996617412e-06,
+      "loss": 0.8538,
+      "step": 6765
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.507742961421479e-06,
+      "loss": 0.7564,
+      "step": 6766
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.505086368820545e-06,
+      "loss": 0.842,
+      "step": 6767
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.502429884130176e-06,
+      "loss": 0.7731,
+      "step": 6768
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.499773507542126e-06,
+      "loss": 0.7404,
+      "step": 6769
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.497117239248142e-06,
+      "loss": 0.8222,
+      "step": 6770
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.49446107943997e-06,
+      "loss": 0.7786,
+      "step": 6771
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.491805028309351e-06,
+      "loss": 0.7527,
+      "step": 6772
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.489149086048e-06,
+      "loss": 0.7855,
+      "step": 6773
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.486493252847653e-06,
+      "loss": 0.7975,
+      "step": 6774
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.483837528900008e-06,
+      "loss": 0.8364,
+      "step": 6775
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.481181914396774e-06,
+      "loss": 0.778,
+      "step": 6776
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.478526409529642e-06,
+      "loss": 0.7868,
+      "step": 6777
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.47587101449031e-06,
+      "loss": 0.8216,
+      "step": 6778
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.473215729470449e-06,
+      "loss": 0.7997,
+      "step": 6779
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.470560554661735e-06,
+      "loss": 0.7329,
+      "step": 6780
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.467905490255832e-06,
+      "loss": 0.8102,
+      "step": 6781
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.465250536444399e-06,
+      "loss": 0.7684,
+      "step": 6782
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.462595693419076e-06,
+      "loss": 0.7257,
+      "step": 6783
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.459940961371512e-06,
+      "loss": 0.8447,
+      "step": 6784
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.457286340493336e-06,
+      "loss": 0.8087,
+      "step": 6785
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.454631830976167e-06,
+      "loss": 0.7342,
+      "step": 6786
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.451977433011632e-06,
+      "loss": 0.6997,
+      "step": 6787
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.449323146791333e-06,
+      "loss": 0.7513,
+      "step": 6788
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.446668972506866e-06,
+      "loss": 0.8609,
+      "step": 6789
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.444014910349828e-06,
+      "loss": 0.8131,
+      "step": 6790
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.441360960511802e-06,
+      "loss": 0.7551,
+      "step": 6791
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.438707123184366e-06,
+      "loss": 0.773,
+      "step": 6792
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.436053398559082e-06,
+      "loss": 0.4476,
+      "step": 6793
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.433399786827516e-06,
+      "loss": 0.7856,
+      "step": 6794
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.430746288181217e-06,
+      "loss": 0.7418,
+      "step": 6795
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.428092902811722e-06,
+      "loss": 0.8133,
+      "step": 6796
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.425439630910575e-06,
+      "loss": 0.7933,
+      "step": 6797
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.422786472669301e-06,
+      "loss": 0.8136,
+      "step": 6798
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.420133428279412e-06,
+      "loss": 0.7338,
+      "step": 6799
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.417480497932428e-06,
+      "loss": 0.7963,
+      "step": 6800
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.414827681819843e-06,
+      "loss": 0.767,
+      "step": 6801
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.412174980133158e-06,
+      "loss": 0.3683,
+      "step": 6802
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.409522393063848e-06,
+      "loss": 0.383,
+      "step": 6803
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.406869920803406e-06,
+      "loss": 0.7681,
+      "step": 6804
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.404217563543288e-06,
+      "loss": 0.7615,
+      "step": 6805
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.401565321474959e-06,
+      "loss": 0.7912,
+      "step": 6806
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.398913194789874e-06,
+      "loss": 0.7558,
+      "step": 6807
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.396261183679475e-06,
+      "loss": 0.822,
+      "step": 6808
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.393609288335194e-06,
+      "loss": 0.8152,
+      "step": 6809
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.390957508948465e-06,
+      "loss": 0.7429,
+      "step": 6810
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.388305845710704e-06,
+      "loss": 0.8,
+      "step": 6811
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.385654298813321e-06,
+      "loss": 0.7818,
+      "step": 6812
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.383002868447719e-06,
+      "loss": 0.739,
+      "step": 6813
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.380351554805294e-06,
+      "loss": 0.7251,
+      "step": 6814
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.377700358077432e-06,
+      "loss": 0.7517,
+      "step": 6815
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.375049278455501e-06,
+      "loss": 0.7532,
+      "step": 6816
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.37239831613088e-06,
+      "loss": 0.8596,
+      "step": 6817
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.369747471294926e-06,
+      "loss": 0.7883,
+      "step": 6818
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.367096744138985e-06,
+      "loss": 0.8289,
+      "step": 6819
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.364446134854411e-06,
+      "loss": 0.788,
+      "step": 6820
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.361795643632532e-06,
+      "loss": 0.4303,
+      "step": 6821
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.359145270664671e-06,
+      "loss": 0.8118,
+      "step": 6822
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.356495016142153e-06,
+      "loss": 0.7808,
+      "step": 6823
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.353844880256281e-06,
+      "loss": 0.7907,
+      "step": 6824
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.351194863198357e-06,
+      "loss": 0.7975,
+      "step": 6825
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.348544965159678e-06,
+      "loss": 0.7403,
+      "step": 6826
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.345895186331521e-06,
+      "loss": 0.8172,
+      "step": 6827
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.343245526905164e-06,
+      "loss": 0.783,
+      "step": 6828
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.340595987071867e-06,
+      "loss": 0.7721,
+      "step": 6829
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.337946567022897e-06,
+      "loss": 0.7321,
+      "step": 6830
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.335297266949494e-06,
+      "loss": 0.7887,
+      "step": 6831
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.332648087042903e-06,
+      "loss": 0.7152,
+      "step": 6832
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.329999027494356e-06,
+      "loss": 0.7715,
+      "step": 6833
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.327350088495074e-06,
+      "loss": 0.7292,
+      "step": 6834
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.324701270236268e-06,
+      "loss": 0.8599,
+      "step": 6835
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.322052572909152e-06,
+      "loss": 0.7272,
+      "step": 6836
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.319403996704916e-06,
+      "loss": 0.7746,
+      "step": 6837
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.316755541814743e-06,
+      "loss": 0.8258,
+      "step": 6838
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.314107208429824e-06,
+      "loss": 0.7728,
+      "step": 6839
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.311458996741325e-06,
+      "loss": 0.8054,
+      "step": 6840
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.3088109069404e-06,
+      "loss": 0.7766,
+      "step": 6841
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.306162939218212e-06,
+      "loss": 0.8256,
+      "step": 6842
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.303515093765899e-06,
+      "loss": 0.7552,
+      "step": 6843
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.3008673707746e-06,
+      "loss": 0.7721,
+      "step": 6844
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.298219770435436e-06,
+      "loss": 0.7852,
+      "step": 6845
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.29557229293953e-06,
+      "loss": 0.8554,
+      "step": 6846
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.29292493847799e-06,
+      "loss": 0.7803,
+      "step": 6847
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.290277707241907e-06,
+      "loss": 0.7684,
+      "step": 6848
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.287630599422385e-06,
+      "loss": 0.7491,
+      "step": 6849
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.284983615210497e-06,
+      "loss": 0.699,
+      "step": 6850
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.282336754797316e-06,
+      "loss": 0.7358,
+      "step": 6851
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.279690018373912e-06,
+      "loss": 0.8241,
+      "step": 6852
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.277043406131334e-06,
+      "loss": 0.7452,
+      "step": 6853
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.274396918260628e-06,
+      "loss": 0.7572,
+      "step": 6854
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.271750554952839e-06,
+      "loss": 0.7597,
+      "step": 6855
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.269104316398987e-06,
+      "loss": 0.802,
+      "step": 6856
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.266458202790097e-06,
+      "loss": 0.8107,
+      "step": 6857
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.26381221431717e-06,
+      "loss": 0.8133,
+      "step": 6858
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.261166351171219e-06,
+      "loss": 0.7948,
+      "step": 6859
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.25852061354323e-06,
+      "loss": 0.3835,
+      "step": 6860
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.255875001624179e-06,
+      "loss": 0.7922,
+      "step": 6861
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.253229515605054e-06,
+      "loss": 0.7357,
+      "step": 6862
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.250584155676809e-06,
+      "loss": 0.7282,
+      "step": 6863
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.247938922030401e-06,
+      "loss": 0.8353,
+      "step": 6864
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.245293814856786e-06,
+      "loss": 0.3851,
+      "step": 6865
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.242648834346892e-06,
+      "loss": 0.7535,
+      "step": 6866
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.240003980691645e-06,
+      "loss": 0.8134,
+      "step": 6867
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.237359254081974e-06,
+      "loss": 0.7601,
+      "step": 6868
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.234714654708783e-06,
+      "loss": 0.7397,
+      "step": 6869
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.232070182762974e-06,
+      "loss": 0.7652,
+      "step": 6870
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.229425838435434e-06,
+      "loss": 0.8209,
+      "step": 6871
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.226781621917054e-06,
+      "loss": 0.7912,
+      "step": 6872
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.224137533398701e-06,
+      "loss": 0.7166,
+      "step": 6873
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.221493573071238e-06,
+      "loss": 0.8298,
+      "step": 6874
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.218849741125527e-06,
+      "loss": 0.7592,
+      "step": 6875
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.216206037752408e-06,
+      "loss": 0.7282,
+      "step": 6876
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.213562463142714e-06,
+      "loss": 0.7109,
+      "step": 6877
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.21091901748728e-06,
+      "loss": 0.7847,
+      "step": 6878
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.208275700976919e-06,
+      "loss": 0.7351,
+      "step": 6879
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.205632513802434e-06,
+      "loss": 0.7866,
+      "step": 6880
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.202989456154636e-06,
+      "loss": 0.7261,
+      "step": 6881
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.200346528224307e-06,
+      "loss": 0.3789,
+      "step": 6882
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.197703730202224e-06,
+      "loss": 0.7553,
+      "step": 6883
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.195061062279162e-06,
+      "loss": 0.7711,
+      "step": 6884
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.192418524645884e-06,
+      "loss": 0.7707,
+      "step": 6885
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.189776117493141e-06,
+      "loss": 0.6774,
+      "step": 6886
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.187133841011671e-06,
+      "loss": 0.796,
+      "step": 6887
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.184491695392215e-06,
+      "loss": 0.7395,
+      "step": 6888
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.18184968082549e-06,
+      "loss": 0.8005,
+      "step": 6889
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.17920779750221e-06,
+      "loss": 0.7542,
+      "step": 6890
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.176566045613086e-06,
+      "loss": 0.747,
+      "step": 6891
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.173924425348806e-06,
+      "loss": 0.3837,
+      "step": 6892
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.171282936900063e-06,
+      "loss": 0.7874,
+      "step": 6893
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.168641580457525e-06,
+      "loss": 0.7811,
+      "step": 6894
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.16600035621187e-06,
+      "loss": 0.7666,
+      "step": 6895
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.163359264353745e-06,
+      "loss": 0.7776,
+      "step": 6896
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.160718305073797e-06,
+      "loss": 0.7583,
+      "step": 6897
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.158077478562674e-06,
+      "loss": 0.7661,
+      "step": 6898
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.155436785010996e-06,
+      "loss": 0.7863,
+      "step": 6899
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.152796224609382e-06,
+      "loss": 0.814,
+      "step": 6900
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.150155797548446e-06,
+      "loss": 0.7942,
+      "step": 6901
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.147515504018785e-06,
+      "loss": 0.7662,
+      "step": 6902
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.144875344210987e-06,
+      "loss": 0.7259,
+      "step": 6903
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.142235318315636e-06,
+      "loss": 0.8294,
+      "step": 6904
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.139595426523302e-06,
+      "loss": 0.8305,
+      "step": 6905
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.136955669024542e-06,
+      "loss": 0.7182,
+      "step": 6906
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.134316046009915e-06,
+      "loss": 0.7789,
+      "step": 6907
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.131676557669956e-06,
+      "loss": 0.8166,
+      "step": 6908
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.1290372041952e-06,
+      "loss": 0.8015,
+      "step": 6909
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.126397985776164e-06,
+      "loss": 0.7886,
+      "step": 6910
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.123758902603367e-06,
+      "loss": 0.84,
+      "step": 6911
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.121119954867309e-06,
+      "loss": 0.8123,
+      "step": 6912
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.11848114275848e-06,
+      "loss": 0.8126,
+      "step": 6913
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.115842466467367e-06,
+      "loss": 0.7393,
+      "step": 6914
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.113203926184445e-06,
+      "loss": 0.7866,
+      "step": 6915
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.11056552210017e-06,
+      "loss": 0.4191,
+      "step": 6916
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.107927254405002e-06,
+      "loss": 0.7779,
+      "step": 6917
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.105289123289384e-06,
+      "loss": 0.778,
+      "step": 6918
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.102651128943745e-06,
+      "loss": 0.7313,
+      "step": 6919
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.100013271558517e-06,
+      "loss": 0.7543,
+      "step": 6920
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.09737555132411e-06,
+      "loss": 0.7208,
+      "step": 6921
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.094737968430924e-06,
+      "loss": 0.772,
+      "step": 6922
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.092100523069357e-06,
+      "loss": 0.8171,
+      "step": 6923
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.089463215429796e-06,
+      "loss": 0.8174,
+      "step": 6924
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.086826045702614e-06,
+      "loss": 0.7565,
+      "step": 6925
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.08418901407817e-06,
+      "loss": 0.7539,
+      "step": 6926
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.081552120746828e-06,
+      "loss": 0.8399,
+      "step": 6927
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.078915365898927e-06,
+      "loss": 0.7658,
+      "step": 6928
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.076278749724797e-06,
+      "loss": 0.777,
+      "step": 6929
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.073642272414774e-06,
+      "loss": 0.7948,
+      "step": 6930
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.071005934159164e-06,
+      "loss": 0.7567,
+      "step": 6931
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.068369735148271e-06,
+      "loss": 0.723,
+      "step": 6932
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.065733675572396e-06,
+      "loss": 0.7615,
+      "step": 6933
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.063097755621818e-06,
+      "loss": 0.7834,
+      "step": 6934
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.060461975486814e-06,
+      "loss": 0.7433,
+      "step": 6935
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.05782633535764e-06,
+      "loss": 0.6997,
+      "step": 6936
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.055190835424565e-06,
+      "loss": 0.7738,
+      "step": 6937
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.052555475877823e-06,
+      "loss": 0.768,
+      "step": 6938
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.049920256907646e-06,
+      "loss": 0.8297,
+      "step": 6939
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.047285178704266e-06,
+      "loss": 0.7806,
+      "step": 6940
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.044650241457891e-06,
+      "loss": 0.7759,
+      "step": 6941
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.042015445358722e-06,
+      "loss": 0.8013,
+      "step": 6942
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.039380790596958e-06,
+      "loss": 0.8691,
+      "step": 6943
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.036746277362778e-06,
+      "loss": 0.7441,
+      "step": 6944
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.034111905846356e-06,
+      "loss": 0.8184,
+      "step": 6945
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.031477676237854e-06,
+      "loss": 0.752,
+      "step": 6946
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.028843588727427e-06,
+      "loss": 0.7819,
+      "step": 6947
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.026209643505217e-06,
+      "loss": 0.8386,
+      "step": 6948
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.023575840761346e-06,
+      "loss": 0.7763,
+      "step": 6949
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.020942180685947e-06,
+      "loss": 0.7948,
+      "step": 6950
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.018308663469128e-06,
+      "loss": 0.7157,
+      "step": 6951
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.015675289300986e-06,
+      "loss": 0.8151,
+      "step": 6952
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.013042058371618e-06,
+      "loss": 0.7775,
+      "step": 6953
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.010408970871098e-06,
+      "loss": 0.7542,
+      "step": 6954
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.007776026989498e-06,
+      "loss": 0.7486,
+      "step": 6955
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.005143226916877e-06,
+      "loss": 0.7558,
+      "step": 6956
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.002510570843287e-06,
+      "loss": 0.7399,
+      "step": 6957
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.999878058958761e-06,
+      "loss": 0.8157,
+      "step": 6958
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.997245691453336e-06,
+      "loss": 0.8124,
+      "step": 6959
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.994613468517024e-06,
+      "loss": 0.7397,
+      "step": 6960
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.991981390339832e-06,
+      "loss": 0.8639,
+      "step": 6961
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.989349457111754e-06,
+      "loss": 0.7573,
+      "step": 6962
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.986717669022786e-06,
+      "loss": 0.823,
+      "step": 6963
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.984086026262894e-06,
+      "loss": 0.7971,
+      "step": 6964
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.98145452902205e-06,
+      "loss": 0.7617,
+      "step": 6965
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.978823177490204e-06,
+      "loss": 0.7292,
+      "step": 6966
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.976191971857308e-06,
+      "loss": 0.4017,
+      "step": 6967
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.973560912313287e-06,
+      "loss": 0.7883,
+      "step": 6968
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.970929999048073e-06,
+      "loss": 0.7256,
+      "step": 6969
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.968299232251574e-06,
+      "loss": 0.3798,
+      "step": 6970
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.965668612113689e-06,
+      "loss": 0.7958,
+      "step": 6971
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.963038138824318e-06,
+      "loss": 0.8338,
+      "step": 6972
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.960407812573338e-06,
+      "loss": 0.8013,
+      "step": 6973
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.957777633550615e-06,
+      "loss": 0.8229,
+      "step": 6974
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.955147601946017e-06,
+      "loss": 0.3826,
+      "step": 6975
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.952517717949389e-06,
+      "loss": 0.8207,
+      "step": 6976
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.949887981750569e-06,
+      "loss": 0.7857,
+      "step": 6977
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.947258393539382e-06,
+      "loss": 0.7669,
+      "step": 6978
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.944628953505655e-06,
+      "loss": 0.8248,
+      "step": 6979
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.941999661839188e-06,
+      "loss": 0.8246,
+      "step": 6980
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.939370518729774e-06,
+      "loss": 0.822,
+      "step": 6981
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.936741524367205e-06,
+      "loss": 0.7821,
+      "step": 6982
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.93411267894125e-06,
+      "loss": 0.794,
+      "step": 6983
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.93148398264167e-06,
+      "loss": 0.7151,
+      "step": 6984
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.92885543565823e-06,
+      "loss": 0.8032,
+      "step": 6985
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.926227038180658e-06,
+      "loss": 0.7415,
+      "step": 6986
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.923598790398694e-06,
+      "loss": 0.7743,
+      "step": 6987
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.920970692502054e-06,
+      "loss": 0.8354,
+      "step": 6988
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.918342744680454e-06,
+      "loss": 0.749,
+      "step": 6989
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.915714947123588e-06,
+      "loss": 0.7618,
+      "step": 6990
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.913087300021138e-06,
+      "loss": 0.7558,
+      "step": 6991
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.910459803562794e-06,
+      "loss": 0.7372,
+      "step": 6992
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.907832457938214e-06,
+      "loss": 0.7834,
+      "step": 6993
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.905205263337052e-06,
+      "loss": 0.8325,
+      "step": 6994
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.902578219948958e-06,
+      "loss": 0.7915,
+      "step": 6995
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.899951327963563e-06,
+      "loss": 0.7643,
+      "step": 6996
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.897324587570486e-06,
+      "loss": 0.873,
+      "step": 6997
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.894697998959344e-06,
+      "loss": 0.7811,
+      "step": 6998
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.892071562319738e-06,
+      "loss": 0.7568,
+      "step": 6999
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.889445277841253e-06,
+      "loss": 0.7223,
+      "step": 7000
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.886819145713473e-06,
+      "loss": 0.7717,
+      "step": 7001
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.884193166125962e-06,
+      "loss": 0.8122,
+      "step": 7002
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.881567339268278e-06,
+      "loss": 0.7727,
+      "step": 7003
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.878941665329964e-06,
+      "loss": 0.765,
+      "step": 7004
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.876316144500558e-06,
+      "loss": 0.7348,
+      "step": 7005
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.873690776969583e-06,
+      "loss": 0.8162,
+      "step": 7006
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.87106556292655e-06,
+      "loss": 0.7884,
+      "step": 7007
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.868440502560963e-06,
+      "loss": 0.8091,
+      "step": 7008
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.865815596062312e-06,
+      "loss": 0.8383,
+      "step": 7009
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.863190843620072e-06,
+      "loss": 0.714,
+      "step": 7010
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.860566245423718e-06,
+      "loss": 0.7592,
+      "step": 7011
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.857941801662703e-06,
+      "loss": 0.7893,
+      "step": 7012
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.85531751252647e-06,
+      "loss": 0.765,
+      "step": 7013
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.85269337820446e-06,
+      "loss": 0.8291,
+      "step": 7014
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.850069398886095e-06,
+      "loss": 0.7677,
+      "step": 7015
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.847445574760783e-06,
+      "loss": 0.6981,
+      "step": 7016
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.844821906017929e-06,
+      "loss": 0.4078,
+      "step": 7017
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.84219839284692e-06,
+      "loss": 0.8285,
+      "step": 7018
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.83957503543714e-06,
+      "loss": 0.7546,
+      "step": 7019
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.836951833977948e-06,
+      "loss": 0.7602,
+      "step": 7020
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.83432878865871e-06,
+      "loss": 0.7596,
+      "step": 7021
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.831705899668767e-06,
+      "loss": 0.7854,
+      "step": 7022
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.829083167197446e-06,
+      "loss": 0.7401,
+      "step": 7023
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.82646059143408e-06,
+      "loss": 0.8233,
+      "step": 7024
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.823838172567974e-06,
+      "loss": 0.7255,
+      "step": 7025
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.821215910788426e-06,
+      "loss": 0.7213,
+      "step": 7026
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.81859380628473e-06,
+      "loss": 0.7814,
+      "step": 7027
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.815971859246158e-06,
+      "loss": 0.8297,
+      "step": 7028
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.81335006986198e-06,
+      "loss": 0.8466,
+      "step": 7029
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.810728438321444e-06,
+      "loss": 0.7547,
+      "step": 7030
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.808106964813802e-06,
+      "loss": 0.7986,
+      "step": 7031
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.80548564952828e-06,
+      "loss": 0.7433,
+      "step": 7032
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.802864492654094e-06,
+      "loss": 0.8004,
+      "step": 7033
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.800243494380459e-06,
+      "loss": 0.782,
+      "step": 7034
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.797622654896572e-06,
+      "loss": 0.8244,
+      "step": 7035
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.795001974391613e-06,
+      "loss": 0.7428,
+      "step": 7036
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.792381453054764e-06,
+      "loss": 0.8003,
+      "step": 7037
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.789761091075181e-06,
+      "loss": 0.7854,
+      "step": 7038
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.787140888642018e-06,
+      "loss": 0.7465,
+      "step": 7039
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.784520845944415e-06,
+      "loss": 0.7063,
+      "step": 7040
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.7819009631715e-06,
+      "loss": 0.7497,
+      "step": 7041
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.779281240512391e-06,
+      "loss": 0.8039,
+      "step": 7042
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.776661678156188e-06,
+      "loss": 0.379,
+      "step": 7043
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.77404227629199e-06,
+      "loss": 0.7596,
+      "step": 7044
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.771423035108878e-06,
+      "loss": 0.794,
+      "step": 7045
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.768803954795917e-06,
+      "loss": 0.3731,
+      "step": 7046
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.766185035542174e-06,
+      "loss": 0.7659,
+      "step": 7047
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.76356627753669e-06,
+      "loss": 0.7203,
+      "step": 7048
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.760947680968502e-06,
+      "loss": 0.7916,
+      "step": 7049
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.758329246026633e-06,
+      "loss": 0.832,
+      "step": 7050
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.7557109729001e-06,
+      "loss": 0.7639,
+      "step": 7051
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.753092861777895e-06,
+      "loss": 0.6807,
+      "step": 7052
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.750474912849015e-06,
+      "loss": 0.7911,
+      "step": 7053
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.747857126302434e-06,
+      "loss": 0.7979,
+      "step": 7054
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.745239502327118e-06,
+      "loss": 0.8084,
+      "step": 7055
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.742622041112014e-06,
+      "loss": 0.752,
+      "step": 7056
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.740004742846072e-06,
+      "loss": 0.7855,
+      "step": 7057
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.737387607718219e-06,
+      "loss": 0.7677,
+      "step": 7058
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.73477063591737e-06,
+      "loss": 0.7335,
+      "step": 7059
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.73215382763244e-06,
+      "loss": 0.7162,
+      "step": 7060
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.729537183052318e-06,
+      "loss": 0.3954,
+      "step": 7061
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.726920702365883e-06,
+      "loss": 0.8284,
+      "step": 7062
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.724304385762015e-06,
+      "loss": 0.7754,
+      "step": 7063
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.721688233429571e-06,
+      "loss": 0.8187,
+      "step": 7064
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.71907224555739e-06,
+      "loss": 0.7181,
+      "step": 7065
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.716456422334318e-06,
+      "loss": 0.7893,
+      "step": 7066
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.713840763949174e-06,
+      "loss": 0.7902,
+      "step": 7067
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.711225270590767e-06,
+      "loss": 0.7546,
+      "step": 7068
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.708609942447898e-06,
+      "loss": 0.8302,
+      "step": 7069
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.705994779709362e-06,
+      "loss": 0.75,
+      "step": 7070
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.703379782563929e-06,
+      "loss": 0.7578,
+      "step": 7071
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.700764951200357e-06,
+      "loss": 0.7924,
+      "step": 7072
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.698150285807409e-06,
+      "loss": 0.7518,
+      "step": 7073
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.695535786573823e-06,
+      "loss": 0.8202,
+      "step": 7074
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.692921453688317e-06,
+      "loss": 0.7654,
+      "step": 7075
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.69030728733962e-06,
+      "loss": 0.8209,
+      "step": 7076
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.68769328771643e-06,
+      "loss": 0.751,
+      "step": 7077
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.685079455007435e-06,
+      "loss": 0.7805,
+      "step": 7078
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.68246578940132e-06,
+      "loss": 0.8027,
+      "step": 7079
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.679852291086756e-06,
+      "loss": 0.7601,
+      "step": 7080
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.677238960252394e-06,
+      "loss": 0.7815,
+      "step": 7081
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.674625797086875e-06,
+      "loss": 0.7789,
+      "step": 7082
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.672012801778838e-06,
+      "loss": 0.7496,
+      "step": 7083
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.669399974516898e-06,
+      "loss": 0.7775,
+      "step": 7084
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.666787315489658e-06,
+      "loss": 0.8228,
+      "step": 7085
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.664174824885723e-06,
+      "loss": 0.7421,
+      "step": 7086
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.66156250289367e-06,
+      "loss": 0.7691,
+      "step": 7087
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.658950349702068e-06,
+      "loss": 0.7841,
+      "step": 7088
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.656338365499478e-06,
+      "loss": 0.8228,
+      "step": 7089
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.653726550474451e-06,
+      "loss": 0.8164,
+      "step": 7090
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.651114904815512e-06,
+      "loss": 0.7625,
+      "step": 7091
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.648503428711194e-06,
+      "loss": 0.7705,
+      "step": 7092
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.64589212235e-06,
+      "loss": 0.7906,
+      "step": 7093
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.643280985920423e-06,
+      "loss": 0.7345,
+      "step": 7094
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.64067001961096e-06,
+      "loss": 0.7847,
+      "step": 7095
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.638059223610078e-06,
+      "loss": 0.793,
+      "step": 7096
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.635448598106233e-06,
+      "loss": 0.7753,
+      "step": 7097
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.63283814328788e-06,
+      "loss": 0.7251,
+      "step": 7098
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.63022785934345e-06,
+      "loss": 0.7582,
+      "step": 7099
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.627617746461376e-06,
+      "loss": 0.7164,
+      "step": 7100
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.625007804830056e-06,
+      "loss": 0.785,
+      "step": 7101
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.6223980346379e-06,
+      "loss": 0.7673,
+      "step": 7102
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.619788436073291e-06,
+      "loss": 0.7363,
+      "step": 7103
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.617179009324599e-06,
+      "loss": 0.6892,
+      "step": 7104
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.614569754580195e-06,
+      "loss": 0.7608,
+      "step": 7105
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.611960672028422e-06,
+      "loss": 0.7976,
+      "step": 7106
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.609351761857614e-06,
+      "loss": 0.787,
+      "step": 7107
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.606743024256104e-06,
+      "loss": 0.7147,
+      "step": 7108
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.604134459412198e-06,
+      "loss": 0.8285,
+      "step": 7109
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.6015260675142e-06,
+      "loss": 0.7672,
+      "step": 7110
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5989178487503905e-06,
+      "loss": 0.8206,
+      "step": 7111
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.596309803309052e-06,
+      "loss": 0.7526,
+      "step": 7112
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5937019313784435e-06,
+      "loss": 0.7867,
+      "step": 7113
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.591094233146808e-06,
+      "loss": 0.7694,
+      "step": 7114
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.588486708802395e-06,
+      "loss": 0.8555,
+      "step": 7115
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.585879358533422e-06,
+      "loss": 0.8247,
+      "step": 7116
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.583272182528097e-06,
+      "loss": 0.7657,
+      "step": 7117
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.580665180974629e-06,
+      "loss": 0.7054,
+      "step": 7118
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.578058354061197e-06,
+      "loss": 0.7584,
+      "step": 7119
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.575451701975979e-06,
+      "loss": 0.8171,
+      "step": 7120
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.572845224907135e-06,
+      "loss": 0.7975,
+      "step": 7121
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5702389230428165e-06,
+      "loss": 0.8266,
+      "step": 7122
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.567632796571158e-06,
+      "loss": 0.7847,
+      "step": 7123
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.565026845680279e-06,
+      "loss": 0.8506,
+      "step": 7124
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.562421070558298e-06,
+      "loss": 0.8426,
+      "step": 7125
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.55981547139331e-06,
+      "loss": 0.76,
+      "step": 7126
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.557210048373395e-06,
+      "loss": 0.8265,
+      "step": 7127
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.554604801686637e-06,
+      "loss": 0.7207,
+      "step": 7128
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.551999731521087e-06,
+      "loss": 0.755,
+      "step": 7129
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.549394838064795e-06,
+      "loss": 0.7846,
+      "step": 7130
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.546790121505797e-06,
+      "loss": 0.8013,
+      "step": 7131
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.544185582032115e-06,
+      "loss": 0.7645,
+      "step": 7132
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.541581219831754e-06,
+      "loss": 0.7872,
+      "step": 7133
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.538977035092717e-06,
+      "loss": 0.7863,
+      "step": 7134
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.536373028002984e-06,
+      "loss": 0.8082,
+      "step": 7135
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.533769198750525e-06,
+      "loss": 0.7975,
+      "step": 7136
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.531165547523294e-06,
+      "loss": 0.7959,
+      "step": 7137
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.528562074509244e-06,
+      "loss": 0.8123,
+      "step": 7138
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5259587798963016e-06,
+      "loss": 0.7698,
+      "step": 7139
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5233556638723885e-06,
+      "loss": 0.7548,
+      "step": 7140
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.520752726625409e-06,
+      "loss": 0.8027,
+      "step": 7141
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.51814996834326e-06,
+      "loss": 0.7503,
+      "step": 7142
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.515547389213815e-06,
+      "loss": 0.7693,
+      "step": 7143
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.512944989424951e-06,
+      "loss": 0.7525,
+      "step": 7144
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.510342769164518e-06,
+      "loss": 0.738,
+      "step": 7145
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.507740728620352e-06,
+      "loss": 0.7394,
+      "step": 7146
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.505138867980292e-06,
+      "loss": 0.7177,
+      "step": 7147
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.5025371874321495e-06,
+      "loss": 0.7626,
+      "step": 7148
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.499935687163724e-06,
+      "loss": 0.7388,
+      "step": 7149
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.497334367362807e-06,
+      "loss": 0.8673,
+      "step": 7150
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.494733228217177e-06,
+      "loss": 0.7699,
+      "step": 7151
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.492132269914597e-06,
+      "loss": 0.7966,
+      "step": 7152
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.4895314926428145e-06,
+      "loss": 0.7651,
+      "step": 7153
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.486930896589573e-06,
+      "loss": 0.8506,
+      "step": 7154
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.484330481942592e-06,
+      "loss": 0.8224,
+      "step": 7155
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.4817302488895814e-06,
+      "loss": 0.7873,
+      "step": 7156
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.479130197618247e-06,
+      "loss": 0.7732,
+      "step": 7157
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.476530328316267e-06,
+      "loss": 0.7801,
+      "step": 7158
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.473930641171314e-06,
+      "loss": 0.8345,
+      "step": 7159
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.471331136371051e-06,
+      "loss": 0.4025,
+      "step": 7160
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.468731814103118e-06,
+      "loss": 0.7408,
+      "step": 7161
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.466132674555153e-06,
+      "loss": 0.7547,
+      "step": 7162
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.463533717914769e-06,
+      "loss": 0.7109,
+      "step": 7163
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.46093494436958e-06,
+      "loss": 0.8451,
+      "step": 7164
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.458336354107174e-06,
+      "loss": 0.786,
+      "step": 7165
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.455737947315127e-06,
+      "loss": 0.782,
+      "step": 7166
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.453139724181013e-06,
+      "loss": 0.7764,
+      "step": 7167
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.450541684892383e-06,
+      "loss": 0.758,
+      "step": 7168
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.4479438296367705e-06,
+      "loss": 0.7315,
+      "step": 7169
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.4453461586017115e-06,
+      "loss": 0.7802,
+      "step": 7170
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.442748671974712e-06,
+      "loss": 0.7953,
+      "step": 7171
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.440151369943276e-06,
+      "loss": 0.8039,
+      "step": 7172
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.4375542526948874e-06,
+      "loss": 0.7844,
+      "step": 7173
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.434957320417024e-06,
+      "loss": 0.6777,
+      "step": 7174
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.432360573297143e-06,
+      "loss": 0.7208,
+      "step": 7175
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.429764011522687e-06,
+      "loss": 0.7775,
+      "step": 7176
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.427167635281095e-06,
+      "loss": 0.8126,
+      "step": 7177
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.424571444759786e-06,
+      "loss": 0.7243,
+      "step": 7178
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.421975440146161e-06,
+      "loss": 0.8284,
+      "step": 7179
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.41937962162762e-06,
+      "loss": 0.7484,
+      "step": 7180
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.4167839893915386e-06,
+      "loss": 0.7726,
+      "step": 7181
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.4141885436252845e-06,
+      "loss": 0.7836,
+      "step": 7182
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.411593284516208e-06,
+      "loss": 0.8946,
+      "step": 7183
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.408998212251654e-06,
+      "loss": 0.7029,
+      "step": 7184
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.406403327018939e-06,
+      "loss": 0.3791,
+      "step": 7185
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.403808629005385e-06,
+      "loss": 0.7283,
+      "step": 7186
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.401214118398284e-06,
+      "loss": 0.8041,
+      "step": 7187
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3986197953849245e-06,
+      "loss": 0.7409,
+      "step": 7188
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.396025660152573e-06,
+      "loss": 0.7412,
+      "step": 7189
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3934317128884944e-06,
+      "loss": 0.7476,
+      "step": 7190
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.390837953779929e-06,
+      "loss": 0.7422,
+      "step": 7191
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.388244383014108e-06,
+      "loss": 0.7831,
+      "step": 7192
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.38565100077825e-06,
+      "loss": 0.826,
+      "step": 7193
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.383057807259559e-06,
+      "loss": 0.7605,
+      "step": 7194
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3804648026452205e-06,
+      "loss": 0.8335,
+      "step": 7195
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.377871987122419e-06,
+      "loss": 0.8153,
+      "step": 7196
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.375279360878314e-06,
+      "loss": 0.4016,
+      "step": 7197
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.372686924100047e-06,
+      "loss": 0.7379,
+      "step": 7198
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3700946769747635e-06,
+      "loss": 0.8017,
+      "step": 7199
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3675026196895836e-06,
+      "loss": 0.7797,
+      "step": 7200
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.36491075243161e-06,
+      "loss": 0.8003,
+      "step": 7201
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.362319075387941e-06,
+      "loss": 0.7324,
+      "step": 7202
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3597275887456555e-06,
+      "loss": 0.8768,
+      "step": 7203
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3571362926918245e-06,
+      "loss": 0.8195,
+      "step": 7204
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.354545187413493e-06,
+      "loss": 0.7796,
+      "step": 7205
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.35195427309771e-06,
+      "loss": 0.7783,
+      "step": 7206
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.349363549931497e-06,
+      "loss": 0.7681,
+      "step": 7207
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.346773018101858e-06,
+      "loss": 0.7712,
+      "step": 7208
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.344182677795804e-06,
+      "loss": 0.7279,
+      "step": 7209
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.341592529200314e-06,
+      "loss": 0.7586,
+      "step": 7210
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.3390025725023515e-06,
+      "loss": 0.7545,
+      "step": 7211
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.336412807888883e-06,
+      "loss": 0.7431,
+      "step": 7212
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.333823235546844e-06,
+      "loss": 0.7519,
+      "step": 7213
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.331233855663169e-06,
+      "loss": 0.7805,
+      "step": 7214
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.328644668424766e-06,
+      "loss": 0.7706,
+      "step": 7215
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.326055674018543e-06,
+      "loss": 0.7108,
+      "step": 7216
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.323466872631384e-06,
+      "loss": 0.7206,
+      "step": 7217
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.320878264450156e-06,
+      "loss": 0.7353,
+      "step": 7218
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.318289849661728e-06,
+      "loss": 0.7934,
+      "step": 7219
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.315701628452943e-06,
+      "loss": 0.3935,
+      "step": 7220
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.313113601010622e-06,
+      "loss": 0.7816,
+      "step": 7221
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.310525767521596e-06,
+      "loss": 0.75,
+      "step": 7222
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.30793812817266e-06,
+      "loss": 0.7858,
+      "step": 7223
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.305350683150605e-06,
+      "loss": 0.7436,
+      "step": 7224
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.302763432642204e-06,
+      "loss": 0.7518,
+      "step": 7225
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.300176376834225e-06,
+      "loss": 0.69,
+      "step": 7226
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.297589515913406e-06,
+      "loss": 0.8107,
+      "step": 7227
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.295002850066488e-06,
+      "loss": 0.8316,
+      "step": 7228
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.292416379480185e-06,
+      "loss": 0.7528,
+      "step": 7229
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.289830104341204e-06,
+      "loss": 0.3979,
+      "step": 7230
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.287244024836231e-06,
+      "loss": 0.8356,
+      "step": 7231
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.284658141151949e-06,
+      "loss": 0.8052,
+      "step": 7232
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.282072453475015e-06,
+      "loss": 0.8014,
+      "step": 7233
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2794869619920805e-06,
+      "loss": 0.8249,
+      "step": 7234
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.276901666889778e-06,
+      "loss": 0.7145,
+      "step": 7235
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.274316568354731e-06,
+      "loss": 0.8054,
+      "step": 7236
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.271731666573538e-06,
+      "loss": 0.7816,
+      "step": 7237
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2691469617328e-06,
+      "loss": 0.7614,
+      "step": 7238
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.266562454019088e-06,
+      "loss": 0.7877,
+      "step": 7239
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.263978143618962e-06,
+      "loss": 0.7214,
+      "step": 7240
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.26139403071898e-06,
+      "loss": 0.8053,
+      "step": 7241
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.258810115505671e-06,
+      "loss": 0.72,
+      "step": 7242
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.256226398165554e-06,
+      "loss": 0.7182,
+      "step": 7243
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.253642878885137e-06,
+      "loss": 0.7143,
+      "step": 7244
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.251059557850913e-06,
+      "loss": 0.7174,
+      "step": 7245
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2484764352493575e-06,
+      "loss": 0.7355,
+      "step": 7246
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.245893511266932e-06,
+      "loss": 0.7327,
+      "step": 7247
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.24331078609009e-06,
+      "loss": 0.7493,
+      "step": 7248
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.240728259905265e-06,
+      "loss": 0.7718,
+      "step": 7249
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.23814593289887e-06,
+      "loss": 0.8528,
+      "step": 7250
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.235563805257321e-06,
+      "loss": 0.3973,
+      "step": 7251
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.232981877167004e-06,
+      "loss": 0.7696,
+      "step": 7252
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.230400148814291e-06,
+      "loss": 0.825,
+      "step": 7253
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.227818620385552e-06,
+      "loss": 0.7424,
+      "step": 7254
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.225237292067135e-06,
+      "loss": 0.7718,
+      "step": 7255
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2226561640453714e-06,
+      "loss": 0.797,
+      "step": 7256
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.220075236506577e-06,
+      "loss": 0.8296,
+      "step": 7257
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.217494509637062e-06,
+      "loss": 0.7334,
+      "step": 7258
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.214913983623113e-06,
+      "loss": 0.754,
+      "step": 7259
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.212333658651005e-06,
+      "loss": 0.7811,
+      "step": 7260
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.209753534907004e-06,
+      "loss": 0.7447,
+      "step": 7261
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.207173612577353e-06,
+      "loss": 0.7766,
+      "step": 7262
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.204593891848279e-06,
+      "loss": 0.7462,
+      "step": 7263
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.20201437290601e-06,
+      "loss": 0.8158,
+      "step": 7264
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.199435055936745e-06,
+      "loss": 0.7883,
+      "step": 7265
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.196855941126666e-06,
+      "loss": 0.7338,
+      "step": 7266
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.194277028661958e-06,
+      "loss": 0.787,
+      "step": 7267
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.191698318728773e-06,
+      "loss": 0.7573,
+      "step": 7268
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1891198115132565e-06,
+      "loss": 0.7549,
+      "step": 7269
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.186541507201535e-06,
+      "loss": 0.8035,
+      "step": 7270
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.183963405979733e-06,
+      "loss": 0.8001,
+      "step": 7271
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.181385508033945e-06,
+      "loss": 0.8044,
+      "step": 7272
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.178807813550253e-06,
+      "loss": 0.8212,
+      "step": 7273
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1762303227147345e-06,
+      "loss": 0.7789,
+      "step": 7274
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.173653035713446e-06,
+      "loss": 0.8279,
+      "step": 7275
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.171075952732425e-06,
+      "loss": 0.7391,
+      "step": 7276
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.168499073957705e-06,
+      "loss": 0.7052,
+      "step": 7277
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.165922399575295e-06,
+      "loss": 0.8151,
+      "step": 7278
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1633459297711885e-06,
+      "loss": 0.3834,
+      "step": 7279
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.160769664731375e-06,
+      "loss": 0.7065,
+      "step": 7280
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.158193604641821e-06,
+      "loss": 0.7612,
+      "step": 7281
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.155617749688478e-06,
+      "loss": 0.7542,
+      "step": 7282
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.153042100057281e-06,
+      "loss": 0.8511,
+      "step": 7283
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.150466655934161e-06,
+      "loss": 0.7654,
+      "step": 7284
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1478914175050225e-06,
+      "loss": 0.3857,
+      "step": 7285
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.145316384955758e-06,
+      "loss": 0.7659,
+      "step": 7286
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1427415584722535e-06,
+      "loss": 0.7126,
+      "step": 7287
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.140166938240369e-06,
+      "loss": 0.8437,
+      "step": 7288
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.137592524445949e-06,
+      "loss": 0.7244,
+      "step": 7289
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.1350183172748365e-06,
+      "loss": 0.7282,
+      "step": 7290
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.132444316912847e-06,
+      "loss": 0.8048,
+      "step": 7291
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.129870523545781e-06,
+      "loss": 0.7383,
+      "step": 7292
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.127296937359437e-06,
+      "loss": 0.7138,
+      "step": 7293
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.124723558539583e-06,
+      "loss": 0.7594,
+      "step": 7294
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.122150387271982e-06,
+      "loss": 0.8018,
+      "step": 7295
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.119577423742374e-06,
+      "loss": 0.7432,
+      "step": 7296
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.117004668136495e-06,
+      "loss": 0.7668,
+      "step": 7297
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.1144321206400575e-06,
+      "loss": 0.808,
+      "step": 7298
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.111859781438758e-06,
+      "loss": 0.8023,
+      "step": 7299
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.109287650718287e-06,
+      "loss": 0.7518,
+      "step": 7300
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.1067157286643105e-06,
+      "loss": 0.7604,
+      "step": 7301
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.10414401546248e-06,
+      "loss": 0.7977,
+      "step": 7302
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.101572511298443e-06,
+      "loss": 0.7414,
+      "step": 7303
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.099001216357817e-06,
+      "loss": 0.7142,
+      "step": 7304
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.096430130826215e-06,
+      "loss": 0.7521,
+      "step": 7305
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0938592548892285e-06,
+      "loss": 0.7309,
+      "step": 7306
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.09128858873244e-06,
+      "loss": 0.7907,
+      "step": 7307
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.088718132541412e-06,
+      "loss": 0.7597,
+      "step": 7308
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.086147886501688e-06,
+      "loss": 0.7635,
+      "step": 7309
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.083577850798811e-06,
+      "loss": 0.719,
+      "step": 7310
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0810080256182945e-06,
+      "loss": 0.7882,
+      "step": 7311
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.078438411145638e-06,
+      "loss": 0.729,
+      "step": 7312
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0758690075663375e-06,
+      "loss": 0.7994,
+      "step": 7313
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.07329981506586e-06,
+      "loss": 0.7799,
+      "step": 7314
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.070730833829663e-06,
+      "loss": 0.8515,
+      "step": 7315
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.068162064043191e-06,
+      "loss": 0.7578,
+      "step": 7316
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.065593505891874e-06,
+      "loss": 0.7358,
+      "step": 7317
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.063025159561115e-06,
+      "loss": 0.7553,
+      "step": 7318
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.06045702523632e-06,
+      "loss": 0.7802,
+      "step": 7319
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.057889103102866e-06,
+      "loss": 0.7625,
+      "step": 7320
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.055321393346118e-06,
+      "loss": 0.7706,
+      "step": 7321
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.052753896151424e-06,
+      "loss": 0.7842,
+      "step": 7322
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.050186611704126e-06,
+      "loss": 0.8038,
+      "step": 7323
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.047619540189538e-06,
+      "loss": 0.7594,
+      "step": 7324
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.045052681792968e-06,
+      "loss": 0.7673,
+      "step": 7325
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.042486036699703e-06,
+      "loss": 0.7912,
+      "step": 7326
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0399196050950185e-06,
+      "loss": 0.8253,
+      "step": 7327
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.037353387164168e-06,
+      "loss": 0.7012,
+      "step": 7328
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.034787383092401e-06,
+      "loss": 0.8021,
+      "step": 7329
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.032221593064943e-06,
+      "loss": 0.7327,
+      "step": 7330
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.029656017266999e-06,
+      "loss": 0.8027,
+      "step": 7331
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.027090655883777e-06,
+      "loss": 0.7949,
+      "step": 7332
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.024525509100452e-06,
+      "loss": 0.7943,
+      "step": 7333
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.021960577102187e-06,
+      "loss": 0.7624,
+      "step": 7334
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.019395860074134e-06,
+      "loss": 0.8097,
+      "step": 7335
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0168313582014294e-06,
+      "loss": 0.7924,
+      "step": 7336
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.014267071669192e-06,
+      "loss": 0.7532,
+      "step": 7337
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0117030006625195e-06,
+      "loss": 0.7956,
+      "step": 7338
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.009139145366511e-06,
+      "loss": 0.8033,
+      "step": 7339
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.006575505966229e-06,
+      "loss": 0.8198,
+      "step": 7340
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.00401208264673e-06,
+      "loss": 0.793,
+      "step": 7341
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.001448875593063e-06,
+      "loss": 0.7441,
+      "step": 7342
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9988858849902485e-06,
+      "loss": 0.8103,
+      "step": 7343
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.996323111023293e-06,
+      "loss": 0.7566,
+      "step": 7344
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.993760553877198e-06,
+      "loss": 0.7551,
+      "step": 7345
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.991198213736936e-06,
+      "loss": 0.8185,
+      "step": 7346
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.988636090787473e-06,
+      "loss": 0.7688,
+      "step": 7347
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9860741852137545e-06,
+      "loss": 0.7185,
+      "step": 7348
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.983512497200716e-06,
+      "loss": 0.8062,
+      "step": 7349
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.980951026933272e-06,
+      "loss": 0.783,
+      "step": 7350
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.978389774596315e-06,
+      "loss": 0.7761,
+      "step": 7351
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.975828740374739e-06,
+      "loss": 0.8146,
+      "step": 7352
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.97326792445341e-06,
+      "loss": 0.781,
+      "step": 7353
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9707073270171745e-06,
+      "loss": 0.7799,
+      "step": 7354
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.96814694825088e-06,
+      "loss": 0.7134,
+      "step": 7355
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.965586788339342e-06,
+      "loss": 0.7105,
+      "step": 7356
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9630268474673646e-06,
+      "loss": 0.8349,
+      "step": 7357
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.960467125819741e-06,
+      "loss": 0.7429,
+      "step": 7358
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.957907623581246e-06,
+      "loss": 0.7161,
+      "step": 7359
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.955348340936632e-06,
+      "loss": 0.4001,
+      "step": 7360
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.952789278070648e-06,
+      "loss": 0.7938,
+      "step": 7361
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.950230435168016e-06,
+      "loss": 0.7535,
+      "step": 7362
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.947671812413448e-06,
+      "loss": 0.8366,
+      "step": 7363
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9451134099916355e-06,
+      "loss": 0.7762,
+      "step": 7364
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.942555228087262e-06,
+      "loss": 0.7436,
+      "step": 7365
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.939997266884986e-06,
+      "loss": 0.7337,
+      "step": 7366
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.937439526569456e-06,
+      "loss": 0.7459,
+      "step": 7367
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.934882007325303e-06,
+      "loss": 0.6802,
+      "step": 7368
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.932324709337146e-06,
+      "loss": 0.8081,
+      "step": 7369
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.929767632789573e-06,
+      "loss": 0.7916,
+      "step": 7370
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.92721077786718e-06,
+      "loss": 0.7637,
+      "step": 7371
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.924654144754526e-06,
+      "loss": 0.7352,
+      "step": 7372
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.922097733636162e-06,
+      "loss": 0.8228,
+      "step": 7373
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9195415446966265e-06,
+      "loss": 0.7472,
+      "step": 7374
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.916985578120438e-06,
+      "loss": 0.7559,
+      "step": 7375
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.914429834092094e-06,
+      "loss": 0.7528,
+      "step": 7376
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.911874312796087e-06,
+      "loss": 0.8264,
+      "step": 7377
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.909319014416884e-06,
+      "loss": 0.781,
+      "step": 7378
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.906763939138945e-06,
+      "loss": 0.7205,
+      "step": 7379
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.9042090871467005e-06,
+      "loss": 0.7852,
+      "step": 7380
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.90165445862458e-06,
+      "loss": 0.7217,
+      "step": 7381
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.899100053756989e-06,
+      "loss": 0.8183,
+      "step": 7382
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.896545872728309e-06,
+      "loss": 0.727,
+      "step": 7383
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8939919157229265e-06,
+      "loss": 0.7069,
+      "step": 7384
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.891438182925192e-06,
+      "loss": 0.7885,
+      "step": 7385
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.888884674519445e-06,
+      "loss": 0.7681,
+      "step": 7386
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.886331390690018e-06,
+      "loss": 0.7188,
+      "step": 7387
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8837783316212135e-06,
+      "loss": 0.7024,
+      "step": 7388
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.88122549749733e-06,
+      "loss": 0.7722,
+      "step": 7389
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.878672888502638e-06,
+      "loss": 0.7857,
+      "step": 7390
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.876120504821404e-06,
+      "loss": 0.7484,
+      "step": 7391
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.873568346637871e-06,
+      "loss": 0.813,
+      "step": 7392
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.871016414136261e-06,
+      "loss": 0.7674,
+      "step": 7393
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.868464707500795e-06,
+      "loss": 0.7682,
+      "step": 7394
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.865913226915662e-06,
+      "loss": 0.8115,
+      "step": 7395
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8633619725650395e-06,
+      "loss": 0.7924,
+      "step": 7396
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.860810944633097e-06,
+      "loss": 0.7766,
+      "step": 7397
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8582601433039756e-06,
+      "loss": 0.8127,
+      "step": 7398
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.855709568761806e-06,
+      "loss": 0.742,
+      "step": 7399
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.853159221190702e-06,
+      "loss": 0.7679,
+      "step": 7400
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.850609100774763e-06,
+      "loss": 0.7625,
+      "step": 7401
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.848059207698069e-06,
+      "loss": 0.7569,
+      "step": 7402
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.845509542144679e-06,
+      "loss": 0.7476,
+      "step": 7403
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.842960104298651e-06,
+      "loss": 0.8136,
+      "step": 7404
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8404108943440086e-06,
+      "loss": 0.8689,
+      "step": 7405
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.837861912464766e-06,
+      "loss": 0.7967,
+      "step": 7406
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.83531315884493e-06,
+      "loss": 0.7844,
+      "step": 7407
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.832764633668475e-06,
+      "loss": 0.7701,
+      "step": 7408
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8302163371193695e-06,
+      "loss": 0.7086,
+      "step": 7409
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.827668269381564e-06,
+      "loss": 0.8306,
+      "step": 7410
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.8251204306389905e-06,
+      "loss": 0.7388,
+      "step": 7411
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.822572821075561e-06,
+      "loss": 0.7625,
+      "step": 7412
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.820025440875182e-06,
+      "loss": 0.7391,
+      "step": 7413
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.817478290221734e-06,
+      "loss": 0.7609,
+      "step": 7414
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.814931369299083e-06,
+      "loss": 0.8023,
+      "step": 7415
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.812384678291073e-06,
+      "loss": 0.8046,
+      "step": 7416
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.809838217381546e-06,
+      "loss": 0.776,
+      "step": 7417
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.807291986754314e-06,
+      "loss": 0.762,
+      "step": 7418
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.80474598659318e-06,
+      "loss": 0.7875,
+      "step": 7419
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.8022002170819234e-06,
+      "loss": 0.8336,
+      "step": 7420
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.799654678404316e-06,
+      "loss": 0.7803,
+      "step": 7421
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.797109370744102e-06,
+      "loss": 0.7713,
+      "step": 7422
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.79456429428502e-06,
+      "loss": 0.7342,
+      "step": 7423
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.792019449210787e-06,
+      "loss": 0.7412,
+      "step": 7424
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.789474835705094e-06,
+      "loss": 0.7158,
+      "step": 7425
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.786930453951637e-06,
+      "loss": 0.7853,
+      "step": 7426
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.784386304134076e-06,
+      "loss": 0.7576,
+      "step": 7427
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.78184238643606e-06,
+      "loss": 0.6862,
+      "step": 7428
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7792987010412224e-06,
+      "loss": 0.7575,
+      "step": 7429
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.776755248133181e-06,
+      "loss": 0.7764,
+      "step": 7430
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7742120278955374e-06,
+      "loss": 0.74,
+      "step": 7431
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7716690405118665e-06,
+      "loss": 0.7289,
+      "step": 7432
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.769126286165744e-06,
+      "loss": 0.7779,
+      "step": 7433
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.766583765040715e-06,
+      "loss": 0.8106,
+      "step": 7434
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.764041477320307e-06,
+      "loss": 0.7601,
+      "step": 7435
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7614994231880435e-06,
+      "loss": 0.7817,
+      "step": 7436
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.758957602827419e-06,
+      "loss": 0.6792,
+      "step": 7437
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.756416016421912e-06,
+      "loss": 0.3691,
+      "step": 7438
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7538746641549946e-06,
+      "loss": 0.8208,
+      "step": 7439
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.751333546210108e-06,
+      "loss": 0.7402,
+      "step": 7440
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.748792662770689e-06,
+      "loss": 0.7752,
+      "step": 7441
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.746252014020145e-06,
+      "loss": 0.7811,
+      "step": 7442
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.74371160014188e-06,
+      "loss": 0.7577,
+      "step": 7443
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.74117142131927e-06,
+      "loss": 0.7816,
+      "step": 7444
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.738631477735675e-06,
+      "loss": 0.7107,
+      "step": 7445
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.736091769574451e-06,
+      "loss": 0.7183,
+      "step": 7446
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.73355229701892e-06,
+      "loss": 0.7621,
+      "step": 7447
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.731013060252391e-06,
+      "loss": 0.7924,
+      "step": 7448
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7284740594581695e-06,
+      "loss": 0.799,
+      "step": 7449
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.725935294819525e-06,
+      "loss": 0.7702,
+      "step": 7450
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.723396766519721e-06,
+      "loss": 0.8843,
+      "step": 7451
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7208584747420045e-06,
+      "loss": 0.734,
+      "step": 7452
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.718320419669599e-06,
+      "loss": 0.8264,
+      "step": 7453
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.715782601485718e-06,
+      "loss": 0.798,
+      "step": 7454
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.713245020373546e-06,
+      "loss": 0.7837,
+      "step": 7455
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.710707676516269e-06,
+      "loss": 0.3718,
+      "step": 7456
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.708170570097041e-06,
+      "loss": 0.7709,
+      "step": 7457
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.705633701298999e-06,
+      "loss": 0.8253,
+      "step": 7458
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.703097070305273e-06,
+      "loss": 0.7508,
+      "step": 7459
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.700560677298973e-06,
+      "loss": 0.7797,
+      "step": 7460
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.698024522463178e-06,
+      "loss": 0.7622,
+      "step": 7461
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.695488605980972e-06,
+      "loss": 0.7829,
+      "step": 7462
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.692952928035406e-06,
+      "loss": 0.7442,
+      "step": 7463
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.690417488809513e-06,
+      "loss": 0.4177,
+      "step": 7464
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.687882288486325e-06,
+      "loss": 0.7917,
+      "step": 7465
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.685347327248839e-06,
+      "loss": 0.7489,
+      "step": 7466
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.682812605280038e-06,
+      "loss": 0.765,
+      "step": 7467
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6802781227629e-06,
+      "loss": 0.8017,
+      "step": 7468
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.67774387988037e-06,
+      "loss": 0.7655,
+      "step": 7469
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6752098768153885e-06,
+      "loss": 0.8095,
+      "step": 7470
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.672676113750866e-06,
+      "loss": 0.7543,
+      "step": 7471
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.670142590869709e-06,
+      "loss": 0.8136,
+      "step": 7472
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.667609308354799e-06,
+      "loss": 0.7102,
+      "step": 7473
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.665076266388995e-06,
+      "loss": 0.7788,
+      "step": 7474
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.662543465155153e-06,
+      "loss": 0.7789,
+      "step": 7475
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.660010904836103e-06,
+      "loss": 0.8231,
+      "step": 7476
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.657478585614649e-06,
+      "loss": 0.7082,
+      "step": 7477
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.654946507673601e-06,
+      "loss": 0.8245,
+      "step": 7478
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.652414671195726e-06,
+      "loss": 0.7432,
+      "step": 7479
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.649883076363789e-06,
+      "loss": 0.757,
+      "step": 7480
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.647351723360533e-06,
+      "loss": 0.8271,
+      "step": 7481
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.644820612368687e-06,
+      "loss": 0.7283,
+      "step": 7482
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.64228974357096e-06,
+      "loss": 0.7287,
+      "step": 7483
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6397591171500334e-06,
+      "loss": 0.7714,
+      "step": 7484
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.637228733288593e-06,
+      "loss": 0.7854,
+      "step": 7485
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.63469859216929e-06,
+      "loss": 0.7846,
+      "step": 7486
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.632168693974758e-06,
+      "loss": 0.7669,
+      "step": 7487
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.629639038887627e-06,
+      "loss": 0.7264,
+      "step": 7488
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.627109627090495e-06,
+      "loss": 0.7075,
+      "step": 7489
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.624580458765949e-06,
+      "loss": 0.7295,
+      "step": 7490
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.622051534096559e-06,
+      "loss": 0.7354,
+      "step": 7491
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6195228532648745e-06,
+      "loss": 0.7076,
+      "step": 7492
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.6169944164534265e-06,
+      "loss": 0.6992,
+      "step": 7493
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.614466223844736e-06,
+      "loss": 0.7809,
+      "step": 7494
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.611938275621297e-06,
+      "loss": 0.7894,
+      "step": 7495
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.609410571965591e-06,
+      "loss": 0.7363,
+      "step": 7496
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.606883113060076e-06,
+      "loss": 0.8112,
+      "step": 7497
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.604355899087207e-06,
+      "loss": 0.7705,
+      "step": 7498
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.601828930229402e-06,
+      "loss": 0.8124,
+      "step": 7499
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.599302206669074e-06,
+      "loss": 0.6821,
+      "step": 7500
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.596775728588616e-06,
+      "loss": 0.7631,
+      "step": 7501
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.594249496170403e-06,
+      "loss": 0.7999,
+      "step": 7502
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.591723509596786e-06,
+      "loss": 0.7729,
+      "step": 7503
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.589197769050111e-06,
+      "loss": 0.3682,
+      "step": 7504
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.586672274712695e-06,
+      "loss": 0.8454,
+      "step": 7505
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.584147026766839e-06,
+      "loss": 0.7845,
+      "step": 7506
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.581622025394834e-06,
+      "loss": 0.7571,
+      "step": 7507
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.579097270778945e-06,
+      "loss": 0.754,
+      "step": 7508
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.576572763101419e-06,
+      "loss": 0.7506,
+      "step": 7509
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.57404850254449e-06,
+      "loss": 0.7633,
+      "step": 7510
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.571524489290373e-06,
+      "loss": 0.8154,
+      "step": 7511
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.569000723521266e-06,
+      "loss": 0.7455,
+      "step": 7512
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.56647720541934e-06,
+      "loss": 0.7967,
+      "step": 7513
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.563953935166764e-06,
+      "loss": 0.3652,
+      "step": 7514
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.56143091294568e-06,
+      "loss": 0.7766,
+      "step": 7515
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.558908138938205e-06,
+      "loss": 0.7165,
+      "step": 7516
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.556385613326452e-06,
+      "loss": 0.8079,
+      "step": 7517
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.553863336292512e-06,
+      "loss": 0.7524,
+      "step": 7518
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.551341308018448e-06,
+      "loss": 0.7684,
+      "step": 7519
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.548819528686321e-06,
+      "loss": 0.6763,
+      "step": 7520
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.54629799847816e-06,
+      "loss": 0.7094,
+      "step": 7521
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.543776717575988e-06,
+      "loss": 0.7918,
+      "step": 7522
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.541255686161796e-06,
+      "loss": 0.6891,
+      "step": 7523
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.538734904417574e-06,
+      "loss": 0.709,
+      "step": 7524
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.53621437252528e-06,
+      "loss": 0.7113,
+      "step": 7525
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.533694090666856e-06,
+      "loss": 0.7941,
+      "step": 7526
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5311740590242364e-06,
+      "loss": 0.7565,
+      "step": 7527
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.528654277779327e-06,
+      "loss": 0.7803,
+      "step": 7528
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.526134747114013e-06,
+      "loss": 0.3739,
+      "step": 7529
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.523615467210176e-06,
+      "loss": 0.7855,
+      "step": 7530
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.521096438249666e-06,
+      "loss": 0.748,
+      "step": 7531
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.518577660414318e-06,
+      "loss": 0.7719,
+      "step": 7532
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5160591338859525e-06,
+      "loss": 0.7062,
+      "step": 7533
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.5135408588463724e-06,
+      "loss": 0.7703,
+      "step": 7534
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.511022835477356e-06,
+      "loss": 0.7963,
+      "step": 7535
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.508505063960664e-06,
+      "loss": 0.782,
+      "step": 7536
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.505987544478052e-06,
+      "loss": 0.6974,
+      "step": 7537
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.503470277211239e-06,
+      "loss": 0.8681,
+      "step": 7538
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.500953262341935e-06,
+      "loss": 0.7856,
+      "step": 7539
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.498436500051837e-06,
+      "loss": 0.413,
+      "step": 7540
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.49591999052261e-06,
+      "loss": 0.4044,
+      "step": 7541
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.493403733935915e-06,
+      "loss": 0.7507,
+      "step": 7542
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.490887730473383e-06,
+      "loss": 0.7797,
+      "step": 7543
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.488371980316639e-06,
+      "loss": 0.718,
+      "step": 7544
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4858564836472744e-06,
+      "loss": 0.7995,
+      "step": 7545
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.483341240646879e-06,
+      "loss": 0.8428,
+      "step": 7546
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.480826251497012e-06,
+      "loss": 0.7436,
+      "step": 7547
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4783115163792185e-06,
+      "loss": 0.7708,
+      "step": 7548
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.475797035475021e-06,
+      "loss": 0.7793,
+      "step": 7549
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.473282808965935e-06,
+      "loss": 0.7081,
+      "step": 7550
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.470768837033445e-06,
+      "loss": 0.7413,
+      "step": 7551
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4682551198590234e-06,
+      "loss": 0.7692,
+      "step": 7552
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4657416576241265e-06,
+      "loss": 0.7119,
+      "step": 7553
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4632284505101865e-06,
+      "loss": 0.7656,
+      "step": 7554
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.460715498698616e-06,
+      "loss": 0.6792,
+      "step": 7555
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.458202802370821e-06,
+      "loss": 0.857,
+      "step": 7556
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.455690361708175e-06,
+      "loss": 0.7946,
+      "step": 7557
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.453178176892037e-06,
+      "loss": 0.7776,
+      "step": 7558
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.450666248103756e-06,
+      "loss": 0.7006,
+      "step": 7559
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.448154575524654e-06,
+      "loss": 0.7196,
+      "step": 7560
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4456431593360325e-06,
+      "loss": 0.7435,
+      "step": 7561
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.44313199971918e-06,
+      "loss": 0.7613,
+      "step": 7562
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.440621096855366e-06,
+      "loss": 0.7322,
+      "step": 7563
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.438110450925843e-06,
+      "loss": 0.7949,
+      "step": 7564
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.435600062111835e-06,
+      "loss": 0.7723,
+      "step": 7565
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.433089930594565e-06,
+      "loss": 0.7931,
+      "step": 7566
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.43058005655522e-06,
+      "loss": 0.8136,
+      "step": 7567
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.428070440174974e-06,
+      "loss": 0.7904,
+      "step": 7568
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4255610816349915e-06,
+      "loss": 0.8101,
+      "step": 7569
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.423051981116406e-06,
+      "loss": 0.76,
+      "step": 7570
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.420543138800334e-06,
+      "loss": 0.7685,
+      "step": 7571
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.418034554867884e-06,
+      "loss": 0.7713,
+      "step": 7572
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.415526229500136e-06,
+      "loss": 0.7556,
+      "step": 7573
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.413018162878151e-06,
+      "loss": 0.6844,
+      "step": 7574
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.410510355182975e-06,
+      "loss": 0.738,
+      "step": 7575
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4080028065956375e-06,
+      "loss": 0.8024,
+      "step": 7576
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.405495517297146e-06,
+      "loss": 0.7824,
+      "step": 7577
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.4029884874684835e-06,
+      "loss": 0.8017,
+      "step": 7578
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.400481717290628e-06,
+      "loss": 0.7795,
+      "step": 7579
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.397975206944529e-06,
+      "loss": 0.7894,
+      "step": 7580
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.395468956611113e-06,
+      "loss": 0.7372,
+      "step": 7581
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.392962966471304e-06,
+      "loss": 0.7594,
+      "step": 7582
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.390457236705989e-06,
+      "loss": 0.6836,
+      "step": 7583
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.38795176749605e-06,
+      "loss": 0.7341,
+      "step": 7584
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.385446559022341e-06,
+      "loss": 0.8047,
+      "step": 7585
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.382941611465705e-06,
+      "loss": 0.8477,
+      "step": 7586
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.380436925006958e-06,
+      "loss": 0.779,
+      "step": 7587
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3779324998269e-06,
+      "loss": 0.743,
+      "step": 7588
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.37542833610632e-06,
+      "loss": 0.732,
+      "step": 7589
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3729244340259776e-06,
+      "loss": 0.7988,
+      "step": 7590
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.370420793766612e-06,
+      "loss": 0.8019,
+      "step": 7591
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.367917415508957e-06,
+      "loss": 0.6876,
+      "step": 7592
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.365414299433717e-06,
+      "loss": 0.7228,
+      "step": 7593
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.362911445721577e-06,
+      "loss": 0.4019,
+      "step": 7594
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3604088545532085e-06,
+      "loss": 0.7272,
+      "step": 7595
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.357906526109262e-06,
+      "loss": 0.7308,
+      "step": 7596
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.355404460570364e-06,
+      "loss": 0.8388,
+      "step": 7597
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.352902658117133e-06,
+      "loss": 0.7819,
+      "step": 7598
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3504011189301585e-06,
+      "loss": 0.7746,
+      "step": 7599
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.347899843190012e-06,
+      "loss": 0.8284,
+      "step": 7600
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3453988310772544e-06,
+      "loss": 0.714,
+      "step": 7601
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.342898082772417e-06,
+      "loss": 0.7813,
+      "step": 7602
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.340397598456018e-06,
+      "loss": 0.7623,
+      "step": 7603
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.337897378308554e-06,
+      "loss": 0.723,
+      "step": 7604
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.335397422510506e-06,
+      "loss": 0.757,
+      "step": 7605
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.332897731242333e-06,
+      "loss": 0.7559,
+      "step": 7606
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.330398304684473e-06,
+      "loss": 0.7723,
+      "step": 7607
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.327899143017353e-06,
+      "loss": 0.7169,
+      "step": 7608
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.325400246421372e-06,
+      "loss": 0.7456,
+      "step": 7609
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3229016150769086e-06,
+      "loss": 0.7674,
+      "step": 7610
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.320403249164335e-06,
+      "loss": 0.7549,
+      "step": 7611
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3179051488639945e-06,
+      "loss": 0.7245,
+      "step": 7612
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.315407314356207e-06,
+      "loss": 0.6959,
+      "step": 7613
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.312909745821285e-06,
+      "loss": 0.8199,
+      "step": 7614
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3104124434395145e-06,
+      "loss": 0.7862,
+      "step": 7615
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.307915407391163e-06,
+      "loss": 0.8063,
+      "step": 7616
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.305418637856476e-06,
+      "loss": 0.7802,
+      "step": 7617
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.302922135015692e-06,
+      "loss": 0.713,
+      "step": 7618
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.300425899049014e-06,
+      "loss": 0.7966,
+      "step": 7619
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.297929930136633e-06,
+      "loss": 0.7092,
+      "step": 7620
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.295434228458726e-06,
+      "loss": 0.775,
+      "step": 7621
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.292938794195442e-06,
+      "loss": 0.7959,
+      "step": 7622
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.290443627526912e-06,
+      "loss": 0.7523,
+      "step": 7623
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.287948728633257e-06,
+      "loss": 0.7788,
+      "step": 7624
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.285454097694565e-06,
+      "loss": 0.693,
+      "step": 7625
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.282959734890912e-06,
+      "loss": 0.7514,
+      "step": 7626
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.28046564040236e-06,
+      "loss": 0.7104,
+      "step": 7627
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.277971814408942e-06,
+      "loss": 0.8294,
+      "step": 7628
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.275478257090674e-06,
+      "loss": 0.7616,
+      "step": 7629
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.27298496862755e-06,
+      "loss": 0.7188,
+      "step": 7630
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.270491949199557e-06,
+      "loss": 0.754,
+      "step": 7631
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.267999198986649e-06,
+      "loss": 0.8177,
+      "step": 7632
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.265506718168765e-06,
+      "loss": 0.7608,
+      "step": 7633
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.263014506925829e-06,
+      "loss": 0.7229,
+      "step": 7634
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2605225654377365e-06,
+      "loss": 0.751,
+      "step": 7635
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.258030893884368e-06,
+      "loss": 0.8118,
+      "step": 7636
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2555394924455946e-06,
+      "loss": 0.747,
+      "step": 7637
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.253048361301251e-06,
+      "loss": 0.7557,
+      "step": 7638
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.2505575006311585e-06,
+      "loss": 0.8704,
+      "step": 7639
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.248066910615125e-06,
+      "loss": 0.6994,
+      "step": 7640
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.245576591432933e-06,
+      "loss": 0.7933,
+      "step": 7641
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.243086543264345e-06,
+      "loss": 0.809,
+      "step": 7642
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.240596766289102e-06,
+      "loss": 0.7555,
+      "step": 7643
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.238107260686939e-06,
+      "loss": 0.7909,
+      "step": 7644
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.235618026637551e-06,
+      "loss": 0.8976,
+      "step": 7645
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.233129064320627e-06,
+      "loss": 0.8396,
+      "step": 7646
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.230640373915837e-06,
+      "loss": 0.7173,
+      "step": 7647
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.228151955602825e-06,
+      "loss": 0.8191,
+      "step": 7648
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.225663809561213e-06,
+      "loss": 0.8034,
+      "step": 7649
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.223175935970618e-06,
+      "loss": 0.7382,
+      "step": 7650
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.22068833501062e-06,
+      "loss": 0.7649,
+      "step": 7651
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.218201006860787e-06,
+      "loss": 0.7368,
+      "step": 7652
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.215713951700672e-06,
+      "loss": 0.7899,
+      "step": 7653
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.213227169709802e-06,
+      "loss": 0.7809,
+      "step": 7654
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.21074066106768e-06,
+      "loss": 0.7843,
+      "step": 7655
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.208254425953799e-06,
+      "loss": 0.8037,
+      "step": 7656
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.205768464547632e-06,
+      "loss": 0.7698,
+      "step": 7657
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.2032827770286255e-06,
+      "loss": 0.7418,
+      "step": 7658
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.200797363576203e-06,
+      "loss": 0.7467,
+      "step": 7659
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1983122243697865e-06,
+      "loss": 0.7152,
+      "step": 7660
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.195827359588759e-06,
+      "loss": 0.7721,
+      "step": 7661
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.193342769412487e-06,
+      "loss": 0.7488,
+      "step": 7662
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.19085845402033e-06,
+      "loss": 0.7826,
+      "step": 7663
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.188374413591611e-06,
+      "loss": 0.7568,
+      "step": 7664
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.185890648305645e-06,
+      "loss": 0.7329,
+      "step": 7665
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1834071583417235e-06,
+      "loss": 0.7891,
+      "step": 7666
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1809239438791166e-06,
+      "loss": 0.7373,
+      "step": 7667
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1784410050970755e-06,
+      "loss": 0.8152,
+      "step": 7668
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.175958342174828e-06,
+      "loss": 0.8511,
+      "step": 7669
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1734759552915915e-06,
+      "loss": 0.7445,
+      "step": 7670
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.170993844626554e-06,
+      "loss": 0.7405,
+      "step": 7671
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.168512010358885e-06,
+      "loss": 0.7095,
+      "step": 7672
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.166030452667741e-06,
+      "loss": 0.782,
+      "step": 7673
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.163549171732251e-06,
+      "loss": 0.7167,
+      "step": 7674
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.161068167731525e-06,
+      "loss": 0.7651,
+      "step": 7675
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.158587440844657e-06,
+      "loss": 0.7929,
+      "step": 7676
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.156106991250721e-06,
+      "loss": 0.8317,
+      "step": 7677
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.153626819128761e-06,
+      "loss": 0.7505,
+      "step": 7678
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.151146924657818e-06,
+      "loss": 0.7068,
+      "step": 7679
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.148667308016898e-06,
+      "loss": 0.729,
+      "step": 7680
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.146187969384994e-06,
+      "loss": 0.7242,
+      "step": 7681
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.143708908941072e-06,
+      "loss": 0.8109,
+      "step": 7682
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.141230126864093e-06,
+      "loss": 0.3979,
+      "step": 7683
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.138751623332981e-06,
+      "loss": 0.812,
+      "step": 7684
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.136273398526649e-06,
+      "loss": 0.7682,
+      "step": 7685
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.133795452623989e-06,
+      "loss": 0.7119,
+      "step": 7686
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.131317785803874e-06,
+      "loss": 0.7826,
+      "step": 7687
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.128840398245147e-06,
+      "loss": 0.7483,
+      "step": 7688
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.126363290126649e-06,
+      "loss": 0.7877,
+      "step": 7689
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1238864616271845e-06,
+      "loss": 0.7231,
+      "step": 7690
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.12140991292554e-06,
+      "loss": 0.7998,
+      "step": 7691
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.118933644200494e-06,
+      "loss": 0.8601,
+      "step": 7692
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.116457655630793e-06,
+      "loss": 0.8119,
+      "step": 7693
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.113981947395163e-06,
+      "loss": 0.7506,
+      "step": 7694
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.111506519672317e-06,
+      "loss": 0.7313,
+      "step": 7695
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.109031372640943e-06,
+      "loss": 0.8545,
+      "step": 7696
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.1065565064797126e-06,
+      "loss": 0.8586,
+      "step": 7697
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.104081921367266e-06,
+      "loss": 0.688,
+      "step": 7698
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.101607617482244e-06,
+      "loss": 0.6871,
+      "step": 7699
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.099133595003247e-06,
+      "loss": 0.7245,
+      "step": 7700
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.09665985410886e-06,
+      "loss": 0.7696,
+      "step": 7701
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.094186394977657e-06,
+      "loss": 0.7175,
+      "step": 7702
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.091713217788183e-06,
+      "loss": 0.8225,
+      "step": 7703
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.089240322718959e-06,
+      "loss": 0.7774,
+      "step": 7704
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.086767709948501e-06,
+      "loss": 0.8353,
+      "step": 7705
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.084295379655287e-06,
+      "loss": 0.8112,
+      "step": 7706
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.081823332017789e-06,
+      "loss": 0.869,
+      "step": 7707
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0793515672144424e-06,
+      "loss": 0.7872,
+      "step": 7708
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.076880085423683e-06,
+      "loss": 0.7322,
+      "step": 7709
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.074408886823911e-06,
+      "loss": 0.7746,
+      "step": 7710
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.071937971593504e-06,
+      "loss": 0.7699,
+      "step": 7711
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.069467339910835e-06,
+      "loss": 0.7399,
+      "step": 7712
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.066996991954244e-06,
+      "loss": 0.8025,
+      "step": 7713
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.064526927902047e-06,
+      "loss": 0.3936,
+      "step": 7714
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.062057147932555e-06,
+      "loss": 0.7161,
+      "step": 7715
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.059587652224044e-06,
+      "loss": 0.7246,
+      "step": 7716
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0571184409547755e-06,
+      "loss": 0.7265,
+      "step": 7717
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.054649514302992e-06,
+      "loss": 0.711,
+      "step": 7718
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0521808724469135e-06,
+      "loss": 0.7717,
+      "step": 7719
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.049712515564734e-06,
+      "loss": 0.7715,
+      "step": 7720
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.047244443834639e-06,
+      "loss": 0.8192,
+      "step": 7721
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.044776657434786e-06,
+      "loss": 0.8089,
+      "step": 7722
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.042309156543308e-06,
+      "loss": 0.8187,
+      "step": 7723
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.03984194133832e-06,
+      "loss": 0.7584,
+      "step": 7724
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.037375011997928e-06,
+      "loss": 0.8099,
+      "step": 7725
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.034908368700198e-06,
+      "loss": 0.7716,
+      "step": 7726
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.032442011623192e-06,
+      "loss": 0.7216,
+      "step": 7727
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.029975940944938e-06,
+      "loss": 0.8278,
+      "step": 7728
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.027510156843459e-06,
+      "loss": 0.7959,
+      "step": 7729
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.025044659496735e-06,
+      "loss": 0.7669,
+      "step": 7730
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.02257944908275e-06,
+      "loss": 0.7849,
+      "step": 7731
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.020114525779453e-06,
+      "loss": 0.7771,
+      "step": 7732
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.017649889764767e-06,
+      "loss": 0.7537,
+      "step": 7733
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.015185541216611e-06,
+      "loss": 0.7536,
+      "step": 7734
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.012721480312873e-06,
+      "loss": 0.8233,
+      "step": 7735
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.010257707231417e-06,
+      "loss": 0.7816,
+      "step": 7736
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.007794222150094e-06,
+      "loss": 0.78,
+      "step": 7737
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0053310252467315e-06,
+      "loss": 0.7332,
+      "step": 7738
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.0028681166991365e-06,
+      "loss": 0.76,
+      "step": 7739
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.000405496685088e-06,
+      "loss": 0.8262,
+      "step": 7740
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.99794316538236e-06,
+      "loss": 0.8061,
+      "step": 7741
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.995481122968692e-06,
+      "loss": 0.7857,
+      "step": 7742
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.993019369621802e-06,
+      "loss": 0.8016,
+      "step": 7743
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9905579055194006e-06,
+      "loss": 0.7257,
+      "step": 7744
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.988096730839166e-06,
+      "loss": 0.7462,
+      "step": 7745
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.985635845758753e-06,
+      "loss": 0.7583,
+      "step": 7746
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.983175250455808e-06,
+      "loss": 0.7931,
+      "step": 7747
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.980714945107946e-06,
+      "loss": 0.7909,
+      "step": 7748
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.978254929892767e-06,
+      "loss": 0.8336,
+      "step": 7749
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9757952049878424e-06,
+      "loss": 0.7922,
+      "step": 7750
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.973335770570735e-06,
+      "loss": 0.793,
+      "step": 7751
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.970876626818975e-06,
+      "loss": 0.8158,
+      "step": 7752
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.968417773910074e-06,
+      "loss": 0.7022,
+      "step": 7753
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.965959212021531e-06,
+      "loss": 0.7725,
+      "step": 7754
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.963500941330814e-06,
+      "loss": 0.7792,
+      "step": 7755
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.96104296201537e-06,
+      "loss": 0.7321,
+      "step": 7756
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.958585274252635e-06,
+      "loss": 0.7468,
+      "step": 7757
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.956127878220015e-06,
+      "loss": 0.6751,
+      "step": 7758
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.953670774094895e-06,
+      "loss": 0.7822,
+      "step": 7759
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.951213962054645e-06,
+      "loss": 0.6857,
+      "step": 7760
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9487574422766114e-06,
+      "loss": 0.7904,
+      "step": 7761
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.946301214938116e-06,
+      "loss": 0.7546,
+      "step": 7762
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.943845280216458e-06,
+      "loss": 0.7526,
+      "step": 7763
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.941389638288927e-06,
+      "loss": 0.7936,
+      "step": 7764
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.938934289332781e-06,
+      "loss": 0.8153,
+      "step": 7765
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.936479233525254e-06,
+      "loss": 0.7389,
+      "step": 7766
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.9340244710435746e-06,
+      "loss": 0.7366,
+      "step": 7767
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.931570002064932e-06,
+      "loss": 0.8039,
+      "step": 7768
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.929115826766507e-06,
+      "loss": 0.7646,
+      "step": 7769
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.926661945325454e-06,
+      "loss": 0.7887,
+      "step": 7770
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.924208357918907e-06,
+      "loss": 0.7279,
+      "step": 7771
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.921755064723973e-06,
+      "loss": 0.7431,
+      "step": 7772
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.919302065917752e-06,
+      "loss": 0.7617,
+      "step": 7773
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.91684936167731e-06,
+      "loss": 0.7603,
+      "step": 7774
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.914396952179698e-06,
+      "loss": 0.8022,
+      "step": 7775
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.911944837601935e-06,
+      "loss": 0.7138,
+      "step": 7776
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.909493018121038e-06,
+      "loss": 0.8118,
+      "step": 7777
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.907041493913987e-06,
+      "loss": 0.7461,
+      "step": 7778
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.904590265157744e-06,
+      "loss": 0.7762,
+      "step": 7779
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.902139332029255e-06,
+      "loss": 0.7974,
+      "step": 7780
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.899688694705441e-06,
+      "loss": 0.7464,
+      "step": 7781
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.897238353363195e-06,
+      "loss": 0.7446,
+      "step": 7782
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.894788308179407e-06,
+      "loss": 0.7479,
+      "step": 7783
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.892338559330926e-06,
+      "loss": 0.6777,
+      "step": 7784
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.889889106994583e-06,
+      "loss": 0.7358,
+      "step": 7785
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8874399513472025e-06,
+      "loss": 0.8507,
+      "step": 7786
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.884991092565574e-06,
+      "loss": 0.777,
+      "step": 7787
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8825425308264635e-06,
+      "loss": 0.7253,
+      "step": 7788
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.880094266306626e-06,
+      "loss": 0.7927,
+      "step": 7789
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.877646299182786e-06,
+      "loss": 0.6963,
+      "step": 7790
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.875198629631658e-06,
+      "loss": 0.3599,
+      "step": 7791
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.872751257829915e-06,
+      "loss": 0.7137,
+      "step": 7792
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8703041839542326e-06,
+      "loss": 0.7655,
+      "step": 7793
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.867857408181249e-06,
+      "loss": 0.7586,
+      "step": 7794
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.865410930687581e-06,
+      "loss": 0.7446,
+      "step": 7795
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8629647516498345e-06,
+      "loss": 0.7346,
+      "step": 7796
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8605188712445846e-06,
+      "loss": 0.7368,
+      "step": 7797
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.858073289648382e-06,
+      "loss": 0.731,
+      "step": 7798
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.855628007037773e-06,
+      "loss": 0.6877,
+      "step": 7799
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.853183023589261e-06,
+      "loss": 0.7516,
+      "step": 7800
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8507383394793436e-06,
+      "loss": 0.7533,
+      "step": 7801
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.848293954884483e-06,
+      "loss": 0.754,
+      "step": 7802
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.845849869981137e-06,
+      "loss": 0.7523,
+      "step": 7803
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.843406084945728e-06,
+      "loss": 0.7635,
+      "step": 7804
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.840962599954655e-06,
+      "loss": 0.786,
+      "step": 7805
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.838519415184309e-06,
+      "loss": 0.8116,
+      "step": 7806
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.836076530811052e-06,
+      "loss": 0.7704,
+      "step": 7807
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.833633947011218e-06,
+      "loss": 0.7287,
+      "step": 7808
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.831191663961132e-06,
+      "loss": 0.7628,
+      "step": 7809
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.828749681837088e-06,
+      "loss": 0.7416,
+      "step": 7810
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.826308000815355e-06,
+      "loss": 0.8009,
+      "step": 7811
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.823866621072196e-06,
+      "loss": 0.7917,
+      "step": 7812
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.821425542783839e-06,
+      "loss": 0.8359,
+      "step": 7813
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.81898476612649e-06,
+      "loss": 0.8004,
+      "step": 7814
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.816544291276336e-06,
+      "loss": 0.7589,
+      "step": 7815
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.81410411840955e-06,
+      "loss": 0.7055,
+      "step": 7816
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.811664247702271e-06,
+      "loss": 0.7685,
+      "step": 7817
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.80922467933062e-06,
+      "loss": 0.7428,
+      "step": 7818
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.806785413470703e-06,
+      "loss": 0.7393,
+      "step": 7819
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.804346450298599e-06,
+      "loss": 0.7857,
+      "step": 7820
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.801907789990356e-06,
+      "loss": 0.7717,
+      "step": 7821
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.799469432722018e-06,
+      "loss": 0.8243,
+      "step": 7822
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.797031378669597e-06,
+      "loss": 0.7365,
+      "step": 7823
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.794593628009077e-06,
+      "loss": 0.8056,
+      "step": 7824
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.792156180916438e-06,
+      "loss": 0.781,
+      "step": 7825
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.789719037567618e-06,
+      "loss": 0.8393,
+      "step": 7826
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.787282198138552e-06,
+      "loss": 0.7658,
+      "step": 7827
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.784845662805134e-06,
+      "loss": 0.7714,
+      "step": 7828
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7824094317432535e-06,
+      "loss": 0.3808,
+      "step": 7829
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7799735051287664e-06,
+      "loss": 0.7317,
+      "step": 7830
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.777537883137507e-06,
+      "loss": 0.7357,
+      "step": 7831
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7751025659453005e-06,
+      "loss": 0.6758,
+      "step": 7832
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.772667553727934e-06,
+      "loss": 0.7686,
+      "step": 7833
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.770232846661177e-06,
+      "loss": 0.8073,
+      "step": 7834
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.767798444920786e-06,
+      "loss": 0.3634,
+      "step": 7835
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.765364348682485e-06,
+      "loss": 0.7468,
+      "step": 7836
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.762930558121975e-06,
+      "loss": 0.7485,
+      "step": 7837
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7604970734149505e-06,
+      "loss": 0.8001,
+      "step": 7838
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.758063894737065e-06,
+      "loss": 0.8335,
+      "step": 7839
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.755631022263959e-06,
+      "loss": 0.7291,
+      "step": 7840
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.753198456171247e-06,
+      "loss": 0.7782,
+      "step": 7841
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.750766196634532e-06,
+      "loss": 0.779,
+      "step": 7842
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.748334243829381e-06,
+      "loss": 0.8138,
+      "step": 7843
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7459025979313435e-06,
+      "loss": 0.7641,
+      "step": 7844
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.743471259115951e-06,
+      "loss": 0.7875,
+      "step": 7845
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.741040227558714e-06,
+      "loss": 0.763,
+      "step": 7846
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.738609503435109e-06,
+      "loss": 0.7919,
+      "step": 7847
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.736179086920605e-06,
+      "loss": 0.7271,
+      "step": 7848
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.733748978190641e-06,
+      "loss": 0.7231,
+      "step": 7849
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.731319177420628e-06,
+      "loss": 0.6968,
+      "step": 7850
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.72888968478597e-06,
+      "loss": 0.84,
+      "step": 7851
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7264605004620366e-06,
+      "loss": 0.7416,
+      "step": 7852
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.724031624624175e-06,
+      "loss": 0.7414,
+      "step": 7853
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.721603057447722e-06,
+      "loss": 0.7308,
+      "step": 7854
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.719174799107981e-06,
+      "loss": 0.759,
+      "step": 7855
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.716746849780235e-06,
+      "loss": 0.7823,
+      "step": 7856
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7143192096397405e-06,
+      "loss": 0.3771,
+      "step": 7857
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7118918788617475e-06,
+      "loss": 0.7291,
+      "step": 7858
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.709464857621467e-06,
+      "loss": 0.7711,
+      "step": 7859
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.7070381460940925e-06,
+      "loss": 0.6761,
+      "step": 7860
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.704611744454804e-06,
+      "loss": 0.7611,
+      "step": 7861
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.702185652878745e-06,
+      "loss": 0.7631,
+      "step": 7862
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.699759871541042e-06,
+      "loss": 0.7687,
+      "step": 7863
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.697334400616808e-06,
+      "loss": 0.8061,
+      "step": 7864
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.694909240281117e-06,
+      "loss": 0.82,
+      "step": 7865
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.692484390709037e-06,
+      "loss": 0.8147,
+      "step": 7866
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6900598520756015e-06,
+      "loss": 0.7934,
+      "step": 7867
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.68763562455583e-06,
+      "loss": 0.8004,
+      "step": 7868
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.685211708324714e-06,
+      "loss": 0.7442,
+      "step": 7869
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6827881035572195e-06,
+      "loss": 0.7175,
+      "step": 7870
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.680364810428304e-06,
+      "loss": 0.7668,
+      "step": 7871
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.677941829112888e-06,
+      "loss": 0.7554,
+      "step": 7872
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.67551915978587e-06,
+      "loss": 0.7558,
+      "step": 7873
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6730968026221415e-06,
+      "loss": 0.6849,
+      "step": 7874
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.6706747577965544e-06,
+      "loss": 0.7336,
+      "step": 7875
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.668253025483941e-06,
+      "loss": 0.7646,
+      "step": 7876
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.665831605859122e-06,
+      "loss": 0.7324,
+      "step": 7877
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.663410499096884e-06,
+      "loss": 0.7477,
+      "step": 7878
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.660989705371993e-06,
+      "loss": 0.8229,
+      "step": 7879
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.658569224859199e-06,
+      "loss": 0.7309,
+      "step": 7880
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.656149057733223e-06,
+      "loss": 0.77,
+      "step": 7881
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.653729204168766e-06,
+      "loss": 0.7809,
+      "step": 7882
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.651309664340498e-06,
+      "loss": 0.3694,
+      "step": 7883
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.648890438423085e-06,
+      "loss": 0.8143,
+      "step": 7884
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.646471526591149e-06,
+      "loss": 0.8181,
+      "step": 7885
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.644052929019308e-06,
+      "loss": 0.796,
+      "step": 7886
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.641634645882142e-06,
+      "loss": 0.7543,
+      "step": 7887
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.639216677354222e-06,
+      "loss": 0.8255,
+      "step": 7888
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.636799023610084e-06,
+      "loss": 0.8138,
+      "step": 7889
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.63438168482425e-06,
+      "loss": 0.7153,
+      "step": 7890
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.631964661171217e-06,
+      "loss": 0.8222,
+      "step": 7891
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.629547952825449e-06,
+      "loss": 0.7577,
+      "step": 7892
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.627131559961409e-06,
+      "loss": 0.7573,
+      "step": 7893
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.624715482753518e-06,
+      "loss": 0.7791,
+      "step": 7894
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.622299721376183e-06,
+      "loss": 0.7731,
+      "step": 7895
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.61988427600378e-06,
+      "loss": 0.7003,
+      "step": 7896
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.617469146810677e-06,
+      "loss": 0.747,
+      "step": 7897
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.615054333971206e-06,
+      "loss": 0.7449,
+      "step": 7898
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.612639837659678e-06,
+      "loss": 0.7887,
+      "step": 7899
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.61022565805039e-06,
+      "loss": 0.7848,
+      "step": 7900
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.607811795317608e-06,
+      "loss": 0.7467,
+      "step": 7901
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.605398249635572e-06,
+      "loss": 0.7867,
+      "step": 7902
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.602985021178511e-06,
+      "loss": 0.7868,
+      "step": 7903
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.600572110120621e-06,
+      "loss": 0.7688,
+      "step": 7904
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.598159516636076e-06,
+      "loss": 0.7693,
+      "step": 7905
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.595747240899036e-06,
+      "loss": 0.6772,
+      "step": 7906
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.593335283083623e-06,
+      "loss": 0.778,
+      "step": 7907
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.590923643363953e-06,
+      "loss": 0.7365,
+      "step": 7908
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.588512321914101e-06,
+      "loss": 0.8061,
+      "step": 7909
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.58610131890814e-06,
+      "loss": 0.7911,
+      "step": 7910
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.583690634520103e-06,
+      "loss": 0.7741,
+      "step": 7911
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5812802689240005e-06,
+      "loss": 0.7513,
+      "step": 7912
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.578870222293835e-06,
+      "loss": 0.7433,
+      "step": 7913
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.576460494803571e-06,
+      "loss": 0.7889,
+      "step": 7914
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.57405108662715e-06,
+      "loss": 0.7619,
+      "step": 7915
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5716419979385065e-06,
+      "loss": 0.7604,
+      "step": 7916
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.569233228911533e-06,
+      "loss": 0.7732,
+      "step": 7917
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.566824779720106e-06,
+      "loss": 0.7052,
+      "step": 7918
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.564416650538086e-06,
+      "loss": 0.7295,
+      "step": 7919
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.562008841539301e-06,
+      "loss": 0.7026,
+      "step": 7920
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.559601352897559e-06,
+      "loss": 0.7218,
+      "step": 7921
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.55719418478664e-06,
+      "loss": 0.751,
+      "step": 7922
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.554787337380316e-06,
+      "loss": 0.7885,
+      "step": 7923
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.552380810852319e-06,
+      "loss": 0.6932,
+      "step": 7924
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5499746053763625e-06,
+      "loss": 0.7586,
+      "step": 7925
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.547568721126144e-06,
+      "loss": 0.7407,
+      "step": 7926
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.545163158275328e-06,
+      "loss": 0.7724,
+      "step": 7927
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5427579169975676e-06,
+      "loss": 0.7714,
+      "step": 7928
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.540352997466476e-06,
+      "loss": 0.7912,
+      "step": 7929
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5379483998556615e-06,
+      "loss": 0.7997,
+      "step": 7930
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.535544124338692e-06,
+      "loss": 0.7382,
+      "step": 7931
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.53314017108913e-06,
+      "loss": 0.7693,
+      "step": 7932
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5307365402804994e-06,
+      "loss": 0.751,
+      "step": 7933
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.528333232086307e-06,
+      "loss": 0.6705,
+      "step": 7934
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5259302466800325e-06,
+      "loss": 0.7577,
+      "step": 7935
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.523527584235143e-06,
+      "loss": 0.768,
+      "step": 7936
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.521125244925072e-06,
+      "loss": 0.7846,
+      "step": 7937
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.518723228923229e-06,
+      "loss": 0.7476,
+      "step": 7938
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.516321536403008e-06,
+      "loss": 0.7762,
+      "step": 7939
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5139201675377785e-06,
+      "loss": 0.6739,
+      "step": 7940
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5115191225008745e-06,
+      "loss": 0.7727,
+      "step": 7941
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5091184014656255e-06,
+      "loss": 0.8238,
+      "step": 7942
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.506718004605323e-06,
+      "loss": 0.7504,
+      "step": 7943
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.504317932093235e-06,
+      "loss": 0.763,
+      "step": 7944
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.501918184102621e-06,
+      "loss": 0.8069,
+      "step": 7945
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.499518760806705e-06,
+      "loss": 0.757,
+      "step": 7946
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.497119662378683e-06,
+      "loss": 0.7482,
+      "step": 7947
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4947208889917405e-06,
+      "loss": 0.7474,
+      "step": 7948
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.492322440819029e-06,
+      "loss": 0.7404,
+      "step": 7949
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.489924318033688e-06,
+      "loss": 0.7541,
+      "step": 7950
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4875265208088166e-06,
+      "loss": 0.7611,
+      "step": 7951
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.48512904931751e-06,
+      "loss": 0.3622,
+      "step": 7952
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.482731903732825e-06,
+      "loss": 0.6884,
+      "step": 7953
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.480335084227796e-06,
+      "loss": 0.7983,
+      "step": 7954
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.477938590975447e-06,
+      "loss": 0.7803,
+      "step": 7955
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.475542424148763e-06,
+      "loss": 0.7832,
+      "step": 7956
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.473146583920709e-06,
+      "loss": 0.7816,
+      "step": 7957
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4707510704642375e-06,
+      "loss": 0.756,
+      "step": 7958
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.468355883952263e-06,
+      "loss": 0.7625,
+      "step": 7959
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4659610245576856e-06,
+      "loss": 0.7401,
+      "step": 7960
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4635664924533715e-06,
+      "loss": 0.833,
+      "step": 7961
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4611722878121795e-06,
+      "loss": 0.7232,
+      "step": 7962
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.458778410806933e-06,
+      "loss": 0.741,
+      "step": 7963
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.456384861610429e-06,
+      "loss": 0.7286,
+      "step": 7964
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.453991640395454e-06,
+      "loss": 0.7726,
+      "step": 7965
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.451598747334759e-06,
+      "loss": 0.7578,
+      "step": 7966
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.449206182601072e-06,
+      "loss": 0.7474,
+      "step": 7967
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.446813946367109e-06,
+      "loss": 0.8629,
+      "step": 7968
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.444422038805543e-06,
+      "loss": 0.7987,
+      "step": 7969
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.442030460089047e-06,
+      "loss": 0.7471,
+      "step": 7970
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.439639210390245e-06,
+      "loss": 0.7716,
+      "step": 7971
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.43724828988176e-06,
+      "loss": 0.8042,
+      "step": 7972
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.434857698736173e-06,
+      "loss": 0.7477,
+      "step": 7973
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.432467437126057e-06,
+      "loss": 0.8127,
+      "step": 7974
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4300775052239495e-06,
+      "loss": 0.784,
+      "step": 7975
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.427687903202366e-06,
+      "loss": 0.7951,
+      "step": 7976
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4252986312338e-06,
+      "loss": 0.7595,
+      "step": 7977
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.422909689490725e-06,
+      "loss": 0.7189,
+      "step": 7978
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.420521078145586e-06,
+      "loss": 0.7461,
+      "step": 7979
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.418132797370803e-06,
+      "loss": 0.7671,
+      "step": 7980
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.415744847338779e-06,
+      "loss": 0.7335,
+      "step": 7981
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.413357228221885e-06,
+      "loss": 0.8056,
+      "step": 7982
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.410969940192469e-06,
+      "loss": 0.7893,
+      "step": 7983
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.408582983422865e-06,
+      "loss": 0.7504,
+      "step": 7984
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.406196358085371e-06,
+      "loss": 0.7723,
+      "step": 7985
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.403810064352264e-06,
+      "loss": 0.7358,
+      "step": 7986
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.401424102395805e-06,
+      "loss": 0.7498,
+      "step": 7987
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.399038472388222e-06,
+      "loss": 0.7835,
+      "step": 7988
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.396653174501718e-06,
+      "loss": 0.7796,
+      "step": 7989
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.394268208908484e-06,
+      "loss": 0.7298,
+      "step": 7990
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.391883575780671e-06,
+      "loss": 0.4132,
+      "step": 7991
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.389499275290422e-06,
+      "loss": 0.705,
+      "step": 7992
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.38711530760984e-06,
+      "loss": 0.7194,
+      "step": 7993
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.384731672911021e-06,
+      "loss": 0.7787,
+      "step": 7994
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.382348371366022e-06,
+      "loss": 0.8059,
+      "step": 7995
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.379965403146881e-06,
+      "loss": 0.7653,
+      "step": 7996
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.377582768425616e-06,
+      "loss": 0.7576,
+      "step": 7997
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.37520046737422e-06,
+      "loss": 0.7652,
+      "step": 7998
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.372818500164652e-06,
+      "loss": 0.7718,
+      "step": 7999
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3704368669688625e-06,
+      "loss": 0.7331,
+      "step": 8000
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.368055567958767e-06,
+      "loss": 0.7664,
+      "step": 8001
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3656746033062605e-06,
+      "loss": 0.7445,
+      "step": 8002
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.363293973183208e-06,
+      "loss": 0.7438,
+      "step": 8003
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.360913677761464e-06,
+      "loss": 0.794,
+      "step": 8004
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.358533717212848e-06,
+      "loss": 0.7692,
+      "step": 8005
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.356154091709151e-06,
+      "loss": 0.7423,
+      "step": 8006
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.353774801422158e-06,
+      "loss": 0.723,
+      "step": 8007
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.351395846523612e-06,
+      "loss": 0.7275,
+      "step": 8008
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.349017227185233e-06,
+      "loss": 0.6731,
+      "step": 8009
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.346638943578734e-06,
+      "loss": 0.7762,
+      "step": 8010
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.344260995875781e-06,
+      "loss": 0.7476,
+      "step": 8011
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.34188338424803e-06,
+      "loss": 0.7827,
+      "step": 8012
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.339506108867114e-06,
+      "loss": 0.7875,
+      "step": 8013
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.337129169904635e-06,
+      "loss": 0.7468,
+      "step": 8014
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.33475256753217e-06,
+      "loss": 0.3972,
+      "step": 8015
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.3323763019212714e-06,
+      "loss": 0.7972,
+      "step": 8016
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.330000373243478e-06,
+      "loss": 0.8063,
+      "step": 8017
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.327624781670293e-06,
+      "loss": 0.698,
+      "step": 8018
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.325249527373193e-06,
+      "loss": 0.7691,
+      "step": 8019
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.322874610523647e-06,
+      "loss": 0.75,
+      "step": 8020
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.3205000312930835e-06,
+      "loss": 0.7946,
+      "step": 8021
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.318125789852908e-06,
+      "loss": 0.7757,
+      "step": 8022
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.315751886374512e-06,
+      "loss": 0.7633,
+      "step": 8023
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.313378321029254e-06,
+      "loss": 0.7583,
+      "step": 8024
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.3110050939884665e-06,
+      "loss": 0.8102,
+      "step": 8025
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.3086322054234675e-06,
+      "loss": 0.7188,
+      "step": 8026
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.30625965550554e-06,
+      "loss": 0.754,
+      "step": 8027
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.303887444405948e-06,
+      "loss": 0.7508,
+      "step": 8028
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.301515572295926e-06,
+      "loss": 0.7523,
+      "step": 8029
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2991440393466965e-06,
+      "loss": 0.7671,
+      "step": 8030
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.296772845729439e-06,
+      "loss": 0.8117,
+      "step": 8031
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.294401991615324e-06,
+      "loss": 0.6985,
+      "step": 8032
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.292031477175497e-06,
+      "loss": 0.7754,
+      "step": 8033
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.289661302581066e-06,
+      "loss": 0.7711,
+      "step": 8034
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2872914680031215e-06,
+      "loss": 0.7409,
+      "step": 8035
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2849219736127385e-06,
+      "loss": 0.7815,
+      "step": 8036
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.282552819580955e-06,
+      "loss": 0.7585,
+      "step": 8037
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.280184006078783e-06,
+      "loss": 0.7174,
+      "step": 8038
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2778155332772254e-06,
+      "loss": 0.8093,
+      "step": 8039
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.275447401347247e-06,
+      "loss": 0.7222,
+      "step": 8040
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2730796104597906e-06,
+      "loss": 0.7638,
+      "step": 8041
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2707121607857716e-06,
+      "loss": 0.7841,
+      "step": 8042
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.268345052496092e-06,
+      "loss": 0.7527,
+      "step": 8043
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.265978285761621e-06,
+      "loss": 0.7397,
+      "step": 8044
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.263611860753196e-06,
+      "loss": 0.757,
+      "step": 8045
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.261245777641648e-06,
+      "loss": 0.7977,
+      "step": 8046
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2588800365977695e-06,
+      "loss": 0.7719,
+      "step": 8047
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.256514637792327e-06,
+      "loss": 0.8117,
+      "step": 8048
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2541495813960756e-06,
+      "loss": 0.793,
+      "step": 8049
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.251784867579728e-06,
+      "loss": 0.7373,
+      "step": 8050
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2494204965139905e-06,
+      "loss": 0.7284,
+      "step": 8051
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.247056468369529e-06,
+      "loss": 0.7529,
+      "step": 8052
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.244692783316996e-06,
+      "loss": 0.7786,
+      "step": 8053
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.242329441527012e-06,
+      "loss": 0.8079,
+      "step": 8054
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.23996644317017e-06,
+      "loss": 0.8105,
+      "step": 8055
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.237603788417055e-06,
+      "loss": 0.794,
+      "step": 8056
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.235241477438209e-06,
+      "loss": 0.7463,
+      "step": 8057
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.23287951040415e-06,
+      "loss": 0.7455,
+      "step": 8058
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.230517887485389e-06,
+      "loss": 0.7605,
+      "step": 8059
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2281566088523925e-06,
+      "loss": 0.7942,
+      "step": 8060
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.225795674675606e-06,
+      "loss": 0.811,
+      "step": 8061
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.223435085125466e-06,
+      "loss": 0.7401,
+      "step": 8062
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.221074840372364e-06,
+      "loss": 0.7326,
+      "step": 8063
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.218714940586671e-06,
+      "loss": 0.7779,
+      "step": 8064
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.216355385938747e-06,
+      "loss": 0.751,
+      "step": 8065
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.21399617659891e-06,
+      "loss": 0.6976,
+      "step": 8066
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.211637312737461e-06,
+      "loss": 0.7893,
+      "step": 8067
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.209278794524671e-06,
+      "loss": 0.7053,
+      "step": 8068
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2069206221307985e-06,
+      "loss": 0.7486,
+      "step": 8069
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2045627957260605e-06,
+      "loss": 0.7492,
+      "step": 8070
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.202205315480664e-06,
+      "loss": 0.6891,
+      "step": 8071
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.199848181564777e-06,
+      "loss": 0.8064,
+      "step": 8072
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.197491394148556e-06,
+      "loss": 0.7473,
+      "step": 8073
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.195134953402121e-06,
+      "loss": 0.8371,
+      "step": 8074
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.192778859495576e-06,
+      "loss": 0.7389,
+      "step": 8075
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.190423112598997e-06,
+      "loss": 0.7908,
+      "step": 8076
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1880677128824255e-06,
+      "loss": 0.7413,
+      "step": 8077
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1857126605158955e-06,
+      "loss": 0.6637,
+      "step": 8078
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.183357955669404e-06,
+      "loss": 0.7389,
+      "step": 8079
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.181003598512926e-06,
+      "loss": 0.8059,
+      "step": 8080
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.178649589216404e-06,
+      "loss": 0.814,
+      "step": 8081
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.176295927949775e-06,
+      "loss": 0.7493,
+      "step": 8082
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.173942614882932e-06,
+      "loss": 0.7445,
+      "step": 8083
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.171589650185744e-06,
+      "loss": 0.7895,
+      "step": 8084
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.16923703402807e-06,
+      "loss": 0.7847,
+      "step": 8085
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.166884766579729e-06,
+      "loss": 0.7341,
+      "step": 8086
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.164532848010516e-06,
+      "loss": 0.7539,
+      "step": 8087
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.162181278490212e-06,
+      "loss": 0.7118,
+      "step": 8088
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1598300581885645e-06,
+      "loss": 0.7505,
+      "step": 8089
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.157479187275289e-06,
+      "loss": 0.7857,
+      "step": 8090
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.155128665920093e-06,
+      "loss": 0.8039,
+      "step": 8091
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.152778494292641e-06,
+      "loss": 0.7898,
+      "step": 8092
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.150428672562589e-06,
+      "loss": 0.7401,
+      "step": 8093
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.14807920089955e-06,
+      "loss": 0.8261,
+      "step": 8094
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.145730079473128e-06,
+      "loss": 0.8,
+      "step": 8095
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.143381308452895e-06,
+      "loss": 0.8161,
+      "step": 8096
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.141032888008389e-06,
+      "loss": 0.7887,
+      "step": 8097
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.138684818309142e-06,
+      "loss": 0.7541,
+      "step": 8098
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.136337099524644e-06,
+      "loss": 0.7084,
+      "step": 8099
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1339897318243625e-06,
+      "loss": 0.7668,
+      "step": 8100
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.131642715377749e-06,
+      "loss": 0.7453,
+      "step": 8101
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.12929605035422e-06,
+      "loss": 0.7987,
+      "step": 8102
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.126949736923168e-06,
+      "loss": 0.7224,
+      "step": 8103
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.124603775253967e-06,
+      "loss": 0.7683,
+      "step": 8104
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.122258165515957e-06,
+      "loss": 0.751,
+      "step": 8105
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.119912907878455e-06,
+      "loss": 0.7553,
+      "step": 8106
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.117568002510758e-06,
+      "loss": 0.7418,
+      "step": 8107
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.115223449582131e-06,
+      "loss": 0.7904,
+      "step": 8108
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.112879249261815e-06,
+      "loss": 0.7832,
+      "step": 8109
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1105354017190255e-06,
+      "loss": 0.7962,
+      "step": 8110
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.108191907122958e-06,
+      "loss": 0.8244,
+      "step": 8111
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.105848765642771e-06,
+      "loss": 0.7547,
+      "step": 8112
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.103505977447614e-06,
+      "loss": 0.7597,
+      "step": 8113
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.101163542706592e-06,
+      "loss": 0.7469,
+      "step": 8114
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.098821461588802e-06,
+      "loss": 0.7052,
+      "step": 8115
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0964797342633e-06,
+      "loss": 0.7473,
+      "step": 8116
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0941383608991326e-06,
+      "loss": 0.7437,
+      "step": 8117
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.091797341665305e-06,
+      "loss": 0.7516,
+      "step": 8118
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.089456676730804e-06,
+      "loss": 0.7871,
+      "step": 8119
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.087116366264597e-06,
+      "loss": 0.7646,
+      "step": 8120
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.084776410435614e-06,
+      "loss": 0.7416,
+      "step": 8121
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.08243680941277e-06,
+      "loss": 0.7365,
+      "step": 8122
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.08009756336494e-06,
+      "loss": 0.8208,
+      "step": 8123
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.077758672460995e-06,
+      "loss": 0.8019,
+      "step": 8124
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0754201368697605e-06,
+      "loss": 0.8167,
+      "step": 8125
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0730819567600435e-06,
+      "loss": 0.7242,
+      "step": 8126
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.0707441323006315e-06,
+      "loss": 0.7274,
+      "step": 8127
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.068406663660277e-06,
+      "loss": 0.7725,
+      "step": 8128
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.066069551007706e-06,
+      "loss": 0.7392,
+      "step": 8129
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.063732794511635e-06,
+      "loss": 0.7399,
+      "step": 8130
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.061396394340734e-06,
+      "loss": 0.3995,
+      "step": 8131
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.059060350663656e-06,
+      "loss": 0.7062,
+      "step": 8132
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.056724663649035e-06,
+      "loss": 0.744,
+      "step": 8133
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.054389333465466e-06,
+      "loss": 0.7722,
+      "step": 8134
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.052054360281531e-06,
+      "loss": 0.7708,
+      "step": 8135
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.049719744265775e-06,
+      "loss": 0.7946,
+      "step": 8136
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.047385485586729e-06,
+      "loss": 0.7239,
+      "step": 8137
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.045051584412888e-06,
+      "loss": 0.7768,
+      "step": 8138
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.042718040912722e-06,
+      "loss": 0.8144,
+      "step": 8139
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.040384855254685e-06,
+      "loss": 0.7658,
+      "step": 8140
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.038052027607195e-06,
+      "loss": 0.7889,
+      "step": 8141
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.035719558138643e-06,
+      "loss": 0.7393,
+      "step": 8142
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.033387447017408e-06,
+      "loss": 0.7876,
+      "step": 8143
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.031055694411829e-06,
+      "loss": 0.7574,
+      "step": 8144
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.028724300490217e-06,
+      "loss": 0.8098,
+      "step": 8145
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0263932654208766e-06,
+      "loss": 0.7806,
+      "step": 8146
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.024062589372067e-06,
+      "loss": 0.7882,
+      "step": 8147
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.02173227251203e-06,
+      "loss": 0.74,
+      "step": 8148
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.019402315008976e-06,
+      "loss": 0.7652,
+      "step": 8149
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.0170727170310995e-06,
+      "loss": 0.7477,
+      "step": 8150
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.01474347874656e-06,
+      "loss": 0.7645,
+      "step": 8151
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.01241460032349e-06,
+      "loss": 0.744,
+      "step": 8152
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.010086081930008e-06,
+      "loss": 0.7764,
+      "step": 8153
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.007757923734189e-06,
+      "loss": 0.7704,
+      "step": 8154
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.005430125904101e-06,
+      "loss": 0.7531,
+      "step": 8155
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.003102688607769e-06,
+      "loss": 0.7277,
+      "step": 8156
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.000775612013207e-06,
+      "loss": 0.8145,
+      "step": 8157
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.998448896288386e-06,
+      "loss": 0.7197,
+      "step": 8158
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.996122541601268e-06,
+      "loss": 0.7439,
+      "step": 8159
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.99379654811978e-06,
+      "loss": 0.7653,
+      "step": 8160
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9914709160118216e-06,
+      "loss": 0.7234,
+      "step": 8161
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.989145645445267e-06,
+      "loss": 0.7974,
+      "step": 8162
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.986820736587971e-06,
+      "loss": 0.7656,
+      "step": 8163
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.984496189607758e-06,
+      "loss": 0.7426,
+      "step": 8164
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.982172004672418e-06,
+      "loss": 0.7462,
+      "step": 8165
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.979848181949733e-06,
+      "loss": 0.7431,
+      "step": 8166
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.977524721607442e-06,
+      "loss": 0.7077,
+      "step": 8167
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9752016238132625e-06,
+      "loss": 0.7158,
+      "step": 8168
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.972878888734894e-06,
+      "loss": 0.8144,
+      "step": 8169
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.970556516540002e-06,
+      "loss": 0.7822,
+      "step": 8170
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.96823450739622e-06,
+      "loss": 0.7739,
+      "step": 8171
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.965912861471174e-06,
+      "loss": 0.7692,
+      "step": 8172
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.963591578932445e-06,
+      "loss": 0.7378,
+      "step": 8173
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.961270659947593e-06,
+      "loss": 0.7654,
+      "step": 8174
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.958950104684163e-06,
+      "loss": 0.7426,
+      "step": 8175
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.956629913309654e-06,
+      "loss": 0.76,
+      "step": 8176
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.954310085991559e-06,
+      "loss": 0.6677,
+      "step": 8177
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.951990622897327e-06,
+      "loss": 0.8011,
+      "step": 8178
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.949671524194395e-06,
+      "loss": 0.7433,
+      "step": 8179
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.947352790050166e-06,
+      "loss": 0.7233,
+      "step": 8180
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.945034420632014e-06,
+      "loss": 0.823,
+      "step": 8181
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.942716416107296e-06,
+      "loss": 0.868,
+      "step": 8182
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.940398776643336e-06,
+      "loss": 0.7553,
+      "step": 8183
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.938081502407428e-06,
+      "loss": 0.7676,
+      "step": 8184
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.935764593566854e-06,
+      "loss": 0.78,
+      "step": 8185
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.933448050288856e-06,
+      "loss": 0.7817,
+      "step": 8186
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.931131872740653e-06,
+      "loss": 0.7188,
+      "step": 8187
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.928816061089436e-06,
+      "loss": 0.7855,
+      "step": 8188
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.92650061550238e-06,
+      "loss": 0.8338,
+      "step": 8189
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.924185536146621e-06,
+      "loss": 0.7573,
+      "step": 8190
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9218708231892695e-06,
+      "loss": 0.7164,
+      "step": 8191
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.919556476797421e-06,
+      "loss": 0.7135,
+      "step": 8192
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.917242497138134e-06,
+      "loss": 0.7419,
+      "step": 8193
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.914928884378439e-06,
+      "loss": 0.7693,
+      "step": 8194
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.912615638685353e-06,
+      "loss": 0.7831,
+      "step": 8195
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.910302760225849e-06,
+      "loss": 0.7367,
+      "step": 8196
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.907990249166891e-06,
+      "loss": 0.7411,
+      "step": 8197
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9056781056754e-06,
+      "loss": 0.7986,
+      "step": 8198
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.903366329918285e-06,
+      "loss": 0.7722,
+      "step": 8199
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9010549220624205e-06,
+      "loss": 0.7237,
+      "step": 8200
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.89874388227465e-06,
+      "loss": 0.7154,
+      "step": 8201
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.896433210721805e-06,
+      "loss": 0.7759,
+      "step": 8202
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.894122907570678e-06,
+      "loss": 0.7191,
+      "step": 8203
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.891812972988033e-06,
+      "loss": 0.8148,
+      "step": 8204
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.889503407140622e-06,
+      "loss": 0.7221,
+      "step": 8205
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.887194210195158e-06,
+      "loss": 0.7001,
+      "step": 8206
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.884885382318324e-06,
+      "loss": 0.7658,
+      "step": 8207
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.882576923676795e-06,
+      "loss": 0.7053,
+      "step": 8208
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.880268834437201e-06,
+      "loss": 0.6977,
+      "step": 8209
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.877961114766148e-06,
+      "loss": 0.7431,
+      "step": 8210
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.875653764830227e-06,
+      "loss": 0.7462,
+      "step": 8211
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8733467847959915e-06,
+      "loss": 0.7236,
+      "step": 8212
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.871040174829969e-06,
+      "loss": 0.7228,
+      "step": 8213
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.868733935098662e-06,
+      "loss": 0.7812,
+      "step": 8214
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.866428065768551e-06,
+      "loss": 0.7919,
+      "step": 8215
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.86412256700608e-06,
+      "loss": 0.734,
+      "step": 8216
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.861817438977675e-06,
+      "loss": 0.725,
+      "step": 8217
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.859512681849736e-06,
+      "loss": 0.7045,
+      "step": 8218
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.857208295788628e-06,
+      "loss": 0.7873,
+      "step": 8219
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.85490428096069e-06,
+      "loss": 0.835,
+      "step": 8220
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.852600637532243e-06,
+      "loss": 0.8192,
+      "step": 8221
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.850297365669577e-06,
+      "loss": 0.7626,
+      "step": 8222
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.847994465538946e-06,
+      "loss": 0.736,
+      "step": 8223
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.845691937306594e-06,
+      "loss": 0.7426,
+      "step": 8224
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8433897811387255e-06,
+      "loss": 0.3835,
+      "step": 8225
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.841087997201518e-06,
+      "loss": 0.7234,
+      "step": 8226
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.838786585661135e-06,
+      "loss": 0.7684,
+      "step": 8227
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.836485546683699e-06,
+      "loss": 0.726,
+      "step": 8228
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.834184880435312e-06,
+      "loss": 0.845,
+      "step": 8229
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.831884587082042e-06,
+      "loss": 0.7903,
+      "step": 8230
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.829584666789946e-06,
+      "loss": 0.7726,
+      "step": 8231
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.827285119725039e-06,
+      "loss": 0.7421,
+      "step": 8232
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8249859460533114e-06,
+      "loss": 0.7598,
+      "step": 8233
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.822687145940735e-06,
+      "loss": 0.7459,
+      "step": 8234
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.820388719553247e-06,
+      "loss": 0.7557,
+      "step": 8235
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.818090667056756e-06,
+      "loss": 0.7447,
+      "step": 8236
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.815792988617151e-06,
+      "loss": 0.749,
+      "step": 8237
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8134956844002914e-06,
+      "loss": 0.7884,
+      "step": 8238
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.811198754572004e-06,
+      "loss": 0.7859,
+      "step": 8239
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.808902199298099e-06,
+      "loss": 0.8084,
+      "step": 8240
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.806606018744351e-06,
+      "loss": 0.6944,
+      "step": 8241
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.804310213076509e-06,
+      "loss": 0.6814,
+      "step": 8242
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.8020147824602925e-06,
+      "loss": 0.7248,
+      "step": 8243
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.799719727061405e-06,
+      "loss": 0.7743,
+      "step": 8244
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.797425047045513e-06,
+      "loss": 0.7901,
+      "step": 8245
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.795130742578252e-06,
+      "loss": 0.7896,
+      "step": 8246
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.792836813825247e-06,
+      "loss": 0.7192,
+      "step": 8247
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.790543260952079e-06,
+      "loss": 0.7959,
+      "step": 8248
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.788250084124307e-06,
+      "loss": 0.7257,
+      "step": 8249
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7859572835074716e-06,
+      "loss": 0.7556,
+      "step": 8250
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.783664859267074e-06,
+      "loss": 0.7795,
+      "step": 8251
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.78137281156859e-06,
+      "loss": 0.749,
+      "step": 8252
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.779081140577479e-06,
+      "loss": 0.7383,
+      "step": 8253
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.776789846459161e-06,
+      "loss": 0.7149,
+      "step": 8254
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.7744989293790324e-06,
+      "loss": 0.8271,
+      "step": 8255
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.772208389502467e-06,
+      "loss": 0.7701,
+      "step": 8256
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.769918226994803e-06,
+      "loss": 0.7287,
+      "step": 8257
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7676284420213625e-06,
+      "loss": 0.7545,
+      "step": 8258
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.765339034747427e-06,
+      "loss": 0.7697,
+      "step": 8259
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.763050005338264e-06,
+      "loss": 0.8012,
+      "step": 8260
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.760761353959106e-06,
+      "loss": 0.7357,
+      "step": 8261
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.758473080775154e-06,
+      "loss": 0.7676,
+      "step": 8262
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.756185185951595e-06,
+      "loss": 0.7831,
+      "step": 8263
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.753897669653578e-06,
+      "loss": 0.7787,
+      "step": 8264
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7516105320462235e-06,
+      "loss": 0.7714,
+      "step": 8265
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.749323773294636e-06,
+      "loss": 0.7309,
+      "step": 8266
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.747037393563883e-06,
+      "loss": 0.7564,
+      "step": 8267
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.744751393019006e-06,
+      "loss": 0.7813,
+      "step": 8268
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.742465771825016e-06,
+      "loss": 0.743,
+      "step": 8269
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7401805301469105e-06,
+      "loss": 0.7974,
+      "step": 8270
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.737895668149644e-06,
+      "loss": 0.6903,
+      "step": 8271
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.735611185998149e-06,
+      "loss": 0.7721,
+      "step": 8272
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.733327083857334e-06,
+      "loss": 0.7141,
+      "step": 8273
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.731043361892078e-06,
+      "loss": 0.7731,
+      "step": 8274
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.728760020267224e-06,
+      "loss": 0.7747,
+      "step": 8275
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.726477059147608e-06,
+      "loss": 0.3739,
+      "step": 8276
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.724194478698013e-06,
+      "loss": 0.8004,
+      "step": 8277
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.721912279083218e-06,
+      "loss": 0.7443,
+      "step": 8278
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.719630460467956e-06,
+      "loss": 0.7804,
+      "step": 8279
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.717349023016947e-06,
+      "loss": 0.7302,
+      "step": 8280
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.715067966894874e-06,
+      "loss": 0.7657,
+      "step": 8281
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.712787292266391e-06,
+      "loss": 0.844,
+      "step": 8282
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.710506999296137e-06,
+      "loss": 0.9061,
+      "step": 8283
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.708227088148711e-06,
+      "loss": 0.7935,
+      "step": 8284
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.705947558988685e-06,
+      "loss": 0.7898,
+      "step": 8285
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.703668411980615e-06,
+      "loss": 0.7436,
+      "step": 8286
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7013896472890185e-06,
+      "loss": 0.7901,
+      "step": 8287
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.699111265078383e-06,
+      "loss": 0.82,
+      "step": 8288
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.696833265513182e-06,
+      "loss": 0.813,
+      "step": 8289
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6945556487578505e-06,
+      "loss": 0.7463,
+      "step": 8290
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.692278414976794e-06,
+      "loss": 0.7269,
+      "step": 8291
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.690001564334402e-06,
+      "loss": 0.8374,
+      "step": 8292
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.687725096995026e-06,
+      "loss": 0.7347,
+      "step": 8293
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.685449013122993e-06,
+      "loss": 0.7655,
+      "step": 8294
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6831733128826e-06,
+      "loss": 0.7494,
+      "step": 8295
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.680897996438124e-06,
+      "loss": 0.8084,
+      "step": 8296
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.678623063953803e-06,
+      "loss": 0.8186,
+      "step": 8297
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.676348515593861e-06,
+      "loss": 0.7262,
+      "step": 8298
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.674074351522478e-06,
+      "loss": 0.6921,
+      "step": 8299
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.671800571903824e-06,
+      "loss": 0.6984,
+      "step": 8300
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.669527176902023e-06,
+      "loss": 0.7464,
+      "step": 8301
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6672541666811874e-06,
+      "loss": 0.8294,
+      "step": 8302
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6649815414053934e-06,
+      "loss": 0.3974,
+      "step": 8303
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.662709301238686e-06,
+      "loss": 0.8036,
+      "step": 8304
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.660437446345093e-06,
+      "loss": 0.695,
+      "step": 8305
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.658165976888607e-06,
+      "loss": 0.7281,
+      "step": 8306
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.655894893033194e-06,
+      "loss": 0.7576,
+      "step": 8307
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.653624194942788e-06,
+      "loss": 0.6996,
+      "step": 8308
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.651353882781308e-06,
+      "loss": 0.7361,
+      "step": 8309
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.649083956712631e-06,
+      "loss": 0.7763,
+      "step": 8310
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.646814416900611e-06,
+      "loss": 0.7449,
+      "step": 8311
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.644545263509082e-06,
+      "loss": 0.3633,
+      "step": 8312
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.642276496701838e-06,
+      "loss": 0.7684,
+      "step": 8313
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.640008116642647e-06,
+      "loss": 0.754,
+      "step": 8314
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.6377401234952615e-06,
+      "loss": 0.7784,
+      "step": 8315
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.63547251742339e-06,
+      "loss": 0.7786,
+      "step": 8316
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.633205298590719e-06,
+      "loss": 0.7586,
+      "step": 8317
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.630938467160915e-06,
+      "loss": 0.7478,
+      "step": 8318
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.628672023297601e-06,
+      "loss": 0.7481,
+      "step": 8319
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.626405967164389e-06,
+      "loss": 0.7718,
+      "step": 8320
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.624140298924846e-06,
+      "loss": 0.7307,
+      "step": 8321
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.621875018742529e-06,
+      "loss": 0.7749,
+      "step": 8322
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.619610126780952e-06,
+      "loss": 0.3925,
+      "step": 8323
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.617345623203605e-06,
+      "loss": 0.7407,
+      "step": 8324
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.615081508173956e-06,
+      "loss": 0.757,
+      "step": 8325
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.612817781855439e-06,
+      "loss": 0.7227,
+      "step": 8326
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.610554444411457e-06,
+      "loss": 0.7232,
+      "step": 8327
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.608291496005396e-06,
+      "loss": 0.7307,
+      "step": 8328
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.606028936800605e-06,
+      "loss": 0.756,
+      "step": 8329
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.603766766960402e-06,
+      "loss": 0.728,
+      "step": 8330
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.601504986648092e-06,
+      "loss": 0.7675,
+      "step": 8331
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.599243596026936e-06,
+      "loss": 0.8047,
+      "step": 8332
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.596982595260172e-06,
+      "loss": 0.7645,
+      "step": 8333
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.594721984511009e-06,
+      "loss": 0.7504,
+      "step": 8334
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.592461763942636e-06,
+      "loss": 0.7114,
+      "step": 8335
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.590201933718203e-06,
+      "loss": 0.7122,
+      "step": 8336
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.587942494000835e-06,
+      "loss": 0.7479,
+      "step": 8337
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.585683444953636e-06,
+      "loss": 0.7953,
+      "step": 8338
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.583424786739666e-06,
+      "loss": 0.7575,
+      "step": 8339
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.581166519521978e-06,
+      "loss": 0.8204,
+      "step": 8340
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.578908643463576e-06,
+      "loss": 0.7881,
+      "step": 8341
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.576651158727451e-06,
+      "loss": 0.7539,
+      "step": 8342
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.574394065476555e-06,
+      "loss": 0.7799,
+      "step": 8343
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.572137363873822e-06,
+      "loss": 0.3549,
+      "step": 8344
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.569881054082151e-06,
+      "loss": 0.7227,
+      "step": 8345
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.567625136264409e-06,
+      "loss": 0.7141,
+      "step": 8346
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5653696105834465e-06,
+      "loss": 0.7897,
+      "step": 8347
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.563114477202076e-06,
+      "loss": 0.8273,
+      "step": 8348
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.560859736283084e-06,
+      "loss": 0.7163,
+      "step": 8349
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.558605387989227e-06,
+      "loss": 0.8004,
+      "step": 8350
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.556351432483243e-06,
+      "loss": 0.7718,
+      "step": 8351
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.554097869927827e-06,
+      "loss": 0.7471,
+      "step": 8352
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.551844700485652e-06,
+      "loss": 0.7243,
+      "step": 8353
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.549591924319372e-06,
+      "loss": 0.7227,
+      "step": 8354
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.547339541591598e-06,
+      "loss": 0.7379,
+      "step": 8355
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.545087552464914e-06,
+      "loss": 0.7618,
+      "step": 8356
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.54283595710189e-06,
+      "loss": 0.7684,
+      "step": 8357
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.540584755665053e-06,
+      "loss": 0.7566,
+      "step": 8358
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.538333948316902e-06,
+      "loss": 0.7531,
+      "step": 8359
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.536083535219921e-06,
+      "loss": 0.7612,
+      "step": 8360
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.533833516536547e-06,
+      "loss": 0.7717,
+      "step": 8361
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.531583892429206e-06,
+      "loss": 0.7673,
+      "step": 8362
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.529334663060281e-06,
+      "loss": 0.7827,
+      "step": 8363
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.52708582859214e-06,
+      "loss": 0.7636,
+      "step": 8364
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.524837389187109e-06,
+      "loss": 0.769,
+      "step": 8365
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.522589345007492e-06,
+      "loss": 0.7179,
+      "step": 8366
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.520341696215571e-06,
+      "loss": 0.6865,
+      "step": 8367
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.518094442973587e-06,
+      "loss": 0.7434,
+      "step": 8368
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.515847585443758e-06,
+      "loss": 0.7921,
+      "step": 8369
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.513601123788278e-06,
+      "loss": 0.8084,
+      "step": 8370
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.511355058169307e-06,
+      "loss": 0.7626,
+      "step": 8371
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5091093887489714e-06,
+      "loss": 0.7534,
+      "step": 8372
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5068641156893835e-06,
+      "loss": 0.7224,
+      "step": 8373
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.504619239152616e-06,
+      "loss": 0.7818,
+      "step": 8374
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.502374759300715e-06,
+      "loss": 0.7531,
+      "step": 8375
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.500130676295694e-06,
+      "loss": 0.8209,
+      "step": 8376
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.497886990299552e-06,
+      "loss": 0.7781,
+      "step": 8377
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.4956437014742425e-06,
+      "loss": 0.7498,
+      "step": 8378
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.493400809981698e-06,
+      "loss": 0.7737,
+      "step": 8379
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.491158315983828e-06,
+      "loss": 0.6939,
+      "step": 8380
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.488916219642498e-06,
+      "loss": 0.7284,
+      "step": 8381
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.486674521119564e-06,
+      "loss": 0.7278,
+      "step": 8382
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.484433220576834e-06,
+      "loss": 0.7758,
+      "step": 8383
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.482192318176105e-06,
+      "loss": 0.7522,
+      "step": 8384
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.479951814079128e-06,
+      "loss": 0.7883,
+      "step": 8385
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.477711708447645e-06,
+      "loss": 0.7585,
+      "step": 8386
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.475472001443352e-06,
+      "loss": 0.7948,
+      "step": 8387
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.473232693227923e-06,
+      "loss": 0.7255,
+      "step": 8388
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.470993783962998e-06,
+      "loss": 0.7916,
+      "step": 8389
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.468755273810202e-06,
+      "loss": 0.7881,
+      "step": 8390
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.466517162931118e-06,
+      "loss": 0.7975,
+      "step": 8391
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.464279451487301e-06,
+      "loss": 0.7756,
+      "step": 8392
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.462042139640287e-06,
+      "loss": 0.7557,
+      "step": 8393
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.459805227551575e-06,
+      "loss": 0.7709,
+      "step": 8394
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.45756871538263e-06,
+      "loss": 0.7322,
+      "step": 8395
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.455332603294904e-06,
+      "loss": 0.7338,
+      "step": 8396
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.453096891449809e-06,
+      "loss": 0.7233,
+      "step": 8397
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4508615800087244e-06,
+      "loss": 0.7884,
+      "step": 8398
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.448626669133014e-06,
+      "loss": 0.7461,
+      "step": 8399
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.446392158984001e-06,
+      "loss": 0.7124,
+      "step": 8400
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.444158049722982e-06,
+      "loss": 0.7623,
+      "step": 8401
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.441924341511229e-06,
+      "loss": 0.7551,
+      "step": 8402
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.439691034509986e-06,
+      "loss": 0.7728,
+      "step": 8403
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.437458128880463e-06,
+      "loss": 0.776,
+      "step": 8404
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.435225624783837e-06,
+      "loss": 0.8007,
+      "step": 8405
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.432993522381268e-06,
+      "loss": 0.7249,
+      "step": 8406
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.430761821833879e-06,
+      "loss": 0.7714,
+      "step": 8407
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.428530523302761e-06,
+      "loss": 0.8173,
+      "step": 8408
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.426299626948987e-06,
+      "loss": 0.7259,
+      "step": 8409
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.424069132933594e-06,
+      "loss": 0.7551,
+      "step": 8410
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.421839041417585e-06,
+      "loss": 0.7162,
+      "step": 8411
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.419609352561945e-06,
+      "loss": 0.7271,
+      "step": 8412
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.417380066527624e-06,
+      "loss": 0.8032,
+      "step": 8413
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.415151183475542e-06,
+      "loss": 0.7829,
+      "step": 8414
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.412922703566587e-06,
+      "loss": 0.7622,
+      "step": 8415
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4106946269616295e-06,
+      "loss": 0.7209,
+      "step": 8416
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.408466953821501e-06,
+      "loss": 0.7958,
+      "step": 8417
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.406239684307001e-06,
+      "loss": 0.7652,
+      "step": 8418
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.404012818578914e-06,
+      "loss": 0.7537,
+      "step": 8419
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.401786356797984e-06,
+      "loss": 0.7621,
+      "step": 8420
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.399560299124922e-06,
+      "loss": 0.6982,
+      "step": 8421
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3973346457204214e-06,
+      "loss": 0.7268,
+      "step": 8422
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.395109396745145e-06,
+      "loss": 0.7392,
+      "step": 8423
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.392884552359716e-06,
+      "loss": 0.7141,
+      "step": 8424
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.390660112724742e-06,
+      "loss": 0.7697,
+      "step": 8425
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.38843607800079e-06,
+      "loss": 0.7293,
+      "step": 8426
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.386212448348404e-06,
+      "loss": 0.7483,
+      "step": 8427
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.38398922392809e-06,
+      "loss": 0.7386,
+      "step": 8428
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.381766404900344e-06,
+      "loss": 0.7798,
+      "step": 8429
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.379543991425612e-06,
+      "loss": 0.752,
+      "step": 8430
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.377321983664319e-06,
+      "loss": 0.6993,
+      "step": 8431
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3751003817768675e-06,
+      "loss": 0.7274,
+      "step": 8432
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.372879185923619e-06,
+      "loss": 0.8359,
+      "step": 8433
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.370658396264909e-06,
+      "loss": 0.7324,
+      "step": 8434
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.368438012961053e-06,
+      "loss": 0.7605,
+      "step": 8435
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.366218036172324e-06,
+      "loss": 0.6698,
+      "step": 8436
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.363998466058968e-06,
+      "loss": 0.727,
+      "step": 8437
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.361779302781215e-06,
+      "loss": 0.7644,
+      "step": 8438
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.359560546499249e-06,
+      "loss": 0.8015,
+      "step": 8439
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.35734219737323e-06,
+      "loss": 0.7039,
+      "step": 8440
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.355124255563294e-06,
+      "loss": 0.7678,
+      "step": 8441
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3529067212295396e-06,
+      "loss": 0.7306,
+      "step": 8442
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.350689594532045e-06,
+      "loss": 0.8089,
+      "step": 8443
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.348472875630848e-06,
+      "loss": 0.7398,
+      "step": 8444
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.346256564685968e-06,
+      "loss": 0.7571,
+      "step": 8445
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.344040661857388e-06,
+      "loss": 0.7005,
+      "step": 8446
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.341825167305058e-06,
+      "loss": 0.7974,
+      "step": 8447
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.339610081188913e-06,
+      "loss": 0.7589,
+      "step": 8448
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.337395403668845e-06,
+      "loss": 0.7964,
+      "step": 8449
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.335181134904716e-06,
+      "loss": 0.7519,
+      "step": 8450
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3329672750563716e-06,
+      "loss": 0.7299,
+      "step": 8451
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.330753824283616e-06,
+      "loss": 0.7063,
+      "step": 8452
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.328540782746227e-06,
+      "loss": 0.6686,
+      "step": 8453
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.32632815060395e-06,
+      "loss": 0.7687,
+      "step": 8454
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.324115928016511e-06,
+      "loss": 0.7271,
+      "step": 8455
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.321904115143597e-06,
+      "loss": 0.7576,
+      "step": 8456
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.319692712144863e-06,
+      "loss": 0.7735,
+      "step": 8457
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.317481719179949e-06,
+      "loss": 0.8159,
+      "step": 8458
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.31527113640845e-06,
+      "loss": 0.7239,
+      "step": 8459
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.3130609639899335e-06,
+      "loss": 0.8089,
+      "step": 8460
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.31085120208395e-06,
+      "loss": 0.7717,
+      "step": 8461
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.308641850850002e-06,
+      "loss": 0.7603,
+      "step": 8462
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.306432910447583e-06,
+      "loss": 0.3457,
+      "step": 8463
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.304224381036135e-06,
+      "loss": 0.8181,
+      "step": 8464
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.30201626277509e-06,
+      "loss": 0.7454,
+      "step": 8465
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.299808555823836e-06,
+      "loss": 0.8431,
+      "step": 8466
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.297601260341733e-06,
+      "loss": 0.7667,
+      "step": 8467
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.295394376488124e-06,
+      "loss": 0.7561,
+      "step": 8468
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2931879044223094e-06,
+      "loss": 0.8,
+      "step": 8469
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.290981844303559e-06,
+      "loss": 0.7471,
+      "step": 8470
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.288776196291126e-06,
+      "loss": 0.7292,
+      "step": 8471
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.28657096054422e-06,
+      "loss": 0.7861,
+      "step": 8472
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.284366137222025e-06,
+      "loss": 0.7628,
+      "step": 8473
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.282161726483701e-06,
+      "loss": 0.797,
+      "step": 8474
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.279957728488372e-06,
+      "loss": 0.7805,
+      "step": 8475
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.277754143395131e-06,
+      "loss": 0.7459,
+      "step": 8476
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.27555097136305e-06,
+      "loss": 0.709,
+      "step": 8477
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.27334821255116e-06,
+      "loss": 0.669,
+      "step": 8478
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.271145867118467e-06,
+      "loss": 0.7691,
+      "step": 8479
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2689439352239535e-06,
+      "loss": 0.739,
+      "step": 8480
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.266742417026564e-06,
+      "loss": 0.7931,
+      "step": 8481
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2645413126852085e-06,
+      "loss": 0.7687,
+      "step": 8482
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2623406223587845e-06,
+      "loss": 0.7531,
+      "step": 8483
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.26014034620614e-06,
+      "loss": 0.7349,
+      "step": 8484
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.257940484386112e-06,
+      "loss": 0.7287,
+      "step": 8485
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.255741037057488e-06,
+      "loss": 0.7435,
+      "step": 8486
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.253542004379043e-06,
+      "loss": 0.7914,
+      "step": 8487
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.251343386509512e-06,
+      "loss": 0.8058,
+      "step": 8488
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.249145183607598e-06,
+      "loss": 0.7665,
+      "step": 8489
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.246947395831986e-06,
+      "loss": 0.7383,
+      "step": 8490
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2447500233413205e-06,
+      "loss": 0.7843,
+      "step": 8491
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.242553066294216e-06,
+      "loss": 0.7726,
+      "step": 8492
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.240356524849265e-06,
+      "loss": 0.7471,
+      "step": 8493
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.238160399165024e-06,
+      "loss": 0.7346,
+      "step": 8494
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.2359646894000205e-06,
+      "loss": 0.7773,
+      "step": 8495
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.233769395712747e-06,
+      "loss": 0.7381,
+      "step": 8496
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.231574518261679e-06,
+      "loss": 0.7985,
+      "step": 8497
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.229380057205251e-06,
+      "loss": 0.7038,
+      "step": 8498
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.227186012701867e-06,
+      "loss": 0.774,
+      "step": 8499
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2249923849099104e-06,
+      "loss": 0.7371,
+      "step": 8500
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2227991739877276e-06,
+      "loss": 0.7743,
+      "step": 8501
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.220606380093628e-06,
+      "loss": 0.8017,
+      "step": 8502
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.218414003385909e-06,
+      "loss": 0.7274,
+      "step": 8503
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.216222044022819e-06,
+      "loss": 0.8075,
+      "step": 8504
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.214030502162594e-06,
+      "loss": 0.7351,
+      "step": 8505
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.211839377963423e-06,
+      "loss": 0.7501,
+      "step": 8506
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.209648671583477e-06,
+      "loss": 0.7619,
+      "step": 8507
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.207458383180893e-06,
+      "loss": 0.8097,
+      "step": 8508
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.20526851291377e-06,
+      "loss": 0.7755,
+      "step": 8509
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.203079060940195e-06,
+      "loss": 0.7903,
+      "step": 8510
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.2008900274182075e-06,
+      "loss": 0.7752,
+      "step": 8511
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.19870141250582e-06,
+      "loss": 0.7605,
+      "step": 8512
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.196513216361027e-06,
+      "loss": 0.7501,
+      "step": 8513
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1943254391417775e-06,
+      "loss": 0.771,
+      "step": 8514
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.192138081005995e-06,
+      "loss": 0.7263,
+      "step": 8515
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.189951142111581e-06,
+      "loss": 0.7108,
+      "step": 8516
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.187764622616396e-06,
+      "loss": 0.7383,
+      "step": 8517
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.18557852267827e-06,
+      "loss": 0.6983,
+      "step": 8518
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.183392842455015e-06,
+      "loss": 0.3724,
+      "step": 8519
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1812075821044015e-06,
+      "loss": 0.7439,
+      "step": 8520
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1790227417841725e-06,
+      "loss": 0.6949,
+      "step": 8521
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.176838321652037e-06,
+      "loss": 0.7283,
+      "step": 8522
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1746543218656844e-06,
+      "loss": 0.7878,
+      "step": 8523
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1724707425827615e-06,
+      "loss": 0.7197,
+      "step": 8524
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1702875839608954e-06,
+      "loss": 0.7021,
+      "step": 8525
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1681048461576734e-06,
+      "loss": 0.7744,
+      "step": 8526
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.165922529330661e-06,
+      "loss": 0.8044,
+      "step": 8527
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.163740633637383e-06,
+      "loss": 0.4098,
+      "step": 8528
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.161559159235348e-06,
+      "loss": 0.7167,
+      "step": 8529
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.159378106282021e-06,
+      "loss": 0.7671,
+      "step": 8530
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.157197474934839e-06,
+      "loss": 0.8019,
+      "step": 8531
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.155017265351217e-06,
+      "loss": 0.7861,
+      "step": 8532
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.152837477688532e-06,
+      "loss": 0.3886,
+      "step": 8533
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.150658112104131e-06,
+      "loss": 0.7616,
+      "step": 8534
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.14847916875533e-06,
+      "loss": 0.6917,
+      "step": 8535
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.14630064779942e-06,
+      "loss": 0.7754,
+      "step": 8536
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.144122549393658e-06,
+      "loss": 0.6845,
+      "step": 8537
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.141944873695265e-06,
+      "loss": 0.7389,
+      "step": 8538
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.139767620861443e-06,
+      "loss": 0.7037,
+      "step": 8539
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.137590791049356e-06,
+      "loss": 0.7486,
+      "step": 8540
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.135414384416135e-06,
+      "loss": 0.7168,
+      "step": 8541
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.13323840111889e-06,
+      "loss": 0.7441,
+      "step": 8542
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.131062841314692e-06,
+      "loss": 0.7384,
+      "step": 8543
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.128887705160579e-06,
+      "loss": 0.6934,
+      "step": 8544
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.126712992813573e-06,
+      "loss": 0.7197,
+      "step": 8545
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1245387044306475e-06,
+      "loss": 0.7634,
+      "step": 8546
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1223648401687595e-06,
+      "loss": 0.7559,
+      "step": 8547
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.120191400184826e-06,
+      "loss": 0.7849,
+      "step": 8548
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.118018384635742e-06,
+      "loss": 0.7269,
+      "step": 8549
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.115845793678363e-06,
+      "loss": 0.7975,
+      "step": 8550
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1136736274695165e-06,
+      "loss": 0.7431,
+      "step": 8551
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.111501886166005e-06,
+      "loss": 0.7519,
+      "step": 8552
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.109330569924594e-06,
+      "loss": 0.7522,
+      "step": 8553
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.107159678902017e-06,
+      "loss": 0.7808,
+      "step": 8554
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.104989213254986e-06,
+      "loss": 0.7129,
+      "step": 8555
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1028191731401745e-06,
+      "loss": 0.7696,
+      "step": 8556
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.100649558714221e-06,
+      "loss": 0.6062,
+      "step": 8557
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0984803701337494e-06,
+      "loss": 0.809,
+      "step": 8558
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.096311607555338e-06,
+      "loss": 0.7452,
+      "step": 8559
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.094143271135537e-06,
+      "loss": 0.7368,
+      "step": 8560
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.09197536103087e-06,
+      "loss": 0.7901,
+      "step": 8561
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.08980787739783e-06,
+      "loss": 0.7257,
+      "step": 8562
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.087640820392875e-06,
+      "loss": 0.6568,
+      "step": 8563
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.085474190172433e-06,
+      "loss": 0.7631,
+      "step": 8564
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.083307986892906e-06,
+      "loss": 0.7893,
+      "step": 8565
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.081142210710657e-06,
+      "loss": 0.7666,
+      "step": 8566
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.078976861782029e-06,
+      "loss": 0.7524,
+      "step": 8567
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.076811940263322e-06,
+      "loss": 0.7638,
+      "step": 8568
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.074647446310817e-06,
+      "loss": 0.7442,
+      "step": 8569
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0724833800807526e-06,
+      "loss": 0.7114,
+      "step": 8570
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.070319741729348e-06,
+      "loss": 0.653,
+      "step": 8571
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.068156531412782e-06,
+      "loss": 0.7367,
+      "step": 8572
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.065993749287208e-06,
+      "loss": 0.7483,
+      "step": 8573
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.063831395508743e-06,
+      "loss": 0.7544,
+      "step": 8574
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.061669470233481e-06,
+      "loss": 0.7604,
+      "step": 8575
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.05950797361748e-06,
+      "loss": 0.7614,
+      "step": 8576
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0573469058167645e-06,
+      "loss": 0.734,
+      "step": 8577
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0551862669873385e-06,
+      "loss": 0.7561,
+      "step": 8578
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.053026057285163e-06,
+      "loss": 0.7801,
+      "step": 8579
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.05086627686617e-06,
+      "loss": 0.7526,
+      "step": 8580
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.048706925886272e-06,
+      "loss": 0.7345,
+      "step": 8581
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.046548004501337e-06,
+      "loss": 0.7597,
+      "step": 8582
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.044389512867205e-06,
+      "loss": 0.8034,
+      "step": 8583
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.042231451139692e-06,
+      "loss": 0.7388,
+      "step": 8584
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.040073819474576e-06,
+      "loss": 0.7619,
+      "step": 8585
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.037916618027603e-06,
+      "loss": 0.6892,
+      "step": 8586
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.035759846954497e-06,
+      "loss": 0.772,
+      "step": 8587
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.033603506410937e-06,
+      "loss": 0.706,
+      "step": 8588
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.031447596552587e-06,
+      "loss": 0.7442,
+      "step": 8589
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.029292117535064e-06,
+      "loss": 0.3971,
+      "step": 8590
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.027137069513969e-06,
+      "loss": 0.6976,
+      "step": 8591
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.024982452644861e-06,
+      "loss": 0.7031,
+      "step": 8592
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.022828267083268e-06,
+      "loss": 0.6939,
+      "step": 8593
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.020674512984697e-06,
+      "loss": 0.7794,
+      "step": 8594
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.018521190504613e-06,
+      "loss": 0.72,
+      "step": 8595
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.016368299798451e-06,
+      "loss": 0.6838,
+      "step": 8596
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0142158410216245e-06,
+      "loss": 0.734,
+      "step": 8597
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0120638143295055e-06,
+      "loss": 0.7632,
+      "step": 8598
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.009912219877435e-06,
+      "loss": 0.803,
+      "step": 8599
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.007761057820734e-06,
+      "loss": 0.7244,
+      "step": 8600
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.00561032831468e-06,
+      "loss": 0.7891,
+      "step": 8601
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.003460031514522e-06,
+      "loss": 0.7914,
+      "step": 8602
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.0013101675754795e-06,
+      "loss": 0.7798,
+      "step": 8603
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.999160736652745e-06,
+      "loss": 0.81,
+      "step": 8604
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.997011738901474e-06,
+      "loss": 0.7619,
+      "step": 8605
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.994863174476787e-06,
+      "loss": 0.8148,
+      "step": 8606
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.992715043533784e-06,
+      "loss": 0.7969,
+      "step": 8607
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9905673462275294e-06,
+      "loss": 0.7257,
+      "step": 8608
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.988420082713049e-06,
+      "loss": 0.7673,
+      "step": 8609
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9862732531453506e-06,
+      "loss": 0.7572,
+      "step": 8610
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.984126857679399e-06,
+      "loss": 0.752,
+      "step": 8611
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.98198089647013e-06,
+      "loss": 0.7955,
+      "step": 8612
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9798353696724555e-06,
+      "loss": 0.7288,
+      "step": 8613
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.9776902774412485e-06,
+      "loss": 0.7264,
+      "step": 8614
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.975545619931352e-06,
+      "loss": 0.7594,
+      "step": 8615
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.973401397297576e-06,
+      "loss": 0.7489,
+      "step": 8616
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.971257609694706e-06,
+      "loss": 0.7168,
+      "step": 8617
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.969114257277492e-06,
+      "loss": 0.7586,
+      "step": 8618
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 3.966971340200646e-06,
+      "loss": 0.7351,
+      "step": 8619
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.964828858618861e-06,
+      "loss": 0.7818,
+      "step": 8620
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9626868126867915e-06,
+      "loss": 0.7236,
+      "step": 8621
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.960545202559056e-06,
+      "loss": 0.8163,
+      "step": 8622
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.958404028390255e-06,
+      "loss": 0.7815,
+      "step": 8623
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.956263290334945e-06,
+      "loss": 0.7972,
+      "step": 8624
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9541229885476525e-06,
+      "loss": 0.7595,
+      "step": 8625
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.951983123182883e-06,
+      "loss": 0.7105,
+      "step": 8626
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.949843694395096e-06,
+      "loss": 0.7678,
+      "step": 8627
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.947704702338732e-06,
+      "loss": 0.7296,
+      "step": 8628
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.945566147168189e-06,
+      "loss": 0.7504,
+      "step": 8629
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.943428029037845e-06,
+      "loss": 0.7552,
+      "step": 8630
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.941290348102037e-06,
+      "loss": 0.8203,
+      "step": 8631
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.939153104515071e-06,
+      "loss": 0.751,
+      "step": 8632
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.93701629843123e-06,
+      "loss": 0.7683,
+      "step": 8633
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.934879930004758e-06,
+      "loss": 0.7176,
+      "step": 8634
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.932743999389864e-06,
+      "loss": 0.3478,
+      "step": 8635
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.930608506740737e-06,
+      "loss": 0.7398,
+      "step": 8636
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9284734522115266e-06,
+      "loss": 0.7594,
+      "step": 8637
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.926338835956347e-06,
+      "loss": 0.6586,
+      "step": 8638
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.924204658129294e-06,
+      "loss": 0.7623,
+      "step": 8639
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9220709188844174e-06,
+      "loss": 0.706,
+      "step": 8640
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.919937618375743e-06,
+      "loss": 0.7209,
+      "step": 8641
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.917804756757262e-06,
+      "loss": 0.7647,
+      "step": 8642
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.915672334182938e-06,
+      "loss": 0.7549,
+      "step": 8643
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9135403508067006e-06,
+      "loss": 0.7121,
+      "step": 8644
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.911408806782442e-06,
+      "loss": 0.7154,
+      "step": 8645
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.909277702264036e-06,
+      "loss": 0.7644,
+      "step": 8646
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.907147037405309e-06,
+      "loss": 0.7492,
+      "step": 8647
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.90501681236007e-06,
+      "loss": 0.733,
+      "step": 8648
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9028870272820814e-06,
+      "loss": 0.7367,
+      "step": 8649
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.900757682325093e-06,
+      "loss": 0.8143,
+      "step": 8650
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.898628777642802e-06,
+      "loss": 0.7585,
+      "step": 8651
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.896500313388891e-06,
+      "loss": 0.7043,
+      "step": 8652
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.894372289717e-06,
+      "loss": 0.7401,
+      "step": 8653
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8922447067807416e-06,
+      "loss": 0.7257,
+      "step": 8654
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.890117564733691e-06,
+      "loss": 0.8097,
+      "step": 8655
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.887990863729404e-06,
+      "loss": 0.798,
+      "step": 8656
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.885864603921393e-06,
+      "loss": 0.7709,
+      "step": 8657
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.883738785463139e-06,
+      "loss": 0.7581,
+      "step": 8658
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.881613408508103e-06,
+      "loss": 0.7142,
+      "step": 8659
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.879488473209698e-06,
+      "loss": 0.7771,
+      "step": 8660
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.877363979721314e-06,
+      "loss": 0.7938,
+      "step": 8661
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8752399281963115e-06,
+      "loss": 0.7027,
+      "step": 8662
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.873116318788015e-06,
+      "loss": 0.704,
+      "step": 8663
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.87099315164971e-06,
+      "loss": 0.7009,
+      "step": 8664
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.868870426934669e-06,
+      "loss": 0.6762,
+      "step": 8665
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.866748144796115e-06,
+      "loss": 0.747,
+      "step": 8666
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8646263053872425e-06,
+      "loss": 0.7867,
+      "step": 8667
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.862504908861224e-06,
+      "loss": 0.7978,
+      "step": 8668
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.860383955371184e-06,
+      "loss": 0.7562,
+      "step": 8669
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.858263445070234e-06,
+      "loss": 0.7684,
+      "step": 8670
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.856143378111433e-06,
+      "loss": 0.8277,
+      "step": 8671
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.854023754647829e-06,
+      "loss": 0.7515,
+      "step": 8672
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.85190457483242e-06,
+      "loss": 0.7195,
+      "step": 8673
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.84978583881818e-06,
+      "loss": 0.8118,
+      "step": 8674
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.847667546758052e-06,
+      "loss": 0.7557,
+      "step": 8675
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.845549698804947e-06,
+      "loss": 0.7444,
+      "step": 8676
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.843432295111735e-06,
+      "loss": 0.7575,
+      "step": 8677
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.841315335831271e-06,
+      "loss": 0.7081,
+      "step": 8678
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.839198821116362e-06,
+      "loss": 0.7556,
+      "step": 8679
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.83708275111979e-06,
+      "loss": 0.7387,
+      "step": 8680
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8349671259942996e-06,
+      "loss": 0.7041,
+      "step": 8681
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.832851945892616e-06,
+      "loss": 0.7799,
+      "step": 8682
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.830737210967419e-06,
+      "loss": 0.7188,
+      "step": 8683
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8286229213713576e-06,
+      "loss": 0.7812,
+      "step": 8684
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8265090772570604e-06,
+      "loss": 0.7249,
+      "step": 8685
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.824395678777111e-06,
+      "loss": 0.391,
+      "step": 8686
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.822282726084061e-06,
+      "loss": 0.765,
+      "step": 8687
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.820170219330444e-06,
+      "loss": 0.7417,
+      "step": 8688
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.81805815866874e-06,
+      "loss": 0.7709,
+      "step": 8689
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.81594654425142e-06,
+      "loss": 0.7779,
+      "step": 8690
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8138353762309023e-06,
+      "loss": 0.7211,
+      "step": 8691
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.811724654759589e-06,
+      "loss": 0.8122,
+      "step": 8692
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.809614379989839e-06,
+      "loss": 0.7589,
+      "step": 8693
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8075045520739794e-06,
+      "loss": 0.7093,
+      "step": 8694
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8053951711643167e-06,
+      "loss": 0.7677,
+      "step": 8695
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.803286237413112e-06,
+      "loss": 0.796,
+      "step": 8696
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.8011777509725942e-06,
+      "loss": 0.7412,
+      "step": 8697
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.799069711994975e-06,
+      "loss": 0.7855,
+      "step": 8698
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.796962120632418e-06,
+      "loss": 0.7921,
+      "step": 8699
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7948549770370568e-06,
+      "loss": 0.797,
+      "step": 8700
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7927482813610027e-06,
+      "loss": 0.7308,
+      "step": 8701
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7906420337563243e-06,
+      "loss": 0.7402,
+      "step": 8702
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7885362343750587e-06,
+      "loss": 0.7379,
+      "step": 8703
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.786430883369219e-06,
+      "loss": 0.7271,
+      "step": 8704
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7843259808907785e-06,
+      "loss": 0.7637,
+      "step": 8705
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.782221527091677e-06,
+      "loss": 0.76,
+      "step": 8706
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.780117522123825e-06,
+      "loss": 0.7679,
+      "step": 8707
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.778013966139106e-06,
+      "loss": 0.6868,
+      "step": 8708
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.775910859289357e-06,
+      "loss": 0.7049,
+      "step": 8709
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7738082017263987e-06,
+      "loss": 0.7828,
+      "step": 8710
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.771705993602006e-06,
+      "loss": 0.7904,
+      "step": 8711
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7696042350679343e-06,
+      "loss": 0.7694,
+      "step": 8712
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.76750292627589e-06,
+      "loss": 0.7852,
+      "step": 8713
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.765402067377565e-06,
+      "loss": 0.7961,
+      "step": 8714
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7633016585246064e-06,
+      "loss": 0.7081,
+      "step": 8715
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7612016998686285e-06,
+      "loss": 0.716,
+      "step": 8716
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.759102191561226e-06,
+      "loss": 0.7313,
+      "step": 8717
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7570031337539457e-06,
+      "loss": 0.6956,
+      "step": 8718
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.754904526598312e-06,
+      "loss": 0.7386,
+      "step": 8719
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7528063702458063e-06,
+      "loss": 0.7715,
+      "step": 8720
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.750708664847893e-06,
+      "loss": 0.714,
+      "step": 8721
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7486114105559912e-06,
+      "loss": 0.7464,
+      "step": 8722
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7465146075214887e-06,
+      "loss": 0.7737,
+      "step": 8723
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.74441825589575e-06,
+      "loss": 0.7218,
+      "step": 8724
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7423223558300983e-06,
+      "loss": 0.7866,
+      "step": 8725
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.74022690747582e-06,
+      "loss": 0.7412,
+      "step": 8726
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.738131910984185e-06,
+      "loss": 0.8359,
+      "step": 8727
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7360373665064166e-06,
+      "loss": 0.7386,
+      "step": 8728
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7339432741937067e-06,
+      "loss": 0.3811,
+      "step": 8729
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7318496341972234e-06,
+      "loss": 0.7161,
+      "step": 8730
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7297564466680913e-06,
+      "loss": 0.7202,
+      "step": 8731
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7276637117574123e-06,
+      "loss": 0.7438,
+      "step": 8732
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.725571429616246e-06,
+      "loss": 0.7768,
+      "step": 8733
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7234796003956297e-06,
+      "loss": 0.7472,
+      "step": 8734
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.721388224246559e-06,
+      "loss": 0.7154,
+      "step": 8735
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.719297301319997e-06,
+      "loss": 0.7586,
+      "step": 8736
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7172068317668842e-06,
+      "loss": 0.7381,
+      "step": 8737
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.715116815738118e-06,
+      "loss": 0.729,
+      "step": 8738
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.713027253384562e-06,
+      "loss": 0.7886,
+      "step": 8739
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.710938144857061e-06,
+      "loss": 0.7028,
+      "step": 8740
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.708849490306412e-06,
+      "loss": 0.7221,
+      "step": 8741
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7067612898833816e-06,
+      "loss": 0.7514,
+      "step": 8742
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.704673543738715e-06,
+      "loss": 0.7684,
+      "step": 8743
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7025862520231113e-06,
+      "loss": 0.7631,
+      "step": 8744
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7004994148872397e-06,
+      "loss": 0.7451,
+      "step": 8745
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6984130324817456e-06,
+      "loss": 0.7773,
+      "step": 8746
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6963271049572312e-06,
+      "loss": 0.7451,
+      "step": 8747
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6942416324642693e-06,
+      "loss": 0.7746,
+      "step": 8748
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6921566151533962e-06,
+      "loss": 0.7271,
+      "step": 8749
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6900720531751277e-06,
+      "loss": 0.7513,
+      "step": 8750
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6879879466799296e-06,
+      "loss": 0.7194,
+      "step": 8751
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6859042958182514e-06,
+      "loss": 0.7846,
+      "step": 8752
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6838211007404932e-06,
+      "loss": 0.7622,
+      "step": 8753
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6817383615970405e-06,
+      "loss": 0.7302,
+      "step": 8754
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6796560785382274e-06,
+      "loss": 0.8242,
+      "step": 8755
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.67757425171437e-06,
+      "loss": 0.7543,
+      "step": 8756
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.675492881275743e-06,
+      "loss": 0.7931,
+      "step": 8757
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6734119673725875e-06,
+      "loss": 0.7657,
+      "step": 8758
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6713315101551193e-06,
+      "loss": 0.7474,
+      "step": 8759
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.669251509773516e-06,
+      "loss": 0.699,
+      "step": 8760
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6671719663779202e-06,
+      "loss": 0.8241,
+      "step": 8761
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6650928801184418e-06,
+      "loss": 0.7329,
+      "step": 8762
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.663014251145167e-06,
+      "loss": 0.67,
+      "step": 8763
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.660936079608137e-06,
+      "loss": 0.812,
+      "step": 8764
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.658858365657364e-06,
+      "loss": 0.7248,
+      "step": 8765
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.656781109442833e-06,
+      "loss": 0.3617,
+      "step": 8766
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6547043111144875e-06,
+      "loss": 0.7512,
+      "step": 8767
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.652627970822239e-06,
+      "loss": 0.7144,
+      "step": 8768
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6505520887159763e-06,
+      "loss": 0.7545,
+      "step": 8769
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6484766649455407e-06,
+      "loss": 0.7226,
+      "step": 8770
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6464016996607465e-06,
+      "loss": 0.7389,
+      "step": 8771
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.644327193011381e-06,
+      "loss": 0.7542,
+      "step": 8772
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6422531451471855e-06,
+      "loss": 0.7923,
+      "step": 8773
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6401795562178834e-06,
+      "loss": 0.7423,
+      "step": 8774
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.638106426373149e-06,
+      "loss": 0.373,
+      "step": 8775
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.636033755762639e-06,
+      "loss": 0.7862,
+      "step": 8776
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6339615445359666e-06,
+      "loss": 0.6971,
+      "step": 8777
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.63188979284271e-06,
+      "loss": 0.7584,
+      "step": 8778
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6298185008324273e-06,
+      "loss": 0.751,
+      "step": 8779
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.62774766865463e-06,
+      "loss": 0.7165,
+      "step": 8780
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6256772964587984e-06,
+      "loss": 0.7554,
+      "step": 8781
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6236073843943896e-06,
+      "loss": 0.7608,
+      "step": 8782
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.621537932610817e-06,
+      "loss": 0.8247,
+      "step": 8783
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6194689412574613e-06,
+      "loss": 0.7821,
+      "step": 8784
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.617400410483679e-06,
+      "loss": 0.8082,
+      "step": 8785
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.615332340438785e-06,
+      "loss": 0.7092,
+      "step": 8786
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6132647312720626e-06,
+      "loss": 0.7945,
+      "step": 8787
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6111975831327593e-06,
+      "loss": 0.681,
+      "step": 8788
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6091308961700988e-06,
+      "loss": 0.8289,
+      "step": 8789
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6070646705332624e-06,
+      "loss": 0.7387,
+      "step": 8790
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6049989063713963e-06,
+      "loss": 0.8175,
+      "step": 8791
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6029336038336273e-06,
+      "loss": 0.7887,
+      "step": 8792
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6008687630690298e-06,
+      "loss": 0.7388,
+      "step": 8793
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5988043842266596e-06,
+      "loss": 0.7516,
+      "step": 8794
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5967404674555374e-06,
+      "loss": 0.6595,
+      "step": 8795
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.594677012904645e-06,
+      "loss": 0.7824,
+      "step": 8796
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5926140207229275e-06,
+      "loss": 0.3609,
+      "step": 8797
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.59055149105931e-06,
+      "loss": 0.7489,
+      "step": 8798
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5884894240626743e-06,
+      "loss": 0.7034,
+      "step": 8799
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.586427819881869e-06,
+      "loss": 0.752,
+      "step": 8800
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5843666786657107e-06,
+      "loss": 0.7553,
+      "step": 8801
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5823060005629874e-06,
+      "loss": 0.7465,
+      "step": 8802
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.580245785722446e-06,
+      "loss": 0.7436,
+      "step": 8803
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.578186034292802e-06,
+      "loss": 0.7046,
+      "step": 8804
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5761267464227446e-06,
+      "loss": 0.8064,
+      "step": 8805
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5740679222609197e-06,
+      "loss": 0.809,
+      "step": 8806
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5720095619559416e-06,
+      "loss": 0.7869,
+      "step": 8807
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5699516656563993e-06,
+      "loss": 0.7281,
+      "step": 8808
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.567894233510839e-06,
+      "loss": 0.6694,
+      "step": 8809
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.565837265667775e-06,
+      "loss": 0.7813,
+      "step": 8810
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5637807622756935e-06,
+      "loss": 0.6842,
+      "step": 8811
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5617247234830397e-06,
+      "loss": 0.7409,
+      "step": 8812
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.559669149438235e-06,
+      "loss": 0.7903,
+      "step": 8813
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5576140402896532e-06,
+      "loss": 0.748,
+      "step": 8814
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5555593961856505e-06,
+      "loss": 0.7971,
+      "step": 8815
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5535052172745388e-06,
+      "loss": 0.7051,
+      "step": 8816
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5514515037045948e-06,
+      "loss": 0.6679,
+      "step": 8817
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5493982556240735e-06,
+      "loss": 0.7815,
+      "step": 8818
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.547345473181185e-06,
+      "loss": 0.745,
+      "step": 8819
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.545293156524108e-06,
+      "loss": 0.8048,
+      "step": 8820
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.543241305800993e-06,
+      "loss": 0.7088,
+      "step": 8821
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5411899211599523e-06,
+      "loss": 0.7254,
+      "step": 8822
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5391390027490613e-06,
+      "loss": 0.7044,
+      "step": 8823
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.537088550716373e-06,
+      "loss": 0.7713,
+      "step": 8824
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5350385652098953e-06,
+      "loss": 0.7675,
+      "step": 8825
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5329890463776075e-06,
+      "loss": 0.7432,
+      "step": 8826
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5309399943674505e-06,
+      "loss": 0.7174,
+      "step": 8827
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5288914093273417e-06,
+      "loss": 0.4009,
+      "step": 8828
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5268432914051577e-06,
+      "loss": 0.7395,
+      "step": 8829
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5247956407487347e-06,
+      "loss": 0.7404,
+      "step": 8830
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5227484575058935e-06,
+      "loss": 0.6961,
+      "step": 8831
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5207017418244005e-06,
+      "loss": 0.7784,
+      "step": 8832
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.518655493852007e-06,
+      "loss": 0.7232,
+      "step": 8833
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.516609713736413e-06,
+      "loss": 0.7543,
+      "step": 8834
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5145644016253023e-06,
+      "loss": 0.8042,
+      "step": 8835
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5125195576663087e-06,
+      "loss": 0.6979,
+      "step": 8836
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.510475182007045e-06,
+      "loss": 0.7788,
+      "step": 8837
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.508431274795082e-06,
+      "loss": 0.7179,
+      "step": 8838
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5063878361779603e-06,
+      "loss": 0.7675,
+      "step": 8839
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.504344866303182e-06,
+      "loss": 0.7183,
+      "step": 8840
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.502302365318225e-06,
+      "loss": 0.7054,
+      "step": 8841
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5002603333705267e-06,
+      "loss": 0.7563,
+      "step": 8842
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4982187706074856e-06,
+      "loss": 0.7409,
+      "step": 8843
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4961776771764788e-06,
+      "loss": 0.7141,
+      "step": 8844
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.494137053224841e-06,
+      "loss": 0.7698,
+      "step": 8845
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4920968988998717e-06,
+      "loss": 0.7574,
+      "step": 8846
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4900572143488453e-06,
+      "loss": 0.7714,
+      "step": 8847
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4880179997189934e-06,
+      "loss": 0.7861,
+      "step": 8848
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4859792551575147e-06,
+      "loss": 0.8074,
+      "step": 8849
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.483940980811582e-06,
+      "loss": 0.7332,
+      "step": 8850
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4819031768283264e-06,
+      "loss": 0.7804,
+      "step": 8851
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4798658433548414e-06,
+      "loss": 0.6981,
+      "step": 8852
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4778289805382014e-06,
+      "loss": 0.8028,
+      "step": 8853
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4757925885254294e-06,
+      "loss": 0.7727,
+      "step": 8854
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4737566674635304e-06,
+      "loss": 0.7686,
+      "step": 8855
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.471721217499461e-06,
+      "loss": 0.7279,
+      "step": 8856
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4696862387801557e-06,
+      "loss": 0.3396,
+      "step": 8857
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4676517314525072e-06,
+      "loss": 0.7846,
+      "step": 8858
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4656176956633723e-06,
+      "loss": 0.7739,
+      "step": 8859
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.4635841315595876e-06,
+      "loss": 0.7895,
+      "step": 8860
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4615510392879416e-06,
+      "loss": 0.7384,
+      "step": 8861
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.459518418995188e-06,
+      "loss": 0.6993,
+      "step": 8862
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.457486270828061e-06,
+      "loss": 0.78,
+      "step": 8863
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4554545949332474e-06,
+      "loss": 0.7221,
+      "step": 8864
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4534233914574013e-06,
+      "loss": 0.7374,
+      "step": 8865
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4513926605471504e-06,
+      "loss": 0.7456,
+      "step": 8866
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.449362402349081e-06,
+      "loss": 0.7534,
+      "step": 8867
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4473326170097477e-06,
+      "loss": 0.7476,
+      "step": 8868
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.445303304675668e-06,
+      "loss": 0.7485,
+      "step": 8869
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4432744654933325e-06,
+      "loss": 0.7365,
+      "step": 8870
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4412460996091935e-06,
+      "loss": 0.77,
+      "step": 8871
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.439218207169662e-06,
+      "loss": 0.788,
+      "step": 8872
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.437190788321131e-06,
+      "loss": 0.7487,
+      "step": 8873
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.435163843209942e-06,
+      "loss": 0.7739,
+      "step": 8874
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.433137371982418e-06,
+      "loss": 0.777,
+      "step": 8875
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4311113747848324e-06,
+      "loss": 0.7344,
+      "step": 8876
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4290858517634396e-06,
+      "loss": 0.748,
+      "step": 8877
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.427060803064445e-06,
+      "loss": 0.7766,
+      "step": 8878
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4250362288340335e-06,
+      "loss": 0.8318,
+      "step": 8879
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.423012129218346e-06,
+      "loss": 0.7143,
+      "step": 8880
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4209885043634938e-06,
+      "loss": 0.7407,
+      "step": 8881
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4189653544155477e-06,
+      "loss": 0.7616,
+      "step": 8882
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.416942679520556e-06,
+      "loss": 0.7293,
+      "step": 8883
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4149204798245226e-06,
+      "loss": 0.7327,
+      "step": 8884
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4128987554734184e-06,
+      "loss": 0.7407,
+      "step": 8885
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.410877506613186e-06,
+      "loss": 0.7366,
+      "step": 8886
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4088567333897272e-06,
+      "loss": 0.7619,
+      "step": 8887
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4068364359489093e-06,
+      "loss": 0.7221,
+      "step": 8888
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4048166144365734e-06,
+      "loss": 0.7581,
+      "step": 8889
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.402797268998519e-06,
+      "loss": 0.8228,
+      "step": 8890
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4007783997805068e-06,
+      "loss": 0.7621,
+      "step": 8891
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3987600069282778e-06,
+      "loss": 0.7361,
+      "step": 8892
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.396742090587526e-06,
+      "loss": 0.694,
+      "step": 8893
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.394724650903911e-06,
+      "loss": 0.805,
+      "step": 8894
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3927076880230703e-06,
+      "loss": 0.7658,
+      "step": 8895
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3906912020905915e-06,
+      "loss": 0.6946,
+      "step": 8896
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.38867519325204e-06,
+      "loss": 0.7238,
+      "step": 8897
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.386659661652937e-06,
+      "loss": 0.7394,
+      "step": 8898
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3846446074387795e-06,
+      "loss": 0.6866,
+      "step": 8899
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3826300307550218e-06,
+      "loss": 0.7408,
+      "step": 8900
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.380615931747082e-06,
+      "loss": 0.8332,
+      "step": 8901
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.378602310560357e-06,
+      "loss": 0.8316,
+      "step": 8902
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.376589167340194e-06,
+      "loss": 0.7089,
+      "step": 8903
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.374576502231911e-06,
+      "loss": 0.7701,
+      "step": 8904
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3725643153807984e-06,
+      "loss": 0.7189,
+      "step": 8905
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3705526069321036e-06,
+      "loss": 0.7745,
+      "step": 8906
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3685413770310405e-06,
+      "loss": 0.7595,
+      "step": 8907
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3665306258227882e-06,
+      "loss": 0.377,
+      "step": 8908
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3645203534524983e-06,
+      "loss": 0.6923,
+      "step": 8909
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3625105600652806e-06,
+      "loss": 0.703,
+      "step": 8910
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3605012458062094e-06,
+      "loss": 0.8005,
+      "step": 8911
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.358492410820331e-06,
+      "loss": 0.7295,
+      "step": 8912
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3564840552526532e-06,
+      "loss": 0.7646,
+      "step": 8913
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3544761792481463e-06,
+      "loss": 0.8304,
+      "step": 8914
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3524687829517535e-06,
+      "loss": 0.8016,
+      "step": 8915
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3504618665083734e-06,
+      "loss": 0.7578,
+      "step": 8916
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3484554300628826e-06,
+      "loss": 0.7459,
+      "step": 8917
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3464494737601096e-06,
+      "loss": 0.7332,
+      "step": 8918
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3444439977448595e-06,
+      "loss": 0.8263,
+      "step": 8919
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.342439002161897e-06,
+      "loss": 0.3822,
+      "step": 8920
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.340434487155948e-06,
+      "loss": 0.7609,
+      "step": 8921
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3384304528717147e-06,
+      "loss": 0.7649,
+      "step": 8922
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3364268994538583e-06,
+      "loss": 0.7051,
+      "step": 8923
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3344238270469997e-06,
+      "loss": 0.8041,
+      "step": 8924
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3324212357957376e-06,
+      "loss": 0.7131,
+      "step": 8925
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.330419125844627e-06,
+      "loss": 0.6942,
+      "step": 8926
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3284174973381877e-06,
+      "loss": 0.7399,
+      "step": 8927
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3264163504209125e-06,
+      "loss": 0.7248,
+      "step": 8928
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.324415685237252e-06,
+      "loss": 0.8115,
+      "step": 8929
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.322415501931622e-06,
+      "loss": 0.7894,
+      "step": 8930
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.320415800648411e-06,
+      "loss": 0.8176,
+      "step": 8931
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.318416581531967e-06,
+      "loss": 0.7694,
+      "step": 8932
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3164178447266006e-06,
+      "loss": 0.8114,
+      "step": 8933
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3144195903765897e-06,
+      "loss": 0.699,
+      "step": 8934
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3124218186261848e-06,
+      "loss": 0.7952,
+      "step": 8935
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.310424529619588e-06,
+      "loss": 0.78,
+      "step": 8936
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.308427723500982e-06,
+      "loss": 0.7751,
+      "step": 8937
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.306431400414499e-06,
+      "loss": 0.7617,
+      "step": 8938
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.304435560504251e-06,
+      "loss": 0.7517,
+      "step": 8939
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3024402039143e-06,
+      "loss": 0.7459,
+      "step": 8940
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.30044533078869e-06,
+      "loss": 0.7886,
+      "step": 8941
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.298450941271416e-06,
+      "loss": 0.7343,
+      "step": 8942
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.296457035506441e-06,
+      "loss": 0.7439,
+      "step": 8943
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2944636136377006e-06,
+      "loss": 0.3724,
+      "step": 8944
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2924706758090887e-06,
+      "loss": 0.8032,
+      "step": 8945
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.290478222164465e-06,
+      "loss": 0.7807,
+      "step": 8946
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2884862528476513e-06,
+      "loss": 0.7285,
+      "step": 8947
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2864947680024453e-06,
+      "loss": 0.7397,
+      "step": 8948
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2845037677725998e-06,
+      "loss": 0.7327,
+      "step": 8949
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.282513252301831e-06,
+      "loss": 0.8447,
+      "step": 8950
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2805232217338334e-06,
+      "loss": 0.6944,
+      "step": 8951
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.278533676212251e-06,
+      "loss": 0.8188,
+      "step": 8952
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.276544615880699e-06,
+      "loss": 0.7552,
+      "step": 8953
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2745560408827635e-06,
+      "loss": 0.7153,
+      "step": 8954
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.272567951361987e-06,
+      "loss": 0.7972,
+      "step": 8955
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2705803474618757e-06,
+      "loss": 0.7378,
+      "step": 8956
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2685932293259127e-06,
+      "loss": 0.7795,
+      "step": 8957
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2666065970975315e-06,
+      "loss": 0.7799,
+      "step": 8958
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.264620450920145e-06,
+      "loss": 0.7418,
+      "step": 8959
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.262634790937116e-06,
+      "loss": 0.7793,
+      "step": 8960
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2606496172917855e-06,
+      "loss": 0.7605,
+      "step": 8961
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.258664930127451e-06,
+      "loss": 0.7283,
+      "step": 8962
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.256680729587375e-06,
+      "loss": 0.7737,
+      "step": 8963
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2546970158147917e-06,
+      "loss": 0.7275,
+      "step": 8964
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.252713788952895e-06,
+      "loss": 0.6751,
+      "step": 8965
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2507310491448396e-06,
+      "loss": 0.7559,
+      "step": 8966
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2487487965337573e-06,
+      "loss": 0.7062,
+      "step": 8967
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.246767031262733e-06,
+      "loss": 0.8084,
+      "step": 8968
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2447857534748184e-06,
+      "loss": 0.7786,
+      "step": 8969
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2428049633130375e-06,
+      "loss": 0.7922,
+      "step": 8970
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.240824660920373e-06,
+      "loss": 0.7429,
+      "step": 8971
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2388448464397715e-06,
+      "loss": 0.7754,
+      "step": 8972
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.236865520014143e-06,
+      "loss": 0.7548,
+      "step": 8973
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2348866817863723e-06,
+      "loss": 0.6858,
+      "step": 8974
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.232908331899298e-06,
+      "loss": 0.799,
+      "step": 8975
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.230930470495727e-06,
+      "loss": 0.7788,
+      "step": 8976
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.228953097718436e-06,
+      "loss": 0.3816,
+      "step": 8977
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2269762137101547e-06,
+      "loss": 0.7074,
+      "step": 8978
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.224999818613589e-06,
+      "loss": 0.7831,
+      "step": 8979
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.223023912571409e-06,
+      "loss": 0.7877,
+      "step": 8980
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2210484957262422e-06,
+      "loss": 0.7741,
+      "step": 8981
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.219073568220681e-06,
+      "loss": 0.7344,
+      "step": 8982
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.217099130197292e-06,
+      "loss": 0.3977,
+      "step": 8983
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.215125181798597e-06,
+      "loss": 0.7848,
+      "step": 8984
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.213151723167084e-06,
+      "loss": 0.781,
+      "step": 8985
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.211178754445212e-06,
+      "loss": 0.7841,
+      "step": 8986
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2092062757753974e-06,
+      "loss": 0.7964,
+      "step": 8987
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.207234287300025e-06,
+      "loss": 0.6723,
+      "step": 8988
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2052627891614373e-06,
+      "loss": 0.7799,
+      "step": 8989
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.203291781501956e-06,
+      "loss": 0.7892,
+      "step": 8990
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2013212644638545e-06,
+      "loss": 0.7314,
+      "step": 8991
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.199351238189371e-06,
+      "loss": 0.7984,
+      "step": 8992
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.19738170282072e-06,
+      "loss": 0.6753,
+      "step": 8993
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1954126585000687e-06,
+      "loss": 0.6999,
+      "step": 8994
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.19344410536955e-06,
+      "loss": 0.7683,
+      "step": 8995
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.191476043571269e-06,
+      "loss": 0.7906,
+      "step": 8996
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.189508473247289e-06,
+      "loss": 0.7137,
+      "step": 8997
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.187541394539637e-06,
+      "loss": 0.705,
+      "step": 8998
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.185574807590308e-06,
+      "loss": 0.8053,
+      "step": 8999
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.183608712541264e-06,
+      "loss": 0.7525,
+      "step": 9000
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.181643109534426e-06,
+      "loss": 0.7279,
+      "step": 9001
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1796779987116756e-06,
+      "loss": 0.7217,
+      "step": 9002
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.177713380214874e-06,
+      "loss": 0.7733,
+      "step": 9003
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1757492541858328e-06,
+      "loss": 0.8095,
+      "step": 9004
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.173785620766329e-06,
+      "loss": 0.3502,
+      "step": 9005
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1718224800981145e-06,
+      "loss": 0.7702,
+      "step": 9006
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.169859832322897e-06,
+      "loss": 0.6956,
+      "step": 9007
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1678976775823454e-06,
+      "loss": 0.7099,
+      "step": 9008
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1659360160181064e-06,
+      "loss": 0.743,
+      "step": 9009
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.163974847771777e-06,
+      "loss": 0.7785,
+      "step": 9010
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.162014172984924e-06,
+      "loss": 0.7781,
+      "step": 9011
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1600539917990846e-06,
+      "loss": 0.7544,
+      "step": 9012
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.15809430435575e-06,
+      "loss": 0.7632,
+      "step": 9013
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.156135110796383e-06,
+      "loss": 0.7511,
+      "step": 9014
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.154176411262403e-06,
+      "loss": 0.7609,
+      "step": 9015
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1522182058952066e-06,
+      "loss": 0.7271,
+      "step": 9016
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1502604948361403e-06,
+      "loss": 0.3952,
+      "step": 9017
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1483032782265278e-06,
+      "loss": 0.7839,
+      "step": 9018
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1463465562076456e-06,
+      "loss": 0.7107,
+      "step": 9019
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1443903289207455e-06,
+      "loss": 0.8086,
+      "step": 9020
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1424345965070326e-06,
+      "loss": 0.7941,
+      "step": 9021
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1404793591076876e-06,
+      "loss": 0.8237,
+      "step": 9022
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.138524616863846e-06,
+      "loss": 0.779,
+      "step": 9023
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1365703699166084e-06,
+      "loss": 0.7575,
+      "step": 9024
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.134616618407048e-06,
+      "loss": 0.7325,
+      "step": 9025
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1326633624761947e-06,
+      "loss": 0.7524,
+      "step": 9026
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1307106022650447e-06,
+      "loss": 0.7905,
+      "step": 9027
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1287583379145537e-06,
+      "loss": 0.7255,
+      "step": 9028
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1268065695656536e-06,
+      "loss": 0.6787,
+      "step": 9029
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1248552973592282e-06,
+      "loss": 0.7521,
+      "step": 9030
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1229045214361296e-06,
+      "loss": 0.8569,
+      "step": 9031
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1209542419371807e-06,
+      "loss": 0.7266,
+      "step": 9032
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.119004459003159e-06,
+      "loss": 0.718,
+      "step": 9033
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1170551727748054e-06,
+      "loss": 0.7714,
+      "step": 9034
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1151063833928386e-06,
+      "loss": 0.7784,
+      "step": 9035
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1131580909979263e-06,
+      "loss": 0.7538,
+      "step": 9036
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1112102957307054e-06,
+      "loss": 0.7143,
+      "step": 9037
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.109262997731782e-06,
+      "loss": 0.7214,
+      "step": 9038
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1073161971417176e-06,
+      "loss": 0.7449,
+      "step": 9039
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.105369894101048e-06,
+      "loss": 0.7659,
+      "step": 9040
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.10342408875026e-06,
+      "loss": 0.8035,
+      "step": 9041
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.10147878122982e-06,
+      "loss": 0.7688,
+      "step": 9042
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0995339716801444e-06,
+      "loss": 0.7899,
+      "step": 9043
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.09758966024162e-06,
+      "loss": 0.8316,
+      "step": 9044
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0956458470546015e-06,
+      "loss": 0.7298,
+      "step": 9045
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0937025322593996e-06,
+      "loss": 0.7445,
+      "step": 9046
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.091759715996291e-06,
+      "loss": 0.7399,
+      "step": 9047
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0898173984055236e-06,
+      "loss": 0.7652,
+      "step": 9048
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.087875579627301e-06,
+      "loss": 0.7483,
+      "step": 9049
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0859342598017906e-06,
+      "loss": 0.7888,
+      "step": 9050
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0839934390691328e-06,
+      "loss": 0.759,
+      "step": 9051
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0820531175694236e-06,
+      "loss": 0.7551,
+      "step": 9052
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.080113295442725e-06,
+      "loss": 0.7517,
+      "step": 9053
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0781739728290607e-06,
+      "loss": 0.7167,
+      "step": 9054
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.076235149868425e-06,
+      "loss": 0.722,
+      "step": 9055
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.074296826700772e-06,
+      "loss": 0.7167,
+      "step": 9056
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0723590034660156e-06,
+      "loss": 0.7922,
+      "step": 9057
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.070421680304043e-06,
+      "loss": 0.7849,
+      "step": 9058
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0684848573546954e-06,
+      "loss": 0.7643,
+      "step": 9059
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.066548534757788e-06,
+      "loss": 0.7713,
+      "step": 9060
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.06461271265309e-06,
+      "loss": 0.7788,
+      "step": 9061
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0626773911803432e-06,
+      "loss": 0.73,
+      "step": 9062
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0607425704792426e-06,
+      "loss": 0.7503,
+      "step": 9063
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.058808250689461e-06,
+      "loss": 0.6967,
+      "step": 9064
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.056874431950625e-06,
+      "loss": 0.7382,
+      "step": 9065
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0549411144023254e-06,
+      "loss": 0.7576,
+      "step": 9066
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.053008298184118e-06,
+      "loss": 0.7037,
+      "step": 9067
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0510759834355284e-06,
+      "loss": 0.7477,
+      "step": 9068
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.049144170296039e-06,
+      "loss": 0.702,
+      "step": 9069
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.047212858905093e-06,
+      "loss": 0.7869,
+      "step": 9070
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0452820494021106e-06,
+      "loss": 0.3862,
+      "step": 9071
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.043351741926465e-06,
+      "loss": 0.7639,
+      "step": 9072
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.041421936617489e-06,
+      "loss": 0.843,
+      "step": 9073
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.039492633614496e-06,
+      "loss": 0.8084,
+      "step": 9074
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.037563833056748e-06,
+      "loss": 0.7336,
+      "step": 9075
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.035635535083472e-06,
+      "loss": 0.7387,
+      "step": 9076
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.033707739833871e-06,
+      "loss": 0.7708,
+      "step": 9077
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.031780447447098e-06,
+      "loss": 0.7452,
+      "step": 9078
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.029853658062273e-06,
+      "loss": 0.699,
+      "step": 9079
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0279273718184876e-06,
+      "loss": 0.804,
+      "step": 9080
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0260015888547835e-06,
+      "loss": 0.8108,
+      "step": 9081
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.024076309310182e-06,
+      "loss": 0.6507,
+      "step": 9082
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0221515333236516e-06,
+      "loss": 0.7236,
+      "step": 9083
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.02022726103414e-06,
+      "loss": 0.7196,
+      "step": 9084
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.018303492580548e-06,
+      "loss": 0.3845,
+      "step": 9085
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0163802281017374e-06,
+      "loss": 0.8002,
+      "step": 9086
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.014457467736549e-06,
+      "loss": 0.7642,
+      "step": 9087
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.012535211623774e-06,
+      "loss": 0.8011,
+      "step": 9088
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0106134599021653e-06,
+      "loss": 0.732,
+      "step": 9089
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0086922127104533e-06,
+      "loss": 0.7337,
+      "step": 9090
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0067714701873186e-06,
+      "loss": 0.7769,
+      "step": 9091
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.004851232471412e-06,
+      "loss": 0.7415,
+      "step": 9092
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0029314997013413e-06,
+      "loss": 0.7202,
+      "step": 9093
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.001012272015691e-06,
+      "loss": 0.345,
+      "step": 9094
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.999093549552996e-06,
+      "loss": 0.7092,
+      "step": 9095
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.997175332451757e-06,
+      "loss": 0.7645,
+      "step": 9096
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.995257620850448e-06,
+      "loss": 0.7832,
+      "step": 9097
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9933404148874945e-06,
+      "loss": 0.7872,
+      "step": 9098
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.991423714701288e-06,
+      "loss": 0.645,
+      "step": 9099
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.9895075204301916e-06,
+      "loss": 0.8076,
+      "step": 9100
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 2.987591832212521e-06,
+      "loss": 0.7588,
+      "step": 9101
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.985676650186565e-06,
+      "loss": 0.7445,
+      "step": 9102
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9837619744905655e-06,
+      "loss": 0.7047,
+      "step": 9103
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.981847805262741e-06,
+      "loss": 0.7374,
+      "step": 9104
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9799341426412586e-06,
+      "loss": 0.7476,
+      "step": 9105
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.978020986764264e-06,
+      "loss": 0.7579,
+      "step": 9106
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.976108337769854e-06,
+      "loss": 0.7197,
+      "step": 9107
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.974196195796094e-06,
+      "loss": 0.6933,
+      "step": 9108
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9722845609810092e-06,
+      "loss": 0.6678,
+      "step": 9109
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9703734334625978e-06,
+      "loss": 0.7177,
+      "step": 9110
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9684628133788106e-06,
+      "loss": 0.7636,
+      "step": 9111
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9665527008675642e-06,
+      "loss": 0.7642,
+      "step": 9112
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.964643096066746e-06,
+      "loss": 0.7843,
+      "step": 9113
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.962733999114198e-06,
+      "loss": 0.7336,
+      "step": 9114
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9608254101477252e-06,
+      "loss": 0.731,
+      "step": 9115
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9589173293051066e-06,
+      "loss": 0.8048,
+      "step": 9116
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.957009756724073e-06,
+      "loss": 0.7304,
+      "step": 9117
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.95510269254232e-06,
+      "loss": 0.7833,
+      "step": 9118
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.953196136897516e-06,
+      "loss": 0.677,
+      "step": 9119
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.951290089927282e-06,
+      "loss": 0.7461,
+      "step": 9120
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9493845517692055e-06,
+      "loss": 0.7526,
+      "step": 9121
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.947479522560841e-06,
+      "loss": 0.743,
+      "step": 9122
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.945575002439699e-06,
+      "loss": 0.7269,
+      "step": 9123
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.943670991543264e-06,
+      "loss": 0.7329,
+      "step": 9124
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.94176749000897e-06,
+      "loss": 0.7791,
+      "step": 9125
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9398644979742275e-06,
+      "loss": 0.713,
+      "step": 9126
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9379620155764034e-06,
+      "loss": 0.7558,
+      "step": 9127
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.936060042952823e-06,
+      "loss": 0.7101,
+      "step": 9128
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.934158580240789e-06,
+      "loss": 0.7279,
+      "step": 9129
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.932257627577553e-06,
+      "loss": 0.7796,
+      "step": 9130
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9303571851003344e-06,
+      "loss": 0.7017,
+      "step": 9131
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.928457252946323e-06,
+      "loss": 0.7694,
+      "step": 9132
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.926557831252662e-06,
+      "loss": 0.7339,
+      "step": 9133
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.924658920156461e-06,
+      "loss": 0.7211,
+      "step": 9134
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9227605197947916e-06,
+      "loss": 0.6669,
+      "step": 9135
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.920862630304695e-06,
+      "loss": 0.7036,
+      "step": 9136
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.918965251823168e-06,
+      "loss": 0.758,
+      "step": 9137
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9170683844871707e-06,
+      "loss": 0.7033,
+      "step": 9138
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9151720284336337e-06,
+      "loss": 0.7494,
+      "step": 9139
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9132761837994426e-06,
+      "loss": 0.7908,
+      "step": 9140
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9113808507214467e-06,
+      "loss": 0.7445,
+      "step": 9141
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.909486029336468e-06,
+      "loss": 0.7245,
+      "step": 9142
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9075917197812766e-06,
+      "loss": 0.6949,
+      "step": 9143
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9056979221926207e-06,
+      "loss": 0.7741,
+      "step": 9144
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9038046367071983e-06,
+      "loss": 0.7726,
+      "step": 9145
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9019118634616817e-06,
+      "loss": 0.7729,
+      "step": 9146
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9000196025926985e-06,
+      "loss": 0.7672,
+      "step": 9147
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8981278542368383e-06,
+      "loss": 0.7892,
+      "step": 9148
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8962366185306657e-06,
+      "loss": 0.7325,
+      "step": 9149
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.894345895610693e-06,
+      "loss": 0.7026,
+      "step": 9150
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.892455685613402e-06,
+      "loss": 0.761,
+      "step": 9151
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8905659886752434e-06,
+      "loss": 0.6834,
+      "step": 9152
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8886768049326207e-06,
+      "loss": 0.7582,
+      "step": 9153
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.886788134521904e-06,
+      "loss": 0.7257,
+      "step": 9154
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8848999775794327e-06,
+      "loss": 0.7547,
+      "step": 9155
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8830123342414995e-06,
+      "loss": 0.7166,
+      "step": 9156
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8811252046443615e-06,
+      "loss": 0.8136,
+      "step": 9157
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.879238588924249e-06,
+      "loss": 0.7172,
+      "step": 9158
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.877352487217343e-06,
+      "loss": 0.7322,
+      "step": 9159
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8754668996597935e-06,
+      "loss": 0.7152,
+      "step": 9160
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8735818263877067e-06,
+      "loss": 0.7435,
+      "step": 9161
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8716972675371645e-06,
+      "loss": 0.7854,
+      "step": 9162
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.869813223244199e-06,
+      "loss": 0.7842,
+      "step": 9163
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8679296936448107e-06,
+      "loss": 0.7554,
+      "step": 9164
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8660466788749675e-06,
+      "loss": 0.7686,
+      "step": 9165
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.864164179070591e-06,
+      "loss": 0.7987,
+      "step": 9166
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8622821943675662e-06,
+      "loss": 0.7925,
+      "step": 9167
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.860400724901752e-06,
+      "loss": 0.7065,
+      "step": 9168
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8585197708089594e-06,
+      "loss": 0.7502,
+      "step": 9169
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8566393322249608e-06,
+      "loss": 0.7334,
+      "step": 9170
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8547594092855037e-06,
+      "loss": 0.726,
+      "step": 9171
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.852880002126287e-06,
+      "loss": 0.7882,
+      "step": 9172
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.851001110882975e-06,
+      "loss": 0.7889,
+      "step": 9173
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8491227356911945e-06,
+      "loss": 0.6929,
+      "step": 9174
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8472448766865403e-06,
+      "loss": 0.7427,
+      "step": 9175
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.845367534004565e-06,
+      "loss": 0.7475,
+      "step": 9176
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8434907077807795e-06,
+      "loss": 0.7677,
+      "step": 9177
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8416143981506714e-06,
+      "loss": 0.7378,
+      "step": 9178
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8397386052496776e-06,
+      "loss": 0.7906,
+      "step": 9179
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8378633292131997e-06,
+      "loss": 0.7606,
+      "step": 9180
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.835988570176612e-06,
+      "loss": 0.7621,
+      "step": 9181
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.834114328275239e-06,
+      "loss": 0.7436,
+      "step": 9182
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.832240603644373e-06,
+      "loss": 0.7341,
+      "step": 9183
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.830367396419269e-06,
+      "loss": 0.7423,
+      "step": 9184
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.82849470673515e-06,
+      "loss": 0.7279,
+      "step": 9185
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8266225347271926e-06,
+      "loss": 0.8263,
+      "step": 9186
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8247508805305367e-06,
+      "loss": 0.8151,
+      "step": 9187
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8228797442802945e-06,
+      "loss": 0.7873,
+      "step": 9188
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8210091261115314e-06,
+      "loss": 0.7496,
+      "step": 9189
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8191390261592733e-06,
+      "loss": 0.7691,
+      "step": 9190
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8172694445585215e-06,
+      "loss": 0.7382,
+      "step": 9191
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8154003814442287e-06,
+      "loss": 0.7634,
+      "step": 9192
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.813531836951311e-06,
+      "loss": 0.7437,
+      "step": 9193
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.811663811214653e-06,
+      "loss": 0.7281,
+      "step": 9194
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.809796304369099e-06,
+      "loss": 0.6974,
+      "step": 9195
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8079293165494502e-06,
+      "loss": 0.7543,
+      "step": 9196
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8060628478904816e-06,
+      "loss": 0.7968,
+      "step": 9197
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.804196898526922e-06,
+      "loss": 0.8287,
+      "step": 9198
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8023314685934644e-06,
+      "loss": 0.7633,
+      "step": 9199
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8004665582247625e-06,
+      "loss": 0.6878,
+      "step": 9200
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.798602167555442e-06,
+      "loss": 0.7596,
+      "step": 9201
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7967382967200774e-06,
+      "loss": 0.7282,
+      "step": 9202
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7948749458532177e-06,
+      "loss": 0.7171,
+      "step": 9203
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.793012115089365e-06,
+      "loss": 0.7323,
+      "step": 9204
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7911498045629924e-06,
+      "loss": 0.7698,
+      "step": 9205
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7892880144085253e-06,
+      "loss": 0.7611,
+      "step": 9206
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.787426744760365e-06,
+      "loss": 0.6786,
+      "step": 9207
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7855659957528624e-06,
+      "loss": 0.7255,
+      "step": 9208
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.783705767520334e-06,
+      "loss": 0.7116,
+      "step": 9209
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7818460601970664e-06,
+      "loss": 0.7754,
+      "step": 9210
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7799868739172996e-06,
+      "loss": 0.7843,
+      "step": 9211
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7781282088152395e-06,
+      "loss": 0.7883,
+      "step": 9212
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7762700650250507e-06,
+      "loss": 0.7142,
+      "step": 9213
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7744124426808706e-06,
+      "loss": 0.71,
+      "step": 9214
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7725553419167883e-06,
+      "loss": 0.7889,
+      "step": 9215
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7706987628668556e-06,
+      "loss": 0.7457,
+      "step": 9216
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.768842705665097e-06,
+      "loss": 0.7525,
+      "step": 9217
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7669871704454874e-06,
+      "loss": 0.7771,
+      "step": 9218
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7651321573419676e-06,
+      "loss": 0.7851,
+      "step": 9219
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7632776664884474e-06,
+      "loss": 0.7325,
+      "step": 9220
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7614236980187903e-06,
+      "loss": 0.7662,
+      "step": 9221
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.7595702520668223e-06,
+      "loss": 0.7355,
+      "step": 9222
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7577173287663404e-06,
+      "loss": 0.7584,
+      "step": 9223
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.755864928251094e-06,
+      "loss": 0.7293,
+      "step": 9224
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7540130506548026e-06,
+      "loss": 0.7316,
+      "step": 9225
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7521616961111384e-06,
+      "loss": 0.7419,
+      "step": 9226
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.750310864753749e-06,
+      "loss": 0.7128,
+      "step": 9227
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.748460556716234e-06,
+      "loss": 0.7098,
+      "step": 9228
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7466107721321543e-06,
+      "loss": 0.7028,
+      "step": 9229
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7447615111350422e-06,
+      "loss": 0.7105,
+      "step": 9230
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7429127738583862e-06,
+      "loss": 0.726,
+      "step": 9231
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7410645604356312e-06,
+      "loss": 0.742,
+      "step": 9232
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.739216871000201e-06,
+      "loss": 0.735,
+      "step": 9233
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7373697056854643e-06,
+      "loss": 0.6885,
+      "step": 9234
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.735523064624759e-06,
+      "loss": 0.779,
+      "step": 9235
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.733676947951389e-06,
+      "loss": 0.7281,
+      "step": 9236
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7318313557986154e-06,
+      "loss": 0.6761,
+      "step": 9237
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7299862882996586e-06,
+      "loss": 0.7206,
+      "step": 9238
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7281417455877103e-06,
+      "loss": 0.7633,
+      "step": 9239
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.726297727795918e-06,
+      "loss": 0.7625,
+      "step": 9240
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.72445423505739e-06,
+      "loss": 0.7949,
+      "step": 9241
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7226112675051987e-06,
+      "loss": 0.7463,
+      "step": 9242
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.720768825272382e-06,
+      "loss": 0.7593,
+      "step": 9243
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7189269084919336e-06,
+      "loss": 0.7277,
+      "step": 9244
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.717085517296817e-06,
+      "loss": 0.7968,
+      "step": 9245
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7152446518199484e-06,
+      "loss": 0.7087,
+      "step": 9246
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7134043121942155e-06,
+      "loss": 0.6903,
+      "step": 9247
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7115644985524592e-06,
+      "loss": 0.788,
+      "step": 9248
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7097252110274906e-06,
+      "loss": 0.7018,
+      "step": 9249
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7078864497520774e-06,
+      "loss": 0.8399,
+      "step": 9250
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7060482148589484e-06,
+      "loss": 0.7035,
+      "step": 9251
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.704210506480801e-06,
+      "loss": 0.7472,
+      "step": 9252
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.702373324750289e-06,
+      "loss": 0.7544,
+      "step": 9253
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.700536669800029e-06,
+      "loss": 0.796,
+      "step": 9254
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6987005417625976e-06,
+      "loss": 0.7109,
+      "step": 9255
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6968649407705415e-06,
+      "loss": 0.7459,
+      "step": 9256
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6950298669563613e-06,
+      "loss": 0.7101,
+      "step": 9257
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6931953204525183e-06,
+      "loss": 0.7619,
+      "step": 9258
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6913613013914464e-06,
+      "loss": 0.6473,
+      "step": 9259
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.689527809905531e-06,
+      "loss": 0.7518,
+      "step": 9260
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.68769484612712e-06,
+      "loss": 0.7493,
+      "step": 9261
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6858624101885324e-06,
+      "loss": 0.7808,
+      "step": 9262
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.684030502222039e-06,
+      "loss": 0.8038,
+      "step": 9263
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6821991223598754e-06,
+      "loss": 0.6813,
+      "step": 9264
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.680368270734244e-06,
+      "loss": 0.6792,
+      "step": 9265
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.678537947477299e-06,
+      "loss": 0.6914,
+      "step": 9266
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6767081527211703e-06,
+      "loss": 0.7357,
+      "step": 9267
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.674878886597935e-06,
+      "loss": 0.7422,
+      "step": 9268
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6730501492396444e-06,
+      "loss": 0.7551,
+      "step": 9269
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6712219407783038e-06,
+      "loss": 0.3648,
+      "step": 9270
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.669394261345879e-06,
+      "loss": 0.7689,
+      "step": 9271
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6675671110743084e-06,
+      "loss": 0.8182,
+      "step": 9272
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.665740490095481e-06,
+      "loss": 0.7817,
+      "step": 9273
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6639143985412498e-06,
+      "loss": 0.779,
+      "step": 9274
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.662088836543436e-06,
+      "loss": 0.711,
+      "step": 9275
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.660263804233817e-06,
+      "loss": 0.7272,
+      "step": 9276
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.658439301744129e-06,
+      "loss": 0.7514,
+      "step": 9277
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6566153292060803e-06,
+      "loss": 0.7268,
+      "step": 9278
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.65479188675133e-06,
+      "loss": 0.7475,
+      "step": 9279
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.652968974511506e-06,
+      "loss": 0.6911,
+      "step": 9280
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6511465926181913e-06,
+      "loss": 0.6885,
+      "step": 9281
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.649324741202941e-06,
+      "loss": 0.7532,
+      "step": 9282
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.647503420397263e-06,
+      "loss": 0.7709,
+      "step": 9283
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6456826303326246e-06,
+      "loss": 0.776,
+      "step": 9284
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6438623711404688e-06,
+      "loss": 0.7377,
+      "step": 9285
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.642042642952183e-06,
+      "loss": 0.6896,
+      "step": 9286
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.640223445899132e-06,
+      "loss": 0.7546,
+      "step": 9287
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.638404780112628e-06,
+      "loss": 0.7734,
+      "step": 9288
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6365866457239574e-06,
+      "loss": 0.7869,
+      "step": 9289
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.634769042864358e-06,
+      "loss": 0.7502,
+      "step": 9290
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6329519716650386e-06,
+      "loss": 0.7247,
+      "step": 9291
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.631135432257161e-06,
+      "loss": 0.8144,
+      "step": 9292
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6293194247718534e-06,
+      "loss": 0.7099,
+      "step": 9293
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.627503949340202e-06,
+      "loss": 0.7809,
+      "step": 9294
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.625689006093263e-06,
+      "loss": 0.7696,
+      "step": 9295
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6238745951620436e-06,
+      "loss": 0.7792,
+      "step": 9296
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.622060716677516e-06,
+      "loss": 0.7191,
+      "step": 9297
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6202473707706213e-06,
+      "loss": 0.7429,
+      "step": 9298
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6184345575722526e-06,
+      "loss": 0.7085,
+      "step": 9299
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6166222772132654e-06,
+      "loss": 0.8349,
+      "step": 9300
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6148105298244853e-06,
+      "loss": 0.7601,
+      "step": 9301
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6129993155366894e-06,
+      "loss": 0.7285,
+      "step": 9302
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.61118863448062e-06,
+      "loss": 0.7013,
+      "step": 9303
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6093784867869856e-06,
+      "loss": 0.7823,
+      "step": 9304
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.607568872586449e-06,
+      "loss": 0.8131,
+      "step": 9305
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6057597920096345e-06,
+      "loss": 0.7287,
+      "step": 9306
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6039512451871376e-06,
+      "loss": 0.7214,
+      "step": 9307
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.602143232249502e-06,
+      "loss": 0.7369,
+      "step": 9308
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6003357533272456e-06,
+      "loss": 0.7,
+      "step": 9309
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5985288085508354e-06,
+      "loss": 0.6441,
+      "step": 9310
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5967223980507116e-06,
+      "loss": 0.7579,
+      "step": 9311
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.594916521957268e-06,
+      "loss": 0.7474,
+      "step": 9312
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.593111180400858e-06,
+      "loss": 0.7724,
+      "step": 9313
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.591306373511807e-06,
+      "loss": 0.7261,
+      "step": 9314
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5895021014203937e-06,
+      "loss": 0.7708,
+      "step": 9315
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.587698364256854e-06,
+      "loss": 0.6668,
+      "step": 9316
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5858951621513984e-06,
+      "loss": 0.7575,
+      "step": 9317
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5840924952341885e-06,
+      "loss": 0.6986,
+      "step": 9318
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.582290363635349e-06,
+      "loss": 0.7257,
+      "step": 9319
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5804887674849655e-06,
+      "loss": 0.7472,
+      "step": 9320
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.578687706913091e-06,
+      "loss": 0.7315,
+      "step": 9321
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.576887182049734e-06,
+      "loss": 0.7706,
+      "step": 9322
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.575087193024861e-06,
+      "loss": 0.709,
+      "step": 9323
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.573287739968412e-06,
+      "loss": 0.7338,
+      "step": 9324
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5714888230102763e-06,
+      "loss": 0.7731,
+      "step": 9325
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.569690442280307e-06,
+      "loss": 0.686,
+      "step": 9326
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5678925979083257e-06,
+      "loss": 0.7911,
+      "step": 9327
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5660952900241052e-06,
+      "loss": 0.7729,
+      "step": 9328
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.56429851875739e-06,
+      "loss": 0.7217,
+      "step": 9329
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5625022842378734e-06,
+      "loss": 0.6717,
+      "step": 9330
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.560706586595223e-06,
+      "loss": 0.6969,
+      "step": 9331
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.55891142595906e-06,
+      "loss": 0.7292,
+      "step": 9332
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5571168024589633e-06,
+      "loss": 0.7484,
+      "step": 9333
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5553227162244855e-06,
+      "loss": 0.7893,
+      "step": 9334
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5535291673851304e-06,
+      "loss": 0.7391,
+      "step": 9335
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.55173615607036e-06,
+      "loss": 0.694,
+      "step": 9336
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5499436824096126e-06,
+      "loss": 0.6686,
+      "step": 9337
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5481517465322726e-06,
+      "loss": 0.7744,
+      "step": 9338
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.546360348567689e-06,
+      "loss": 0.4079,
+      "step": 9339
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.54456948864518e-06,
+      "loss": 0.756,
+      "step": 9340
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5427791668940162e-06,
+      "loss": 0.7987,
+      "step": 9341
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.5409893834434294e-06,
+      "loss": 0.7352,
+      "step": 9342
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5392001384226208e-06,
+      "loss": 0.7532,
+      "step": 9343
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5374114319607447e-06,
+      "loss": 0.7438,
+      "step": 9344
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5356232641869194e-06,
+      "loss": 0.7229,
+      "step": 9345
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5338356352302206e-06,
+      "loss": 0.7404,
+      "step": 9346
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5320485452196952e-06,
+      "loss": 0.8153,
+      "step": 9347
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5302619942843385e-06,
+      "loss": 0.7408,
+      "step": 9348
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.528475982553118e-06,
+      "loss": 0.7559,
+      "step": 9349
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.526690510154951e-06,
+      "loss": 0.7462,
+      "step": 9350
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.52490557721873e-06,
+      "loss": 0.7731,
+      "step": 9351
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5231211838732926e-06,
+      "loss": 0.7943,
+      "step": 9352
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.521337330247453e-06,
+      "loss": 0.7128,
+      "step": 9353
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.519554016469975e-06,
+      "loss": 0.7265,
+      "step": 9354
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5177712426695854e-06,
+      "loss": 0.837,
+      "step": 9355
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.515989008974978e-06,
+      "loss": 0.7807,
+      "step": 9356
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.514207315514803e-06,
+      "loss": 0.8012,
+      "step": 9357
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5124261624176682e-06,
+      "loss": 0.7802,
+      "step": 9358
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.510645549812152e-06,
+      "loss": 0.771,
+      "step": 9359
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.508865477826784e-06,
+      "loss": 0.7873,
+      "step": 9360
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.507085946590062e-06,
+      "loss": 0.7738,
+      "step": 9361
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.505306956230437e-06,
+      "loss": 0.7828,
+      "step": 9362
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5035285068763315e-06,
+      "loss": 0.8055,
+      "step": 9363
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5017505986561207e-06,
+      "loss": 0.759,
+      "step": 9364
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4999732316981396e-06,
+      "loss": 0.7246,
+      "step": 9365
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.498196406130695e-06,
+      "loss": 0.7972,
+      "step": 9366
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4964201220820426e-06,
+      "loss": 0.6757,
+      "step": 9367
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4946443796804008e-06,
+      "loss": 0.7404,
+      "step": 9368
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4928691790539573e-06,
+      "loss": 0.7861,
+      "step": 9369
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4910945203308547e-06,
+      "loss": 0.7483,
+      "step": 9370
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4893204036391927e-06,
+      "loss": 0.7498,
+      "step": 9371
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4875468291070436e-06,
+      "loss": 0.7637,
+      "step": 9372
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4857737968624276e-06,
+      "loss": 0.7618,
+      "step": 9373
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4840013070333325e-06,
+      "loss": 0.7833,
+      "step": 9374
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.482229359747703e-06,
+      "loss": 0.7587,
+      "step": 9375
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.480457955133453e-06,
+      "loss": 0.8018,
+      "step": 9376
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.478687093318448e-06,
+      "loss": 0.7125,
+      "step": 9377
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4769167744305155e-06,
+      "loss": 0.7188,
+      "step": 9378
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.475146998597453e-06,
+      "loss": 0.7285,
+      "step": 9379
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4733777659470083e-06,
+      "loss": 0.7078,
+      "step": 9380
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4716090766068903e-06,
+      "loss": 0.786,
+      "step": 9381
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.469840930704779e-06,
+      "loss": 0.7505,
+      "step": 9382
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4680733283683047e-06,
+      "loss": 0.7391,
+      "step": 9383
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4663062697250583e-06,
+      "loss": 0.7231,
+      "step": 9384
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.464539754902603e-06,
+      "loss": 0.3899,
+      "step": 9385
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4627737840284503e-06,
+      "loss": 0.7422,
+      "step": 9386
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.461008357230078e-06,
+      "loss": 0.6831,
+      "step": 9387
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4592434746349203e-06,
+      "loss": 0.722,
+      "step": 9388
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4574791363703786e-06,
+      "loss": 0.7322,
+      "step": 9389
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.455715342563815e-06,
+      "loss": 0.7706,
+      "step": 9390
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.453952093342543e-06,
+      "loss": 0.7614,
+      "step": 9391
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.452189388833849e-06,
+      "loss": 0.7283,
+      "step": 9392
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4504272291649722e-06,
+      "loss": 0.7751,
+      "step": 9393
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4486656144631093e-06,
+      "loss": 0.3959,
+      "step": 9394
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.44690454485543e-06,
+      "loss": 0.7982,
+      "step": 9395
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4451440204690548e-06,
+      "loss": 0.6515,
+      "step": 9396
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.443384041431063e-06,
+      "loss": 0.7722,
+      "step": 9397
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4416246078685057e-06,
+      "loss": 0.8192,
+      "step": 9398
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4398657199083852e-06,
+      "loss": 0.3521,
+      "step": 9399
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4381073776776666e-06,
+      "loss": 0.716,
+      "step": 9400
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4363495813032734e-06,
+      "loss": 0.7493,
+      "step": 9401
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.434592330912099e-06,
+      "loss": 0.79,
+      "step": 9402
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.432835626630986e-06,
+      "loss": 0.7497,
+      "step": 9403
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.431079468586741e-06,
+      "loss": 0.7382,
+      "step": 9404
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.429323856906137e-06,
+      "loss": 0.7134,
+      "step": 9405
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.427568791715902e-06,
+      "loss": 0.7638,
+      "step": 9406
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4258142731427214e-06,
+      "loss": 0.7325,
+      "step": 9407
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4240603013132515e-06,
+      "loss": 0.7533,
+      "step": 9408
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.422306876354097e-06,
+      "loss": 0.6997,
+      "step": 9409
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4205539983918357e-06,
+      "loss": 0.6936,
+      "step": 9410
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4188016675529925e-06,
+      "loss": 0.7447,
+      "step": 9411
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4170498839640668e-06,
+      "loss": 0.7045,
+      "step": 9412
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.415298647751506e-06,
+      "loss": 0.7436,
+      "step": 9413
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4135479590417234e-06,
+      "loss": 0.7587,
+      "step": 9414
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.411797817961097e-06,
+      "loss": 0.749,
+      "step": 9415
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.410048224635958e-06,
+      "loss": 0.7535,
+      "step": 9416
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.408299179192599e-06,
+      "loss": 0.7844,
+      "step": 9417
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.40655068175728e-06,
+      "loss": 0.8022,
+      "step": 9418
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.404802732456214e-06,
+      "loss": 0.7213,
+      "step": 9419
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4030553314155736e-06,
+      "loss": 0.6999,
+      "step": 9420
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4013084787615028e-06,
+      "loss": 0.7228,
+      "step": 9421
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3995621746200926e-06,
+      "loss": 0.6791,
+      "step": 9422
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3978164191174e-06,
+      "loss": 0.677,
+      "step": 9423
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.396071212379447e-06,
+      "loss": 0.7422,
+      "step": 9424
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.394326554532209e-06,
+      "loss": 0.6879,
+      "step": 9425
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3925824457016244e-06,
+      "loss": 0.7226,
+      "step": 9426
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.390838886013589e-06,
+      "loss": 0.7198,
+      "step": 9427
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3890958755939676e-06,
+      "loss": 0.7835,
+      "step": 9428
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.387353414568574e-06,
+      "loss": 0.678,
+      "step": 9429
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.385611503063194e-06,
+      "loss": 0.7723,
+      "step": 9430
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3838701412035625e-06,
+      "loss": 0.7605,
+      "step": 9431
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3821293291153837e-06,
+      "loss": 0.7463,
+      "step": 9432
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3803890669243146e-06,
+      "loss": 0.7699,
+      "step": 9433
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.378649354755982e-06,
+      "loss": 0.693,
+      "step": 9434
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.376910192735963e-06,
+      "loss": 0.7502,
+      "step": 9435
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.375171580989797e-06,
+      "loss": 0.7314,
+      "step": 9436
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3734335196429926e-06,
+      "loss": 0.7538,
+      "step": 9437
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3716960088210083e-06,
+      "loss": 0.7627,
+      "step": 9438
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3699590486492663e-06,
+      "loss": 0.7339,
+      "step": 9439
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.368222639253147e-06,
+      "loss": 0.7026,
+      "step": 9440
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3664867807579984e-06,
+      "loss": 0.3796,
+      "step": 9441
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.364751473289122e-06,
+      "loss": 0.6935,
+      "step": 9442
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3630167169717778e-06,
+      "loss": 0.7309,
+      "step": 9443
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3612825119311945e-06,
+      "loss": 0.7037,
+      "step": 9444
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.359548858292554e-06,
+      "loss": 0.6954,
+      "step": 9445
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3578157561809966e-06,
+      "loss": 0.6904,
+      "step": 9446
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3560832057216333e-06,
+      "loss": 0.739,
+      "step": 9447
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.354351207039526e-06,
+      "loss": 0.7841,
+      "step": 9448
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3526197602596946e-06,
+      "loss": 0.7632,
+      "step": 9449
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3508888655071304e-06,
+      "loss": 0.7031,
+      "step": 9450
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3491585229067725e-06,
+      "loss": 0.3644,
+      "step": 9451
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.347428732583533e-06,
+      "loss": 0.7417,
+      "step": 9452
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.345699494662269e-06,
+      "loss": 0.7443,
+      "step": 9453
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.343970809267814e-06,
+      "loss": 0.7413,
+      "step": 9454
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3422426765249483e-06,
+      "loss": 0.7827,
+      "step": 9455
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3405150965584158e-06,
+      "loss": 0.6982,
+      "step": 9456
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3387880694929276e-06,
+      "loss": 0.7329,
+      "step": 9457
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3370615954531463e-06,
+      "loss": 0.7109,
+      "step": 9458
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3353356745636944e-06,
+      "loss": 0.8161,
+      "step": 9459
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3336103069491655e-06,
+      "loss": 0.7216,
+      "step": 9460
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3318854927341005e-06,
+      "loss": 0.7398,
+      "step": 9461
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3301612320430036e-06,
+      "loss": 0.8153,
+      "step": 9462
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3284375250003457e-06,
+      "loss": 0.6972,
+      "step": 9463
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3267143717305505e-06,
+      "loss": 0.7353,
+      "step": 9464
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.324991772358004e-06,
+      "loss": 0.7408,
+      "step": 9465
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3232697270070505e-06,
+      "loss": 0.7275,
+      "step": 9466
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.321548235801999e-06,
+      "loss": 0.7534,
+      "step": 9467
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3198272988671156e-06,
+      "loss": 0.3941,
+      "step": 9468
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3181069163266222e-06,
+      "loss": 0.7645,
+      "step": 9469
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3163870883047113e-06,
+      "loss": 0.7708,
+      "step": 9470
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3146678149255232e-06,
+      "loss": 0.6627,
+      "step": 9471
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3129490963131673e-06,
+      "loss": 0.7006,
+      "step": 9472
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.311230932591707e-06,
+      "loss": 0.7382,
+      "step": 9473
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.309513323885173e-06,
+      "loss": 0.763,
+      "step": 9474
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3077962703175437e-06,
+      "loss": 0.7103,
+      "step": 9475
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.306079772012774e-06,
+      "loss": 0.7567,
+      "step": 9476
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.304363829094762e-06,
+      "loss": 0.7414,
+      "step": 9477
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.302648441687375e-06,
+      "loss": 0.7932,
+      "step": 9478
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.300933609914442e-06,
+      "loss": 0.7088,
+      "step": 9479
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.299219333899746e-06,
+      "loss": 0.7634,
+      "step": 9480
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2975056137670315e-06,
+      "loss": 0.3816,
+      "step": 9481
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.295792449640001e-06,
+      "loss": 0.7593,
+      "step": 9482
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.294079841642326e-06,
+      "loss": 0.7543,
+      "step": 9483
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.292367789897628e-06,
+      "loss": 0.7292,
+      "step": 9484
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2906562945294896e-06,
+      "loss": 0.7026,
+      "step": 9485
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2889453556614596e-06,
+      "loss": 0.7112,
+      "step": 9486
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2872349734170405e-06,
+      "loss": 0.6747,
+      "step": 9487
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.285525147919694e-06,
+      "loss": 0.7219,
+      "step": 9488
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2838158792928487e-06,
+      "loss": 0.745,
+      "step": 9489
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.282107167659886e-06,
+      "loss": 0.7378,
+      "step": 9490
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2803990131441467e-06,
+      "loss": 0.7375,
+      "step": 9491
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2786914158689412e-06,
+      "loss": 0.7637,
+      "step": 9492
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.276984375957524e-06,
+      "loss": 0.7524,
+      "step": 9493
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2752778935331266e-06,
+      "loss": 0.6696,
+      "step": 9494
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.273571968718924e-06,
+      "loss": 0.7603,
+      "step": 9495
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2718666016380653e-06,
+      "loss": 0.727,
+      "step": 9496
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.27016179241365e-06,
+      "loss": 0.7181,
+      "step": 9497
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.268457541168737e-06,
+      "loss": 0.7467,
+      "step": 9498
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.266753848026353e-06,
+      "loss": 0.7201,
+      "step": 9499
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2650507131094777e-06,
+      "loss": 0.7914,
+      "step": 9500
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.263348136541048e-06,
+      "loss": 0.8171,
+      "step": 9501
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.261646118443971e-06,
+      "loss": 0.7117,
+      "step": 9502
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2599446589411033e-06,
+      "loss": 0.8106,
+      "step": 9503
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.258243758155264e-06,
+      "loss": 0.3749,
+      "step": 9504
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.256543416209237e-06,
+      "loss": 0.7462,
+      "step": 9505
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2548436332257607e-06,
+      "loss": 0.7755,
+      "step": 9506
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2531444093275313e-06,
+      "loss": 0.7133,
+      "step": 9507
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2514457446372073e-06,
+      "loss": 0.7069,
+      "step": 9508
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.249747639277411e-06,
+      "loss": 0.7734,
+      "step": 9509
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2480500933707195e-06,
+      "loss": 0.7167,
+      "step": 9510
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.246353107039665e-06,
+      "loss": 0.7907,
+      "step": 9511
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2446566804067527e-06,
+      "loss": 0.7731,
+      "step": 9512
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2429608135944325e-06,
+      "loss": 0.764,
+      "step": 9513
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2412655067251268e-06,
+      "loss": 0.7247,
+      "step": 9514
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.239570759921206e-06,
+      "loss": 0.69,
+      "step": 9515
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2378765733050113e-06,
+      "loss": 0.7368,
+      "step": 9516
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2361829469988315e-06,
+      "loss": 0.7311,
+      "step": 9517
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.234489881124927e-06,
+      "loss": 0.7362,
+      "step": 9518
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.232797375805511e-06,
+      "loss": 0.7275,
+      "step": 9519
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2311054311627557e-06,
+      "loss": 0.7948,
+      "step": 9520
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.229414047318791e-06,
+      "loss": 0.7939,
+      "step": 9521
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.227723224395716e-06,
+      "loss": 0.745,
+      "step": 9522
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2260329625155808e-06,
+      "loss": 0.7468,
+      "step": 9523
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2243432618003933e-06,
+      "loss": 0.7513,
+      "step": 9524
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.222654122372131e-06,
+      "loss": 0.386,
+      "step": 9525
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2209655443527223e-06,
+      "loss": 0.7164,
+      "step": 9526
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.219277527864053e-06,
+      "loss": 0.7503,
+      "step": 9527
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.217590073027981e-06,
+      "loss": 0.7069,
+      "step": 9528
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2159031799663113e-06,
+      "loss": 0.6627,
+      "step": 9529
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2142168488008087e-06,
+      "loss": 0.6948,
+      "step": 9530
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2125310796532083e-06,
+      "loss": 0.7399,
+      "step": 9531
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2108458726451943e-06,
+      "loss": 0.7016,
+      "step": 9532
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.209161227898411e-06,
+      "loss": 0.6808,
+      "step": 9533
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.207477145534469e-06,
+      "loss": 0.7246,
+      "step": 9534
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2057936256749313e-06,
+      "loss": 0.663,
+      "step": 9535
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.204110668441326e-06,
+      "loss": 0.7674,
+      "step": 9536
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.202428273955133e-06,
+      "loss": 0.7031,
+      "step": 9537
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2007464423378023e-06,
+      "loss": 0.7511,
+      "step": 9538
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1990651737107327e-06,
+      "loss": 0.7412,
+      "step": 9539
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.197384468195286e-06,
+      "loss": 0.7407,
+      "step": 9540
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1957043259127885e-06,
+      "loss": 0.7748,
+      "step": 9541
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.19402474698452e-06,
+      "loss": 0.7217,
+      "step": 9542
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1923457315317166e-06,
+      "loss": 0.6905,
+      "step": 9543
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1906672796755846e-06,
+      "loss": 0.7676,
+      "step": 9544
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1889893915372818e-06,
+      "loss": 0.7987,
+      "step": 9545
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1873120672379244e-06,
+      "loss": 0.6457,
+      "step": 9546
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.185635306898589e-06,
+      "loss": 0.7374,
+      "step": 9547
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.183959110640318e-06,
+      "loss": 0.7615,
+      "step": 9548
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.182283478584106e-06,
+      "loss": 0.7259,
+      "step": 9549
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1806084108509053e-06,
+      "loss": 0.7404,
+      "step": 9550
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1789339075616357e-06,
+      "loss": 0.7957,
+      "step": 9551
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1772599688371688e-06,
+      "loss": 0.7926,
+      "step": 9552
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.175586594798337e-06,
+      "loss": 0.7536,
+      "step": 9553
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1739137855659366e-06,
+      "loss": 0.6736,
+      "step": 9554
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1722415412607146e-06,
+      "loss": 0.752,
+      "step": 9555
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1705698620033866e-06,
+      "loss": 0.7341,
+      "step": 9556
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1688987479146227e-06,
+      "loss": 0.6806,
+      "step": 9557
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1672281991150534e-06,
+      "loss": 0.7001,
+      "step": 9558
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.165558215725264e-06,
+      "loss": 0.6835,
+      "step": 9559
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.163888797865803e-06,
+      "loss": 0.7147,
+      "step": 9560
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1622199456571803e-06,
+      "loss": 0.7458,
+      "step": 9561
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1605516592198607e-06,
+      "loss": 0.7145,
+      "step": 9562
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1588839386742668e-06,
+      "loss": 0.7637,
+      "step": 9563
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.157216784140789e-06,
+      "loss": 0.7024,
+      "step": 9564
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1555501957397685e-06,
+      "loss": 0.7355,
+      "step": 9565
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.153884173591505e-06,
+      "loss": 0.7054,
+      "step": 9566
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1522187178162657e-06,
+      "loss": 0.7928,
+      "step": 9567
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.15055382853427e-06,
+      "loss": 0.7155,
+      "step": 9568
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1488895058656946e-06,
+      "loss": 0.7626,
+      "step": 9569
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1472257499306847e-06,
+      "loss": 0.7654,
+      "step": 9570
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.145562560849337e-06,
+      "loss": 0.7066,
+      "step": 9571
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.143899938741707e-06,
+      "loss": 0.7872,
+      "step": 9572
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.14223788372781e-06,
+      "loss": 0.7775,
+      "step": 9573
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.140576395927624e-06,
+      "loss": 0.7135,
+      "step": 9574
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1389154754610865e-06,
+      "loss": 0.7476,
+      "step": 9575
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1372551224480863e-06,
+      "loss": 0.7407,
+      "step": 9576
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1355953370084816e-06,
+      "loss": 0.8029,
+      "step": 9577
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.13393611926208e-06,
+      "loss": 0.7244,
+      "step": 9578
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.132277469328652e-06,
+      "loss": 0.6682,
+      "step": 9579
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.13061938732793e-06,
+      "loss": 0.3881,
+      "step": 9580
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1289618733796037e-06,
+      "loss": 0.7153,
+      "step": 9581
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1273049276033163e-06,
+      "loss": 0.7669,
+      "step": 9582
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1256485501186797e-06,
+      "loss": 0.7424,
+      "step": 9583
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.123992741045259e-06,
+      "loss": 0.7229,
+      "step": 9584
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.122337500502576e-06,
+      "loss": 0.7434,
+      "step": 9585
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1206828286101143e-06,
+      "loss": 0.7445,
+      "step": 9586
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.119028725487321e-06,
+      "loss": 0.7813,
+      "step": 9587
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1173751912535955e-06,
+      "loss": 0.7958,
+      "step": 9588
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1157222260282963e-06,
+      "loss": 0.7262,
+      "step": 9589
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.114069829930746e-06,
+      "loss": 0.7976,
+      "step": 9590
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.112418003080222e-06,
+      "loss": 0.7523,
+      "step": 9591
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1107667455959603e-06,
+      "loss": 0.7177,
+      "step": 9592
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1091160575971603e-06,
+      "loss": 0.8078,
+      "step": 9593
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.107465939202972e-06,
+      "loss": 0.7383,
+      "step": 9594
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.105816390532516e-06,
+      "loss": 0.6514,
+      "step": 9595
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1041674117048595e-06,
+      "loss": 0.7559,
+      "step": 9596
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1025190028390385e-06,
+      "loss": 0.6724,
+      "step": 9597
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1008711640540425e-06,
+      "loss": 0.728,
+      "step": 9598
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.099223895468817e-06,
+      "loss": 0.7071,
+      "step": 9599
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.097577197202277e-06,
+      "loss": 0.7704,
+      "step": 9600
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.095931069373286e-06,
+      "loss": 0.7055,
+      "step": 9601
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0942855121006677e-06,
+      "loss": 0.7748,
+      "step": 9602
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.092640525503212e-06,
+      "loss": 0.7687,
+      "step": 9603
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.09099610969966e-06,
+      "loss": 0.7245,
+      "step": 9604
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.089352264808712e-06,
+      "loss": 0.7593,
+      "step": 9605
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0877089909490332e-06,
+      "loss": 0.8088,
+      "step": 9606
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0860662882392423e-06,
+      "loss": 0.6961,
+      "step": 9607
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0844241567979138e-06,
+      "loss": 0.7826,
+      "step": 9608
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.082782596743592e-06,
+      "loss": 0.7608,
+      "step": 9609
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.08114160819477e-06,
+      "loss": 0.7075,
+      "step": 9610
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0795011912698994e-06,
+      "loss": 0.7574,
+      "step": 9611
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.077861346087401e-06,
+      "loss": 0.7099,
+      "step": 9612
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0762220727656425e-06,
+      "loss": 0.7162,
+      "step": 9613
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0745833714229536e-06,
+      "loss": 0.7143,
+      "step": 9614
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0729452421776297e-06,
+      "loss": 0.6499,
+      "step": 9615
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.071307685147913e-06,
+      "loss": 0.7433,
+      "step": 9616
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0696707004520177e-06,
+      "loss": 0.3685,
+      "step": 9617
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0680342882081027e-06,
+      "loss": 0.6817,
+      "step": 9618
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0663984485343005e-06,
+      "loss": 0.7735,
+      "step": 9619
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.064763181548689e-06,
+      "loss": 0.7779,
+      "step": 9620
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.063128487369308e-06,
+      "loss": 0.7969,
+      "step": 9621
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.061494366114165e-06,
+      "loss": 0.7559,
+      "step": 9622
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.059860817901216e-06,
+      "loss": 0.6757,
+      "step": 9623
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.058227842848376e-06,
+      "loss": 0.6842,
+      "step": 9624
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0565954410735256e-06,
+      "loss": 0.7559,
+      "step": 9625
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0549636126945005e-06,
+      "loss": 0.7733,
+      "step": 9626
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.053332357829091e-06,
+      "loss": 0.7712,
+      "step": 9627
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0517016765950495e-06,
+      "loss": 0.7015,
+      "step": 9628
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0500715691100903e-06,
+      "loss": 0.7171,
+      "step": 9629
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.048442035491881e-06,
+      "loss": 0.7523,
+      "step": 9630
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0468130758580475e-06,
+      "loss": 0.7938,
+      "step": 9631
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.045184690326182e-06,
+      "loss": 0.7251,
+      "step": 9632
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.043556879013826e-06,
+      "loss": 0.7757,
+      "step": 9633
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0419296420384816e-06,
+      "loss": 0.7462,
+      "step": 9634
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0403029795176156e-06,
+      "loss": 0.6886,
+      "step": 9635
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.038676891568645e-06,
+      "loss": 0.7501,
+      "step": 9636
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.037051378308953e-06,
+      "loss": 0.7733,
+      "step": 9637
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.035426439855872e-06,
+      "loss": 0.7602,
+      "step": 9638
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.033802076326705e-06,
+      "loss": 0.772,
+      "step": 9639
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0321782878387044e-06,
+      "loss": 0.7487,
+      "step": 9640
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0305550745090795e-06,
+      "loss": 0.7691,
+      "step": 9641
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.028932436455009e-06,
+      "loss": 0.7362,
+      "step": 9642
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0273103737936194e-06,
+      "loss": 0.719,
+      "step": 9643
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0256888866419977e-06,
+      "loss": 0.7409,
+      "step": 9644
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0240679751171965e-06,
+      "loss": 0.7183,
+      "step": 9645
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.022447639336218e-06,
+      "loss": 0.7811,
+      "step": 9646
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0208278794160243e-06,
+      "loss": 0.7066,
+      "step": 9647
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0192086954735435e-06,
+      "loss": 0.3512,
+      "step": 9648
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.017590087625654e-06,
+      "loss": 0.7673,
+      "step": 9649
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.015972055989193e-06,
+      "loss": 0.7454,
+      "step": 9650
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.014354600680961e-06,
+      "loss": 0.7302,
+      "step": 9651
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0127377218177156e-06,
+      "loss": 0.6934,
+      "step": 9652
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0111214195161687e-06,
+      "loss": 0.7074,
+      "step": 9653
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0095056938929915e-06,
+      "loss": 0.6901,
+      "step": 9654
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0078905450648202e-06,
+      "loss": 0.7621,
+      "step": 9655
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0062759731482407e-06,
+      "loss": 0.7352,
+      "step": 9656
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.004661978259804e-06,
+      "loss": 0.7374,
+      "step": 9657
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0030485605160143e-06,
+      "loss": 0.7063,
+      "step": 9658
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0014357200333388e-06,
+      "loss": 0.7047,
+      "step": 9659
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.999823456928197e-06,
+      "loss": 0.7513,
+      "step": 9660
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9982117713169756e-06,
+      "loss": 0.8082,
+      "step": 9661
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9966006633160107e-06,
+      "loss": 0.7719,
+      "step": 9662
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9949901330415987e-06,
+      "loss": 0.7905,
+      "step": 9663
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9933801806100005e-06,
+      "loss": 0.755,
+      "step": 9664
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9917708061374297e-06,
+      "loss": 0.7166,
+      "step": 9665
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9901620097400567e-06,
+      "loss": 0.7884,
+      "step": 9666
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9885537915340124e-06,
+      "loss": 0.735,
+      "step": 9667
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9869461516353893e-06,
+      "loss": 0.767,
+      "step": 9668
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9853390901602353e-06,
+      "loss": 0.7134,
+      "step": 9669
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.983732607224551e-06,
+      "loss": 0.7108,
+      "step": 9670
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.982126702944308e-06,
+      "loss": 0.7092,
+      "step": 9671
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9805213774354248e-06,
+      "loss": 0.7462,
+      "step": 9672
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.97891663081378e-06,
+      "loss": 0.6805,
+      "step": 9673
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9773124631952166e-06,
+      "loss": 0.6966,
+      "step": 9674
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.975708874695531e-06,
+      "loss": 0.7076,
+      "step": 9675
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9741058654304747e-06,
+      "loss": 0.726,
+      "step": 9676
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9725034355157656e-06,
+      "loss": 0.7194,
+      "step": 9677
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.970901585067072e-06,
+      "loss": 0.7343,
+      "step": 9678
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9693003142000277e-06,
+      "loss": 0.7864,
+      "step": 9679
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.967699623030216e-06,
+      "loss": 0.8052,
+      "step": 9680
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.966099511673187e-06,
+      "loss": 0.7238,
+      "step": 9681
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9644999802444443e-06,
+      "loss": 0.6776,
+      "step": 9682
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9629010288594464e-06,
+      "loss": 0.7074,
+      "step": 9683
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.961302657633619e-06,
+      "loss": 0.7746,
+      "step": 9684
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9597048666823394e-06,
+      "loss": 0.7732,
+      "step": 9685
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.95810765612094e-06,
+      "loss": 0.7091,
+      "step": 9686
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.956511026064721e-06,
+      "loss": 0.7213,
+      "step": 9687
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9549149766289353e-06,
+      "loss": 0.7331,
+      "step": 9688
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.953319507928789e-06,
+      "loss": 0.7957,
+      "step": 9689
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.951724620079457e-06,
+      "loss": 0.7708,
+      "step": 9690
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9501303131960648e-06,
+      "loss": 0.7548,
+      "step": 9691
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9485365873936957e-06,
+      "loss": 0.8079,
+      "step": 9692
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9469434427873923e-06,
+      "loss": 0.7206,
+      "step": 9693
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9453508794921594e-06,
+      "loss": 0.6752,
+      "step": 9694
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.943758897622956e-06,
+      "loss": 0.7908,
+      "step": 9695
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.942167497294696e-06,
+      "loss": 0.7515,
+      "step": 9696
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.940576678622259e-06,
+      "loss": 0.7941,
+      "step": 9697
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.938986441720474e-06,
+      "loss": 0.8053,
+      "step": 9698
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9373967867041388e-06,
+      "loss": 0.7839,
+      "step": 9699
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.935807713687996e-06,
+      "loss": 0.7695,
+      "step": 9700
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9342192227867595e-06,
+      "loss": 0.7405,
+      "step": 9701
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9326313141150887e-06,
+      "loss": 0.6641,
+      "step": 9702
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.931043987787613e-06,
+      "loss": 0.8164,
+      "step": 9703
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.9294572439189095e-06,
+      "loss": 0.7921,
+      "step": 9704
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.92787108262352e-06,
+      "loss": 0.7669,
+      "step": 9705
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.926285504015938e-06,
+      "loss": 0.732,
+      "step": 9706
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.924700508210623e-06,
+      "loss": 0.7315,
+      "step": 9707
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.923116095321987e-06,
+      "loss": 0.6585,
+      "step": 9708
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9215322654643987e-06,
+      "loss": 0.7432,
+      "step": 9709
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.919949018752191e-06,
+      "loss": 0.7996,
+      "step": 9710
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.918366355299648e-06,
+      "loss": 0.77,
+      "step": 9711
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.916784275221013e-06,
+      "loss": 0.7249,
+      "step": 9712
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.915202778630494e-06,
+      "loss": 0.7138,
+      "step": 9713
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9136218656422477e-06,
+      "loss": 0.7164,
+      "step": 9714
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9120415363703914e-06,
+      "loss": 0.7324,
+      "step": 9715
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9104617909290057e-06,
+      "loss": 0.7487,
+      "step": 9716
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.908882629432123e-06,
+      "loss": 0.7257,
+      "step": 9717
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9073040519937303e-06,
+      "loss": 0.8204,
+      "step": 9718
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9057260587277858e-06,
+      "loss": 0.7414,
+      "step": 9719
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9041486497481898e-06,
+      "loss": 0.7505,
+      "step": 9720
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.902571825168814e-06,
+      "loss": 0.7565,
+      "step": 9721
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9009955851034768e-06,
+      "loss": 0.666,
+      "step": 9722
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8994199296659633e-06,
+      "loss": 0.7912,
+      "step": 9723
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8978448589700115e-06,
+      "loss": 0.6909,
+      "step": 9724
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8962703731293141e-06,
+      "loss": 0.7242,
+      "step": 9725
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8946964722575302e-06,
+      "loss": 0.7526,
+      "step": 9726
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8931231564682716e-06,
+      "loss": 0.7192,
+      "step": 9727
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.891550425875105e-06,
+      "loss": 0.7083,
+      "step": 9728
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8899782805915622e-06,
+      "loss": 0.779,
+      "step": 9729
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8884067207311262e-06,
+      "loss": 0.734,
+      "step": 9730
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8868357464072396e-06,
+      "loss": 0.7375,
+      "step": 9731
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.885265357733308e-06,
+      "loss": 0.7291,
+      "step": 9732
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8836955548226854e-06,
+      "loss": 0.7604,
+      "step": 9733
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.882126337788691e-06,
+      "loss": 0.7833,
+      "step": 9734
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8805577067445946e-06,
+      "loss": 0.7368,
+      "step": 9735
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8789896618036342e-06,
+      "loss": 0.7399,
+      "step": 9736
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8774222030789958e-06,
+      "loss": 0.777,
+      "step": 9737
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8758553306838257e-06,
+      "loss": 0.6793,
+      "step": 9738
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8742890447312324e-06,
+      "loss": 0.7539,
+      "step": 9739
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.872723345334274e-06,
+      "loss": 0.7108,
+      "step": 9740
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.871158232605973e-06,
+      "loss": 0.785,
+      "step": 9741
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8695937066593096e-06,
+      "loss": 0.7483,
+      "step": 9742
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8680297676072178e-06,
+      "loss": 0.8112,
+      "step": 9743
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8664664155625877e-06,
+      "loss": 0.7633,
+      "step": 9744
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8649036506382745e-06,
+      "loss": 0.7221,
+      "step": 9745
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8633414729470856e-06,
+      "loss": 0.3539,
+      "step": 9746
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8617798826017851e-06,
+      "loss": 0.6813,
+      "step": 9747
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8602188797150955e-06,
+      "loss": 0.7379,
+      "step": 9748
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.858658464399703e-06,
+      "loss": 0.6897,
+      "step": 9749
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8570986367682442e-06,
+      "loss": 0.7417,
+      "step": 9750
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8555393969333113e-06,
+      "loss": 0.7068,
+      "step": 9751
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8539807450074644e-06,
+      "loss": 0.7607,
+      "step": 9752
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8524226811032131e-06,
+      "loss": 0.7715,
+      "step": 9753
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8508652053330223e-06,
+      "loss": 0.7313,
+      "step": 9754
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8493083178093252e-06,
+      "loss": 0.7542,
+      "step": 9755
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.847752018644503e-06,
+      "loss": 0.743,
+      "step": 9756
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.846196307950895e-06,
+      "loss": 0.7043,
+      "step": 9757
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8446411858408042e-06,
+      "loss": 0.7654,
+      "step": 9758
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8430866524264869e-06,
+      "loss": 0.7507,
+      "step": 9759
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8415327078201528e-06,
+      "loss": 0.7454,
+      "step": 9760
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8399793521339771e-06,
+      "loss": 0.7603,
+      "step": 9761
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.838426585480092e-06,
+      "loss": 0.6951,
+      "step": 9762
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8368744079705803e-06,
+      "loss": 0.7611,
+      "step": 9763
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8353228197174855e-06,
+      "loss": 0.7318,
+      "step": 9764
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8337718208328125e-06,
+      "loss": 0.7379,
+      "step": 9765
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8322214114285187e-06,
+      "loss": 0.7628,
+      "step": 9766
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8306715916165185e-06,
+      "loss": 0.7759,
+      "step": 9767
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8291223615086906e-06,
+      "loss": 0.7833,
+      "step": 9768
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8275737212168632e-06,
+      "loss": 0.7658,
+      "step": 9769
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8260256708528223e-06,
+      "loss": 0.783,
+      "step": 9770
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.824478210528321e-06,
+      "loss": 0.7627,
+      "step": 9771
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8229313403550596e-06,
+      "loss": 0.6914,
+      "step": 9772
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.821385060444698e-06,
+      "loss": 0.7894,
+      "step": 9773
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8198393709088534e-06,
+      "loss": 0.7095,
+      "step": 9774
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8182942718591069e-06,
+      "loss": 0.7379,
+      "step": 9775
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8167497634069886e-06,
+      "loss": 0.6772,
+      "step": 9776
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8152058456639865e-06,
+      "loss": 0.7301,
+      "step": 9777
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8136625187415535e-06,
+      "loss": 0.7398,
+      "step": 9778
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8121197827510895e-06,
+      "loss": 0.7224,
+      "step": 9779
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.810577637803963e-06,
+      "loss": 0.7498,
+      "step": 9780
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8090360840114885e-06,
+      "loss": 0.7335,
+      "step": 9781
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.807495121484949e-06,
+      "loss": 0.7639,
+      "step": 9782
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8059547503355734e-06,
+      "loss": 0.8034,
+      "step": 9783
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8044149706745583e-06,
+      "loss": 0.6709,
+      "step": 9784
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8028757826130504e-06,
+      "loss": 0.7607,
+      "step": 9785
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8013371862621576e-06,
+      "loss": 0.728,
+      "step": 9786
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7997991817329397e-06,
+      "loss": 0.7411,
+      "step": 9787
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7982617691364235e-06,
+      "loss": 0.7646,
+      "step": 9788
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7967249485835847e-06,
+      "loss": 0.6912,
+      "step": 9789
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7951887201853557e-06,
+      "loss": 0.8023,
+      "step": 9790
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7936530840526367e-06,
+      "loss": 0.7256,
+      "step": 9791
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7921180402962724e-06,
+      "loss": 0.7636,
+      "step": 9792
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7905835890270696e-06,
+      "loss": 0.7737,
+      "step": 9793
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7890497303557963e-06,
+      "loss": 0.7998,
+      "step": 9794
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.787516464393173e-06,
+      "loss": 0.815,
+      "step": 9795
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7859837912498767e-06,
+      "loss": 0.7299,
+      "step": 9796
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7844517110365479e-06,
+      "loss": 0.8219,
+      "step": 9797
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7829202238637766e-06,
+      "loss": 0.784,
+      "step": 9798
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7813893298421125e-06,
+      "loss": 0.7458,
+      "step": 9799
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7798590290820684e-06,
+      "loss": 0.7505,
+      "step": 9800
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7783293216941034e-06,
+      "loss": 0.744,
+      "step": 9801
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7768002077886437e-06,
+      "loss": 0.7629,
+      "step": 9802
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7752716874760656e-06,
+      "loss": 0.7749,
+      "step": 9803
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7737437608667108e-06,
+      "loss": 0.7145,
+      "step": 9804
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7722164280708677e-06,
+      "loss": 0.7103,
+      "step": 9805
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7706896891987879e-06,
+      "loss": 0.6789,
+      "step": 9806
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7691635443606813e-06,
+      "loss": 0.7626,
+      "step": 9807
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.767637993666712e-06,
+      "loss": 0.7436,
+      "step": 9808
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7661130372269996e-06,
+      "loss": 0.7978,
+      "step": 9809
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7645886751516283e-06,
+      "loss": 0.6966,
+      "step": 9810
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.763064907550631e-06,
+      "loss": 0.7182,
+      "step": 9811
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7615417345340024e-06,
+      "loss": 0.7093,
+      "step": 9812
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7600191562116887e-06,
+      "loss": 0.7639,
+      "step": 9813
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.758497172693604e-06,
+      "loss": 0.6973,
+      "step": 9814
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7569757840896107e-06,
+      "loss": 0.7257,
+      "step": 9815
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7554549905095265e-06,
+      "loss": 0.7533,
+      "step": 9816
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7539347920631356e-06,
+      "loss": 0.7945,
+      "step": 9817
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7524151888601715e-06,
+      "loss": 0.7799,
+      "step": 9818
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7508961810103241e-06,
+      "loss": 0.7631,
+      "step": 9819
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7493777686232493e-06,
+      "loss": 0.6836,
+      "step": 9820
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7478599518085481e-06,
+      "loss": 0.7425,
+      "step": 9821
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7463427306757886e-06,
+      "loss": 0.7159,
+      "step": 9822
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7448261053344884e-06,
+      "loss": 0.7504,
+      "step": 9823
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.743310075894129e-06,
+      "loss": 0.7312,
+      "step": 9824
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.741794642464142e-06,
+      "loss": 0.704,
+      "step": 9825
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7402798051539182e-06,
+      "loss": 0.7286,
+      "step": 9826
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7387655640728118e-06,
+      "loss": 0.7599,
+      "step": 9827
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7372519193301242e-06,
+      "loss": 0.7783,
+      "step": 9828
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7357388710351163e-06,
+      "loss": 0.7647,
+      "step": 9829
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7342264192970137e-06,
+      "loss": 0.7234,
+      "step": 9830
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7327145642249887e-06,
+      "loss": 0.6574,
+      "step": 9831
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7312033059281741e-06,
+      "loss": 0.7096,
+      "step": 9832
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7296926445156636e-06,
+      "loss": 0.804,
+      "step": 9833
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7281825800965035e-06,
+      "loss": 0.7176,
+      "step": 9834
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7266731127796943e-06,
+      "loss": 0.685,
+      "step": 9835
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7251642426742021e-06,
+      "loss": 0.3699,
+      "step": 9836
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7236559698889443e-06,
+      "loss": 0.774,
+      "step": 9837
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.722148294532794e-06,
+      "loss": 0.738,
+      "step": 9838
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7206412167145804e-06,
+      "loss": 0.7611,
+      "step": 9839
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7191347365430976e-06,
+      "loss": 0.7518,
+      "step": 9840
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7176288541270868e-06,
+      "loss": 0.7111,
+      "step": 9841
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7161235695752542e-06,
+      "loss": 0.7908,
+      "step": 9842
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7146188829962551e-06,
+      "loss": 0.7327,
+      "step": 9843
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7131147944987091e-06,
+      "loss": 0.6966,
+      "step": 9844
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7116113041911853e-06,
+      "loss": 0.7746,
+      "step": 9845
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7101084121822175e-06,
+      "loss": 0.7528,
+      "step": 9846
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7086061185802904e-06,
+      "loss": 0.6997,
+      "step": 9847
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.707104423493845e-06,
+      "loss": 0.7632,
+      "step": 9848
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7056033270312844e-06,
+      "loss": 0.7819,
+      "step": 9849
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7041028293009653e-06,
+      "loss": 0.6663,
+      "step": 9850
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7026029304112013e-06,
+      "loss": 0.7222,
+      "step": 9851
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7011036304702589e-06,
+      "loss": 0.6874,
+      "step": 9852
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6996049295863704e-06,
+      "loss": 0.7256,
+      "step": 9853
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.698106827867717e-06,
+      "loss": 0.789,
+      "step": 9854
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.696609325422439e-06,
+      "loss": 0.7871,
+      "step": 9855
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6951124223586368e-06,
+      "loss": 0.7715,
+      "step": 9856
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6936161187843635e-06,
+      "loss": 0.7848,
+      "step": 9857
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6921204148076264e-06,
+      "loss": 0.7394,
+      "step": 9858
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6906253105363979e-06,
+      "loss": 0.764,
+      "step": 9859
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6891308060786004e-06,
+      "loss": 0.7,
+      "step": 9860
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6876369015421124e-06,
+      "loss": 0.7408,
+      "step": 9861
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6861435970347772e-06,
+      "loss": 0.7623,
+      "step": 9862
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6846508926643835e-06,
+      "loss": 0.7931,
+      "step": 9863
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.683158788538687e-06,
+      "loss": 0.7045,
+      "step": 9864
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6816672847653902e-06,
+      "loss": 0.7256,
+      "step": 9865
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6801763814521644e-06,
+      "loss": 0.7526,
+      "step": 9866
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6786860787066262e-06,
+      "loss": 0.7736,
+      "step": 9867
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.677196376636353e-06,
+      "loss": 0.6904,
+      "step": 9868
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6757072753488811e-06,
+      "loss": 0.6938,
+      "step": 9869
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6742187749517024e-06,
+      "loss": 0.7844,
+      "step": 9870
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6727308755522587e-06,
+      "loss": 0.7556,
+      "step": 9871
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6712435772579615e-06,
+      "loss": 0.7214,
+      "step": 9872
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6697568801761688e-06,
+      "loss": 0.7451,
+      "step": 9873
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6682707844141954e-06,
+      "loss": 0.6923,
+      "step": 9874
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6667852900793192e-06,
+      "loss": 0.7211,
+      "step": 9875
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6653003972787706e-06,
+      "loss": 0.7627,
+      "step": 9876
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6638161061197322e-06,
+      "loss": 0.7167,
+      "step": 9877
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6623324167093535e-06,
+      "loss": 0.6945,
+      "step": 9878
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6608493291547334e-06,
+      "loss": 0.7032,
+      "step": 9879
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6593668435629263e-06,
+      "loss": 0.6908,
+      "step": 9880
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6578849600409453e-06,
+      "loss": 0.716,
+      "step": 9881
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6564036786957649e-06,
+      "loss": 0.7211,
+      "step": 9882
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6549229996343075e-06,
+      "loss": 0.7893,
+      "step": 9883
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6534429229634584e-06,
+      "loss": 0.7713,
+      "step": 9884
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.651963448790055e-06,
+      "loss": 0.766,
+      "step": 9885
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6504845772208966e-06,
+      "loss": 0.7402,
+      "step": 9886
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6490063083627317e-06,
+      "loss": 0.7088,
+      "step": 9887
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6475286423222736e-06,
+      "loss": 0.7594,
+      "step": 9888
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.646051579206186e-06,
+      "loss": 0.7263,
+      "step": 9889
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.644575119121088e-06,
+      "loss": 0.7353,
+      "step": 9890
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.643099262173563e-06,
+      "loss": 0.7201,
+      "step": 9891
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.641624008470144e-06,
+      "loss": 0.7333,
+      "step": 9892
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6401493581173222e-06,
+      "loss": 0.6913,
+      "step": 9893
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6386753112215425e-06,
+      "loss": 0.7404,
+      "step": 9894
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6372018678892155e-06,
+      "loss": 0.7694,
+      "step": 9895
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.635729028226698e-06,
+      "loss": 0.7519,
+      "step": 9896
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6342567923403052e-06,
+      "loss": 0.7898,
+      "step": 9897
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6327851603363153e-06,
+      "loss": 0.754,
+      "step": 9898
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6313141323209558e-06,
+      "loss": 0.7878,
+      "step": 9899
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6298437084004115e-06,
+      "loss": 0.6951,
+      "step": 9900
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6283738886808287e-06,
+      "loss": 0.6754,
+      "step": 9901
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6269046732683059e-06,
+      "loss": 0.6729,
+      "step": 9902
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6254360622688947e-06,
+      "loss": 0.7003,
+      "step": 9903
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6239680557886117e-06,
+      "loss": 0.782,
+      "step": 9904
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6225006539334221e-06,
+      "loss": 0.7257,
+      "step": 9905
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6210338568092532e-06,
+      "loss": 0.7536,
+      "step": 9906
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6195676645219827e-06,
+      "loss": 0.7282,
+      "step": 9907
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6181020771774513e-06,
+      "loss": 0.721,
+      "step": 9908
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6166370948814526e-06,
+      "loss": 0.65,
+      "step": 9909
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6151727177397315e-06,
+      "loss": 0.7905,
+      "step": 9910
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6137089458580002e-06,
+      "loss": 0.7622,
+      "step": 9911
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6122457793419188e-06,
+      "loss": 0.6554,
+      "step": 9912
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.610783218297105e-06,
+      "loss": 0.8297,
+      "step": 9913
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6093212628291367e-06,
+      "loss": 0.7994,
+      "step": 9914
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6078599130435445e-06,
+      "loss": 0.688,
+      "step": 9915
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6063991690458125e-06,
+      "loss": 0.737,
+      "step": 9916
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6049390309413903e-06,
+      "loss": 0.7609,
+      "step": 9917
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.603479498835676e-06,
+      "loss": 0.7005,
+      "step": 9918
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6020205728340265e-06,
+      "loss": 0.3806,
+      "step": 9919
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6005622530417508e-06,
+      "loss": 0.7606,
+      "step": 9920
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5991045395641236e-06,
+      "loss": 0.7397,
+      "step": 9921
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5976474325063674e-06,
+      "loss": 0.6843,
+      "step": 9922
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5961909319736624e-06,
+      "loss": 0.7641,
+      "step": 9923
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.594735038071149e-06,
+      "loss": 0.6883,
+      "step": 9924
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5932797509039176e-06,
+      "loss": 0.7641,
+      "step": 9925
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5918250705770211e-06,
+      "loss": 0.7339,
+      "step": 9926
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5903709971954683e-06,
+      "loss": 0.7371,
+      "step": 9927
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.588917530864218e-06,
+      "loss": 0.7773,
+      "step": 9928
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.587464671688187e-06,
+      "loss": 0.7504,
+      "step": 9929
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.586012419772256e-06,
+      "loss": 0.7507,
+      "step": 9930
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5845607752212522e-06,
+      "loss": 0.7364,
+      "step": 9931
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.583109738139964e-06,
+      "loss": 0.7553,
+      "step": 9932
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.581659308633131e-06,
+      "loss": 0.7323,
+      "step": 9933
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5802094868054585e-06,
+      "loss": 0.6563,
+      "step": 9934
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5787602727615992e-06,
+      "loss": 0.7262,
+      "step": 9935
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5773116666061627e-06,
+      "loss": 0.7611,
+      "step": 9936
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5758636684437211e-06,
+      "loss": 0.7533,
+      "step": 9937
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5744162783787975e-06,
+      "loss": 0.8018,
+      "step": 9938
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5729694965158682e-06,
+      "loss": 0.7638,
+      "step": 9939
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.571523322959374e-06,
+      "loss": 0.6821,
+      "step": 9940
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.570077757813705e-06,
+      "loss": 0.7505,
+      "step": 9941
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5686328011832075e-06,
+      "loss": 0.741,
+      "step": 9942
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5671884531721903e-06,
+      "loss": 0.721,
+      "step": 9943
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.5657447138849125e-06,
+      "loss": 0.6262,
+      "step": 9944
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.564301583425587e-06,
+      "loss": 0.7607,
+      "step": 9945
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5628590618983885e-06,
+      "loss": 0.708,
+      "step": 9946
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5614171494074494e-06,
+      "loss": 0.7089,
+      "step": 9947
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5599758460568515e-06,
+      "loss": 0.7977,
+      "step": 9948
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.558535151950633e-06,
+      "loss": 0.7615,
+      "step": 9949
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5570950671927943e-06,
+      "loss": 0.7985,
+      "step": 9950
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5556555918872873e-06,
+      "loss": 0.7845,
+      "step": 9951
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.554216726138018e-06,
+      "loss": 0.7077,
+      "step": 9952
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5527784700488558e-06,
+      "loss": 0.7784,
+      "step": 9953
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5513408237236194e-06,
+      "loss": 0.7156,
+      "step": 9954
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5499037872660817e-06,
+      "loss": 0.7826,
+      "step": 9955
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5484673607799817e-06,
+      "loss": 0.7355,
+      "step": 9956
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5470315443690042e-06,
+      "loss": 0.7294,
+      "step": 9957
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5455963381367955e-06,
+      "loss": 0.6795,
+      "step": 9958
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5441617421869536e-06,
+      "loss": 0.7122,
+      "step": 9959
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.542727756623038e-06,
+      "loss": 0.7536,
+      "step": 9960
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5412943815485604e-06,
+      "loss": 0.7434,
+      "step": 9961
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5398616170669866e-06,
+      "loss": 0.8043,
+      "step": 9962
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5384294632817454e-06,
+      "loss": 0.8103,
+      "step": 9963
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5369979202962148e-06,
+      "loss": 0.7524,
+      "step": 9964
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.535566988213728e-06,
+      "loss": 0.7316,
+      "step": 9965
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5341366671375802e-06,
+      "loss": 0.8187,
+      "step": 9966
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5327069571710207e-06,
+      "loss": 0.7437,
+      "step": 9967
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5312778584172505e-06,
+      "loss": 0.7451,
+      "step": 9968
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5298493709794315e-06,
+      "loss": 0.7063,
+      "step": 9969
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.528421494960679e-06,
+      "loss": 0.3732,
+      "step": 9970
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.526994230464064e-06,
+      "loss": 0.7647,
+      "step": 9971
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5255675775926104e-06,
+      "loss": 0.694,
+      "step": 9972
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5241415364493072e-06,
+      "loss": 0.7533,
+      "step": 9973
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5227161071370899e-06,
+      "loss": 0.8105,
+      "step": 9974
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5212912897588528e-06,
+      "loss": 0.6714,
+      "step": 9975
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.519867084417449e-06,
+      "loss": 0.6722,
+      "step": 9976
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5184434912156843e-06,
+      "loss": 0.7544,
+      "step": 9977
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5170205102563184e-06,
+      "loss": 0.7558,
+      "step": 9978
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5155981416420729e-06,
+      "loss": 0.6897,
+      "step": 9979
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.514176385475621e-06,
+      "loss": 0.7102,
+      "step": 9980
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5127552418595893e-06,
+      "loss": 0.6917,
+      "step": 9981
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.511334710896567e-06,
+      "loss": 0.75,
+      "step": 9982
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5099147926890946e-06,
+      "loss": 0.7523,
+      "step": 9983
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5084954873396662e-06,
+      "loss": 0.7605,
+      "step": 9984
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5070767949507382e-06,
+      "loss": 0.7927,
+      "step": 9985
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.505658715624716e-06,
+      "loss": 0.7643,
+      "step": 9986
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.504241249463968e-06,
+      "loss": 0.8014,
+      "step": 9987
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.502824396570809e-06,
+      "loss": 0.7348,
+      "step": 9988
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5014081570475192e-06,
+      "loss": 0.7663,
+      "step": 9989
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4999925309963293e-06,
+      "loss": 0.775,
+      "step": 9990
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.498577518519423e-06,
+      "loss": 0.7705,
+      "step": 9991
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4971631197189484e-06,
+      "loss": 0.7986,
+      "step": 9992
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.495749334697001e-06,
+      "loss": 0.7886,
+      "step": 9993
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4943361635556318e-06,
+      "loss": 0.7662,
+      "step": 9994
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4929236063968589e-06,
+      "loss": 0.7667,
+      "step": 9995
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4915116633226424e-06,
+      "loss": 0.7636,
+      "step": 9996
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.490100334434902e-06,
+      "loss": 0.7479,
+      "step": 9997
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4886896198355205e-06,
+      "loss": 0.7307,
+      "step": 9998
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4872795196263267e-06,
+      "loss": 0.7442,
+      "step": 9999
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4858700339091093e-06,
+      "loss": 0.7916,
+      "step": 10000
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4844611627856109e-06,
+      "loss": 0.7692,
+      "step": 10001
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.483052906357535e-06,
+      "loss": 0.7933,
+      "step": 10002
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4816452647265345e-06,
+      "loss": 0.6858,
+      "step": 10003
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4802382379942182e-06,
+      "loss": 0.7356,
+      "step": 10004
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.478831826262156e-06,
+      "loss": 0.7324,
+      "step": 10005
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4774260296318666e-06,
+      "loss": 0.7256,
+      "step": 10006
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4760208482048322e-06,
+      "loss": 0.7486,
+      "step": 10007
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4746162820824805e-06,
+      "loss": 0.7697,
+      "step": 10008
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.473212331366205e-06,
+      "loss": 0.8015,
+      "step": 10009
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4718089961573468e-06,
+      "loss": 0.7994,
+      "step": 10010
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4704062765572092e-06,
+      "loss": 0.7874,
+      "step": 10011
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4690041726670456e-06,
+      "loss": 0.7659,
+      "step": 10012
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4676026845880676e-06,
+      "loss": 0.7932,
+      "step": 10013
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4662018124214394e-06,
+      "loss": 0.6938,
+      "step": 10014
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4648015562682872e-06,
+      "loss": 0.7356,
+      "step": 10015
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4634019162296876e-06,
+      "loss": 0.7727,
+      "step": 10016
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4620028924066699e-06,
+      "loss": 0.7845,
+      "step": 10017
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4606044849002277e-06,
+      "loss": 0.7612,
+      "step": 10018
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4592066938113049e-06,
+      "loss": 0.7075,
+      "step": 10019
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.457809519240796e-06,
+      "loss": 0.7043,
+      "step": 10020
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.456412961289564e-06,
+      "loss": 0.7351,
+      "step": 10021
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4550170200584145e-06,
+      "loss": 0.709,
+      "step": 10022
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4536216956481132e-06,
+      "loss": 0.7413,
+      "step": 10023
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4522269881593852e-06,
+      "loss": 0.7375,
+      "step": 10024
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4508328976929066e-06,
+      "loss": 0.7482,
+      "step": 10025
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.449439424349307e-06,
+      "loss": 0.7381,
+      "step": 10026
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4480465682291788e-06,
+      "loss": 0.7067,
+      "step": 10027
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4466543294330626e-06,
+      "loss": 0.365,
+      "step": 10028
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.445262708061459e-06,
+      "loss": 0.7762,
+      "step": 10029
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.443871704214821e-06,
+      "loss": 0.7078,
+      "step": 10030
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4424813179935604e-06,
+      "loss": 0.7726,
+      "step": 10031
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4410915494980427e-06,
+      "loss": 0.7967,
+      "step": 10032
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.439702398828583e-06,
+      "loss": 0.7606,
+      "step": 10033
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4383138660854645e-06,
+      "loss": 0.7754,
+      "step": 10034
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.436925951368916e-06,
+      "loss": 0.6892,
+      "step": 10035
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4355386547791205e-06,
+      "loss": 0.6748,
+      "step": 10036
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4341519764162272e-06,
+      "loss": 0.7763,
+      "step": 10037
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4327659163803298e-06,
+      "loss": 0.721,
+      "step": 10038
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.431380474771481e-06,
+      "loss": 0.7257,
+      "step": 10039
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4299956516896873e-06,
+      "loss": 0.6991,
+      "step": 10040
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4286114472349177e-06,
+      "loss": 0.7894,
+      "step": 10041
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4272278615070877e-06,
+      "loss": 0.7605,
+      "step": 10042
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4258448946060698e-06,
+      "loss": 0.7757,
+      "step": 10043
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4244625466316985e-06,
+      "loss": 0.7422,
+      "step": 10044
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4230808176837563e-06,
+      "loss": 0.7447,
+      "step": 10045
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.421699707861981e-06,
+      "loss": 0.68,
+      "step": 10046
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4203192172660729e-06,
+      "loss": 0.7374,
+      "step": 10047
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4189393459956792e-06,
+      "loss": 0.7545,
+      "step": 10048
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4175600941504086e-06,
+      "loss": 0.8125,
+      "step": 10049
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4161814618298209e-06,
+      "loss": 0.7567,
+      "step": 10050
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4148034491334351e-06,
+      "loss": 0.7515,
+      "step": 10051
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4134260561607215e-06,
+      "loss": 0.8017,
+      "step": 10052
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4120492830111054e-06,
+      "loss": 0.7201,
+      "step": 10053
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4106731297839749e-06,
+      "loss": 0.7499,
+      "step": 10054
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4092975965786636e-06,
+      "loss": 0.6753,
+      "step": 10055
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4079226834944637e-06,
+      "loss": 0.7027,
+      "step": 10056
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4065483906306277e-06,
+      "loss": 0.7072,
+      "step": 10057
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4051747180863562e-06,
+      "loss": 0.7824,
+      "step": 10058
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4038016659608066e-06,
+      "loss": 0.7088,
+      "step": 10059
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4024292343530976e-06,
+      "loss": 0.7676,
+      "step": 10060
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4010574233622953e-06,
+      "loss": 0.7004,
+      "step": 10061
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3996862330874217e-06,
+      "loss": 0.8009,
+      "step": 10062
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3983156636274619e-06,
+      "loss": 0.78,
+      "step": 10063
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.396945715081347e-06,
+      "loss": 0.6968,
+      "step": 10064
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.3955763875479677e-06,
+      "loss": 0.7766,
+      "step": 10065
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3942076811261674e-06,
+      "loss": 0.7139,
+      "step": 10066
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.39283959591475e-06,
+      "loss": 0.7771,
+      "step": 10067
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.391472132012468e-06,
+      "loss": 0.743,
+      "step": 10068
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3901052895180333e-06,
+      "loss": 0.7611,
+      "step": 10069
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3887390685301105e-06,
+      "loss": 0.368,
+      "step": 10070
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3873734691473228e-06,
+      "loss": 0.7305,
+      "step": 10071
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3860084914682438e-06,
+      "loss": 0.7911,
+      "step": 10072
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3846441355914064e-06,
+      "loss": 0.742,
+      "step": 10073
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3832804016152956e-06,
+      "loss": 0.7908,
+      "step": 10074
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3819172896383515e-06,
+      "loss": 0.7262,
+      "step": 10075
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3805547997589752e-06,
+      "loss": 0.7651,
+      "step": 10076
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3791929320755137e-06,
+      "loss": 0.7184,
+      "step": 10077
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3778316866862762e-06,
+      "loss": 0.7692,
+      "step": 10078
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3764710636895206e-06,
+      "loss": 0.7411,
+      "step": 10079
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3751110631834686e-06,
+      "loss": 0.7638,
+      "step": 10080
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.373751685266289e-06,
+      "loss": 0.7255,
+      "step": 10081
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3723929300361083e-06,
+      "loss": 0.7817,
+      "step": 10082
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3710347975910109e-06,
+      "loss": 0.3614,
+      "step": 10083
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3696772880290333e-06,
+      "loss": 0.7812,
+      "step": 10084
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3683204014481644e-06,
+      "loss": 0.7146,
+      "step": 10085
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.366964137946355e-06,
+      "loss": 0.3998,
+      "step": 10086
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3656084976215067e-06,
+      "loss": 0.7299,
+      "step": 10087
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3642534805714735e-06,
+      "loss": 0.7342,
+      "step": 10088
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3628990868940706e-06,
+      "loss": 0.7686,
+      "step": 10089
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.361545316687063e-06,
+      "loss": 0.7183,
+      "step": 10090
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3601921700481768e-06,
+      "loss": 0.7069,
+      "step": 10091
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.358839647075083e-06,
+      "loss": 0.7218,
+      "step": 10092
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3574877478654191e-06,
+      "loss": 0.6806,
+      "step": 10093
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3561364725167703e-06,
+      "loss": 0.7581,
+      "step": 10094
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3547858211266762e-06,
+      "loss": 0.7347,
+      "step": 10095
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.353435793792638e-06,
+      "loss": 0.7042,
+      "step": 10096
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.352086390612105e-06,
+      "loss": 0.7198,
+      "step": 10097
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3507376116824832e-06,
+      "loss": 0.7479,
+      "step": 10098
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3493894571011367e-06,
+      "loss": 0.3299,
+      "step": 10099
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3480419269653822e-06,
+      "loss": 0.7245,
+      "step": 10100
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3466950213724882e-06,
+      "loss": 0.7887,
+      "step": 10101
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3453487404196862e-06,
+      "loss": 0.7144,
+      "step": 10102
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3440030842041541e-06,
+      "loss": 0.7296,
+      "step": 10103
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3426580528230293e-06,
+      "loss": 0.6584,
+      "step": 10104
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.341313646373401e-06,
+      "loss": 0.7279,
+      "step": 10105
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3399698649523196e-06,
+      "loss": 0.6968,
+      "step": 10106
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.338626708656784e-06,
+      "loss": 0.7849,
+      "step": 10107
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3372841775837476e-06,
+      "loss": 0.7282,
+      "step": 10108
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3359422718301263e-06,
+      "loss": 0.7386,
+      "step": 10109
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3346009914927805e-06,
+      "loss": 0.6929,
+      "step": 10110
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.333260336668536e-06,
+      "loss": 0.7667,
+      "step": 10111
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.331920307454162e-06,
+      "loss": 0.7403,
+      "step": 10112
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3305809039463957e-06,
+      "loss": 0.7696,
+      "step": 10113
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3292421262419165e-06,
+      "loss": 0.7893,
+      "step": 10114
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3279039744373678e-06,
+      "loss": 0.7367,
+      "step": 10115
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.326566448629344e-06,
+      "loss": 0.7278,
+      "step": 10116
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3252295489143895e-06,
+      "loss": 0.7613,
+      "step": 10117
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3238932753890166e-06,
+      "loss": 0.7671,
+      "step": 10118
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.32255762814968e-06,
+      "loss": 0.7888,
+      "step": 10119
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3212226072927936e-06,
+      "loss": 0.719,
+      "step": 10120
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3198882129147239e-06,
+      "loss": 0.7284,
+      "step": 10121
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3185544451117992e-06,
+      "loss": 0.7022,
+      "step": 10122
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3172213039802962e-06,
+      "loss": 0.7455,
+      "step": 10123
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.315888789616444e-06,
+      "loss": 0.7558,
+      "step": 10124
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3145569021164361e-06,
+      "loss": 0.3806,
+      "step": 10125
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3132256415764133e-06,
+      "loss": 0.7559,
+      "step": 10126
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3118950080924687e-06,
+      "loss": 0.7262,
+      "step": 10127
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3105650017606597e-06,
+      "loss": 0.8182,
+      "step": 10128
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3092356226769908e-06,
+      "loss": 0.7083,
+      "step": 10129
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.307906870937422e-06,
+      "loss": 0.7107,
+      "step": 10130
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.30657874663787e-06,
+      "loss": 0.7254,
+      "step": 10131
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3052512498742086e-06,
+      "loss": 0.771,
+      "step": 10132
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.303924380742262e-06,
+      "loss": 0.7406,
+      "step": 10133
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3025981393378062e-06,
+      "loss": 0.7405,
+      "step": 10134
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3012725257565829e-06,
+      "loss": 0.779,
+      "step": 10135
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2999475400942785e-06,
+      "loss": 0.7579,
+      "step": 10136
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2986231824465346e-06,
+      "loss": 0.7501,
+      "step": 10137
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2972994529089544e-06,
+      "loss": 0.7705,
+      "step": 10138
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2959763515770906e-06,
+      "loss": 0.7505,
+      "step": 10139
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2946538785464492e-06,
+      "loss": 0.7231,
+      "step": 10140
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2933320339124955e-06,
+      "loss": 0.7368,
+      "step": 10141
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.292010817770647e-06,
+      "loss": 0.7696,
+      "step": 10142
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2906902302162726e-06,
+      "loss": 0.7668,
+      "step": 10143
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2893702713447042e-06,
+      "loss": 0.8196,
+      "step": 10144
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2880509412512198e-06,
+      "loss": 0.806,
+      "step": 10145
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2867322400310566e-06,
+      "loss": 0.7165,
+      "step": 10146
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.285414167779403e-06,
+      "loss": 0.6943,
+      "step": 10147
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2840967245914082e-06,
+      "loss": 0.7893,
+      "step": 10148
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2827799105621696e-06,
+      "loss": 0.771,
+      "step": 10149
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2814637257867402e-06,
+      "loss": 0.73,
+      "step": 10150
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2801481703601303e-06,
+      "loss": 0.79,
+      "step": 10151
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2788332443773066e-06,
+      "loss": 0.7308,
+      "step": 10152
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2775189479331818e-06,
+      "loss": 0.7298,
+      "step": 10153
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2762052811226344e-06,
+      "loss": 0.7495,
+      "step": 10154
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2748922440404876e-06,
+      "loss": 0.672,
+      "step": 10155
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2735798367815222e-06,
+      "loss": 0.7558,
+      "step": 10156
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.272268059440479e-06,
+      "loss": 0.7731,
+      "step": 10157
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2709569121120457e-06,
+      "loss": 0.7,
+      "step": 10158
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2696463948908688e-06,
+      "loss": 0.7529,
+      "step": 10159
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2683365078715458e-06,
+      "loss": 0.6911,
+      "step": 10160
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2670272511486337e-06,
+      "loss": 0.7229,
+      "step": 10161
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.265718624816642e-06,
+      "loss": 0.7708,
+      "step": 10162
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2644106289700297e-06,
+      "loss": 0.7467,
+      "step": 10163
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2631032637032204e-06,
+      "loss": 0.7225,
+      "step": 10164
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2617965291105826e-06,
+      "loss": 0.7836,
+      "step": 10165
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2604904252864436e-06,
+      "loss": 0.7505,
+      "step": 10166
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2591849523250866e-06,
+      "loss": 0.818,
+      "step": 10167
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.257880110320746e-06,
+      "loss": 0.7234,
+      "step": 10168
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2565758993676103e-06,
+      "loss": 0.7618,
+      "step": 10169
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2552723195598281e-06,
+      "loss": 0.7307,
+      "step": 10170
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2539693709914936e-06,
+      "loss": 0.7035,
+      "step": 10171
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2526670537566665e-06,
+      "loss": 0.789,
+      "step": 10172
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2513653679493476e-06,
+      "loss": 0.7304,
+      "step": 10173
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.250064313663506e-06,
+      "loss": 0.8058,
+      "step": 10174
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2487638909930567e-06,
+      "loss": 0.7793,
+      "step": 10175
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2474641000318655e-06,
+      "loss": 0.7104,
+      "step": 10176
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2461649408737652e-06,
+      "loss": 0.7142,
+      "step": 10177
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2448664136125333e-06,
+      "loss": 0.714,
+      "step": 10178
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.243568518341901e-06,
+      "loss": 0.7663,
+      "step": 10179
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2422712551555626e-06,
+      "loss": 0.7567,
+      "step": 10180
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2409746241471577e-06,
+      "loss": 0.7051,
+      "step": 10181
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2396786254102823e-06,
+      "loss": 0.7261,
+      "step": 10182
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2383832590384926e-06,
+      "loss": 0.7432,
+      "step": 10183
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2370885251252929e-06,
+      "loss": 0.7464,
+      "step": 10184
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2357944237641439e-06,
+      "loss": 0.7805,
+      "step": 10185
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2345009550484577e-06,
+      "loss": 0.7477,
+      "step": 10186
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2332081190716084e-06,
+      "loss": 0.6898,
+      "step": 10187
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2319159159269167e-06,
+      "loss": 0.7125,
+      "step": 10188
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2306243457076584e-06,
+      "loss": 0.7085,
+      "step": 10189
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2293334085070706e-06,
+      "loss": 0.7704,
+      "step": 10190
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.228043104418336e-06,
+      "loss": 0.7561,
+      "step": 10191
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2267534335345977e-06,
+      "loss": 0.7102,
+      "step": 10192
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2254643959489487e-06,
+      "loss": 0.7087,
+      "step": 10193
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2241759917544405e-06,
+      "loss": 0.7537,
+      "step": 10194
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.222888221044074e-06,
+      "loss": 0.7981,
+      "step": 10195
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2216010839108117e-06,
+      "loss": 0.7012,
+      "step": 10196
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2203145804475624e-06,
+      "loss": 0.3612,
+      "step": 10197
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2190287107471932e-06,
+      "loss": 0.7678,
+      "step": 10198
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2177434749025218e-06,
+      "loss": 0.6982,
+      "step": 10199
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2164588730063298e-06,
+      "loss": 0.7485,
+      "step": 10200
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2151749051513417e-06,
+      "loss": 0.7619,
+      "step": 10201
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2138915714302391e-06,
+      "loss": 0.8034,
+      "step": 10202
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2126088719356656e-06,
+      "loss": 0.6576,
+      "step": 10203
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.211326806760209e-06,
+      "loss": 0.694,
+      "step": 10204
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2100453759964138e-06,
+      "loss": 0.7399,
+      "step": 10205
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2087645797367854e-06,
+      "loss": 0.723,
+      "step": 10206
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2074844180737743e-06,
+      "loss": 0.7904,
+      "step": 10207
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2062048910997892e-06,
+      "loss": 0.7221,
+      "step": 10208
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.204925998907197e-06,
+      "loss": 0.7014,
+      "step": 10209
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2036477415883108e-06,
+      "loss": 0.7666,
+      "step": 10210
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2023701192354009e-06,
+      "loss": 0.6975,
+      "step": 10211
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.201093131940696e-06,
+      "loss": 0.7064,
+      "step": 10212
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1998167797963722e-06,
+      "loss": 0.7194,
+      "step": 10213
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1985410628945681e-06,
+      "loss": 0.7729,
+      "step": 10214
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1972659813273656e-06,
+      "loss": 0.7231,
+      "step": 10215
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1959915351868122e-06,
+      "loss": 0.7751,
+      "step": 10216
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1947177245649012e-06,
+      "loss": 0.6788,
+      "step": 10217
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1934445495535818e-06,
+      "loss": 0.7755,
+      "step": 10218
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1921720102447598e-06,
+      "loss": 0.7293,
+      "step": 10219
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.190900106730295e-06,
+      "loss": 0.7193,
+      "step": 10220
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.189628839101995e-06,
+      "loss": 0.743,
+      "step": 10221
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.188358207451632e-06,
+      "loss": 0.7507,
+      "step": 10222
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1870882118709247e-06,
+      "loss": 0.7339,
+      "step": 10223
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.185818852451548e-06,
+      "loss": 0.7464,
+      "step": 10224
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1845501292851281e-06,
+      "loss": 0.7678,
+      "step": 10225
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.183282042463253e-06,
+      "loss": 0.7611,
+      "step": 10226
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1820145920774561e-06,
+      "loss": 0.6728,
+      "step": 10227
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1807477782192279e-06,
+      "loss": 0.7748,
+      "step": 10228
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.179481600980017e-06,
+      "loss": 0.7886,
+      "step": 10229
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1782160604512204e-06,
+      "loss": 0.7261,
+      "step": 10230
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.176951156724191e-06,
+      "loss": 0.7818,
+      "step": 10231
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1756868898902374e-06,
+      "loss": 0.7508,
+      "step": 10232
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.174423260040618e-06,
+      "loss": 0.7306,
+      "step": 10233
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1731602672665532e-06,
+      "loss": 0.7913,
+      "step": 10234
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1718979116592067e-06,
+      "loss": 0.7982,
+      "step": 10235
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1706361933097065e-06,
+      "loss": 0.7772,
+      "step": 10236
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1693751123091256e-06,
+      "loss": 0.3494,
+      "step": 10237
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1681146687485011e-06,
+      "loss": 0.7495,
+      "step": 10238
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1668548627188137e-06,
+      "loss": 0.7516,
+      "step": 10239
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1655956943110046e-06,
+      "loss": 0.6624,
+      "step": 10240
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1643371636159639e-06,
+      "loss": 0.7578,
+      "step": 10241
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.163079270724543e-06,
+      "loss": 0.7842,
+      "step": 10242
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1618220157275418e-06,
+      "loss": 0.7791,
+      "step": 10243
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1605653987157117e-06,
+      "loss": 0.7821,
+      "step": 10244
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1593094197797684e-06,
+      "loss": 0.7019,
+      "step": 10245
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1580540790103712e-06,
+      "loss": 0.3954,
+      "step": 10246
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1567993764981356e-06,
+      "loss": 0.7476,
+      "step": 10247
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1555453123336357e-06,
+      "loss": 0.694,
+      "step": 10248
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.154291886607396e-06,
+      "loss": 0.7639,
+      "step": 10249
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1530390994098905e-06,
+      "loss": 0.7531,
+      "step": 10250
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1517869508315583e-06,
+      "loss": 0.7139,
+      "step": 10251
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1505354409627834e-06,
+      "loss": 0.7097,
+      "step": 10252
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1492845698939037e-06,
+      "loss": 0.7222,
+      "step": 10253
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.148034337715218e-06,
+      "loss": 0.7559,
+      "step": 10254
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.14678474451697e-06,
+      "loss": 0.7137,
+      "step": 10255
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1455357903893661e-06,
+      "loss": 0.7335,
+      "step": 10256
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1442874754225574e-06,
+      "loss": 0.6833,
+      "step": 10257
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.143039799706659e-06,
+      "loss": 0.7723,
+      "step": 10258
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1417927633317317e-06,
+      "loss": 0.6896,
+      "step": 10259
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1405463663877914e-06,
+      "loss": 0.7111,
+      "step": 10260
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1393006089648128e-06,
+      "loss": 0.7744,
+      "step": 10261
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1380554911527209e-06,
+      "loss": 0.7736,
+      "step": 10262
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1368110130413902e-06,
+      "loss": 0.7224,
+      "step": 10263
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1355671747206598e-06,
+      "loss": 0.7463,
+      "step": 10264
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1343239762803127e-06,
+      "loss": 0.7799,
+      "step": 10265
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.133081417810089e-06,
+      "loss": 0.7302,
+      "step": 10266
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1318394993996828e-06,
+      "loss": 0.7889,
+      "step": 10267
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1305982211387457e-06,
+      "loss": 0.7157,
+      "step": 10268
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.129357583116877e-06,
+      "loss": 0.7024,
+      "step": 10269
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1281175854236304e-06,
+      "loss": 0.7368,
+      "step": 10270
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1268782281485191e-06,
+      "loss": 0.7468,
+      "step": 10271
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1256395113810049e-06,
+      "loss": 0.7961,
+      "step": 10272
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1244014352105025e-06,
+      "loss": 0.769,
+      "step": 10273
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1231639997263865e-06,
+      "loss": 0.6633,
+      "step": 10274
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1219272050179775e-06,
+      "loss": 0.7089,
+      "step": 10275
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1206910511745571e-06,
+      "loss": 0.7341,
+      "step": 10276
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1194555382853534e-06,
+      "loss": 0.8191,
+      "step": 10277
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1182206664395577e-06,
+      "loss": 0.7504,
+      "step": 10278
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.116986435726305e-06,
+      "loss": 0.7442,
+      "step": 10279
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1157528462346879e-06,
+      "loss": 0.7267,
+      "step": 10280
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1145198980537574e-06,
+      "loss": 0.7888,
+      "step": 10281
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1132875912725117e-06,
+      "loss": 0.7127,
+      "step": 10282
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1120559259799024e-06,
+      "loss": 0.6913,
+      "step": 10283
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1108249022648421e-06,
+      "loss": 0.737,
+      "step": 10284
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1095945202161894e-06,
+      "loss": 0.6686,
+      "step": 10285
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1083647799227593e-06,
+      "loss": 0.7042,
+      "step": 10286
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1071356814733236e-06,
+      "loss": 0.7458,
+      "step": 10287
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1059072249566039e-06,
+      "loss": 0.7677,
+      "step": 10288
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1046794104612734e-06,
+      "loss": 0.7644,
+      "step": 10289
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.103452238075967e-06,
+      "loss": 0.7384,
+      "step": 10290
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1022257078892662e-06,
+      "loss": 0.7517,
+      "step": 10291
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1009998199897076e-06,
+      "loss": 0.7263,
+      "step": 10292
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0997745744657806e-06,
+      "loss": 0.7809,
+      "step": 10293
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0985499714059334e-06,
+      "loss": 0.7665,
+      "step": 10294
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0973260108985594e-06,
+      "loss": 0.7469,
+      "step": 10295
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0961026930320163e-06,
+      "loss": 0.743,
+      "step": 10296
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.094880017894605e-06,
+      "loss": 0.7495,
+      "step": 10297
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0936579855745877e-06,
+      "loss": 0.7785,
+      "step": 10298
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0924365961601723e-06,
+      "loss": 0.3585,
+      "step": 10299
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.091215849739531e-06,
+      "loss": 0.6869,
+      "step": 10300
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0899957464007804e-06,
+      "loss": 0.7493,
+      "step": 10301
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0887762862319917e-06,
+      "loss": 0.7305,
+      "step": 10302
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0875574693211955e-06,
+      "loss": 0.743,
+      "step": 10303
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0863392957563712e-06,
+      "loss": 0.7612,
+      "step": 10304
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0851217656254532e-06,
+      "loss": 0.7031,
+      "step": 10305
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0839048790163254e-06,
+      "loss": 0.7924,
+      "step": 10306
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.0826886360168342e-06,
+      "loss": 0.8073,
+      "step": 10307
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0814730367147719e-06,
+      "loss": 0.671,
+      "step": 10308
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0802580811978846e-06,
+      "loss": 0.7133,
+      "step": 10309
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.079043769553879e-06,
+      "loss": 0.7147,
+      "step": 10310
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0778301018704074e-06,
+      "loss": 0.74,
+      "step": 10311
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0766170782350759e-06,
+      "loss": 0.7603,
+      "step": 10312
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0754046987354527e-06,
+      "loss": 0.6795,
+      "step": 10313
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0741929634590497e-06,
+      "loss": 0.7311,
+      "step": 10314
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0729818724933362e-06,
+      "loss": 0.7085,
+      "step": 10315
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0717714259257372e-06,
+      "loss": 0.7686,
+      "step": 10316
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0705616238436245e-06,
+      "loss": 0.7876,
+      "step": 10317
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0693524663343347e-06,
+      "loss": 0.738,
+      "step": 10318
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0681439534851446e-06,
+      "loss": 0.7081,
+      "step": 10319
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0669360853832943e-06,
+      "loss": 0.8027,
+      "step": 10320
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0657288621159744e-06,
+      "loss": 0.6945,
+      "step": 10321
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.064522283770325e-06,
+      "loss": 0.7221,
+      "step": 10322
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0633163504334476e-06,
+      "loss": 0.6827,
+      "step": 10323
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0621110621923903e-06,
+      "loss": 0.7387,
+      "step": 10324
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0609064191341556e-06,
+      "loss": 0.7919,
+      "step": 10325
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0597024213457042e-06,
+      "loss": 0.7147,
+      "step": 10326
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.058499068913944e-06,
+      "loss": 0.7399,
+      "step": 10327
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0572963619257393e-06,
+      "loss": 0.7456,
+      "step": 10328
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0560943004679103e-06,
+      "loss": 0.7622,
+      "step": 10329
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0548928846272267e-06,
+      "loss": 0.705,
+      "step": 10330
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0536921144904121e-06,
+      "loss": 0.775,
+      "step": 10331
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0524919901441422e-06,
+      "loss": 0.7217,
+      "step": 10332
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0512925116750527e-06,
+      "loss": 0.7676,
+      "step": 10333
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.050093679169727e-06,
+      "loss": 0.7551,
+      "step": 10334
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0488954927146989e-06,
+      "loss": 0.698,
+      "step": 10335
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0476979523964636e-06,
+      "loss": 0.7105,
+      "step": 10336
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0465010583014657e-06,
+      "loss": 0.7658,
+      "step": 10337
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0453048105161013e-06,
+      "loss": 0.715,
+      "step": 10338
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0441092091267247e-06,
+      "loss": 0.6899,
+      "step": 10339
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.042914254219639e-06,
+      "loss": 0.7485,
+      "step": 10340
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0417199458810989e-06,
+      "loss": 0.7722,
+      "step": 10341
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.040526284197322e-06,
+      "loss": 0.7601,
+      "step": 10342
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0393332692544688e-06,
+      "loss": 0.7578,
+      "step": 10343
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.038140901138658e-06,
+      "loss": 0.7062,
+      "step": 10344
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0369491799359589e-06,
+      "loss": 0.7239,
+      "step": 10345
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0357581057324007e-06,
+      "loss": 0.7542,
+      "step": 10346
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.034567678613958e-06,
+      "loss": 0.7199,
+      "step": 10347
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.033377898666561e-06,
+      "loss": 0.7071,
+      "step": 10348
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0321887659760975e-06,
+      "loss": 0.8198,
+      "step": 10349
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0310002806284037e-06,
+      "loss": 0.696,
+      "step": 10350
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0298124427092693e-06,
+      "loss": 0.7534,
+      "step": 10351
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0286252523044405e-06,
+      "loss": 0.7441,
+      "step": 10352
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0274387094996141e-06,
+      "loss": 0.6342,
+      "step": 10353
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0262528143804385e-06,
+      "loss": 0.7444,
+      "step": 10354
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0250675670325228e-06,
+      "loss": 0.7393,
+      "step": 10355
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0238829675414186e-06,
+      "loss": 0.7554,
+      "step": 10356
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.022699015992642e-06,
+      "loss": 0.7442,
+      "step": 10357
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0215157124716514e-06,
+      "loss": 0.7358,
+      "step": 10358
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0203330570638681e-06,
+      "loss": 0.8074,
+      "step": 10359
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0191510498546597e-06,
+      "loss": 0.6939,
+      "step": 10360
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0179696909293501e-06,
+      "loss": 0.7488,
+      "step": 10361
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0167889803732167e-06,
+      "loss": 0.7929,
+      "step": 10362
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.015608918271489e-06,
+      "loss": 0.7936,
+      "step": 10363
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0144295047093467e-06,
+      "loss": 0.6582,
+      "step": 10364
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0132507397719316e-06,
+      "loss": 0.735,
+      "step": 10365
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0120726235443302e-06,
+      "loss": 0.7118,
+      "step": 10366
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0108951561115832e-06,
+      "loss": 0.7743,
+      "step": 10367
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0097183375586893e-06,
+      "loss": 0.747,
+      "step": 10368
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0085421679705966e-06,
+      "loss": 0.7894,
+      "step": 10369
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0073666474322042e-06,
+      "loss": 0.7381,
+      "step": 10370
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0061917760283724e-06,
+      "loss": 0.8147,
+      "step": 10371
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0050175538439054e-06,
+      "loss": 0.7799,
+      "step": 10372
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.003843980963567e-06,
+      "loss": 0.6747,
+      "step": 10373
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0026710574720676e-06,
+      "loss": 0.7351,
+      "step": 10374
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0014987834540802e-06,
+      "loss": 0.7919,
+      "step": 10375
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.000327158994221e-06,
+      "loss": 0.6949,
+      "step": 10376
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.991561841770693e-07,
+      "loss": 0.718,
+      "step": 10377
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.97985859087145e-07,
+      "loss": 0.7573,
+      "step": 10378
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.968161838089352e-07,
+      "loss": 0.73,
+      "step": 10379
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.956471584268678e-07,
+      "loss": 0.7351,
+      "step": 10380
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.944787830253332e-07,
+      "loss": 0.7528,
+      "step": 10381
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.933110576886683e-07,
+      "loss": 0.8226,
+      "step": 10382
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.921439825011647e-07,
+      "loss": 0.6872,
+      "step": 10383
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.909775575470703e-07,
+      "loss": 0.7384,
+      "step": 10384
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.898117829105835e-07,
+      "loss": 0.6842,
+      "step": 10385
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.886466586758537e-07,
+      "loss": 0.7123,
+      "step": 10386
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.874821849269855e-07,
+      "loss": 0.7328,
+      "step": 10387
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.863183617480398e-07,
+      "loss": 0.6603,
+      "step": 10388
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.85155189223026e-07,
+      "loss": 0.7421,
+      "step": 10389
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.839926674359045e-07,
+      "loss": 0.777,
+      "step": 10390
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.828307964705986e-07,
+      "loss": 0.7945,
+      "step": 10391
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.81669576410973e-07,
+      "loss": 0.7862,
+      "step": 10392
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.805090073408496e-07,
+      "loss": 0.7009,
+      "step": 10393
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.793490893440095e-07,
+      "loss": 0.3823,
+      "step": 10394
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.781898225041785e-07,
+      "loss": 0.713,
+      "step": 10395
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.77031206905037e-07,
+      "loss": 0.385,
+      "step": 10396
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.758732426302232e-07,
+      "loss": 0.7725,
+      "step": 10397
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.747159297633214e-07,
+      "loss": 0.7602,
+      "step": 10398
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.735592683878758e-07,
+      "loss": 0.761,
+      "step": 10399
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.724032585873767e-07,
+      "loss": 0.3623,
+      "step": 10400
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.712479004452747e-07,
+      "loss": 0.7651,
+      "step": 10401
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.700931940449676e-07,
+      "loss": 0.7071,
+      "step": 10402
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.689391394698068e-07,
+      "loss": 0.7172,
+      "step": 10403
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.677857368030997e-07,
+      "loss": 0.7129,
+      "step": 10404
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.666329861281042e-07,
+      "loss": 0.6904,
+      "step": 10405
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.654808875280307e-07,
+      "loss": 0.6729,
+      "step": 10406
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.64329441086046e-07,
+      "loss": 0.7233,
+      "step": 10407
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.631786468852667e-07,
+      "loss": 0.8044,
+      "step": 10408
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.620285050087607e-07,
+      "loss": 0.7974,
+      "step": 10409
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.608790155395543e-07,
+      "loss": 0.7255,
+      "step": 10410
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.597301785606228e-07,
+      "loss": 0.7625,
+      "step": 10411
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.585819941548935e-07,
+      "loss": 0.7356,
+      "step": 10412
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.574344624052479e-07,
+      "loss": 0.7672,
+      "step": 10413
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.56287583394524e-07,
+      "loss": 0.7402,
+      "step": 10414
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.55141357205508e-07,
+      "loss": 0.7286,
+      "step": 10415
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.53995783920938e-07,
+      "loss": 0.7157,
+      "step": 10416
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.528508636235112e-07,
+      "loss": 0.71,
+      "step": 10417
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.517065963958705e-07,
+      "loss": 0.3416,
+      "step": 10418
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.505629823206175e-07,
+      "loss": 0.7482,
+      "step": 10419
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.494200214803017e-07,
+      "loss": 0.7627,
+      "step": 10420
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.482777139574318e-07,
+      "loss": 0.7634,
+      "step": 10421
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.471360598344604e-07,
+      "loss": 0.7415,
+      "step": 10422
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.459950591938027e-07,
+      "loss": 0.7973,
+      "step": 10423
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.448547121178197e-07,
+      "loss": 0.7497,
+      "step": 10424
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.437150186888289e-07,
+      "loss": 0.7338,
+      "step": 10425
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.425759789890942e-07,
+      "loss": 0.7381,
+      "step": 10426
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.414375931008446e-07,
+      "loss": 0.733,
+      "step": 10427
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.402998611062508e-07,
+      "loss": 0.728,
+      "step": 10428
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.391627830874384e-07,
+      "loss": 0.727,
+      "step": 10429
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.380263591264915e-07,
+      "loss": 0.7814,
+      "step": 10430
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.368905893054414e-07,
+      "loss": 0.7445,
+      "step": 10431
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.357554737062713e-07,
+      "loss": 0.7661,
+      "step": 10432
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.346210124109245e-07,
+      "loss": 0.7036,
+      "step": 10433
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.3348720550129e-07,
+      "loss": 0.6931,
+      "step": 10434
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.323540530592101e-07,
+      "loss": 0.7641,
+      "step": 10435
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.312215551664838e-07,
+      "loss": 0.6716,
+      "step": 10436
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.300897119048601e-07,
+      "loss": 0.7375,
+      "step": 10437
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.289585233560406e-07,
+      "loss": 0.7773,
+      "step": 10438
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.278279896016817e-07,
+      "loss": 0.7679,
+      "step": 10439
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.266981107233886e-07,
+      "loss": 0.7797,
+      "step": 10440
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.255688868027246e-07,
+      "loss": 0.723,
+      "step": 10441
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.244403179212002e-07,
+      "loss": 0.7003,
+      "step": 10442
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.233124041602848e-07,
+      "loss": 0.7772,
+      "step": 10443
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.221851456013952e-07,
+      "loss": 0.7822,
+      "step": 10444
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.210585423259011e-07,
+      "loss": 0.7819,
+      "step": 10445
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.199325944151294e-07,
+      "loss": 0.7007,
+      "step": 10446
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.188073019503552e-07,
+      "loss": 0.7165,
+      "step": 10447
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.176826650128057e-07,
+      "loss": 0.7334,
+      "step": 10448
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.165586836836682e-07,
+      "loss": 0.7003,
+      "step": 10449
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.154353580440745e-07,
+      "loss": 0.7296,
+      "step": 10450
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.143126881751118e-07,
+      "loss": 0.7761,
+      "step": 10451
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.131906741578201e-07,
+      "loss": 0.7634,
+      "step": 10452
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.120693160731931e-07,
+      "loss": 0.7653,
+      "step": 10453
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.109486140021761e-07,
+      "loss": 0.7265,
+      "step": 10454
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.098285680256647e-07,
+      "loss": 0.7539,
+      "step": 10455
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.087091782245138e-07,
+      "loss": 0.7177,
+      "step": 10456
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.075904446795237e-07,
+      "loss": 0.7872,
+      "step": 10457
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.064723674714504e-07,
+      "loss": 0.73,
+      "step": 10458
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.053549466810041e-07,
+      "loss": 0.7359,
+      "step": 10459
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.042381823888446e-07,
+      "loss": 0.7051,
+      "step": 10460
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.031220746755876e-07,
+      "loss": 0.7279,
+      "step": 10461
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.020066236217962e-07,
+      "loss": 0.7199,
+      "step": 10462
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.008918293079926e-07,
+      "loss": 0.7684,
+      "step": 10463
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.997776918146484e-07,
+      "loss": 0.713,
+      "step": 10464
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.986642112221833e-07,
+      "loss": 0.7346,
+      "step": 10465
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.975513876109798e-07,
+      "loss": 0.78,
+      "step": 10466
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.964392210613649e-07,
+      "loss": 0.7231,
+      "step": 10467
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.953277116536174e-07,
+      "loss": 0.7348,
+      "step": 10468
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.942168594679779e-07,
+      "loss": 0.7611,
+      "step": 10469
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.931066645846298e-07,
+      "loss": 0.7415,
+      "step": 10470
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.919971270837114e-07,
+      "loss": 0.7519,
+      "step": 10471
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.908882470453173e-07,
+      "loss": 0.7053,
+      "step": 10472
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.897800245494925e-07,
+      "loss": 0.7224,
+      "step": 10473
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.886724596762309e-07,
+      "loss": 0.6574,
+      "step": 10474
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.87565552505486e-07,
+      "loss": 0.746,
+      "step": 10475
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.864593031171587e-07,
+      "loss": 0.8378,
+      "step": 10476
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.853537115911037e-07,
+      "loss": 0.6801,
+      "step": 10477
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.842487780071263e-07,
+      "loss": 0.7509,
+      "step": 10478
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.831445024449903e-07,
+      "loss": 0.692,
+      "step": 10479
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.820408849844031e-07,
+      "loss": 0.811,
+      "step": 10480
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.809379257050344e-07,
+      "loss": 0.7109,
+      "step": 10481
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.79835624686498e-07,
+      "loss": 0.7864,
+      "step": 10482
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.787339820083662e-07,
+      "loss": 0.7681,
+      "step": 10483
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.776329977501586e-07,
+      "loss": 0.7018,
+      "step": 10484
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.765326719913525e-07,
+      "loss": 0.6998,
+      "step": 10485
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.754330048113746e-07,
+      "loss": 0.7806,
+      "step": 10486
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.743339962896013e-07,
+      "loss": 0.6834,
+      "step": 10487
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.732356465053693e-07,
+      "loss": 0.6944,
+      "step": 10488
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.721379555379617e-07,
+      "loss": 0.7135,
+      "step": 10489
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.710409234666117e-07,
+      "loss": 0.8163,
+      "step": 10490
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.69944550370515e-07,
+      "loss": 0.7429,
+      "step": 10491
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.688488363288095e-07,
+      "loss": 0.754,
+      "step": 10492
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.677537814205905e-07,
+      "loss": 0.6928,
+      "step": 10493
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.666593857249028e-07,
+      "loss": 0.8015,
+      "step": 10494
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.655656493207488e-07,
+      "loss": 0.7473,
+      "step": 10495
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.644725722870784e-07,
+      "loss": 0.7583,
+      "step": 10496
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.633801547027942e-07,
+      "loss": 0.7699,
+      "step": 10497
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.622883966467554e-07,
+      "loss": 0.7168,
+      "step": 10498
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.61197298197769e-07,
+      "loss": 0.7454,
+      "step": 10499
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.60106859434594e-07,
+      "loss": 0.7194,
+      "step": 10500
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.590170804359488e-07,
+      "loss": 0.674,
+      "step": 10501
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.579279612804947e-07,
+      "loss": 0.3531,
+      "step": 10502
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.568395020468512e-07,
+      "loss": 0.7094,
+      "step": 10503
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.55751702813592e-07,
+      "loss": 0.7354,
+      "step": 10504
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.546645636592376e-07,
+      "loss": 0.7398,
+      "step": 10505
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.535780846622632e-07,
+      "loss": 0.7213,
+      "step": 10506
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.524922659010948e-07,
+      "loss": 0.7558,
+      "step": 10507
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.514071074541153e-07,
+      "loss": 0.7832,
+      "step": 10508
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.503226093996564e-07,
+      "loss": 0.7047,
+      "step": 10509
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.492387718160011e-07,
+      "loss": 0.7147,
+      "step": 10510
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.481555947813879e-07,
+      "loss": 0.7504,
+      "step": 10511
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.470730783740066e-07,
+      "loss": 0.7423,
+      "step": 10512
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.459912226719958e-07,
+      "loss": 0.7337,
+      "step": 10513
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.449100277534528e-07,
+      "loss": 0.6991,
+      "step": 10514
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.438294936964231e-07,
+      "loss": 0.7431,
+      "step": 10515
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.427496205789021e-07,
+      "loss": 0.7609,
+      "step": 10516
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.416704084788463e-07,
+      "loss": 0.71,
+      "step": 10517
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.405918574741545e-07,
+      "loss": 0.7648,
+      "step": 10518
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.395139676426833e-07,
+      "loss": 0.6821,
+      "step": 10519
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.384367390622384e-07,
+      "loss": 0.7346,
+      "step": 10520
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.373601718105829e-07,
+      "loss": 0.7094,
+      "step": 10521
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.362842659654257e-07,
+      "loss": 0.7265,
+      "step": 10522
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.35209021604434e-07,
+      "loss": 0.7087,
+      "step": 10523
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.341344388052253e-07,
+      "loss": 0.7694,
+      "step": 10524
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.330605176453655e-07,
+      "loss": 0.8412,
+      "step": 10525
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.319872582023769e-07,
+      "loss": 0.7144,
+      "step": 10526
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.309146605537333e-07,
+      "loss": 0.7094,
+      "step": 10527
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.298427247768604e-07,
+      "loss": 0.7314,
+      "step": 10528
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.28771450949134e-07,
+      "loss": 0.7145,
+      "step": 10529
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.27700839147888e-07,
+      "loss": 0.7968,
+      "step": 10530
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.266308894504027e-07,
+      "loss": 0.7205,
+      "step": 10531
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.255616019339119e-07,
+      "loss": 0.7375,
+      "step": 10532
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.244929766756005e-07,
+      "loss": 0.7031,
+      "step": 10533
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.234250137526123e-07,
+      "loss": 0.6826,
+      "step": 10534
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.223577132420357e-07,
+      "loss": 0.7609,
+      "step": 10535
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.212910752209114e-07,
+      "loss": 0.7211,
+      "step": 10536
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.202250997662398e-07,
+      "loss": 0.7746,
+      "step": 10537
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.19159786954965e-07,
+      "loss": 0.6768,
+      "step": 10538
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.180951368639867e-07,
+      "loss": 0.7539,
+      "step": 10539
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.170311495701599e-07,
+      "loss": 0.7281,
+      "step": 10540
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.159678251502845e-07,
+      "loss": 0.7592,
+      "step": 10541
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.149051636811211e-07,
+      "loss": 0.7368,
+      "step": 10542
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.138431652393741e-07,
+      "loss": 0.7472,
+      "step": 10543
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.127818299017076e-07,
+      "loss": 0.7231,
+      "step": 10544
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.117211577447326e-07,
+      "loss": 0.7975,
+      "step": 10545
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.106611488450111e-07,
+      "loss": 0.7684,
+      "step": 10546
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.096018032790653e-07,
+      "loss": 0.6582,
+      "step": 10547
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.085431211233607e-07,
+      "loss": 0.7442,
+      "step": 10548
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.074851024543185e-07,
+      "loss": 0.6564,
+      "step": 10549
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.06427747348314e-07,
+      "loss": 0.7261,
+      "step": 10550
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.053710558816718e-07,
+      "loss": 0.7106,
+      "step": 10551
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.043150281306666e-07,
+      "loss": 0.6992,
+      "step": 10552
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.032596641715318e-07,
+      "loss": 0.789,
+      "step": 10553
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.022049640804474e-07,
+      "loss": 0.6805,
+      "step": 10554
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.01150927933545e-07,
+      "loss": 0.7103,
+      "step": 10555
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.000975558069135e-07,
+      "loss": 0.7381,
+      "step": 10556
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.990448477765911e-07,
+      "loss": 0.736,
+      "step": 10557
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.979928039185647e-07,
+      "loss": 0.7731,
+      "step": 10558
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.969414243087769e-07,
+      "loss": 0.7236,
+      "step": 10559
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.958907090231238e-07,
+      "loss": 0.7084,
+      "step": 10560
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.948406581374479e-07,
+      "loss": 0.6911,
+      "step": 10561
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.93791271727552e-07,
+      "loss": 0.7003,
+      "step": 10562
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.92742549869182e-07,
+      "loss": 0.7423,
+      "step": 10563
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.916944926380432e-07,
+      "loss": 0.7324,
+      "step": 10564
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.906471001097859e-07,
+      "loss": 0.7394,
+      "step": 10565
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.89600372360021e-07,
+      "loss": 0.3903,
+      "step": 10566
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.885543094643044e-07,
+      "loss": 0.7308,
+      "step": 10567
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.875089114981448e-07,
+      "loss": 0.7361,
+      "step": 10568
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.864641785370075e-07,
+      "loss": 0.7333,
+      "step": 10569
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.854201106563053e-07,
+      "loss": 0.6983,
+      "step": 10570
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.843767079314035e-07,
+      "loss": 0.7265,
+      "step": 10571
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.833339704376197e-07,
+      "loss": 0.7436,
+      "step": 10572
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.82291898250227e-07,
+      "loss": 0.6984,
+      "step": 10573
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.812504914444464e-07,
+      "loss": 0.7516,
+      "step": 10574
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.802097500954486e-07,
+      "loss": 0.7309,
+      "step": 10575
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.79169674278365e-07,
+      "loss": 0.7479,
+      "step": 10576
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.781302640682709e-07,
+      "loss": 0.7378,
+      "step": 10577
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.770915195401951e-07,
+      "loss": 0.6917,
+      "step": 10578
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.760534407691223e-07,
+      "loss": 0.7222,
+      "step": 10579
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.750160278299845e-07,
+      "loss": 0.6801,
+      "step": 10580
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.739792807976676e-07,
+      "loss": 0.7504,
+      "step": 10581
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.729431997470105e-07,
+      "loss": 0.6841,
+      "step": 10582
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.719077847528011e-07,
+      "loss": 0.6837,
+      "step": 10583
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.708730358897831e-07,
+      "loss": 0.7814,
+      "step": 10584
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.698389532326477e-07,
+      "loss": 0.7525,
+      "step": 10585
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.688055368560432e-07,
+      "loss": 0.7936,
+      "step": 10586
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.677727868345652e-07,
+      "loss": 0.7573,
+      "step": 10587
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.66740703242761e-07,
+      "loss": 0.7061,
+      "step": 10588
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.657092861551352e-07,
+      "loss": 0.7353,
+      "step": 10589
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.646785356461395e-07,
+      "loss": 0.7291,
+      "step": 10590
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.636484517901766e-07,
+      "loss": 0.7344,
+      "step": 10591
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.626190346616069e-07,
+      "loss": 0.7906,
+      "step": 10592
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.615902843347378e-07,
+      "loss": 0.7455,
+      "step": 10593
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.605622008838276e-07,
+      "loss": 0.7138,
+      "step": 10594
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.595347843830914e-07,
+      "loss": 0.7662,
+      "step": 10595
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.585080349066931e-07,
+      "loss": 0.7212,
+      "step": 10596
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.57481952528748e-07,
+      "loss": 0.7843,
+      "step": 10597
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.564565373233212e-07,
+      "loss": 0.7522,
+      "step": 10598
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.554317893644381e-07,
+      "loss": 0.7027,
+      "step": 10599
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.54407708726067e-07,
+      "loss": 0.7566,
+      "step": 10600
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.533842954821303e-07,
+      "loss": 0.7312,
+      "step": 10601
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.523615497065062e-07,
+      "loss": 0.6983,
+      "step": 10602
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.513394714730182e-07,
+      "loss": 0.7511,
+      "step": 10603
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.503180608554494e-07,
+      "loss": 0.7276,
+      "step": 10604
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.492973179275265e-07,
+      "loss": 0.7391,
+      "step": 10605
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.482772427629359e-07,
+      "loss": 0.664,
+      "step": 10606
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.472578354353078e-07,
+      "loss": 0.726,
+      "step": 10607
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.46239096018232e-07,
+      "loss": 0.7081,
+      "step": 10608
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.452210245852443e-07,
+      "loss": 0.7097,
+      "step": 10609
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.442036212098325e-07,
+      "loss": 0.7701,
+      "step": 10610
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.431868859654423e-07,
+      "loss": 0.7873,
+      "step": 10611
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.42170818925465e-07,
+      "loss": 0.7431,
+      "step": 10612
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.411554201632443e-07,
+      "loss": 0.7395,
+      "step": 10613
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.401406897520769e-07,
+      "loss": 0.7123,
+      "step": 10614
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.391266277652143e-07,
+      "loss": 0.694,
+      "step": 10615
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.381132342758546e-07,
+      "loss": 0.6992,
+      "step": 10616
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.37100509357147e-07,
+      "loss": 0.7394,
+      "step": 10617
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.36088453082201e-07,
+      "loss": 0.7964,
+      "step": 10618
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.35077065524068e-07,
+      "loss": 0.6762,
+      "step": 10619
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.340663467557552e-07,
+      "loss": 0.7123,
+      "step": 10620
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.330562968502231e-07,
+      "loss": 0.778,
+      "step": 10621
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.320469158803822e-07,
+      "loss": 0.7816,
+      "step": 10622
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.310382039190933e-07,
+      "loss": 0.7284,
+      "step": 10623
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.300301610391724e-07,
+      "loss": 0.7374,
+      "step": 10624
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.290227873133826e-07,
+      "loss": 0.7275,
+      "step": 10625
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.280160828144455e-07,
+      "loss": 0.7402,
+      "step": 10626
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.270100476150244e-07,
+      "loss": 0.7187,
+      "step": 10627
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.260046817877464e-07,
+      "loss": 0.7568,
+      "step": 10628
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.249999854051803e-07,
+      "loss": 0.7355,
+      "step": 10629
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.239959585398493e-07,
+      "loss": 0.7667,
+      "step": 10630
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.229926012642329e-07,
+      "loss": 0.729,
+      "step": 10631
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.219899136507558e-07,
+      "loss": 0.7305,
+      "step": 10632
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.209878957717964e-07,
+      "loss": 0.7157,
+      "step": 10633
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.199865476996892e-07,
+      "loss": 0.7399,
+      "step": 10634
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.189858695067131e-07,
+      "loss": 0.8037,
+      "step": 10635
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.179858612651036e-07,
+      "loss": 0.7389,
+      "step": 10636
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.169865230470463e-07,
+      "loss": 0.7483,
+      "step": 10637
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.159878549246801e-07,
+      "loss": 0.7839,
+      "step": 10638
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.149898569700919e-07,
+      "loss": 0.7527,
+      "step": 10639
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.139925292553218e-07,
+      "loss": 0.7227,
+      "step": 10640
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.129958718523644e-07,
+      "loss": 0.7035,
+      "step": 10641
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.119998848331622e-07,
+      "loss": 0.6712,
+      "step": 10642
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.110045682696087e-07,
+      "loss": 0.7859,
+      "step": 10643
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.100099222335566e-07,
+      "loss": 0.7009,
+      "step": 10644
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.090159467967983e-07,
+      "loss": 0.7638,
+      "step": 10645
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.080226420310888e-07,
+      "loss": 0.7124,
+      "step": 10646
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.070300080081261e-07,
+      "loss": 0.8188,
+      "step": 10647
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.060380447995663e-07,
+      "loss": 0.7168,
+      "step": 10648
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.050467524770133e-07,
+      "loss": 0.39,
+      "step": 10649
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.040561311120254e-07,
+      "loss": 0.7533,
+      "step": 10650
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.030661807761096e-07,
+      "loss": 0.7656,
+      "step": 10651
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.020769015407247e-07,
+      "loss": 0.6682,
+      "step": 10652
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.010882934772811e-07,
+      "loss": 0.3339,
+      "step": 10653
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.001003566571452e-07,
+      "loss": 0.76,
+      "step": 10654
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.991130911516298e-07,
+      "loss": 0.7392,
+      "step": 10655
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.981264970319979e-07,
+      "loss": 0.7501,
+      "step": 10656
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.971405743694715e-07,
+      "loss": 0.7488,
+      "step": 10657
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.96155323235217e-07,
+      "loss": 0.7637,
+      "step": 10658
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.95170743700354e-07,
+      "loss": 0.7807,
+      "step": 10659
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.941868358359583e-07,
+      "loss": 0.7219,
+      "step": 10660
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.932035997130504e-07,
+      "loss": 0.7271,
+      "step": 10661
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.922210354026037e-07,
+      "loss": 0.7888,
+      "step": 10662
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.912391429755494e-07,
+      "loss": 0.831,
+      "step": 10663
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.902579225027628e-07,
+      "loss": 0.7673,
+      "step": 10664
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.892773740550717e-07,
+      "loss": 0.7098,
+      "step": 10665
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.882974977032619e-07,
+      "loss": 0.7106,
+      "step": 10666
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.873182935180612e-07,
+      "loss": 0.732,
+      "step": 10667
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 6.863397615701573e-07,
+      "loss": 0.7292,
+      "step": 10668
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.853619019301816e-07,
+      "loss": 0.7714,
+      "step": 10669
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.843847146687266e-07,
+      "loss": 0.6895,
+      "step": 10670
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.834081998563269e-07,
+      "loss": 0.6748,
+      "step": 10671
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.824323575634706e-07,
+      "loss": 0.7416,
+      "step": 10672
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.814571878606036e-07,
+      "loss": 0.7801,
+      "step": 10673
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.804826908181172e-07,
+      "loss": 0.7453,
+      "step": 10674
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.79508866506352e-07,
+      "loss": 0.7185,
+      "step": 10675
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.785357149956095e-07,
+      "loss": 0.697,
+      "step": 10676
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.775632363561335e-07,
+      "loss": 0.7632,
+      "step": 10677
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.765914306581223e-07,
+      "loss": 0.7268,
+      "step": 10678
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.756202979717253e-07,
+      "loss": 0.7284,
+      "step": 10679
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.746498383670463e-07,
+      "loss": 0.769,
+      "step": 10680
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.736800519141373e-07,
+      "loss": 0.672,
+      "step": 10681
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.727109386829999e-07,
+      "loss": 0.5985,
+      "step": 10682
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.717424987435927e-07,
+      "loss": 0.7522,
+      "step": 10683
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.707747321658231e-07,
+      "loss": 0.7307,
+      "step": 10684
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.698076390195452e-07,
+      "loss": 0.7861,
+      "step": 10685
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.688412193745741e-07,
+      "loss": 0.7346,
+      "step": 10686
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.678754733006665e-07,
+      "loss": 0.7314,
+      "step": 10687
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.669104008675376e-07,
+      "loss": 0.6502,
+      "step": 10688
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.659460021448505e-07,
+      "loss": 0.7657,
+      "step": 10689
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.649822772022207e-07,
+      "loss": 0.7203,
+      "step": 10690
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.64019226109216e-07,
+      "loss": 0.6868,
+      "step": 10691
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.630568489353506e-07,
+      "loss": 0.7366,
+      "step": 10692
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.620951457500979e-07,
+      "loss": 0.7482,
+      "step": 10693
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.61134116622878e-07,
+      "loss": 0.716,
+      "step": 10694
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.601737616230585e-07,
+      "loss": 0.7617,
+      "step": 10695
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.592140808199688e-07,
+      "loss": 0.369,
+      "step": 10696
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.582550742828808e-07,
+      "loss": 0.7778,
+      "step": 10697
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.572967420810194e-07,
+      "loss": 0.7325,
+      "step": 10698
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.563390842835649e-07,
+      "loss": 0.7541,
+      "step": 10699
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.553821009596439e-07,
+      "loss": 0.3921,
+      "step": 10700
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.544257921783359e-07,
+      "loss": 0.6904,
+      "step": 10701
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.534701580086733e-07,
+      "loss": 0.6927,
+      "step": 10702
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.525151985196399e-07,
+      "loss": 0.7049,
+      "step": 10703
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.51560913780167e-07,
+      "loss": 0.7133,
+      "step": 10704
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.506073038591398e-07,
+      "loss": 0.771,
+      "step": 10705
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.496543688253976e-07,
+      "loss": 0.6538,
+      "step": 10706
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.487021087477252e-07,
+      "loss": 0.7284,
+      "step": 10707
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.477505236948623e-07,
+      "loss": 0.7853,
+      "step": 10708
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.467996137355026e-07,
+      "loss": 0.7099,
+      "step": 10709
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.458493789382836e-07,
+      "loss": 0.6715,
+      "step": 10710
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.448998193717982e-07,
+      "loss": 0.7198,
+      "step": 10711
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.439509351045936e-07,
+      "loss": 0.6603,
+      "step": 10712
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.43002726205163e-07,
+      "loss": 0.7593,
+      "step": 10713
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.420551927419516e-07,
+      "loss": 0.7637,
+      "step": 10714
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.411083347833603e-07,
+      "loss": 0.8019,
+      "step": 10715
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.401621523977375e-07,
+      "loss": 0.81,
+      "step": 10716
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.392166456533821e-07,
+      "loss": 0.7148,
+      "step": 10717
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.382718146185452e-07,
+      "loss": 0.7843,
+      "step": 10718
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.373276593614319e-07,
+      "loss": 0.6741,
+      "step": 10719
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.363841799501958e-07,
+      "loss": 0.4114,
+      "step": 10720
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.354413764529388e-07,
+      "loss": 0.7882,
+      "step": 10721
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.344992489377222e-07,
+      "loss": 0.8069,
+      "step": 10722
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.335577974725504e-07,
+      "loss": 0.7768,
+      "step": 10723
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.326170221253825e-07,
+      "loss": 0.6926,
+      "step": 10724
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.316769229641295e-07,
+      "loss": 0.7775,
+      "step": 10725
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.307375000566529e-07,
+      "loss": 0.7565,
+      "step": 10726
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.297987534707628e-07,
+      "loss": 0.7972,
+      "step": 10727
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.288606832742239e-07,
+      "loss": 0.7518,
+      "step": 10728
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.279232895347531e-07,
+      "loss": 0.7356,
+      "step": 10729
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.269865723200141e-07,
+      "loss": 0.7778,
+      "step": 10730
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.26050531697624e-07,
+      "loss": 0.8466,
+      "step": 10731
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.251151677351519e-07,
+      "loss": 0.7702,
+      "step": 10732
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.241804805001173e-07,
+      "loss": 0.7452,
+      "step": 10733
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.232464700599883e-07,
+      "loss": 0.7005,
+      "step": 10734
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.2231313648219e-07,
+      "loss": 0.6765,
+      "step": 10735
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.213804798340927e-07,
+      "loss": 0.7299,
+      "step": 10736
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.204485001830207e-07,
+      "loss": 0.6784,
+      "step": 10737
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.19517197596251e-07,
+      "loss": 0.676,
+      "step": 10738
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.18586572141009e-07,
+      "loss": 0.7285,
+      "step": 10739
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.176566238844695e-07,
+      "loss": 0.7325,
+      "step": 10740
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.167273528937645e-07,
+      "loss": 0.7634,
+      "step": 10741
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.157987592359726e-07,
+      "loss": 0.6995,
+      "step": 10742
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.148708429781224e-07,
+      "loss": 0.7379,
+      "step": 10743
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.139436041871983e-07,
+      "loss": 0.7515,
+      "step": 10744
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.130170429301318e-07,
+      "loss": 0.8107,
+      "step": 10745
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.120911592738055e-07,
+      "loss": 0.7104,
+      "step": 10746
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.111659532850589e-07,
+      "loss": 0.7379,
+      "step": 10747
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.10241425030672e-07,
+      "loss": 0.7359,
+      "step": 10748
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.093175745773883e-07,
+      "loss": 0.7125,
+      "step": 10749
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.083944019918908e-07,
+      "loss": 0.7547,
+      "step": 10750
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.07471907340823e-07,
+      "loss": 0.7257,
+      "step": 10751
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.065500906907729e-07,
+      "loss": 0.7812,
+      "step": 10752
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.056289521082815e-07,
+      "loss": 0.6874,
+      "step": 10753
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.047084916598433e-07,
+      "loss": 0.6883,
+      "step": 10754
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.037887094119021e-07,
+      "loss": 0.7704,
+      "step": 10755
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.028696054308492e-07,
+      "loss": 0.6863,
+      "step": 10756
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.019511797830335e-07,
+      "loss": 0.7831,
+      "step": 10757
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.0103343253475e-07,
+      "loss": 0.7356,
+      "step": 10758
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.001163637522478e-07,
+      "loss": 0.7142,
+      "step": 10759
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.99199973501724e-07,
+      "loss": 0.7498,
+      "step": 10760
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.982842618493301e-07,
+      "loss": 0.7031,
+      "step": 10761
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.973692288611654e-07,
+      "loss": 0.3782,
+      "step": 10762
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.964548746032817e-07,
+      "loss": 0.761,
+      "step": 10763
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.955411991416837e-07,
+      "loss": 0.7261,
+      "step": 10764
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.946282025423233e-07,
+      "loss": 0.767,
+      "step": 10765
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.937158848711044e-07,
+      "loss": 0.7095,
+      "step": 10766
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.928042461938866e-07,
+      "loss": 0.7379,
+      "step": 10767
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.918932865764726e-07,
+      "loss": 0.7482,
+      "step": 10768
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.909830060846222e-07,
+      "loss": 0.7185,
+      "step": 10769
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.900734047840439e-07,
+      "loss": 0.6656,
+      "step": 10770
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.891644827403975e-07,
+      "loss": 0.7012,
+      "step": 10771
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.882562400192937e-07,
+      "loss": 0.7745,
+      "step": 10772
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.873486766862935e-07,
+      "loss": 0.7449,
+      "step": 10773
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.864417928069099e-07,
+      "loss": 0.7822,
+      "step": 10774
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.855355884466074e-07,
+      "loss": 0.7368,
+      "step": 10775
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.84630063670798e-07,
+      "loss": 0.6733,
+      "step": 10776
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.837252185448506e-07,
+      "loss": 0.7176,
+      "step": 10777
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.828210531340783e-07,
+      "loss": 0.6575,
+      "step": 10778
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.819175675037492e-07,
+      "loss": 0.7328,
+      "step": 10779
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.810147617190842e-07,
+      "loss": 0.744,
+      "step": 10780
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.801126358452491e-07,
+      "loss": 0.7691,
+      "step": 10781
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.79211189947364e-07,
+      "loss": 0.7623,
+      "step": 10782
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.783104240905035e-07,
+      "loss": 0.816,
+      "step": 10783
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.774103383396868e-07,
+      "loss": 0.6945,
+      "step": 10784
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.765109327598872e-07,
+      "loss": 0.7301,
+      "step": 10785
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.756122074160264e-07,
+      "loss": 0.7175,
+      "step": 10786
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.747141623729824e-07,
+      "loss": 0.7533,
+      "step": 10787
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.738167976955777e-07,
+      "loss": 0.7809,
+      "step": 10788
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.729201134485928e-07,
+      "loss": 0.7433,
+      "step": 10789
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.720241096967504e-07,
+      "loss": 0.7262,
+      "step": 10790
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.711287865047322e-07,
+      "loss": 0.7497,
+      "step": 10791
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.70234143937164e-07,
+      "loss": 0.7278,
+      "step": 10792
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.693401820586286e-07,
+      "loss": 0.7756,
+      "step": 10793
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.684469009336568e-07,
+      "loss": 0.716,
+      "step": 10794
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.67554300626727e-07,
+      "loss": 0.7188,
+      "step": 10795
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.666623812022753e-07,
+      "loss": 0.7199,
+      "step": 10796
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.657711427246837e-07,
+      "loss": 0.3813,
+      "step": 10797
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.648805852582851e-07,
+      "loss": 0.6825,
+      "step": 10798
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.639907088673646e-07,
+      "loss": 0.7266,
+      "step": 10799
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.63101513616161e-07,
+      "loss": 0.8085,
+      "step": 10800
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.622129995688586e-07,
+      "loss": 0.7306,
+      "step": 10801
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.613251667895947e-07,
+      "loss": 0.7251,
+      "step": 10802
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.604380153424582e-07,
+      "loss": 0.7691,
+      "step": 10803
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.595515452914902e-07,
+      "loss": 0.7224,
+      "step": 10804
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.586657567006759e-07,
+      "loss": 0.7269,
+      "step": 10805
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.57780649633961e-07,
+      "loss": 0.7743,
+      "step": 10806
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.568962241552344e-07,
+      "loss": 0.6846,
+      "step": 10807
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.560124803283385e-07,
+      "loss": 0.685,
+      "step": 10808
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.551294182170686e-07,
+      "loss": 0.7015,
+      "step": 10809
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.542470378851661e-07,
+      "loss": 0.7144,
+      "step": 10810
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.533653393963267e-07,
+      "loss": 0.7519,
+      "step": 10811
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.524843228141963e-07,
+      "loss": 0.7128,
+      "step": 10812
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.516039882023716e-07,
+      "loss": 0.7527,
+      "step": 10813
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.507243356243996e-07,
+      "loss": 0.7632,
+      "step": 10814
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.498453651437763e-07,
+      "loss": 0.7478,
+      "step": 10815
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.489670768239519e-07,
+      "loss": 0.7461,
+      "step": 10816
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.480894707283269e-07,
+      "loss": 0.7613,
+      "step": 10817
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.472125469202472e-07,
+      "loss": 0.6679,
+      "step": 10818
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.463363054630189e-07,
+      "loss": 0.7211,
+      "step": 10819
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.454607464198913e-07,
+      "loss": 0.7832,
+      "step": 10820
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.445858698540652e-07,
+      "loss": 0.7373,
+      "step": 10821
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.437116758286965e-07,
+      "loss": 0.7169,
+      "step": 10822
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.428381644068881e-07,
+      "loss": 0.6958,
+      "step": 10823
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.41965335651693e-07,
+      "loss": 0.7172,
+      "step": 10824
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.410931896261174e-07,
+      "loss": 0.6982,
+      "step": 10825
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.402217263931186e-07,
+      "loss": 0.7255,
+      "step": 10826
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.393509460156021e-07,
+      "loss": 0.6918,
+      "step": 10827
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.384808485564241e-07,
+      "loss": 0.6682,
+      "step": 10828
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.376114340783956e-07,
+      "loss": 0.7295,
+      "step": 10829
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.367427026442718e-07,
+      "loss": 0.6919,
+      "step": 10830
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.358746543167659e-07,
+      "loss": 0.7445,
+      "step": 10831
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.350072891585356e-07,
+      "loss": 0.7639,
+      "step": 10832
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.341406072321942e-07,
+      "loss": 0.7601,
+      "step": 10833
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.332746086002993e-07,
+      "loss": 0.7175,
+      "step": 10834
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.324092933253677e-07,
+      "loss": 0.7686,
+      "step": 10835
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.315446614698605e-07,
+      "loss": 0.7601,
+      "step": 10836
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.3068071309619e-07,
+      "loss": 0.7018,
+      "step": 10837
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.298174482667207e-07,
+      "loss": 0.7592,
+      "step": 10838
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.289548670437683e-07,
+      "loss": 0.7966,
+      "step": 10839
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.280929694895997e-07,
+      "loss": 0.7323,
+      "step": 10840
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.272317556664275e-07,
+      "loss": 0.7589,
+      "step": 10841
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.263712256364218e-07,
+      "loss": 0.7066,
+      "step": 10842
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.255113794616995e-07,
+      "loss": 0.6612,
+      "step": 10843
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.246522172043267e-07,
+      "loss": 0.7666,
+      "step": 10844
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.237937389263248e-07,
+      "loss": 0.6826,
+      "step": 10845
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.22935944689662e-07,
+      "loss": 0.7743,
+      "step": 10846
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.220788345562578e-07,
+      "loss": 0.7446,
+      "step": 10847
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.212224085879835e-07,
+      "loss": 0.707,
+      "step": 10848
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.203666668466611e-07,
+      "loss": 0.7185,
+      "step": 10849
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.1951160939406e-07,
+      "loss": 0.7379,
+      "step": 10850
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.186572362919051e-07,
+      "loss": 0.7208,
+      "step": 10851
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.178035476018672e-07,
+      "loss": 0.6679,
+      "step": 10852
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.169505433855726e-07,
+      "loss": 0.7102,
+      "step": 10853
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.160982237045931e-07,
+      "loss": 0.7282,
+      "step": 10854
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.15246588620455e-07,
+      "loss": 0.7283,
+      "step": 10855
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.143956381946347e-07,
+      "loss": 0.7368,
+      "step": 10856
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.135453724885542e-07,
+      "loss": 0.7748,
+      "step": 10857
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.126957915635955e-07,
+      "loss": 0.7621,
+      "step": 10858
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.118468954810818e-07,
+      "loss": 0.7037,
+      "step": 10859
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.109986843022908e-07,
+      "loss": 0.7863,
+      "step": 10860
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.101511580884521e-07,
+      "loss": 0.7754,
+      "step": 10861
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.093043169007461e-07,
+      "loss": 0.7216,
+      "step": 10862
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.08458160800298e-07,
+      "loss": 0.7565,
+      "step": 10863
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.076126898481903e-07,
+      "loss": 0.7749,
+      "step": 10864
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.067679041054552e-07,
+      "loss": 0.7327,
+      "step": 10865
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.059238036330705e-07,
+      "loss": 0.7497,
+      "step": 10866
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.050803884919664e-07,
+      "loss": 0.7046,
+      "step": 10867
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.042376587430298e-07,
+      "loss": 0.7429,
+      "step": 10868
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.033956144470909e-07,
+      "loss": 0.6879,
+      "step": 10869
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.02554255664931e-07,
+      "loss": 0.6456,
+      "step": 10870
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.017135824572872e-07,
+      "loss": 0.8229,
+      "step": 10871
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.008735948848409e-07,
+      "loss": 0.8124,
+      "step": 10872
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.00034293008228e-07,
+      "loss": 0.7347,
+      "step": 10873
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.991956768880324e-07,
+      "loss": 0.7547,
+      "step": 10874
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.983577465847922e-07,
+      "loss": 0.6458,
+      "step": 10875
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.975205021589901e-07,
+      "loss": 0.7608,
+      "step": 10876
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.966839436710658e-07,
+      "loss": 0.7127,
+      "step": 10877
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.958480711814062e-07,
+      "loss": 0.7526,
+      "step": 10878
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.950128847503466e-07,
+      "loss": 0.7108,
+      "step": 10879
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.941783844381753e-07,
+      "loss": 0.7471,
+      "step": 10880
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.933445703051343e-07,
+      "loss": 0.7362,
+      "step": 10881
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.925114424114086e-07,
+      "loss": 0.7381,
+      "step": 10882
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.916790008171391e-07,
+      "loss": 0.7488,
+      "step": 10883
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.908472455824154e-07,
+      "loss": 0.3756,
+      "step": 10884
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.900161767672795e-07,
+      "loss": 0.7712,
+      "step": 10885
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.891857944317202e-07,
+      "loss": 0.7147,
+      "step": 10886
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.883560986356794e-07,
+      "loss": 0.7238,
+      "step": 10887
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.875270894390504e-07,
+      "loss": 0.712,
+      "step": 10888
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.866987669016709e-07,
+      "loss": 0.7744,
+      "step": 10889
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.858711310833398e-07,
+      "loss": 0.671,
+      "step": 10890
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.850441820437957e-07,
+      "loss": 0.752,
+      "step": 10891
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.842179198427322e-07,
+      "loss": 0.6929,
+      "step": 10892
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.833923445397936e-07,
+      "loss": 0.7216,
+      "step": 10893
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.825674561945759e-07,
+      "loss": 0.75,
+      "step": 10894
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.817432548666234e-07,
+      "loss": 0.8154,
+      "step": 10895
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.809197406154287e-07,
+      "loss": 0.7033,
+      "step": 10896
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.800969135004397e-07,
+      "loss": 0.7123,
+      "step": 10897
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.792747735810521e-07,
+      "loss": 0.6688,
+      "step": 10898
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.784533209166109e-07,
+      "loss": 0.7724,
+      "step": 10899
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.77632555566414e-07,
+      "loss": 0.7299,
+      "step": 10900
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.768124775897087e-07,
+      "loss": 0.6946,
+      "step": 10901
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.759930870456897e-07,
+      "loss": 0.7365,
+      "step": 10902
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7517438399350856e-07,
+      "loss": 0.7129,
+      "step": 10903
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7435636849226143e-07,
+      "loss": 0.733,
+      "step": 10904
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7353904060099767e-07,
+      "loss": 0.7194,
+      "step": 10905
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7272240037871456e-07,
+      "loss": 0.7307,
+      "step": 10906
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.719064478843638e-07,
+      "loss": 0.7542,
+      "step": 10907
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.710911831768439e-07,
+      "loss": 0.7241,
+      "step": 10908
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.7027660631500326e-07,
+      "loss": 0.7245,
+      "step": 10909
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.694627173576449e-07,
+      "loss": 0.7315,
+      "step": 10910
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6864951636351964e-07,
+      "loss": 0.7758,
+      "step": 10911
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6783700339132485e-07,
+      "loss": 0.7371,
+      "step": 10912
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6702517849971376e-07,
+      "loss": 0.7481,
+      "step": 10913
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6621404174729157e-07,
+      "loss": 0.7687,
+      "step": 10914
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6540359319260485e-07,
+      "loss": 0.6606,
+      "step": 10915
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6459383289416015e-07,
+      "loss": 0.7318,
+      "step": 10916
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.637847609104085e-07,
+      "loss": 0.684,
+      "step": 10917
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6297637729975307e-07,
+      "loss": 0.7786,
+      "step": 10918
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.621686821205451e-07,
+      "loss": 0.7295,
+      "step": 10919
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.613616754310923e-07,
+      "loss": 0.7658,
+      "step": 10920
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6055535728964486e-07,
+      "loss": 0.7041,
+      "step": 10921
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.597497277544083e-07,
+      "loss": 0.6909,
+      "step": 10922
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.589447868835384e-07,
+      "loss": 0.7598,
+      "step": 10923
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5814053473513754e-07,
+      "loss": 0.7051,
+      "step": 10924
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5733697136726154e-07,
+      "loss": 0.6756,
+      "step": 10925
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.565340968379173e-07,
+      "loss": 0.6252,
+      "step": 10926
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5573191120505955e-07,
+      "loss": 0.7343,
+      "step": 10927
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5493041452659316e-07,
+      "loss": 0.7001,
+      "step": 10928
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.541296068603751e-07,
+      "loss": 0.7366,
+      "step": 10929
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5332948826421254e-07,
+      "loss": 0.7612,
+      "step": 10930
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5253005879585923e-07,
+      "loss": 0.7477,
+      "step": 10931
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.517313185130268e-07,
+      "loss": 0.7368,
+      "step": 10932
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.50933267473368e-07,
+      "loss": 0.7855,
+      "step": 10933
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.501359057344934e-07,
+      "loss": 0.6643,
+      "step": 10934
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.493392333539581e-07,
+      "loss": 0.7199,
+      "step": 10935
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4854325038927173e-07,
+      "loss": 0.6913,
+      "step": 10936
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4774795689789375e-07,
+      "loss": 0.715,
+      "step": 10937
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.469533529372283e-07,
+      "loss": 0.7464,
+      "step": 10938
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.461594385646384e-07,
+      "loss": 0.7246,
+      "step": 10939
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.453662138374304e-07,
+      "loss": 0.7608,
+      "step": 10940
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4457367881286297e-07,
+      "loss": 0.7367,
+      "step": 10941
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4378183354814694e-07,
+      "loss": 0.7665,
+      "step": 10942
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.429906781004423e-07,
+      "loss": 0.7174,
+      "step": 10943
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.422002125268565e-07,
+      "loss": 0.6822,
+      "step": 10944
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4141043688444965e-07,
+      "loss": 0.6983,
+      "step": 10945
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4062135123023373e-07,
+      "loss": 0.6893,
+      "step": 10946
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.398329556211689e-07,
+      "loss": 0.6956,
+      "step": 10947
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3904525011416176e-07,
+      "loss": 0.7232,
+      "step": 10948
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.38258234766078e-07,
+      "loss": 0.7185,
+      "step": 10949
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3747190963372655e-07,
+      "loss": 0.6745,
+      "step": 10950
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.366862747738665e-07,
+      "loss": 0.6907,
+      "step": 10951
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.359013302432125e-07,
+      "loss": 0.8264,
+      "step": 10952
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3511707609842134e-07,
+      "loss": 0.6443,
+      "step": 10953
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.343335123961101e-07,
+      "loss": 0.7382,
+      "step": 10954
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.335506391928357e-07,
+      "loss": 0.7857,
+      "step": 10955
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.327684565451118e-07,
+      "loss": 0.7823,
+      "step": 10956
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3198696450940216e-07,
+      "loss": 0.7406,
+      "step": 10957
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3120616314211494e-07,
+      "loss": 0.7266,
+      "step": 10958
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3042605249961624e-07,
+      "loss": 0.7165,
+      "step": 10959
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.296466326382165e-07,
+      "loss": 0.75,
+      "step": 10960
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.288679036141774e-07,
+      "loss": 0.759,
+      "step": 10961
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.28089865483714e-07,
+      "loss": 0.7358,
+      "step": 10962
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2731251830298803e-07,
+      "loss": 0.7511,
+      "step": 10963
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2653586212811016e-07,
+      "loss": 0.7186,
+      "step": 10964
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.257598970151478e-07,
+      "loss": 0.7845,
+      "step": 10965
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2498462302011155e-07,
+      "loss": 0.6986,
+      "step": 10966
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.242100401989635e-07,
+      "loss": 0.3509,
+      "step": 10967
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2343614860761886e-07,
+      "loss": 0.7749,
+      "step": 10968
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.226629483019418e-07,
+      "loss": 0.7485,
+      "step": 10969
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.218904393377443e-07,
+      "loss": 0.7081,
+      "step": 10970
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.2111862177078963e-07,
+      "loss": 0.691,
+      "step": 10971
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.203474956567943e-07,
+      "loss": 0.8025,
+      "step": 10972
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1957706105141827e-07,
+      "loss": 0.7457,
+      "step": 10973
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.188073180102792e-07,
+      "loss": 0.7299,
+      "step": 10974
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.180382665889371e-07,
+      "loss": 0.7716,
+      "step": 10975
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.172699068429109e-07,
+      "loss": 0.7345,
+      "step": 10976
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1650223882766074e-07,
+      "loss": 0.7517,
+      "step": 10977
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.157352625986033e-07,
+      "loss": 0.7319,
+      "step": 10978
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1496897821110217e-07,
+      "loss": 0.7504,
+      "step": 10979
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.142033857204697e-07,
+      "loss": 0.8,
+      "step": 10980
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1343848518197395e-07,
+      "loss": 0.692,
+      "step": 10981
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.126742766508274e-07,
+      "loss": 0.6286,
+      "step": 10982
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1191076018219477e-07,
+      "loss": 0.7912,
+      "step": 10983
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.111479358311887e-07,
+      "loss": 0.7066,
+      "step": 10984
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.103858036528774e-07,
+      "loss": 0.7802,
+      "step": 10985
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.096243637022734e-07,
+      "loss": 0.739,
+      "step": 10986
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.088636160343396e-07,
+      "loss": 0.7966,
+      "step": 10987
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0810356070399513e-07,
+      "loss": 0.7918,
+      "step": 10988
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.073441977661019e-07,
+      "loss": 0.7514,
+      "step": 10989
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.065855272754737e-07,
+      "loss": 0.7503,
+      "step": 10990
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0582754928687797e-07,
+      "loss": 0.7683,
+      "step": 10991
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0507026385502747e-07,
+      "loss": 0.7541,
+      "step": 10992
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.043136710345863e-07,
+      "loss": 0.7245,
+      "step": 10993
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.035577708801719e-07,
+      "loss": 0.6892,
+      "step": 10994
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0280256344634725e-07,
+      "loss": 0.7802,
+      "step": 10995
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.020480487876277e-07,
+      "loss": 0.7443,
+      "step": 10996
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.012942269584763e-07,
+      "loss": 0.7656,
+      "step": 10997
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.005410980133118e-07,
+      "loss": 0.7603,
+      "step": 10998
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.99788662006495e-07,
+      "loss": 0.7267,
+      "step": 10999
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9903691899234155e-07,
+      "loss": 0.6969,
+      "step": 11000
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.982858690251168e-07,
+      "loss": 0.7212,
+      "step": 11001
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.975355121590363e-07,
+      "loss": 0.7411,
+      "step": 11002
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.967858484482623e-07,
+      "loss": 0.6919,
+      "step": 11003
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9603687794691146e-07,
+      "loss": 0.6446,
+      "step": 11004
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.952886007090473e-07,
+      "loss": 0.7977,
+      "step": 11005
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.945410167886832e-07,
+      "loss": 0.7367,
+      "step": 11006
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.93794126239786e-07,
+      "loss": 0.6887,
+      "step": 11007
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.930479291162692e-07,
+      "loss": 0.3891,
+      "step": 11008
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.923024254719965e-07,
+      "loss": 0.7638,
+      "step": 11009
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9155761536078365e-07,
+      "loss": 0.7885,
+      "step": 11010
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.908134988363932e-07,
+      "loss": 0.7398,
+      "step": 11011
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9007007595253996e-07,
+      "loss": 0.7664,
+      "step": 11012
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8932734676288776e-07,
+      "loss": 0.7438,
+      "step": 11013
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8858531132105137e-07,
+      "loss": 0.7094,
+      "step": 11014
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.878439696805936e-07,
+      "loss": 0.722,
+      "step": 11015
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.871033218950293e-07,
+      "loss": 0.6945,
+      "step": 11016
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.863633680178214e-07,
+      "loss": 0.7107,
+      "step": 11017
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.856241081023859e-07,
+      "loss": 0.7669,
+      "step": 11018
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.848855422020825e-07,
+      "loss": 0.6947,
+      "step": 11019
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.841476703702274e-07,
+      "loss": 0.6671,
+      "step": 11020
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8341049266008457e-07,
+      "loss": 0.7673,
+      "step": 11021
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8267400912486484e-07,
+      "loss": 0.7258,
+      "step": 11022
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8193821981773237e-07,
+      "loss": 0.7765,
+      "step": 11023
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.8120312479180243e-07,
+      "loss": 0.7435,
+      "step": 11024
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.804687241001348e-07,
+      "loss": 0.7269,
+      "step": 11025
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7973501779574263e-07,
+      "loss": 0.759,
+      "step": 11026
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.790020059315902e-07,
+      "loss": 0.7463,
+      "step": 11027
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7826968856058964e-07,
+      "loss": 0.7177,
+      "step": 11028
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7753806573560206e-07,
+      "loss": 0.7416,
+      "step": 11029
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.7680713750944176e-07,
+      "loss": 0.7826,
+      "step": 11030
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7607690393487106e-07,
+      "loss": 0.7404,
+      "step": 11031
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.753473650645978e-07,
+      "loss": 0.7593,
+      "step": 11032
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.746185209512898e-07,
+      "loss": 0.7544,
+      "step": 11033
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7389037164755504e-07,
+      "loss": 0.7582,
+      "step": 11034
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7316291720595477e-07,
+      "loss": 0.6643,
+      "step": 11035
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.724361576790025e-07,
+      "loss": 0.7074,
+      "step": 11036
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7171009311915864e-07,
+      "loss": 0.6924,
+      "step": 11037
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.709847235788344e-07,
+      "loss": 0.7309,
+      "step": 11038
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.7026004911038915e-07,
+      "loss": 0.6749,
+      "step": 11039
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6953606976613653e-07,
+      "loss": 0.7283,
+      "step": 11040
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.688127855983348e-07,
+      "loss": 0.7377,
+      "step": 11041
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.680901966591943e-07,
+      "loss": 0.3771,
+      "step": 11042
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6736830300087787e-07,
+      "loss": 0.7939,
+      "step": 11043
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.666471046754927e-07,
+      "loss": 0.7335,
+      "step": 11044
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.659266017350982e-07,
+      "loss": 0.735,
+      "step": 11045
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6520679423170723e-07,
+      "loss": 0.7349,
+      "step": 11046
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6448768221727603e-07,
+      "loss": 0.7003,
+      "step": 11047
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6376926574371527e-07,
+      "loss": 0.7698,
+      "step": 11048
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6305154486288353e-07,
+      "loss": 0.729,
+      "step": 11049
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6233451962659037e-07,
+      "loss": 0.7078,
+      "step": 11050
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.616181900865945e-07,
+      "loss": 0.6962,
+      "step": 11051
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.6090255629460115e-07,
+      "loss": 0.7845,
+      "step": 11052
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.601876183022723e-07,
+      "loss": 0.7081,
+      "step": 11053
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5947337616121456e-07,
+      "loss": 0.7585,
+      "step": 11054
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.587598299229844e-07,
+      "loss": 0.7325,
+      "step": 11055
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5804697963909176e-07,
+      "loss": 0.7162,
+      "step": 11056
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5733482536099095e-07,
+      "loss": 0.7277,
+      "step": 11057
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5662336714009095e-07,
+      "loss": 0.7256,
+      "step": 11058
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5591260502774725e-07,
+      "loss": 0.688,
+      "step": 11059
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5520253907526783e-07,
+      "loss": 0.7925,
+      "step": 11060
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.544931693339071e-07,
+      "loss": 0.6936,
+      "step": 11061
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5378449585487195e-07,
+      "loss": 0.721,
+      "step": 11062
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5307651868931923e-07,
+      "loss": 0.7159,
+      "step": 11063
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5236923788835366e-07,
+      "loss": 0.6985,
+      "step": 11064
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.516626535030276e-07,
+      "loss": 0.7515,
+      "step": 11065
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5095676558435043e-07,
+      "loss": 0.7083,
+      "step": 11066
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5025157418327347e-07,
+      "loss": 0.7441,
+      "step": 11067
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4954707935070165e-07,
+      "loss": 0.6678,
+      "step": 11068
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.48843281137492e-07,
+      "loss": 0.7241,
+      "step": 11069
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.481401795944439e-07,
+      "loss": 0.7563,
+      "step": 11070
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4743777477231344e-07,
+      "loss": 0.7521,
+      "step": 11071
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.467360667218034e-07,
+      "loss": 0.7525,
+      "step": 11072
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.460350554935665e-07,
+      "loss": 0.7242,
+      "step": 11073
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.453347411382046e-07,
+      "loss": 0.7364,
+      "step": 11074
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.446351237062717e-07,
+      "loss": 0.6836,
+      "step": 11075
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4393620324826847e-07,
+      "loss": 0.6788,
+      "step": 11076
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4323797981464677e-07,
+      "loss": 0.7627,
+      "step": 11077
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4254045345580854e-07,
+      "loss": 0.7188,
+      "step": 11078
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.418436242221035e-07,
+      "loss": 0.7478,
+      "step": 11079
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.411474921638336e-07,
+      "loss": 0.7764,
+      "step": 11080
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.4045205733124865e-07,
+      "loss": 0.7398,
+      "step": 11081
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3975731977455073e-07,
+      "loss": 0.7109,
+      "step": 11082
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.390632795438864e-07,
+      "loss": 0.6836,
+      "step": 11083
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.383699366893567e-07,
+      "loss": 0.7026,
+      "step": 11084
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3767729126101046e-07,
+      "loss": 0.8041,
+      "step": 11085
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.369853433088477e-07,
+      "loss": 0.7686,
+      "step": 11086
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.362940928828129e-07,
+      "loss": 0.7274,
+      "step": 11087
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.356035400328084e-07,
+      "loss": 0.7711,
+      "step": 11088
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3491368480868093e-07,
+      "loss": 0.7033,
+      "step": 11089
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3422452726022625e-07,
+      "loss": 0.7828,
+      "step": 11090
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3353606743719016e-07,
+      "loss": 0.7183,
+      "step": 11091
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.328483053892739e-07,
+      "loss": 0.7143,
+      "step": 11092
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.32161241166119e-07,
+      "loss": 0.771,
+      "step": 11093
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.314748748173235e-07,
+      "loss": 0.6737,
+      "step": 11094
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3078920639243337e-07,
+      "loss": 0.7165,
+      "step": 11095
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.301042359409423e-07,
+      "loss": 0.7408,
+      "step": 11096
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2941996351229523e-07,
+      "loss": 0.6997,
+      "step": 11097
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2873638915588593e-07,
+      "loss": 0.7606,
+      "step": 11098
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2805351292106046e-07,
+      "loss": 0.7448,
+      "step": 11099
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.273713348571106e-07,
+      "loss": 0.7025,
+      "step": 11100
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2668985501328023e-07,
+      "loss": 0.7,
+      "step": 11101
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.260090734387622e-07,
+      "loss": 0.7595,
+      "step": 11102
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.253289901826984e-07,
+      "loss": 0.743,
+      "step": 11103
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2464960529417945e-07,
+      "loss": 0.7528,
+      "step": 11104
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.239709188222484e-07,
+      "loss": 0.714,
+      "step": 11105
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.232929308158972e-07,
+      "loss": 0.7158,
+      "step": 11106
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.226156413240633e-07,
+      "loss": 0.7664,
+      "step": 11107
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.21939050395641e-07,
+      "loss": 0.7392,
+      "step": 11108
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2126315807946786e-07,
+      "loss": 0.7386,
+      "step": 11109
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.205879644243326e-07,
+      "loss": 0.7278,
+      "step": 11110
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1991346947897516e-07,
+      "loss": 0.7566,
+      "step": 11111
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.192396732920855e-07,
+      "loss": 0.775,
+      "step": 11112
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.18566575912298e-07,
+      "loss": 0.7211,
+      "step": 11113
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1789417738820384e-07,
+      "loss": 0.7008,
+      "step": 11114
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.172224777683386e-07,
+      "loss": 0.8038,
+      "step": 11115
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1655147710118907e-07,
+      "loss": 0.7543,
+      "step": 11116
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1588117543519205e-07,
+      "loss": 0.718,
+      "step": 11117
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1521157281873216e-07,
+      "loss": 0.7636,
+      "step": 11118
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1454266930014634e-07,
+      "loss": 0.7383,
+      "step": 11119
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.138744649277181e-07,
+      "loss": 0.7134,
+      "step": 11120
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.132069597496834e-07,
+      "loss": 0.6598,
+      "step": 11121
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1254015381422587e-07,
+      "loss": 0.7154,
+      "step": 11122
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.11874047169477e-07,
+      "loss": 0.7399,
+      "step": 11123
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.112086398635228e-07,
+      "loss": 0.7695,
+      "step": 11124
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.105439319443937e-07,
+      "loss": 0.6997,
+      "step": 11125
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0987992346007135e-07,
+      "loss": 0.7666,
+      "step": 11126
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0921661445848963e-07,
+      "loss": 0.7004,
+      "step": 11127
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.08554004987528e-07,
+      "loss": 0.7327,
+      "step": 11128
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0789209509501594e-07,
+      "loss": 0.6743,
+      "step": 11129
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.072308848287364e-07,
+      "loss": 0.7372,
+      "step": 11130
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.065703742364168e-07,
+      "loss": 0.6763,
+      "step": 11131
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0591056336573665e-07,
+      "loss": 0.697,
+      "step": 11132
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.052514522643246e-07,
+      "loss": 0.6998,
+      "step": 11133
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0459304097975816e-07,
+      "loss": 0.7828,
+      "step": 11134
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0393532955956707e-07,
+      "loss": 0.7648,
+      "step": 11135
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.032783180512244e-07,
+      "loss": 0.8184,
+      "step": 11136
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.026220065021601e-07,
+      "loss": 0.7874,
+      "step": 11137
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0196639495974846e-07,
+      "loss": 0.7609,
+      "step": 11138
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.01311483471316e-07,
+      "loss": 0.7372,
+      "step": 11139
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.00657272084135e-07,
+      "loss": 0.7708,
+      "step": 11140
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.0000376084543425e-07,
+      "loss": 0.7508,
+      "step": 11141
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9935094980238276e-07,
+      "loss": 0.6873,
+      "step": 11142
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.986988390021073e-07,
+      "loss": 0.7438,
+      "step": 11143
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.980474284916801e-07,
+      "loss": 0.7101,
+      "step": 11144
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9739671831812144e-07,
+      "loss": 0.7433,
+      "step": 11145
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.967467085284037e-07,
+      "loss": 0.7145,
+      "step": 11146
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9609739916944936e-07,
+      "loss": 0.7604,
+      "step": 11147
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9544879028812754e-07,
+      "loss": 0.797,
+      "step": 11148
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.9480088193125755e-07,
+      "loss": 0.7465,
+      "step": 11149
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 2.941536741456108e-07,
+      "loss": 0.6954,
+      "step": 11150
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.935071669779044e-07,
+      "loss": 0.8335,
+      "step": 11151
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.928613604748065e-07,
+      "loss": 0.7213,
+      "step": 11152
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.922162546829355e-07,
+      "loss": 0.7024,
+      "step": 11153
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.915718496488595e-07,
+      "loss": 0.8365,
+      "step": 11154
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9092814541909265e-07,
+      "loss": 0.7499,
+      "step": 11155
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.9028514204010203e-07,
+      "loss": 0.705,
+      "step": 11156
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8964283955830283e-07,
+      "loss": 0.3652,
+      "step": 11157
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8900123802006017e-07,
+      "loss": 0.7003,
+      "step": 11158
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.88360337471687e-07,
+      "loss": 0.6517,
+      "step": 11159
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.877201379594474e-07,
+      "loss": 0.7487,
+      "step": 11160
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8708063952955556e-07,
+      "loss": 0.7451,
+      "step": 11161
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.864418422281723e-07,
+      "loss": 0.6943,
+      "step": 11162
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.858037461014096e-07,
+      "loss": 0.704,
+      "step": 11163
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8516635119532845e-07,
+      "loss": 0.7567,
+      "step": 11164
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8452965755593863e-07,
+      "loss": 0.7568,
+      "step": 11165
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8389366522920234e-07,
+      "loss": 0.7616,
+      "step": 11166
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.832583742610262e-07,
+      "loss": 0.8096,
+      "step": 11167
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8262378469726904e-07,
+      "loss": 0.6962,
+      "step": 11168
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8198989658374085e-07,
+      "loss": 0.7919,
+      "step": 11169
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8135670996619735e-07,
+      "loss": 0.7442,
+      "step": 11170
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8072422489034636e-07,
+      "loss": 0.6952,
+      "step": 11171
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.8009244140184135e-07,
+      "loss": 0.4033,
+      "step": 11172
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.794613595462892e-07,
+      "loss": 0.7796,
+      "step": 11173
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.788309793692467e-07,
+      "loss": 0.7446,
+      "step": 11174
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7820130091621325e-07,
+      "loss": 0.6855,
+      "step": 11175
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.775723242326478e-07,
+      "loss": 0.7409,
+      "step": 11176
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.769440493639497e-07,
+      "loss": 0.8056,
+      "step": 11177
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7631647635547046e-07,
+      "loss": 0.8049,
+      "step": 11178
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7568960525251376e-07,
+      "loss": 0.7016,
+      "step": 11179
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7506343610032904e-07,
+      "loss": 0.697,
+      "step": 11180
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.744379689441168e-07,
+      "loss": 0.666,
+      "step": 11181
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.738132038290264e-07,
+      "loss": 0.7086,
+      "step": 11182
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.731891408001575e-07,
+      "loss": 0.694,
+      "step": 11183
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.725657799025583e-07,
+      "loss": 0.722,
+      "step": 11184
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.719431211812229e-07,
+      "loss": 0.7091,
+      "step": 11185
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.71321164681102e-07,
+      "loss": 0.7432,
+      "step": 11186
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7069991044708977e-07,
+      "loss": 0.7459,
+      "step": 11187
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.7007935852403133e-07,
+      "loss": 0.7355,
+      "step": 11188
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.69459508956722e-07,
+      "loss": 0.6954,
+      "step": 11189
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6884036178990603e-07,
+      "loss": 0.7671,
+      "step": 11190
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6822191706827426e-07,
+      "loss": 0.7057,
+      "step": 11191
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.676041748364733e-07,
+      "loss": 0.7526,
+      "step": 11192
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.669871351390918e-07,
+      "loss": 0.7472,
+      "step": 11193
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.663707980206709e-07,
+      "loss": 0.6904,
+      "step": 11194
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6575516352570385e-07,
+      "loss": 0.749,
+      "step": 11195
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6514023169862737e-07,
+      "loss": 0.741,
+      "step": 11196
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6452600258383144e-07,
+      "loss": 0.7205,
+      "step": 11197
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.639124762256529e-07,
+      "loss": 0.6967,
+      "step": 11198
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.632996526683818e-07,
+      "loss": 0.7362,
+      "step": 11199
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6268753195625273e-07,
+      "loss": 0.6899,
+      "step": 11200
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.620761141334538e-07,
+      "loss": 0.7072,
+      "step": 11201
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.614653992441185e-07,
+      "loss": 0.7857,
+      "step": 11202
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6085538733233276e-07,
+      "loss": 0.7387,
+      "step": 11203
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.602460784421279e-07,
+      "loss": 0.7027,
+      "step": 11204
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5963747261748993e-07,
+      "loss": 0.7304,
+      "step": 11205
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.590295699023504e-07,
+      "loss": 0.7522,
+      "step": 11206
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.584223703405886e-07,
+      "loss": 0.7546,
+      "step": 11207
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.578158739760395e-07,
+      "loss": 0.7612,
+      "step": 11208
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5721008085247914e-07,
+      "loss": 0.733,
+      "step": 11209
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5660499101363925e-07,
+      "loss": 0.7536,
+      "step": 11210
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5600060450319595e-07,
+      "loss": 0.7244,
+      "step": 11211
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5539692136478e-07,
+      "loss": 0.7528,
+      "step": 11212
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5479394164196645e-07,
+      "loss": 0.7398,
+      "step": 11213
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5419166537828055e-07,
+      "loss": 0.7612,
+      "step": 11214
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.535900926171997e-07,
+      "loss": 0.7242,
+      "step": 11215
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5298922340214915e-07,
+      "loss": 0.8179,
+      "step": 11216
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.523890577764998e-07,
+      "loss": 0.6907,
+      "step": 11217
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5178959578357696e-07,
+      "loss": 0.7578,
+      "step": 11218
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.511908374666516e-07,
+      "loss": 0.7515,
+      "step": 11219
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5059278286894585e-07,
+      "loss": 0.7247,
+      "step": 11220
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4999543203363175e-07,
+      "loss": 0.73,
+      "step": 11221
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4939878500382487e-07,
+      "loss": 0.6844,
+      "step": 11222
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4880284182259963e-07,
+      "loss": 0.7207,
+      "step": 11223
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.482076025329705e-07,
+      "loss": 0.7276,
+      "step": 11224
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4761306717790645e-07,
+      "loss": 0.7439,
+      "step": 11225
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.470192358003243e-07,
+      "loss": 0.7099,
+      "step": 11226
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.464261084430886e-07,
+      "loss": 0.7068,
+      "step": 11227
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.458336851490162e-07,
+      "loss": 0.7246,
+      "step": 11228
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.452419659608696e-07,
+      "loss": 0.7401,
+      "step": 11229
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.446509509213624e-07,
+      "loss": 0.8045,
+      "step": 11230
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.440606400731582e-07,
+      "loss": 0.6781,
+      "step": 11231
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.434710334588675e-07,
+      "loss": 0.728,
+      "step": 11232
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.428821311210505e-07,
+      "loss": 0.7947,
+      "step": 11233
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4229393310221894e-07,
+      "loss": 0.7092,
+      "step": 11234
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4170643944483207e-07,
+      "loss": 0.7039,
+      "step": 11235
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4111965019129713e-07,
+      "loss": 0.7702,
+      "step": 11236
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.405335653839713e-07,
+      "loss": 0.7553,
+      "step": 11237
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3994818506516194e-07,
+      "loss": 0.7258,
+      "step": 11238
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3936350927712516e-07,
+      "loss": 0.7655,
+      "step": 11239
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.387795380620639e-07,
+      "loss": 0.7054,
+      "step": 11240
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3819627146213443e-07,
+      "loss": 0.6991,
+      "step": 11241
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3761370951943862e-07,
+      "loss": 0.7469,
+      "step": 11242
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3703185227603064e-07,
+      "loss": 0.7434,
+      "step": 11243
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3645069977390912e-07,
+      "loss": 0.7736,
+      "step": 11244
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3587025205502712e-07,
+      "loss": 0.7205,
+      "step": 11245
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3529050916128336e-07,
+      "loss": 0.6836,
+      "step": 11246
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3471147113452662e-07,
+      "loss": 0.6504,
+      "step": 11247
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3413313801655568e-07,
+      "loss": 0.6755,
+      "step": 11248
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.335555098491149e-07,
+      "loss": 0.7379,
+      "step": 11249
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3297858667390427e-07,
+      "loss": 0.7435,
+      "step": 11250
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3240236853256604e-07,
+      "loss": 0.3305,
+      "step": 11251
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3182685546669692e-07,
+      "loss": 0.7154,
+      "step": 11252
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3125204751783703e-07,
+      "loss": 0.775,
+      "step": 11253
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3067794472748317e-07,
+      "loss": 0.7643,
+      "step": 11254
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.3010454713707442e-07,
+      "loss": 0.7315,
+      "step": 11255
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.295318547880021e-07,
+      "loss": 0.3747,
+      "step": 11256
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.289598677216065e-07,
+      "loss": 0.6727,
+      "step": 11257
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2838858597917567e-07,
+      "loss": 0.7482,
+      "step": 11258
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2781800960194776e-07,
+      "loss": 0.771,
+      "step": 11259
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2724813863111207e-07,
+      "loss": 0.7028,
+      "step": 11260
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.266789731078023e-07,
+      "loss": 0.7664,
+      "step": 11261
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2611051307310338e-07,
+      "loss": 0.7394,
+      "step": 11262
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2554275856805253e-07,
+      "loss": 0.7934,
+      "step": 11263
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2497570963363024e-07,
+      "loss": 0.7473,
+      "step": 11264
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2440936631077158e-07,
+      "loss": 0.6676,
+      "step": 11265
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2384372864035608e-07,
+      "loss": 0.7945,
+      "step": 11266
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.232787966632155e-07,
+      "loss": 0.7492,
+      "step": 11267
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2271457042012945e-07,
+      "loss": 0.7612,
+      "step": 11268
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2215104995182647e-07,
+      "loss": 0.6869,
+      "step": 11269
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2158823529898399e-07,
+      "loss": 0.734,
+      "step": 11270
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.2102612650222954e-07,
+      "loss": 0.7696,
+      "step": 11271
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2046472360213845e-07,
+      "loss": 0.7184,
+      "step": 11272
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1990402663923604e-07,
+      "loss": 0.6527,
+      "step": 11273
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1934403565399666e-07,
+      "loss": 0.6623,
+      "step": 11274
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1878475068684234e-07,
+      "loss": 0.7992,
+      "step": 11275
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1822617177814532e-07,
+      "loss": 0.7221,
+      "step": 11276
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1766829896822884e-07,
+      "loss": 0.7556,
+      "step": 11277
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.171111322973596e-07,
+      "loss": 0.8206,
+      "step": 11278
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1655467180575872e-07,
+      "loss": 0.7646,
+      "step": 11279
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1599891753359414e-07,
+      "loss": 0.7333,
+      "step": 11280
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1544386952098372e-07,
+      "loss": 0.7279,
+      "step": 11281
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.14889527807991e-07,
+      "loss": 0.7285,
+      "step": 11282
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1433589243463505e-07,
+      "loss": 0.7239,
+      "step": 11283
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1378296344087724e-07,
+      "loss": 0.7131,
+      "step": 11284
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.132307408666312e-07,
+      "loss": 0.7266,
+      "step": 11285
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1267922475176174e-07,
+      "loss": 0.4214,
+      "step": 11286
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1212841513607807e-07,
+      "loss": 0.7424,
+      "step": 11287
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1157831205933843e-07,
+      "loss": 0.7255,
+      "step": 11288
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.110289155612566e-07,
+      "loss": 0.7725,
+      "step": 11289
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.1048022568148863e-07,
+      "loss": 0.7522,
+      "step": 11290
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.099322424596406e-07,
+      "loss": 0.8488,
+      "step": 11291
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.093849659352698e-07,
+      "loss": 0.7684,
+      "step": 11292
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0883839614788127e-07,
+      "loss": 0.7337,
+      "step": 11293
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.082925331369312e-07,
+      "loss": 0.7548,
+      "step": 11294
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0774737694181924e-07,
+      "loss": 0.7477,
+      "step": 11295
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0720292760189942e-07,
+      "loss": 0.7577,
+      "step": 11296
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0665918515647365e-07,
+      "loss": 0.6847,
+      "step": 11297
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0611614964479055e-07,
+      "loss": 0.7468,
+      "step": 11298
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0557382110605096e-07,
+      "loss": 0.7128,
+      "step": 11299
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0503219957940245e-07,
+      "loss": 0.6728,
+      "step": 11300
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0449128510393934e-07,
+      "loss": 0.7835,
+      "step": 11301
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0395107771871146e-07,
+      "loss": 0.7515,
+      "step": 11302
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0341157746271101e-07,
+      "loss": 0.7486,
+      "step": 11303
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0287278437488345e-07,
+      "loss": 0.7225,
+      "step": 11304
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0233469849412102e-07,
+      "loss": 0.7803,
+      "step": 11305
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.01797319859266e-07,
+      "loss": 0.7477,
+      "step": 11306
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0126064850910954e-07,
+      "loss": 0.7751,
+      "step": 11307
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.007246844823896e-07,
+      "loss": 0.7269,
+      "step": 11308
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.0018942781779626e-07,
+      "loss": 0.7051,
+      "step": 11309
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.996548785539676e-07,
+      "loss": 0.6909,
+      "step": 11310
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9912103672948714e-07,
+      "loss": 0.7969,
+      "step": 11311
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9858790238289404e-07,
+      "loss": 0.7117,
+      "step": 11312
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.980554755526709e-07,
+      "loss": 0.6992,
+      "step": 11313
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.975237562772503e-07,
+      "loss": 0.6747,
+      "step": 11314
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9699274459501594e-07,
+      "loss": 0.7328,
+      "step": 11315
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.964624405442994e-07,
+      "loss": 0.7061,
+      "step": 11316
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.95932844163379e-07,
+      "loss": 0.788,
+      "step": 11317
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9540395549048297e-07,
+      "loss": 0.7077,
+      "step": 11318
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9487577456379302e-07,
+      "loss": 0.7399,
+      "step": 11319
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9434830142143314e-07,
+      "loss": 0.7544,
+      "step": 11320
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9382153610147837e-07,
+      "loss": 0.7325,
+      "step": 11321
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9329547864195497e-07,
+      "loss": 0.6907,
+      "step": 11322
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9277012908083592e-07,
+      "loss": 0.7742,
+      "step": 11323
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9224548745604532e-07,
+      "loss": 0.7403,
+      "step": 11324
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9172155380545178e-07,
+      "loss": 0.7572,
+      "step": 11325
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9119832816687723e-07,
+      "loss": 0.6945,
+      "step": 11326
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9067581057808927e-07,
+      "loss": 0.716,
+      "step": 11327
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.901540010768077e-07,
+      "loss": 0.7487,
+      "step": 11328
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.896328997006991e-07,
+      "loss": 0.7655,
+      "step": 11329
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8911250648737888e-07,
+      "loss": 0.7813,
+      "step": 11330
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.885928214744115e-07,
+      "loss": 0.7599,
+      "step": 11331
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8807384469931023e-07,
+      "loss": 0.7192,
+      "step": 11332
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8755557619953845e-07,
+      "loss": 0.7112,
+      "step": 11333
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8703801601250516e-07,
+      "loss": 0.7529,
+      "step": 11334
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8652116417557375e-07,
+      "loss": 0.7527,
+      "step": 11335
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.860050207260522e-07,
+      "loss": 0.7804,
+      "step": 11336
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8548958570119625e-07,
+      "loss": 0.7373,
+      "step": 11337
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.849748591382161e-07,
+      "loss": 0.7495,
+      "step": 11338
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8446084107426542e-07,
+      "loss": 0.7236,
+      "step": 11339
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8394753154644784e-07,
+      "loss": 0.713,
+      "step": 11340
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8343493059181817e-07,
+      "loss": 0.6498,
+      "step": 11341
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8292303824737789e-07,
+      "loss": 0.7501,
+      "step": 11342
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8241185455007858e-07,
+      "loss": 0.7356,
+      "step": 11343
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8190137953681963e-07,
+      "loss": 0.6837,
+      "step": 11344
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8139161324444933e-07,
+      "loss": 0.7639,
+      "step": 11345
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.808825557097671e-07,
+      "loss": 0.7013,
+      "step": 11346
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.803742069695158e-07,
+      "loss": 0.7254,
+      "step": 11347
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7986656706039495e-07,
+      "loss": 0.776,
+      "step": 11348
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7935963601904637e-07,
+      "loss": 0.7719,
+      "step": 11349
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.78853413882063e-07,
+      "loss": 0.3741,
+      "step": 11350
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7834790068598674e-07,
+      "loss": 0.7238,
+      "step": 11351
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7784309646730836e-07,
+      "loss": 0.7669,
+      "step": 11352
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.773390012624654e-07,
+      "loss": 0.8565,
+      "step": 11353
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7683561510784875e-07,
+      "loss": 0.82,
+      "step": 11354
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7633293803979489e-07,
+      "loss": 0.7588,
+      "step": 11355
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7583097009458815e-07,
+      "loss": 0.3614,
+      "step": 11356
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.75329711308464e-07,
+      "loss": 0.8097,
+      "step": 11357
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7482916171760677e-07,
+      "loss": 0.739,
+      "step": 11358
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7432932135814763e-07,
+      "loss": 0.7441,
+      "step": 11359
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7383019026616654e-07,
+      "loss": 0.7426,
+      "step": 11360
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7333176847769584e-07,
+      "loss": 0.7534,
+      "step": 11361
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.728340560287134e-07,
+      "loss": 0.7456,
+      "step": 11362
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7233705295514492e-07,
+      "loss": 0.7277,
+      "step": 11363
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7184075929286949e-07,
+      "loss": 0.7042,
+      "step": 11364
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7134517507770842e-07,
+      "loss": 0.7894,
+      "step": 11365
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.708503003454398e-07,
+      "loss": 0.772,
+      "step": 11366
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7035613513178285e-07,
+      "loss": 0.7359,
+      "step": 11367
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.698626794724112e-07,
+      "loss": 0.7229,
+      "step": 11368
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6936993340294306e-07,
+      "loss": 0.7954,
+      "step": 11369
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6887789695894996e-07,
+      "loss": 0.7545,
+      "step": 11370
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6838657017594795e-07,
+      "loss": 0.7099,
+      "step": 11371
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.678959530894031e-07,
+      "loss": 0.7119,
+      "step": 11372
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.674060457347304e-07,
+      "loss": 0.7412,
+      "step": 11373
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6691684814729603e-07,
+      "loss": 0.7416,
+      "step": 11374
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6642836036241172e-07,
+      "loss": 0.7124,
+      "step": 11375
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6594058241533706e-07,
+      "loss": 0.6914,
+      "step": 11376
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6545351434128498e-07,
+      "loss": 0.7393,
+      "step": 11377
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6496715617541405e-07,
+      "loss": 0.7717,
+      "step": 11378
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.644815079528317e-07,
+      "loss": 0.6976,
+      "step": 11379
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6399656970859434e-07,
+      "loss": 0.71,
+      "step": 11380
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6351234147770846e-07,
+      "loss": 0.791,
+      "step": 11381
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6302882329512493e-07,
+      "loss": 0.7359,
+      "step": 11382
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.625460151957514e-07,
+      "loss": 0.6979,
+      "step": 11383
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.620639172144356e-07,
+      "loss": 0.7254,
+      "step": 11384
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6158252938597852e-07,
+      "loss": 0.7395,
+      "step": 11385
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6110185174513126e-07,
+      "loss": 0.7351,
+      "step": 11386
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6062188432658943e-07,
+      "loss": 0.6896,
+      "step": 11387
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6014262716500083e-07,
+      "loss": 0.6879,
+      "step": 11388
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5966408029495894e-07,
+      "loss": 0.7679,
+      "step": 11389
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5918624375101055e-07,
+      "loss": 0.3645,
+      "step": 11390
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.5870911756764585e-07,
+      "loss": 0.3605,
+      "step": 11391
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5823270177930728e-07,
+      "loss": 0.7547,
+      "step": 11392
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.577569964203851e-07,
+      "loss": 0.7249,
+      "step": 11393
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5728200152521855e-07,
+      "loss": 0.7811,
+      "step": 11394
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.568077171280935e-07,
+      "loss": 0.6952,
+      "step": 11395
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5633414326324814e-07,
+      "loss": 0.7353,
+      "step": 11396
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5586127996486622e-07,
+      "loss": 0.7697,
+      "step": 11397
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5538912726708267e-07,
+      "loss": 0.7591,
+      "step": 11398
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.549176852039791e-07,
+      "loss": 0.7086,
+      "step": 11399
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5444695380958607e-07,
+      "loss": 0.7746,
+      "step": 11400
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.539769331178842e-07,
+      "loss": 0.7424,
+      "step": 11401
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5350762316280189e-07,
+      "loss": 0.7273,
+      "step": 11402
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5303902397821646e-07,
+      "loss": 0.6261,
+      "step": 11403
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.525711355979542e-07,
+      "loss": 0.6782,
+      "step": 11404
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5210395805578705e-07,
+      "loss": 0.7213,
+      "step": 11405
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5163749138544237e-07,
+      "loss": 0.7641,
+      "step": 11406
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5117173562058884e-07,
+      "loss": 0.7384,
+      "step": 11407
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5070669079484956e-07,
+      "loss": 0.7389,
+      "step": 11408
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5024235694179102e-07,
+      "loss": 0.7927,
+      "step": 11409
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4977873409493415e-07,
+      "loss": 0.751,
+      "step": 11410
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4931582228774443e-07,
+      "loss": 0.7104,
+      "step": 11411
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4885362155363735e-07,
+      "loss": 0.7345,
+      "step": 11412
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4839213192597623e-07,
+      "loss": 0.7212,
+      "step": 11413
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4793135343807442e-07,
+      "loss": 0.8096,
+      "step": 11414
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4747128612319307e-07,
+      "loss": 0.6958,
+      "step": 11415
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.470119300145423e-07,
+      "loss": 0.7438,
+      "step": 11416
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.465532851452811e-07,
+      "loss": 0.744,
+      "step": 11417
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.460953515485164e-07,
+      "loss": 0.7831,
+      "step": 11418
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4563812925730393e-07,
+      "loss": 0.6906,
+      "step": 11419
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.451816183046495e-07,
+      "loss": 0.6848,
+      "step": 11420
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4472581872350454e-07,
+      "loss": 0.7164,
+      "step": 11421
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4427073054677277e-07,
+      "loss": 0.8109,
+      "step": 11422
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4381635380730451e-07,
+      "loss": 0.7301,
+      "step": 11423
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4336268853789913e-07,
+      "loss": 0.7094,
+      "step": 11424
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4290973477130266e-07,
+      "loss": 0.7145,
+      "step": 11425
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4245749254021334e-07,
+      "loss": 0.7403,
+      "step": 11426
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4200596187727623e-07,
+      "loss": 0.7789,
+      "step": 11427
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.415551428150852e-07,
+      "loss": 0.7299,
+      "step": 11428
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.4110503538618092e-07,
+      "loss": 0.7151,
+      "step": 11429
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.406556396230574e-07,
+      "loss": 0.793,
+      "step": 11430
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.40206955558152e-07,
+      "loss": 0.7582,
+      "step": 11431
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.397589832238555e-07,
+      "loss": 0.7871,
+      "step": 11432
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3931172265250203e-07,
+      "loss": 0.7157,
+      "step": 11433
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.38865173876378e-07,
+      "loss": 0.7206,
+      "step": 11434
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3841933692771982e-07,
+      "loss": 0.7272,
+      "step": 11435
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.379742118387073e-07,
+      "loss": 0.7678,
+      "step": 11436
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3752979864147364e-07,
+      "loss": 0.7651,
+      "step": 11437
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3708609736809764e-07,
+      "loss": 0.7287,
+      "step": 11438
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.366431080506081e-07,
+      "loss": 0.7864,
+      "step": 11439
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.362008307209839e-07,
+      "loss": 0.7114,
+      "step": 11440
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3575926541114947e-07,
+      "loss": 0.7113,
+      "step": 11441
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3531841215297937e-07,
+      "loss": 0.7493,
+      "step": 11442
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3487827097829698e-07,
+      "loss": 0.7379,
+      "step": 11443
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3443884191887247e-07,
+      "loss": 0.6873,
+      "step": 11444
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3400012500642934e-07,
+      "loss": 0.6901,
+      "step": 11445
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.335621202726345e-07,
+      "loss": 0.776,
+      "step": 11446
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3312482774910374e-07,
+      "loss": 0.7299,
+      "step": 11447
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3268824746740628e-07,
+      "loss": 0.7152,
+      "step": 11448
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3225237945905357e-07,
+      "loss": 0.7816,
+      "step": 11449
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3181722375551152e-07,
+      "loss": 0.7202,
+      "step": 11450
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3138278038819062e-07,
+      "loss": 0.6994,
+      "step": 11451
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3094904938845244e-07,
+      "loss": 0.6416,
+      "step": 11452
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3051603078760411e-07,
+      "loss": 0.7619,
+      "step": 11453
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3008372461690288e-07,
+      "loss": 0.725,
+      "step": 11454
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2965213090755713e-07,
+      "loss": 0.7644,
+      "step": 11455
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2922124969072082e-07,
+      "loss": 0.8098,
+      "step": 11456
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.287910809974957e-07,
+      "loss": 0.7377,
+      "step": 11457
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2836162485893477e-07,
+      "loss": 0.7172,
+      "step": 11458
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2793288130603764e-07,
+      "loss": 0.7599,
+      "step": 11459
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.27504850369754e-07,
+      "loss": 0.7165,
+      "step": 11460
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.270775320809814e-07,
+      "loss": 0.3837,
+      "step": 11461
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.266509264705651e-07,
+      "loss": 0.7452,
+      "step": 11462
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.262250335693005e-07,
+      "loss": 0.75,
+      "step": 11463
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2579985340792966e-07,
+      "loss": 0.6906,
+      "step": 11464
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2537538601714582e-07,
+      "loss": 0.7137,
+      "step": 11465
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2495163142758782e-07,
+      "loss": 0.7132,
+      "step": 11466
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.245285896698445e-07,
+      "loss": 0.7356,
+      "step": 11467
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2410626077445475e-07,
+      "loss": 0.7635,
+      "step": 11468
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2368464477190202e-07,
+      "loss": 0.7663,
+      "step": 11469
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.23263741692623e-07,
+      "loss": 0.7434,
+      "step": 11470
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.22843551566999e-07,
+      "loss": 0.7866,
+      "step": 11471
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2242407442536352e-07,
+      "loss": 0.7462,
+      "step": 11472
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2200531029799344e-07,
+      "loss": 0.6909,
+      "step": 11473
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2158725921512017e-07,
+      "loss": 0.734,
+      "step": 11474
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2116992120692062e-07,
+      "loss": 0.7703,
+      "step": 11475
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2075329630351852e-07,
+      "loss": 0.3427,
+      "step": 11476
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2033738453498867e-07,
+      "loss": 0.7269,
+      "step": 11477
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1992218593135485e-07,
+      "loss": 0.7606,
+      "step": 11478
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1950770052258752e-07,
+      "loss": 0.7097,
+      "step": 11479
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.19093928338605e-07,
+      "loss": 0.76,
+      "step": 11480
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1868086940927781e-07,
+      "loss": 0.7659,
+      "step": 11481
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.182685237644221e-07,
+      "loss": 0.7551,
+      "step": 11482
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1785689143380186e-07,
+      "loss": 0.7392,
+      "step": 11483
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1744597244713107e-07,
+      "loss": 0.8033,
+      "step": 11484
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1703576683407381e-07,
+      "loss": 0.774,
+      "step": 11485
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1662627462423859e-07,
+      "loss": 0.7607,
+      "step": 11486
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1621749584718622e-07,
+      "loss": 0.7726,
+      "step": 11487
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1580943053242421e-07,
+      "loss": 0.6949,
+      "step": 11488
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1540207870940678e-07,
+      "loss": 0.7537,
+      "step": 11489
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1499544040754151e-07,
+      "loss": 0.7775,
+      "step": 11490
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1458951565618049e-07,
+      "loss": 0.6324,
+      "step": 11491
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1418430448462581e-07,
+      "loss": 0.7296,
+      "step": 11492
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1377980692212631e-07,
+      "loss": 0.7263,
+      "step": 11493
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1337602299788308e-07,
+      "loss": 0.7486,
+      "step": 11494
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.129729527410417e-07,
+      "loss": 0.7276,
+      "step": 11495
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1257059618069665e-07,
+      "loss": 0.7693,
+      "step": 11496
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1216895334589472e-07,
+      "loss": 0.7312,
+      "step": 11497
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1176802426562827e-07,
+      "loss": 0.7932,
+      "step": 11498
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1136780896883636e-07,
+      "loss": 0.7441,
+      "step": 11499
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1096830748441034e-07,
+      "loss": 0.7358,
+      "step": 11500
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1056951984118824e-07,
+      "loss": 0.3642,
+      "step": 11501
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1017144606795482e-07,
+      "loss": 0.7086,
+      "step": 11502
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.097740861934471e-07,
+      "loss": 0.7475,
+      "step": 11503
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0937744024634767e-07,
+      "loss": 0.7882,
+      "step": 11504
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.089815082552892e-07,
+      "loss": 0.7431,
+      "step": 11505
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0858629024885104e-07,
+      "loss": 0.7568,
+      "step": 11506
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.081917862555626e-07,
+      "loss": 0.7012,
+      "step": 11507
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0779799630390108e-07,
+      "loss": 0.7467,
+      "step": 11508
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0740492042229267e-07,
+      "loss": 0.7415,
+      "step": 11509
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0701255863911131e-07,
+      "loss": 0.6645,
+      "step": 11510
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.066209109826799e-07,
+      "loss": 0.7473,
+      "step": 11511
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.0622997748126807e-07,
+      "loss": 0.7691,
+      "step": 11512
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0583975816309877e-07,
+      "loss": 0.7484,
+      "step": 11513
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0545025305633727e-07,
+      "loss": 0.7173,
+      "step": 11514
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0506146218909996e-07,
+      "loss": 0.7394,
+      "step": 11515
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0467338558945328e-07,
+      "loss": 0.7365,
+      "step": 11516
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0428602328541037e-07,
+      "loss": 0.7391,
+      "step": 11517
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0389937530493221e-07,
+      "loss": 0.7692,
+      "step": 11518
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.035134416759287e-07,
+      "loss": 0.8157,
+      "step": 11519
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0312822242625976e-07,
+      "loss": 0.7796,
+      "step": 11520
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0274371758373203e-07,
+      "loss": 0.7983,
+      "step": 11521
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0235992717609999e-07,
+      "loss": 0.7946,
+      "step": 11522
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0197685123106927e-07,
+      "loss": 0.7207,
+      "step": 11523
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0159448977629216e-07,
+      "loss": 0.6963,
+      "step": 11524
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0121284283936773e-07,
+      "loss": 0.7737,
+      "step": 11525
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0083191044784723e-07,
+      "loss": 0.7932,
+      "step": 11526
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0045169262922761e-07,
+      "loss": 0.7141,
+      "step": 11527
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0007218941095464e-07,
+      "loss": 0.7784,
+      "step": 11528
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.969340082042312e-08,
+      "loss": 0.7068,
+      "step": 11529
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.931532688497447e-08,
+      "loss": 0.7125,
+      "step": 11530
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.893796763190355e-08,
+      "loss": 0.7842,
+      "step": 11531
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.856132308844635e-08,
+      "loss": 0.7203,
+      "step": 11532
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.818539328179333e-08,
+      "loss": 0.6547,
+      "step": 11533
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.781017823908057e-08,
+      "loss": 0.7513,
+      "step": 11534
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.743567798739084e-08,
+      "loss": 0.7408,
+      "step": 11535
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.706189255376141e-08,
+      "loss": 0.7613,
+      "step": 11536
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.668882196517182e-08,
+      "loss": 0.7205,
+      "step": 11537
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.631646624855162e-08,
+      "loss": 0.8042,
+      "step": 11538
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.594482543078043e-08,
+      "loss": 0.7551,
+      "step": 11539
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.557389953868567e-08,
+      "loss": 0.8046,
+      "step": 11540
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.520368859904039e-08,
+      "loss": 0.6932,
+      "step": 11541
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.483419263857096e-08,
+      "loss": 0.6885,
+      "step": 11542
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.44654116839483e-08,
+      "loss": 0.6632,
+      "step": 11543
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.40973457617933e-08,
+      "loss": 0.788,
+      "step": 11544
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.372999489867363e-08,
+      "loss": 0.7479,
+      "step": 11545
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.336335912110916e-08,
+      "loss": 0.7308,
+      "step": 11546
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.29974384555632e-08,
+      "loss": 0.741,
+      "step": 11547
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.263223292844903e-08,
+      "loss": 0.7888,
+      "step": 11548
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.226774256613225e-08,
+      "loss": 0.7171,
+      "step": 11549
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.19039673949207e-08,
+      "loss": 0.791,
+      "step": 11550
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.154090744107558e-08,
+      "loss": 0.7551,
+      "step": 11551
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.117856273080372e-08,
+      "loss": 0.7261,
+      "step": 11552
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.081693329026087e-08,
+      "loss": 0.7723,
+      "step": 11553
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.045601914555058e-08,
+      "loss": 0.7415,
+      "step": 11554
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.009582032272757e-08,
+      "loss": 0.7496,
+      "step": 11555
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.973633684779103e-08,
+      "loss": 0.6823,
+      "step": 11556
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.937756874669024e-08,
+      "loss": 0.3844,
+      "step": 11557
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.901951604532445e-08,
+      "loss": 0.7312,
+      "step": 11558
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.866217876953853e-08,
+      "loss": 0.7032,
+      "step": 11559
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.830555694512632e-08,
+      "loss": 0.6959,
+      "step": 11560
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.794965059783277e-08,
+      "loss": 0.6967,
+      "step": 11561
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.759445975334735e-08,
+      "loss": 0.7508,
+      "step": 11562
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.723998443730952e-08,
+      "loss": 0.7029,
+      "step": 11563
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.688622467530772e-08,
+      "loss": 0.7408,
+      "step": 11564
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.653318049287818e-08,
+      "loss": 0.7729,
+      "step": 11565
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.618085191550607e-08,
+      "loss": 0.7815,
+      "step": 11566
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.582923896862217e-08,
+      "loss": 0.8006,
+      "step": 11567
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.54783416776106e-08,
+      "loss": 0.7223,
+      "step": 11568
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.512816006779889e-08,
+      "loss": 0.7139,
+      "step": 11569
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.47786941644646e-08,
+      "loss": 0.7593,
+      "step": 11570
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.442994399283532e-08,
+      "loss": 0.6958,
+      "step": 11571
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.408190957808426e-08,
+      "loss": 0.7853,
+      "step": 11572
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.373459094533687e-08,
+      "loss": 0.6583,
+      "step": 11573
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.338798811966197e-08,
+      "loss": 0.7513,
+      "step": 11574
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.304210112607958e-08,
+      "loss": 0.7098,
+      "step": 11575
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.269692998955859e-08,
+      "loss": 0.7611,
+      "step": 11576
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.235247473501351e-08,
+      "loss": 0.7319,
+      "step": 11577
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.200873538731113e-08,
+      "loss": 0.7312,
+      "step": 11578
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.166571197126271e-08,
+      "loss": 0.7958,
+      "step": 11579
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.132340451162957e-08,
+      "loss": 0.7498,
+      "step": 11580
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.098181303312302e-08,
+      "loss": 0.6825,
+      "step": 11581
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.064093756039782e-08,
+      "loss": 0.7578,
+      "step": 11582
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.030077811806203e-08,
+      "loss": 0.7173,
+      "step": 11583
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.996133473067046e-08,
+      "loss": 0.7255,
+      "step": 11584
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.962260742272466e-08,
+      "loss": 0.7414,
+      "step": 11585
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.928459621867613e-08,
+      "loss": 0.757,
+      "step": 11586
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.894730114292426e-08,
+      "loss": 0.6827,
+      "step": 11587
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.861072221981736e-08,
+      "loss": 0.7135,
+      "step": 11588
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.827485947365043e-08,
+      "loss": 0.7456,
+      "step": 11589
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.793971292866854e-08,
+      "loss": 0.7732,
+      "step": 11590
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.760528260906452e-08,
+      "loss": 0.3858,
+      "step": 11591
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.7271568538978e-08,
+      "loss": 0.6945,
+      "step": 11592
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.693857074249966e-08,
+      "loss": 0.7328,
+      "step": 11593
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.660628924366476e-08,
+      "loss": 0.7151,
+      "step": 11594
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.627472406646186e-08,
+      "loss": 0.8059,
+      "step": 11595
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.594387523482405e-08,
+      "loss": 0.7348,
+      "step": 11596
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.561374277263334e-08,
+      "loss": 0.7117,
+      "step": 11597
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.528432670372066e-08,
+      "loss": 0.7124,
+      "step": 11598
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.495562705186365e-08,
+      "loss": 0.7614,
+      "step": 11599
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.462764384079114e-08,
+      "loss": 0.7463,
+      "step": 11600
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.43003770941797e-08,
+      "loss": 0.743,
+      "step": 11601
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.397382683565047e-08,
+      "loss": 0.8021,
+      "step": 11602
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.364799308877568e-08,
+      "loss": 0.696,
+      "step": 11603
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.332287587707877e-08,
+      "loss": 0.7177,
+      "step": 11604
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.299847522402537e-08,
+      "loss": 0.7405,
+      "step": 11605
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.267479115303233e-08,
+      "loss": 0.7085,
+      "step": 11606
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.235182368746652e-08,
+      "loss": 0.7246,
+      "step": 11607
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.202957285064149e-08,
+      "loss": 0.7027,
+      "step": 11608
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.170803866581754e-08,
+      "loss": 0.7497,
+      "step": 11609
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.1387221156205e-08,
+      "loss": 0.7845,
+      "step": 11610
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.10671203449631e-08,
+      "loss": 0.659,
+      "step": 11611
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.074773625519671e-08,
+      "loss": 0.7294,
+      "step": 11612
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.042906890996293e-08,
+      "loss": 0.7463,
+      "step": 11613
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.011111833226226e-08,
+      "loss": 0.6965,
+      "step": 11614
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.979388454504744e-08,
+      "loss": 0.8195,
+      "step": 11615
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.947736757121793e-08,
+      "loss": 0.7419,
+      "step": 11616
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.916156743362212e-08,
+      "loss": 0.7225,
+      "step": 11617
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.884648415505513e-08,
+      "loss": 0.7259,
+      "step": 11618
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.853211775826096e-08,
+      "loss": 0.7249,
+      "step": 11619
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.82184682659337e-08,
+      "loss": 0.7161,
+      "step": 11620
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.790553570071302e-08,
+      "loss": 0.7189,
+      "step": 11621
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.759332008518749e-08,
+      "loss": 0.7487,
+      "step": 11622
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.728182144189577e-08,
+      "loss": 0.7379,
+      "step": 11623
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.69710397933232e-08,
+      "loss": 0.7651,
+      "step": 11624
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.666097516190296e-08,
+      "loss": 0.7537,
+      "step": 11625
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.635162757001712e-08,
+      "loss": 0.7895,
+      "step": 11626
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.604299703999673e-08,
+      "loss": 0.6778,
+      "step": 11627
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.573508359412062e-08,
+      "loss": 0.7347,
+      "step": 11628
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.542788725461324e-08,
+      "loss": 0.8127,
+      "step": 11629
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.512140804365241e-08,
+      "loss": 0.7124,
+      "step": 11630
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.481564598335933e-08,
+      "loss": 0.7813,
+      "step": 11631
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.451060109580743e-08,
+      "loss": 0.6948,
+      "step": 11632
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.420627340301466e-08,
+      "loss": 0.7573,
+      "step": 11633
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.390266292694903e-08,
+      "loss": 0.7289,
+      "step": 11634
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.359976968952852e-08,
+      "loss": 0.6747,
+      "step": 11635
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.329759371261457e-08,
+      "loss": 0.7016,
+      "step": 11636
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.299613501802304e-08,
+      "loss": 0.8154,
+      "step": 11637
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.26953936275132e-08,
+      "loss": 0.753,
+      "step": 11638
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.239536956279324e-08,
+      "loss": 0.7383,
+      "step": 11639
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.20960628455225e-08,
+      "loss": 0.7259,
+      "step": 11640
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.179747349730481e-08,
+      "loss": 0.7377,
+      "step": 11641
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.149960153969403e-08,
+      "loss": 0.7427,
+      "step": 11642
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.120244699419298e-08,
+      "loss": 0.7453,
+      "step": 11643
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.090600988225226e-08,
+      "loss": 0.7673,
+      "step": 11644
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.061029022526699e-08,
+      "loss": 0.7277,
+      "step": 11645
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.031528804458786e-08,
+      "loss": 0.7112,
+      "step": 11646
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.002100336150784e-08,
+      "loss": 0.73,
+      "step": 11647
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.972743619726883e-08,
+      "loss": 0.7576,
+      "step": 11648
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.943458657306278e-08,
+      "loss": 0.7883,
+      "step": 11649
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.914245451003053e-08,
+      "loss": 0.7618,
+      "step": 11650
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.8851040029258564e-08,
+      "loss": 0.6939,
+      "step": 11651
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.856034315178227e-08,
+      "loss": 0.6633,
+      "step": 11652
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.827036389858598e-08,
+      "loss": 0.7343,
+      "step": 11653
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.798110229060072e-08,
+      "loss": 0.7416,
+      "step": 11654
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.7692558348708685e-08,
+      "loss": 0.7154,
+      "step": 11655
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.740473209373876e-08,
+      "loss": 0.7305,
+      "step": 11656
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.7117623546466546e-08,
+      "loss": 0.7975,
+      "step": 11657
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.68312327276177e-08,
+      "loss": 0.7382,
+      "step": 11658
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.654555965786457e-08,
+      "loss": 0.7505,
+      "step": 11659
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.626060435782843e-08,
+      "loss": 0.7613,
+      "step": 11660
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.59763668480795e-08,
+      "loss": 0.7364,
+      "step": 11661
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.569284714913581e-08,
+      "loss": 0.7151,
+      "step": 11662
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.54100452814621e-08,
+      "loss": 0.7051,
+      "step": 11663
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.512796126547426e-08,
+      "loss": 0.7001,
+      "step": 11664
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.484659512153157e-08,
+      "loss": 0.7561,
+      "step": 11665
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.456594686994776e-08,
+      "loss": 0.7205,
+      "step": 11666
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.428601653097998e-08,
+      "loss": 0.7799,
+      "step": 11667
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.400680412483428e-08,
+      "loss": 0.8033,
+      "step": 11668
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.3728309671667864e-08,
+      "loss": 0.7285,
+      "step": 11669
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.3450533191581336e-08,
+      "loss": 0.7604,
+      "step": 11670
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.3173474704627526e-08,
+      "loss": 0.7437,
+      "step": 11671
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.2897134230806e-08,
+      "loss": 0.6647,
+      "step": 11672
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.262151179006414e-08,
+      "loss": 0.7066,
+      "step": 11673
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.2346607402297136e-08,
+      "loss": 0.7046,
+      "step": 11674
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.207242108734911e-08,
+      "loss": 0.8111,
+      "step": 11675
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.179895286501424e-08,
+      "loss": 0.7769,
+      "step": 11676
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.152620275503006e-08,
+      "loss": 0.7616,
+      "step": 11677
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.12541707770875e-08,
+      "loss": 0.7174,
+      "step": 11678
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.0982856950820835e-08,
+      "loss": 0.8144,
+      "step": 11679
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.0712261295817745e-08,
+      "loss": 0.6636,
+      "step": 11680
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.0442383831608155e-08,
+      "loss": 0.7588,
+      "step": 11681
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.017322457767537e-08,
+      "loss": 0.6992,
+      "step": 11682
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.99047835534483e-08,
+      "loss": 0.7423,
+      "step": 11683
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.9637060778302546e-08,
+      "loss": 0.6618,
+      "step": 11684
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.9370056271567104e-08,
+      "loss": 0.7172,
+      "step": 11685
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.910377005251321e-08,
+      "loss": 0.7633,
+      "step": 11686
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.883820214036328e-08,
+      "loss": 0.7773,
+      "step": 11687
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.857335255428752e-08,
+      "loss": 0.75,
+      "step": 11688
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.8309221313405096e-08,
+      "loss": 0.7525,
+      "step": 11689
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.804580843678075e-08,
+      "loss": 0.6844,
+      "step": 11690
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.778311394342927e-08,
+      "loss": 0.7056,
+      "step": 11691
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.752113785231438e-08,
+      "loss": 0.7979,
+      "step": 11692
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.72598801823454e-08,
+      "loss": 0.7453,
+      "step": 11693
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.699934095238279e-08,
+      "loss": 0.6684,
+      "step": 11694
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.673952018123151e-08,
+      "loss": 0.7194,
+      "step": 11695
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.64804178876499e-08,
+      "loss": 0.7685,
+      "step": 11696
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.622203409033854e-08,
+      "loss": 0.7584,
+      "step": 11697
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.5964368807950297e-08,
+      "loss": 0.3832,
+      "step": 11698
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.570742205908363e-08,
+      "loss": 0.7312,
+      "step": 11699
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.5451193862287024e-08,
+      "loss": 0.7408,
+      "step": 11700
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.519568423605791e-08,
+      "loss": 0.742,
+      "step": 11701
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.49408931988371e-08,
+      "loss": 0.7173,
+      "step": 11702
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.4686820769018755e-08,
+      "loss": 0.6324,
+      "step": 11703
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.443346696494266e-08,
+      "loss": 0.7541,
+      "step": 11704
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.418083180489752e-08,
+      "loss": 0.7219,
+      "step": 11705
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.3928915307119845e-08,
+      "loss": 0.7527,
+      "step": 11706
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.367771748979288e-08,
+      "loss": 0.7304,
+      "step": 11707
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.342723837105101e-08,
+      "loss": 0.7264,
+      "step": 11708
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.31774779689742e-08,
+      "loss": 0.7175,
+      "step": 11709
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.292843630159249e-08,
+      "loss": 0.664,
+      "step": 11710
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.268011338688149e-08,
+      "loss": 0.7691,
+      "step": 11711
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.243250924276798e-08,
+      "loss": 0.8095,
+      "step": 11712
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.218562388712433e-08,
+      "loss": 0.7233,
+      "step": 11713
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.1939457337771824e-08,
+      "loss": 0.7855,
+      "step": 11714
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.169400961248071e-08,
+      "loss": 0.739,
+      "step": 11715
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.1449280728969034e-08,
+      "loss": 0.7099,
+      "step": 11716
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.120527070490043e-08,
+      "loss": 0.8011,
+      "step": 11717
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.096197955789194e-08,
+      "loss": 0.7822,
+      "step": 11718
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.071940730550284e-08,
+      "loss": 0.7088,
+      "step": 11719
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.047755396524467e-08,
+      "loss": 0.6923,
+      "step": 11720
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.023641955457569e-08,
+      "loss": 0.7388,
+      "step": 11721
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.9996004090901985e-08,
+      "loss": 0.7827,
+      "step": 11722
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.975630759157745e-08,
+      "loss": 0.7495,
+      "step": 11723
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.9517330073904904e-08,
+      "loss": 0.7336,
+      "step": 11724
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.9279071555134994e-08,
+      "loss": 0.7548,
+      "step": 11725
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.904153205246619e-08,
+      "loss": 0.6642,
+      "step": 11726
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.8804711583046996e-08,
+      "loss": 0.7006,
+      "step": 11727
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.85686101639704e-08,
+      "loss": 0.7485,
+      "step": 11728
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.8333227812279436e-08,
+      "loss": 0.7784,
+      "step": 11729
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.809856454496608e-08,
+      "loss": 0.6768,
+      "step": 11730
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.78646203789701e-08,
+      "loss": 0.6722,
+      "step": 11731
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.763139533117688e-08,
+      "loss": 0.7154,
+      "step": 11732
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.739888941842296e-08,
+      "loss": 0.7781,
+      "step": 11733
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.7167102657491575e-08,
+      "loss": 0.7291,
+      "step": 11734
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.693603506511489e-08,
+      "loss": 0.7712,
+      "step": 11735
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.670568665797181e-08,
+      "loss": 0.7181,
+      "step": 11736
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.647605745269012e-08,
+      "loss": 0.788,
+      "step": 11737
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.6247147465846565e-08,
+      "loss": 0.8231,
+      "step": 11738
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.6018956713963495e-08,
+      "loss": 0.7479,
+      "step": 11739
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.5791485213514386e-08,
+      "loss": 0.715,
+      "step": 11740
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.556473298091945e-08,
+      "loss": 0.3681,
+      "step": 11741
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.533870003254447e-08,
+      "loss": 0.6912,
+      "step": 11742
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.511338638470862e-08,
+      "loss": 0.7049,
+      "step": 11743
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.488879205367335e-08,
+      "loss": 0.7129,
+      "step": 11744
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.466491705565345e-08,
+      "loss": 0.7329,
+      "step": 11745
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.444176140680822e-08,
+      "loss": 0.7059,
+      "step": 11746
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.421932512324699e-08,
+      "loss": 0.733,
+      "step": 11747
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3997608221024716e-08,
+      "loss": 0.7299,
+      "step": 11748
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3776610716147464e-08,
+      "loss": 0.7754,
+      "step": 11749
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.3556332624568035e-08,
+      "loss": 0.7138,
+      "step": 11750
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.333677396218593e-08,
+      "loss": 0.7044,
+      "step": 11751
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.311793474485181e-08,
+      "loss": 0.7425,
+      "step": 11752
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.2899814988360815e-08,
+      "loss": 0.7125,
+      "step": 11753
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.268241470846034e-08,
+      "loss": 0.7766,
+      "step": 11754
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.246573392084007e-08,
+      "loss": 0.7222,
+      "step": 11755
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.2249772641144153e-08,
+      "loss": 0.7189,
+      "step": 11756
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.2034530884960116e-08,
+      "loss": 0.7782,
+      "step": 11757
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.182000866782664e-08,
+      "loss": 0.7169,
+      "step": 11758
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.16062060052269e-08,
+      "loss": 0.7335,
+      "step": 11759
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.139312291259633e-08,
+      "loss": 0.7379,
+      "step": 11760
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.118075940531595e-08,
+      "loss": 0.7507,
+      "step": 11761
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.096911549871462e-08,
+      "loss": 0.7611,
+      "step": 11762
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.0758191208069e-08,
+      "loss": 0.7643,
+      "step": 11763
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.0547986548606914e-08,
+      "loss": 0.7695,
+      "step": 11764
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.033850153550066e-08,
+      "loss": 0.7341,
+      "step": 11765
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.01297361838726e-08,
+      "loss": 0.7484,
+      "step": 11766
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.992169050879179e-08,
+      "loss": 0.7465,
+      "step": 11767
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9714364525276207e-08,
+      "loss": 0.3703,
+      "step": 11768
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9507758248292774e-08,
+      "loss": 0.7164,
+      "step": 11769
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9301871692754004e-08,
+      "loss": 0.7373,
+      "step": 11770
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.9096704873521342e-08,
+      "loss": 0.699,
+      "step": 11771
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8892257805406275e-08,
+      "loss": 0.7876,
+      "step": 11772
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8688530503167e-08,
+      "loss": 0.698,
+      "step": 11773
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.848552298150842e-08,
+      "loss": 0.7345,
+      "step": 11774
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.8283235255085472e-08,
+      "loss": 0.7867,
+      "step": 11775
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.808166733849982e-08,
+      "loss": 0.6802,
+      "step": 11776
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7880819246300927e-08,
+      "loss": 0.3348,
+      "step": 11777
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.768069099298942e-08,
+      "loss": 0.763,
+      "step": 11778
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7481282593009305e-08,
+      "loss": 0.7627,
+      "step": 11779
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.728259406075684e-08,
+      "loss": 0.6904,
+      "step": 11780
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7084625410571663e-08,
+      "loss": 0.7678,
+      "step": 11781
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.688737665674679e-08,
+      "loss": 0.7274,
+      "step": 11782
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6690847813519716e-08,
+      "loss": 0.7353,
+      "step": 11783
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6495038895076873e-08,
+      "loss": 0.7694,
+      "step": 11784
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6299949915552514e-08,
+      "loss": 0.7678,
+      "step": 11785
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.6105580889029814e-08,
+      "loss": 0.687,
+      "step": 11786
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.591193182953866e-08,
+      "loss": 0.7529,
+      "step": 11787
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5719002751057875e-08,
+      "loss": 0.6792,
+      "step": 11788
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5526793667514092e-08,
+      "loss": 0.7783,
+      "step": 11789
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.5335304592782885e-08,
+      "loss": 0.7766,
+      "step": 11790
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.51445355406843e-08,
+      "loss": 0.7935,
+      "step": 11791
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.495448652499177e-08,
+      "loss": 0.7122,
+      "step": 11792
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4765157559423215e-08,
+      "loss": 0.7201,
+      "step": 11793
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.457654865764436e-08,
+      "loss": 0.7305,
+      "step": 11794
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.438865983327099e-08,
+      "loss": 0.7084,
+      "step": 11795
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4201491099865582e-08,
+      "loss": 0.3765,
+      "step": 11796
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.4015042470939553e-08,
+      "loss": 0.6481,
+      "step": 11797
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.382931395994992e-08,
+      "loss": 0.8003,
+      "step": 11798
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.364430558030595e-08,
+      "loss": 0.7295,
+      "step": 11799
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.34600173453603e-08,
+      "loss": 0.7817,
+      "step": 11800
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.327644926841788e-08,
+      "loss": 0.7604,
+      "step": 11801
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.3093601362728092e-08,
+      "loss": 0.7826,
+      "step": 11802
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2911473641490378e-08,
+      "loss": 0.7755,
+      "step": 11803
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.273006611785089e-08,
+      "loss": 0.732,
+      "step": 11804
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.254937880490582e-08,
+      "loss": 0.7508,
+      "step": 11805
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2369411715698065e-08,
+      "loss": 0.7054,
+      "step": 11806
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2190164863217234e-08,
+      "loss": 0.7314,
+      "step": 11807
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.2011638260402978e-08,
+      "loss": 0.7222,
+      "step": 11808
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1833831920143876e-08,
+      "loss": 0.7742,
+      "step": 11809
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.165674585527189e-08,
+      "loss": 0.71,
+      "step": 11810
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.148038007857123e-08,
+      "loss": 0.7142,
+      "step": 11811
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1304734602773936e-08,
+      "loss": 0.7283,
+      "step": 11812
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.1129809440558757e-08,
+      "loss": 0.7084,
+      "step": 11813
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.095560460455115e-08,
+      "loss": 0.7671,
+      "step": 11814
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.078212010732772e-08,
+      "loss": 0.7672,
+      "step": 11815
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.060935596141067e-08,
+      "loss": 0.7589,
+      "step": 11816
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.043731217927114e-08,
+      "loss": 0.6957,
+      "step": 11817
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0265988773329193e-08,
+      "loss": 0.6598,
+      "step": 11818
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.0095385755949383e-08,
+      "loss": 0.7534,
+      "step": 11819
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.992550313944963e-08,
+      "loss": 0.6993,
+      "step": 11820
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9756340936090134e-08,
+      "loss": 0.7482,
+      "step": 11821
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9587899158084456e-08,
+      "loss": 0.7234,
+      "step": 11822
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9420177817589537e-08,
+      "loss": 0.7732,
+      "step": 11823
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.925317692671347e-08,
+      "loss": 0.7466,
+      "step": 11824
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.9086896497511052e-08,
+      "loss": 0.6901,
+      "step": 11825
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8921336541984913e-08,
+      "loss": 0.756,
+      "step": 11826
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.87564970720866e-08,
+      "loss": 0.7355,
+      "step": 11827
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.859237809971548e-08,
+      "loss": 0.3855,
+      "step": 11828
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8428979636716525e-08,
+      "loss": 0.7725,
+      "step": 11829
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8266301694886966e-08,
+      "loss": 0.7047,
+      "step": 11830
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8104344285968522e-08,
+      "loss": 0.7625,
+      "step": 11831
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7943107421651838e-08,
+      "loss": 0.7651,
+      "step": 11832
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7782591113576497e-08,
+      "loss": 0.7063,
+      "step": 11833
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.762279537332878e-08,
+      "loss": 0.6949,
+      "step": 11834
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7463720212445022e-08,
+      "loss": 0.7187,
+      "step": 11835
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7305365642407147e-08,
+      "loss": 0.7611,
+      "step": 11836
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.7147731674644897e-08,
+      "loss": 0.7568,
+      "step": 11837
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.699081832053806e-08,
+      "loss": 0.7599,
+      "step": 11838
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6834625591414244e-08,
+      "loss": 0.6832,
+      "step": 11839
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.667915349854665e-08,
+      "loss": 0.7418,
+      "step": 11840
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6524402053159638e-08,
+      "loss": 0.7579,
+      "step": 11841
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.637037126642316e-08,
+      "loss": 0.7052,
+      "step": 11842
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6217061149456093e-08,
+      "loss": 0.7285,
+      "step": 11843
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.6064471713325146e-08,
+      "loss": 0.7991,
+      "step": 11844
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5912602969044844e-08,
+      "loss": 0.6578,
+      "step": 11845
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5761454927577526e-08,
+      "loss": 0.7212,
+      "step": 11846
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.561102759983446e-08,
+      "loss": 0.7455,
+      "step": 11847
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5461320996674746e-08,
+      "loss": 0.7541,
+      "step": 11848
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5312335128903068e-08,
+      "loss": 0.7798,
+      "step": 11849
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5164070007276375e-08,
+      "loss": 0.7705,
+      "step": 11850
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5016525642495007e-08,
+      "loss": 0.8017,
+      "step": 11851
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4869702045211543e-08,
+      "loss": 0.721,
+      "step": 11852
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4723599226023067e-08,
+      "loss": 0.6853,
+      "step": 11853
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4578217195476696e-08,
+      "loss": 0.3779,
+      "step": 11854
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.443355596406626e-08,
+      "loss": 0.7014,
+      "step": 11855
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4289615542234514e-08,
+      "loss": 0.7254,
+      "step": 11856
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4146395940370927e-08,
+      "loss": 0.6659,
+      "step": 11857
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4003897168816116e-08,
+      "loss": 0.7617,
+      "step": 11858
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3862119237854076e-08,
+      "loss": 0.7338,
+      "step": 11859
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3721062157719956e-08,
+      "loss": 0.7237,
+      "step": 11860
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3580725938595607e-08,
+      "loss": 0.7697,
+      "step": 11861
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3441110590612928e-08,
+      "loss": 0.7348,
+      "step": 11862
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3302216123847189e-08,
+      "loss": 0.6624,
+      "step": 11863
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3164042548327038e-08,
+      "loss": 0.6957,
+      "step": 11864
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3026589874024499e-08,
+      "loss": 0.6954,
+      "step": 11865
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.288985811086385e-08,
+      "loss": 0.7886,
+      "step": 11866
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.275384726871276e-08,
+      "loss": 0.724,
+      "step": 11867
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2618557357391147e-08,
+      "loss": 0.7767,
+      "step": 11868
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2483988386663426e-08,
+      "loss": 0.7937,
+      "step": 11869
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2350140366245156e-08,
+      "loss": 0.7191,
+      "step": 11870
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2217013305796387e-08,
+      "loss": 0.7858,
+      "step": 11871
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2084607214928323e-08,
+      "loss": 0.7204,
+      "step": 11872
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1952922103197762e-08,
+      "loss": 0.7338,
+      "step": 11873
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1821957980110432e-08,
+      "loss": 0.6714,
+      "step": 11874
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1691714855120995e-08,
+      "loss": 0.7427,
+      "step": 11875
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1562192737629706e-08,
+      "loss": 0.725,
+      "step": 11876
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1433391636986868e-08,
+      "loss": 0.7732,
+      "step": 11877
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1305311562490595e-08,
+      "loss": 0.731,
+      "step": 11878
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1177952523385715e-08,
+      "loss": 0.7301,
+      "step": 11879
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1051314528864876e-08,
+      "loss": 0.7507,
+      "step": 11880
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0925397588070763e-08,
+      "loss": 0.7336,
+      "step": 11881
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0800201710092772e-08,
+      "loss": 0.7292,
+      "step": 11882
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0675726903967009e-08,
+      "loss": 0.7083,
+      "step": 11883
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.055197317867962e-08,
+      "loss": 0.7524,
+      "step": 11884
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0428940543162347e-08,
+      "loss": 0.7418,
+      "step": 11885
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0306629006299196e-08,
+      "loss": 0.8147,
+      "step": 11886
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0185038576916439e-08,
+      "loss": 0.7187,
+      "step": 11887
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0064169263792612e-08,
+      "loss": 0.7801,
+      "step": 11888
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.944021075651844e-09,
+      "loss": 0.7063,
+      "step": 11889
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.82459402116831e-09,
+      "loss": 0.7326,
+      "step": 11890
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.70588810896067e-09,
+      "loss": 0.7893,
+      "step": 11891
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.587903347599847e-09,
+      "loss": 0.7265,
+      "step": 11892
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.470639745602361e-09,
+      "loss": 0.7518,
+      "step": 11893
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.354097311431443e-09,
+      "loss": 0.7362,
+      "step": 11894
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.238276053500361e-09,
+      "loss": 0.801,
+      "step": 11895
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.123175980171318e-09,
+      "loss": 0.6168,
+      "step": 11896
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.008797099749888e-09,
+      "loss": 0.7521,
+      "step": 11897
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.895139420495025e-09,
+      "loss": 0.74,
+      "step": 11898
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.782202950610163e-09,
+      "loss": 0.7349,
+      "step": 11899
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.669987698247673e-09,
+      "loss": 0.7696,
+      "step": 11900
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.558493671507739e-09,
+      "loss": 0.7008,
+      "step": 11901
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.44772087843837e-09,
+      "loss": 0.8419,
+      "step": 11902
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.337669327036501e-09,
+      "loss": 0.7885,
+      "step": 11903
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.228339025245779e-09,
+      "loss": 0.7671,
+      "step": 11904
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.11972998095767e-09,
+      "loss": 0.7641,
+      "step": 11905
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.011842202013675e-09,
+      "loss": 0.7567,
+      "step": 11906
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.904675696199793e-09,
+      "loss": 0.7579,
+      "step": 11907
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.798230471253165e-09,
+      "loss": 0.7424,
+      "step": 11908
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.692506534856537e-09,
+      "loss": 0.6827,
+      "step": 11909
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.58750389464269e-09,
+      "loss": 0.7092,
+      "step": 11910
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.483222558191117e-09,
+      "loss": 0.7617,
+      "step": 11911
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.37966253302802e-09,
+      "loss": 0.7145,
+      "step": 11912
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.276823826630752e-09,
+      "loss": 0.7516,
+      "step": 11913
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.174706446421154e-09,
+      "loss": 0.7831,
+      "step": 11914
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.0733103997722155e-09,
+      "loss": 0.729,
+      "step": 11915
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.972635694001417e-09,
+      "loss": 0.7374,
+      "step": 11916
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.87268233637739e-09,
+      "loss": 0.7683,
+      "step": 11917
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.773450334114362e-09,
+      "loss": 0.7723,
+      "step": 11918
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.674939694376603e-09,
+      "loss": 0.793,
+      "step": 11919
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.5771504242739814e-09,
+      "loss": 0.7815,
+      "step": 11920
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.480082530865295e-09,
+      "loss": 0.7272,
+      "step": 11921
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.3837360211582714e-09,
+      "loss": 0.6587,
+      "step": 11922
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.288110902107347e-09,
+      "loss": 0.6683,
+      "step": 11923
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.1932071806147795e-09,
+      "loss": 0.7026,
+      "step": 11924
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.099024863531755e-09,
+      "loss": 0.7498,
+      "step": 11925
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.005563957657279e-09,
+      "loss": 0.7589,
+      "step": 11926
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.912824469737066e-09,
+      "loss": 0.7318,
+      "step": 11927
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.820806406465762e-09,
+      "loss": 0.7517,
+      "step": 11928
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.729509774485831e-09,
+      "loss": 0.7577,
+      "step": 11929
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.638934580386446e-09,
+      "loss": 0.7887,
+      "step": 11930
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.549080830707931e-09,
+      "loss": 0.7247,
+      "step": 11931
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.45994853193399e-09,
+      "loss": 0.7122,
+      "step": 11932
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.371537690500583e-09,
+      "loss": 0.7575,
+      "step": 11933
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.283848312789275e-09,
+      "loss": 0.7309,
+      "step": 11934
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.1968804051294455e-09,
+      "loss": 0.6827,
+      "step": 11935
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.110633973798296e-09,
+      "loss": 0.7802,
+      "step": 11936
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.0251090250230675e-09,
+      "loss": 0.6993,
+      "step": 11937
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.940305564976599e-09,
+      "loss": 0.6771,
+      "step": 11938
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.856223599779553e-09,
+      "loss": 0.7708,
+      "step": 11939
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.772863135503736e-09,
+      "loss": 0.7227,
+      "step": 11940
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.690224178164337e-09,
+      "loss": 0.7108,
+      "step": 11941
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.608306733726586e-09,
+      "loss": 0.7099,
+      "step": 11942
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.52711080810575e-09,
+      "loss": 0.7868,
+      "step": 11943
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.4466364071604765e-09,
+      "loss": 0.7902,
+      "step": 11944
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.366883536702782e-09,
+      "loss": 0.7445,
+      "step": 11945
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.2878522024869525e-09,
+      "loss": 0.7149,
+      "step": 11946
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.209542410218426e-09,
+      "loss": 0.7271,
+      "step": 11947
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.131954165551566e-09,
+      "loss": 0.6938,
+      "step": 11948
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.055087474085228e-09,
+      "loss": 0.7015,
+      "step": 11949
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.978942341369419e-09,
+      "loss": 0.6934,
+      "step": 11950
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.903518772898629e-09,
+      "loss": 0.7233,
+      "step": 11951
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.828816774120725e-09,
+      "loss": 0.7215,
+      "step": 11952
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.754836350424729e-09,
+      "loss": 0.7503,
+      "step": 11953
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.6815775071530336e-09,
+      "loss": 0.7852,
+      "step": 11954
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.6090402495925216e-09,
+      "loss": 0.6501,
+      "step": 11955
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.5372245829801142e-09,
+      "loss": 0.771,
+      "step": 11956
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.4661305124994436e-09,
+      "loss": 0.7313,
+      "step": 11957
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3957580432830706e-09,
+      "loss": 0.7101,
+      "step": 11958
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.326107180410265e-09,
+      "loss": 0.7397,
+      "step": 11959
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.2571779289092277e-09,
+      "loss": 0.7385,
+      "step": 11960
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.188970293754867e-09,
+      "loss": 0.7219,
+      "step": 11961
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.121484279871023e-09,
+      "loss": 0.6881,
+      "step": 11962
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.0547198921304646e-09,
+      "loss": 0.7566,
+      "step": 11963
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.9886771353504485e-09,
+      "loss": 0.7222,
+      "step": 11964
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.9233560142993832e-09,
+      "loss": 0.6997,
+      "step": 11965
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.858756533692386e-09,
+      "loss": 0.7484,
+      "step": 11966
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.794878698193504e-09,
+      "loss": 0.6914,
+      "step": 11967
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.7317225124123825e-09,
+      "loss": 0.7177,
+      "step": 11968
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.669287980907598e-09,
+      "loss": 0.6546,
+      "step": 11969
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.6075751081877653e-09,
+      "loss": 0.7563,
+      "step": 11970
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.5465838987071003e-09,
+      "loss": 0.6806,
+      "step": 11971
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.486314356867636e-09,
+      "loss": 0.7803,
+      "step": 11972
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.4267664870192275e-09,
+      "loss": 0.7609,
+      "step": 11973
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.367940293461768e-09,
+      "loss": 0.7923,
+      "step": 11974
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.3098357804418602e-09,
+      "loss": 0.7377,
+      "step": 11975
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.252452952152817e-09,
+      "loss": 0.8086,
+      "step": 11976
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1957918127368804e-09,
+      "loss": 0.7209,
+      "step": 11977
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1398523662841118e-09,
+      "loss": 0.7281,
+      "step": 11978
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.084634616833503e-09,
+      "loss": 0.7263,
+      "step": 11979
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0301385683696438e-09,
+      "loss": 0.7141,
+      "step": 11980
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9763642248271653e-09,
+      "loss": 0.7312,
+      "step": 11981
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.923311590087407e-09,
+      "loss": 0.7109,
+      "step": 11982
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8709806679806374e-09,
+      "loss": 0.765,
+      "step": 11983
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8193714622838366e-09,
+      "loss": 0.7116,
+      "step": 11984
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.768483976721802e-09,
+      "loss": 0.7016,
+      "step": 11985
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.718318214968262e-09,
+      "loss": 0.7004,
+      "step": 11986
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.668874180645874e-09,
+      "loss": 0.791,
+      "step": 11987
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.620151877321785e-09,
+      "loss": 0.797,
+      "step": 11988
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5721513085131811e-09,
+      "loss": 0.7466,
+      "step": 11989
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.524872477686179e-09,
+      "loss": 0.7897,
+      "step": 11990
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4783153882536038e-09,
+      "loss": 0.6769,
+      "step": 11991
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4324800435749908e-09,
+      "loss": 0.7601,
+      "step": 11992
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3873664469599146e-09,
+      "loss": 0.7645,
+      "step": 11993
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3429746016646594e-09,
+      "loss": 0.7022,
+      "step": 11994
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2993045108933289e-09,
+      "loss": 0.7232,
+      "step": 11995
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2563561777978462e-09,
+      "loss": 0.6672,
+      "step": 11996
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2141296054801742e-09,
+      "loss": 0.7226,
+      "step": 11997
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1726247969867654e-09,
+      "loss": 0.7591,
+      "step": 11998
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1318417553141114e-09,
+      "loss": 0.7193,
+      "step": 11999
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0917804834065237e-09,
+      "loss": 0.7436,
+      "step": 12000
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0524409841561334e-09,
+      "loss": 0.7317,
+      "step": 12001
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0138232604017806e-09,
+      "loss": 0.7137,
+      "step": 12002
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.759273149312353e-10,
+      "loss": 0.7538,
+      "step": 12003
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.387531504800872e-10,
+      "loss": 0.7426,
+      "step": 12004
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.023007697317454e-10,
+      "loss": 0.7465,
+      "step": 12005
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.665701753185485e-10,
+      "loss": 0.7104,
+      "step": 12006
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.315613698184343e-10,
+      "loss": 0.7239,
+      "step": 12007
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.972743557582707e-10,
+      "loss": 0.7352,
+      "step": 12008
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.637091356149651e-10,
+      "loss": 0.7755,
+      "step": 12009
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.308657118088036e-10,
+      "loss": 0.6934,
+      "step": 12010
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.987440867134432e-10,
+      "loss": 0.6946,
+      "step": 12011
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.67344262644809e-10,
+      "loss": 0.7019,
+      "step": 12012
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.366662418721969e-10,
+      "loss": 0.6896,
+      "step": 12013
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.067100266071713e-10,
+      "loss": 0.767,
+      "step": 12014
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.774756190146669e-10,
+      "loss": 0.7282,
+      "step": 12015
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.489630212041075e-10,
+      "loss": 0.7676,
+      "step": 12016
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.211722352338466e-10,
+      "loss": 0.7163,
+      "step": 12017
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.941032631089471e-10,
+      "loss": 0.7991,
+      "step": 12018
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.677561067845115e-10,
+      "loss": 0.7353,
+      "step": 12019
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.421307681612419e-10,
+      "loss": 0.7227,
+      "step": 12020
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.1722724908987987e-10,
+      "loss": 0.6975,
+      "step": 12021
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.930455513689868e-10,
+      "loss": 0.7132,
+      "step": 12022
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.6958567674161284e-10,
+      "loss": 0.7422,
+      "step": 12023
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.4684762690417873e-10,
+      "loss": 0.7416,
+      "step": 12024
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.2483140349537367e-10,
+      "loss": 0.7231,
+      "step": 12025
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.035370081061473e-10,
+      "loss": 0.7571,
+      "step": 12026
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.82964442271938e-10,
+      "loss": 0.711,
+      "step": 12027
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.6311370748044463e-10,
+      "loss": 0.732,
+      "step": 12028
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.4398480516163447e-10,
+      "loss": 0.7572,
+      "step": 12029
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.2557773669884542e-10,
+      "loss": 0.7062,
+      "step": 12030
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.0789250341879397e-10,
+      "loss": 0.7533,
+      "step": 12031
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9092910659934681e-10,
+      "loss": 0.6771,
+      "step": 12032
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.7468754746507998e-10,
+      "loss": 0.6487,
+      "step": 12033
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5916782718727875e-10,
+      "loss": 0.7677,
+      "step": 12034
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4436994688837858e-10,
+      "loss": 0.7741,
+      "step": 12035
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3029390763308336e-10,
+      "loss": 0.7024,
+      "step": 12036
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.1693971044057784e-10,
+      "loss": 0.7793,
+      "step": 12037
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0430735627342536e-10,
+      "loss": 0.7073,
+      "step": 12038
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.239684604422927e-11,
+      "loss": 0.8048,
+      "step": 12039
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.120818061230218e-11,
+      "loss": 0.7194,
+      "step": 12040
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.074136078477622e-11,
+      "loss": 0.7325,
+      "step": 12041
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.099638731882352e-11,
+      "loss": 0.3666,
+      "step": 12042
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.1973260916104994e-11,
+      "loss": 0.7897,
+      "step": 12043
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.367198222832159e-11,
+      "loss": 0.7762,
+      "step": 12044
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.609255185610394e-11,
+      "loss": 0.7132,
+      "step": 12045
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.923497034457157e-11,
+      "loss": 0.7479,
+      "step": 12046
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.3099238189994155e-11,
+      "loss": 0.6813,
+      "step": 12047
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.768535583535069e-11,
+      "loss": 0.801,
+      "step": 12048
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.299332367032946e-11,
+      "loss": 0.7363,
+      "step": 12049
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.023142034658706e-12,
+      "loss": 0.7791,
+      "step": 12050
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.774811214775966e-12,
+      "loss": 0.7411,
+      "step": 12051
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.248331444938302e-12,
+      "loss": 0.6914,
+      "step": 12052
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.4437029083325117e-12,
+      "loss": 0.7105,
+      "step": 12053
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.6092573374446605e-13,
+      "loss": 0.5137,
+      "step": 12054
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 0.6864,
+      "step": 12055
+    },
+    {
+      "epoch": 1.0,
+      "step": 12055,
+      "total_flos": 1.4340413752475648e+16,
+      "train_loss": 0.7935676227783771,
+      "train_runtime": 87601.1782,
+      "train_samples_per_second": 17.614,
+      "train_steps_per_second": 0.138
+    }
+  ],
+  "max_steps": 12055,
+  "num_train_epochs": 1,
+  "total_flos": 1.4340413752475648e+16,
+  "trial_name": null,
+  "trial_params": null
+}